Pueblo Floodhazard Finalreport

U.S.
Department of the Interior

Bureau of Reclamation June 2006

Flood Hazard Study
Pueblo Dam, Colorado

Final Report

Mission Statements

The mission of the Department of the Interior is to protect and
provide access to our Nations natural and cultural heritage and
honor our trust responsibilities to Indian Tribes and our
commitments to island communities.

The mission of the Bureau of Reclamation is to manage, develop,
and protect water and related resources in an environmentally and
economically sound manner in the interest of the American public.

Cover Photo: Overflow spillway in the center of the concrete massive head buttress dam section at
Pueblo Dam. Photograph by John F. England, Jr. on July 26, 2004.

U.S. Department of the Interior
Bureau of Reclamation
Technical Service Center
Flood Hydrology Group
Denver, Colorado June 2006

Flood Hazard Study
Pueblo Dam, Colorado

Final Report

by

John F. England, Jr., Ph.D., P.E., P.H.
Jeanne E. Klawon, M.S.
Ralph E. Klinger, Ph.D.
Travis R. Bauer, P.E.

Peer Review Certification: This document has been peer reviewed and is believed to be in
accordance with the service agreement and standards of the profession.
ii
EXECUTIVE SUMMARY
Pueblo Dam cannot successfully pass the Probable Maximum Flood (PMF). The purpose
of this study was to assess the flood hazard for Pueblo Dam near Pueblo, Colorado. The three
main objectives of the flood hazard study were: (1) significantly extend the time base and spatial
distribution of the existing peak-flow observations by collecting historical and paleoflood data in
the Arkansas River basin; (2) estimate constraints on storm intensity, duration and spatial
distribution and flood durations and volumes for the Arkansas River basin upstream from Pueblo
Dam; and (3) develop probabilistic flood hydrographs (peak, volume, and duration) for
exceedance probabilities from 1 in 100 to greater than 1 in 10,000 for a risk analysis. These
objectives were designed to address Safety of Dams (SOD) recommendation 2000-SOD-E for
Pueblo Dam: prepare a final hydrology report, suitable for use in a risk analysis study. In
2000, the hydrologic risk at Pueblo Dam exceeded guidelines. The objectives also included
improving the theoretical basis and data for making extreme flood probability statements; and
reconciling results from some of the previous flood studies. In order to meet the objectives, the
following nine main tasks were completed: (1) radar-based hydrometeorology study (Princeton
University); (2) loss of life study; (3) limited paleoflood study; (4) stochastic storm generation;
(5) rainfall-runoff modeling; (6) flood frequency analysis; (7) flood routings; (8) a report; and
(9) risk analysis support.
Investigations into the flood hydrology and hydrometeorology within the Arkansas River
basin and surrounding region were based on streamflow, paleoflood, and radar data. The largest
peak flows in the region resulted from three extreme storms: June 1921, June 1965, and July
1976. A very clear break appears in the scaling relationships between peak flows and drainage
areas at scales larger than about 1,000 mi
2
. Extreme storms, radar data, and flood data suggest
that this is due to partial-area storms and rainfall-runoff mechanisms within the watershed. Peak
flow magnitudes within the basin are dramatically reduced with elevation. Upstream of
Parkdale, snowmelt runoff is the dominant streamflow mechanism and the cause of the largest
peaks. Downstream of Parkdale, peaks and maximum mean daily flows are caused by large
rainfall-runoff events, and sometimes snowmelt runoff. Warm-season radar data and analyses
from the period 1995 through 2003 showed a decrease in rainfall accumulations with elevation.
Because of the storms' dynamics and evolution, heavy rainfall is distributed mainly on the lower
basin. The upper basin (above Wellsville) rarely experiences extreme rainfall as evidenced by
the geographical distribution of storm total rainfall and storm activity.
Detailed information was collected from four sites to characterize paleofloods and
nonexceedance bounds for Pueblo Dam: at Adobe Park, the Loma Linda recreation area,
Parkdale, and Pueblo State Park. Nine soil/stratigraphic descriptions and seven radiocarbon
iii
dates of key deposits were used in conjunction with geomorphic mapping and HEC-RAS flow
modeling to determine age estimates of each soil and peak discharges required to inundate the
surfaces. Paleohydrologic bounds spanned 550 to 10,000 years within the watershed and
provided substantially longer record lengths for frequency analysis. Nonexceedance bounds for
similar age surfaces generally increased in the downstream direction and change markedly
between Loma Linda and Pueblo State Park from approximately 14,000 ft
3
/s to 150,000 ft
3
/s for
Holocene alluvium between the ages of 700 and 2,000 years. The paleoflood data confirmed that
the upper basin is characterized by snowmelt runoff, and the lower basin (near Parkdale and
downstream) is dominated by rainfall-runoff. These data place limits on the spatial distribution
of extreme rainstorms. At Pueblo State Park, the data suggests that floods slightly larger in
magnitude than June 1921 have not been experienced in the past 800 years. Flood frequency
curves in the lower watershed at Pueblo and Parkdale reflected extreme floods from rainfall, and
were much larger than at the two upstream snowmelt locations. The estimated 10,000-year peak
flow at Pueblo was about 177,000 ft
3
/s. The frequency curves were used to constrain flood
frequency curves estimated with the Two-dimensional Runoff, Erosion, and eXport (TREX)
watershed rainfall-runoff model.
The TREX rainfall-runoff model was applied to the watershed to estimate extreme flood
hydrographs. A stochastic storm transposition (SST) model was implemented to estimate basin-
average rainfall depths and probabilities based on depth-area-duration (DAD) data for runoff
model input. Fifteen storms were used in the simulation to estimate extreme basin average
rainfall depth probabilities. Basin-average rainfall depths for a 36-hour storm ranged from 4.0 to
10.9 inches. The model was calibrated to the record June 1921 flood and was validated with the
May 1894 flood. Rainfall frequency and TREX flood frequency curve extreme upper tails were
relatively flat compared to paleoflood frequency curves. A peak flow estimate at Pueblo for the
10,000-year return period was about 230,000 ft
3
/s. A good match between runoff model and
paleoflood frequency curves was obtained by restricting storm area and storm center locations.
TREX model hydrographs were used in reservoir routing to determine maximum
elevations for extreme floods. Based on routing of extreme flood hydrographs, a best estimate
return period for the design maximum reservoir elevation (4,919 feet) is 50,000 years, with an
upper estimate equal to 5,000 years. The maximum reservoir elevations were in the range of
4,919 and 4,920 feet based on routing the hydrographs, and return periods were about 100,000
years (best estimate) to 20,000 years (upper estimate). These results suggested that the
overtopping probability is remote. Reservoir elevation frequency curves presented in this report
are comparable to those previously published.
iv
TABLE OF CONTENTS
EXECUTIVE SUMMARY............................................................................................................iii
LIST OF TABLES........................................................................................................................viii
LIST OF FIGURES........................................................................................................................ix
1.0 INTRODUCTION...............................................................................................................1
1.1 Purpose and Scope...................................................................................................1
1.2 Pueblo Dam and Reservoir.......................................................................................3
1.3 Flood Hydrology Background and Dam Safety Issues............................................4
1.4 Loss of Life Estimates..............................................................................................7
1.5 Acknowledgments....................................................................................................7
2.0 FLOOD HYDROLOGY AND HYDROMETEOROLOGY.............................................11
2.1 Arkansas River Basin Overview............................................................................11
2.2 Extreme Floods in the Arkansas River Basin and Region.....................................14
2.2.1 Flood Process Typology............................................................................16
2.2.2 Largest Recorded Peak Discharges............................................................18
2.2.3 Seasonality and Process Relationships......................................................24
2.2.4 Historical Information and Floods.............................................................31
2.3 Regional Extreme Storms and Hydrometeorology................................................33
2.3.1 Extreme Storm Database...........................................................................34
2.3.2 Radar Data and Flood Zones......................................................................38
2.4 Summary................................................................................................................41
3.0 PALEOFLOOD HYDROLOGY.......................................................................................43
3.1 Introduction............................................................................................................43
3.2 Objective................................................................................................................43
3.3 Geologic Setting.....................................................................................................43
3.4 Previous Work........................................................................................................44
3.5 Site Selection..........................................................................................................46
3.6 Methods..................................................................................................................47
3.6.1 Paleoflood Hydrology and Non-exceedance Bounds................................47
3.6.2 Data Collection..........................................................................................50
3.6.3 Hydraulic Modeling...................................................................................51
v
3.7 Geomorphic Data and HEC-RAS Modeling..........................................................52
3.7.1 Pueblo State Park.......................................................................................53
3.7.2 Parkdale.....................................................................................................57
3.7.3 Loma Linda................................................................................................62
3.7.4 Adobe Park................................................................................................66
3.8 Discussion..............................................................................................................71
3.8.1 General Observations.................................................................................71
3.8.2 Development of Paleoflood Nonexceedance bounds for Pueblo Dam......71
3.8.3 Data Uncertainty........................................................................................74
3.9 Summary................................................................................................................75
4.0 FLOOD FREQUENCY ANALYSIS WITH PALEOFLOOD DATA..............................77
4.1 Flood Frequency Methods.....................................................................................77
4.2 Arkansas River at Pueblo State Park.....................................................................79
4.3 Arkansas River at Parkdale....................................................................................82
4.4 Arkansas River at Loma Linda..............................................................................85
4.5 Arkansas River at Adobe Park...............................................................................88
4.6 Regional Flood Frequency ....................................................................................90
4.7 Summary................................................................................................................92
5.0 EXTREME STORM MODELING....................................................................................93
5.1 Storm Transposition Overview..............................................................................93
5.2 Stochastic Storm Transposition Model..................................................................95
5.2.1 Overview of Probability Concepts.............................................................95
5.2.2 Stochastic Storm Transposition Theory.....................................................96
5.2.3 Storm Spatial Distribution.........................................................................98
5.2.4 Storm Center Distribution..........................................................................99
5.2.5 Storm Temporal Distribution...................................................................100
5.3 Extreme Storm Modeling Results and Discussion..............................................100
5.4 Summary..............................................................................................................108
6.0 FLOOD FREQUENCY WITH THE TREX MODEL....................................................111
6.1 TREX Model and Inputs......................................................................................111
6.2 Calibration and Validation to Largest Observed Floods.....................................116
6.3 Peak-Flow Frequency Estimation with Extreme Storms.....................................121
6.4 Model and Paleoflood Data-Based Peak-Flow Frequency Comparisons............124
6.5 Sensitivity Analysis.............................................................................................125
6.5.1 Initial Conditions........................................................................................125
6.5.2 Spatial Distributions of Storm Rainfall......................................................128
vi
6.5.3 Storm Duration and Temporal Distribution................................................129
6.5.4 Model and Peak-Flow Frequency Revisited...............................................131
6.6 Summary..............................................................................................................135
7.0 RESERVOIR ELEVATION FREQUENCY AND RISK ANALYSIS INPUTS...........137
7.1 Initial Reservoir Levels........................................................................................137
7.2 Hydrograph Routings and Elevation Probabilities...............................................139
7.3 Hydrologic Risk Inputs........................................................................................141
7.4 Summary..............................................................................................................143
8.0 CONCLUSIONS.............................................................................................................145
9.0 REFERENCES................................................................................................................149
vii
LIST OF TABLES
1.1 Pueblo Flood Hazard Study Major Work Tasks..................................................................2
1.2 Pueblo Dam Pertinent Data.................................................................................................4
1.3 Pueblo Dam Design Flood and Probable Maximum Flood Summary................................5
1.4 Loss of Life Resulting from Various Causes of Dam Failure.............................................8
2.1 Streamflow Gaging Stations Analyzed..............................................................................16
2.2 Largest Peak Discharge Estimates in the Arkansas River Basin.......................................19
2.3 Largest Observed Peak Discharge Estimates that Define a Drainage Area-Peak Discharge
Relation..............................................................................................................................23
2.4 Largest Observed Unit Discharge Estimates that Define an Elevation-Unit Discharge
Relation..............................................................................................................................25
2.5 Historical Record Start Date and Floods at Select Locations Within the Arkansas River
Basin...................................................................................................................................33
2.6 Example Depth-Area Duration Data Table........................................................................35
2.7 Extreme Storms from DAD Catalog Considered for Transposition to Pueblo..................37
3.1 Study Reach Characteristics..............................................................................................52
3.2 Pleistocene Map Units of Van Alstine...............................................................................68
3.3 Range of Nonexceedance Bound Peak Discharge Estimates.............................................72
3.4 Summary of Paleoflood Nonexceedance Bounds, Upper Arkansas River........................75
4.1 Peak discharge records from the Arkansas River near Pueblo State Park.........................79
4.2 Arkansas River at Pueblo State Park Peak Discharge Frequency Results.........................81
4.3 Arkansas River at Parkdale Peak Discharge Frequency Results.......................................85
4.4 Arkansas River at Loma Linda Peak Discharge Frequency Results..................................87
4.5 Arkansas River at Adobe Park Peak Discharge Frequency Results..................................89
5.1 Fifteen Extreme Storms from DAD Catalog Transposed to Arkansas Watershed..........102
5.2 Sensitivity of Effective Area Estimates for Restricted Storm Centers/Areas..................108
6.1 Channel Width and Bank Height Measurements in the Arkansas River Basin...............113
6.2 Calibration Results for the June 1921 Flood....................................................................117
6.3 Calibrated Manning n Estimates for Overland Flow Grid Cells.....................................118
6.4 Calibrated Green-Ampt Infiltration Parameters for Overland Flow Grid Cells..............118
6.5 Validation Results for the May 1894 Flood.....................................................................121
7.1 Summary of Best Estimate Extreme Flood Hydrographs from the TREX Model.......140
7.2 Reservoir Elevation Frequency Summary.......................................................................140
7.3 Exceedance Probability Estimates for Specified Reservoir Elevations...........................143
7.4 Reservoir Elevation Probability Estimates for Risk Analysis Event Tree.......................143
viii
LIST OF FIGURES
1.1 Location map of Pueblo Dam and Reservoir......................................................................3
1.2 Pueblo volume-critical spillway design flood and current General Storm PMF
hydrographs.........................................................................................................................6
2.1 Arkansas River Basin study watershed..............................................................................12
2.2 Arkansas River study watershed main GIS data layers: (a) DEM; (b) landuse; and (c)
soils....................................................................................................................................13
2.3 Locations of elevations greater than and less than or equal to 7,500 ft within the Arkansas
River watershed.................................................................................................................15
2.4 Maximum peak discharge data and drainage-area envelope curve...................................22
2.5 Maximum unit peak discharge data and elevation envelope curve...................................25
2.6 Histograms of annual peaks and 1-day maxima versus month for six locations within the
Arkansas River basin.........................................................................................................27
2.7 Record snowmelt flood within the Arkansas River basin during June 1957.....................28
2.8 Recent snowmelt flood within the Arkansas River basin during June 1995.....................28
2.9 Largest recorded snowmelt (June 1957) and rainfall-generated flood (June 1921)
hydrographs in the Arkansas River Basin upstream of Pueblo.........................................29
2.10 Peak-flow correlation relationships between upstream and downstream locations..........30
2.11 Peak discharge-maximum mean daily flow relationships.................................................31
2.12 Locations of extreme storms near the Arkansas River watershed.....................................36
2.13 Geographical distribution of storm activity (percent) of the sixty-six storms events from
1995-2003 based on KPUX radar......................................................................................39
2.14 Storm activity (percent) as a function of elevation of the sixty-six storms events............39
2.15 Storm accumulations as a function of elevation of the sixty-six storms events................40
2.16 Estimated extreme flood zones in the Arkansas River basin.............................................40
3.1 Map of the upper Arkansas River basin showing the location of stream gages and
paleoflood study sites.........................................................................................................46
3.2 Idealized channel cross-section illustrating the fluvial landforms important to the concept
of a nonexceedance bound.................................................................................................49
3.3 Geomorphic map of the Arkansas River in Pueblo State Park reach upstream of the dam
and reservoir......................................................................................................................53
3.4 Schematic of soil profile at site AR1.................................................................................54
3.5 Schematic profiles of soils at sites AR2 and AR9.............................................................55
3.6 Aerial photograph comparison of the Pueblo State Park reach between 1937 and 1999. .56
3.7 Schematic cross section showing the stage of the preferred nonexceedance bound peak
discharge of modeled flows and its relationship to geomorphic surfaces and stratigraphic
sites in the Pueblo State Park reach...................................................................................58
ix
3.8 Computed water-surface profiles for estimated nonexceedance bound peak discharge in
the Pueblo State Park reach................................................................................................58
3.9 Surficial photogeologic map showing stratigraphic sites and the approximate location
HEC-RAS cross sections 3 through 8 in the Parkdale reach.............................................59
3.10 Soil exposure at site AR3...................................................................................................60
3.11 Photograph at site AR5 showing the historical flood deposit............................................61
sites in the Parkdale reach..................................................................................................62
3.13 Computed water-surface profiles for the range of preferred nonexceedance peak
discharges in the Parkdale reach........................................................................................63
3.14 Surficial photogeologic map of the Loma Linda reach showing stratigraphic sites and the
approximate location of HEC-RAS cross sections............................................................64
3.15 Vegetated slackwater deposit at site AR7..........................................................................65
3.16 Soil profile at site AR7......................................................................................................66
sites in the Loma Linda reach............................................................................................67
3.18 Computed water-surface profiles for nonexceedance bound peak discharges in the Loma
Linda reach.........................................................................................................................67
3.19 Surficial photogeologic map of the Adobe Park reach showing the approximate location
of HEC-RAS cross sections and stratigraphic site AR8....................................................69
3.20 Soil profile at site AR8......................................................................................................70
sites in the Adobe Park reach.............................................................................................70
3.22 Computed water-surface profiles for nonexceedance bound peak discharges in the Adobe
Park reach...........................................................................................................................71
4.1 Locations of four paleoflood study sites within the Arkansas River basin.......................78
4.2 Approximate unregulated peak discharge, historical and paleoflood estimates, Arkansas
River at Pueblo...................................................................................................................80
4.3 Approximate peak discharge frequency curve, Arkansas River at Pueblo........................81
4.4 Approximate unregulated peak discharge and historical flood estimates, Arkansas River
at Canon City.....................................................................................................................83
River at Parkdale................................................................................................................83
4.6 Approximate peak discharge frequency curve, Arkansas River at Parkdale.....................84
River at Loma Linda..........................................................................................................86
x
4.8 Approximate peak discharge frequency curve, Arkansas River at Loma Linda...............87
River at Adobe Park...........................................................................................................89
4.10 Approximate peak discharge frequency curve, Arkansas River at Adobe Park................90
4.11 Approximate unregulated non-dimensional peak discharge frequency curves for the four
sites within the Arkansas River basin................................................................................91
5.1 Typical extreme storm total rainfall accumulation estimates in space..............................94
5.2 Storm transposition concepts - transposition area, watershed area, storm area................95
5.3 Storm spatial representation with storm center location, orientation and cartesian
coordiante system...............................................................................................................99
5.4 Normalized mass curves from DAD data for June 1921 and May 1894.........................101
5.5 Storm transposition region and spatial distribution of the 15 extreme storms................103
5.6 Example effective storm area...........................................................................................104
5.7 Annual exceedance probability of the average rainfall depth over the Arkansas River
watershed.........................................................................................................................105
5.8 Example restricted effective storm area...........................................................................106
5.9 Sensitivity of the annual exceedance probability of the average rainfall depth over the
Arkansas River watershed...............................................................................................107
5.10 Postulated alternative tail (Normal) distributions for basin-average rainfall depth
probability curves.............................................................................................................109
6.1 Elevation grid (960 m) and channel cells for modeling the Arkansas River basin upstream
of Pueblo..........................................................................................................................112
6.2 Spatial channel width estimation from Arkansas River basin data.................................114
6.3 Spatial channel bank height estimation from Arkansas River basin data........................114
6.4 Spatial Manning n overland flow index map...................................................................115
6.5 Spatial Green-Ampt parameter index map......................................................................115
6.6 Mass curves for the June 1921 storm...............................................................................116
6.7 June 1921 extreme flood hydrograph and TREX model calibration...............................117
6.8 Cumulative rainfall, surface depth, and cumulative infiltration results at 3.5 hrs...........119
6.9 Cumulative rainfall, surface depth, and cumulative infiltration results at 10.3 hours (at
peak).................................................................................................................................119
6.10 Cumulative rainfall, surface depth, and cumulative infiltration results at 24 hrs............119
6.11 Mass curves for the May 1894 storm...............................................................................120
6.12 May 1894 extreme flood hydrographs and TREX model validation...............................121
6.13 Spatial storm pattern for TREX model runs and flood frequency...................................122
6.14 Flood frequency curve at Pueblo from TREX with corresponding SST basin-average
depth curve with no restrictions.......................................................................................123
6.15 Flood frequency curves at Pueblo, Parkdale, Wellsville and Salida from TREX...........123
xi
6.16 Runoff hydrographs for the largest simulated rainfall depth at four locations in the
watershed.........................................................................................................................124
6.17 TREX model flood frequency and streamflow/paleoflood frequency curves at (a) Pueblo,
(b) Parkdale, (c) Wellsville and (d) Salida.......................................................................126
6.18 TREX model flood frequency curves at Pueblo with varying initial soil saturation.......128
6.19 TREX model flood frequency curve at Pueblo with varying basin-average depth rainfall
frequency..........................................................................................................................130
6.20 Rearranged temporal distribution of the 36-hour NP2-23 storm of June 1964...............130
6.21 TREX model flood frequency curves at Pueblo with rearranged temporal distributions
and stretched durations....................................................................................................131
6.22 Restricted spatial storm pattern for TREX model runs and flood frequency..................132
6.23 TREX model flood frequency and streamflow/paleoflood frequency curves.................133
7.1 Annual exceedance probability of maximum reservoir levels at Pueblo Dam for the flood
season...............................................................................................................................138
7.2 Daily reservoir elevation percentiles for the flood season...............................................138
7.3 Fifteen extreme flood hydrographs from the TREX model.............................................139
7.4 Reservoir elevation frequency curves..............................................................................141
7.5 Extrapolated TREX reservoir elevation frequency curves..............................................142
APPENDICES
APPENDIX A LOSS OF LIFE STUDY
APPENDIX B RADAR DATA STUDY
APPENDIX C FIELD DESCRIPTIONS OF SOILS PROPERTIES
APPENDIX D MACROFLORAL ID AND RADIOCARBON DATA
APPENDIX E HEC-RAS MODEL CROSS SECTION DATA
APPENDIX F PEAK FLOW FREQUENCY INPUT FILES
APPENDIX G RAINFALL-RUNOFF MODEL INPUT
xii
1.0 Introduction
1.1 Purpose and Scope
The purpose of this study is to assess the flood hazard for Pueblo Dam near Pueblo,
Colorado. The three main objectives of the flood hazard study are: (1) significantly extend the
time base and spatial distribution of the existing peak-flow observations by collecting historical
and paleoflood data in the Arkansas River basin and surrounding region to extend and verify the
findings of the preliminary reconnaissance work (Levish, 1998); (2) estimate constraints on (a)
storm intensity, duration and spatial distribution and (b) flood durations and volumes for the
Arkansas River basin upstream from Pueblo Dam; and (3) develop probabilistic flood
hydrographs (peak, volume, and duration) for exceedance probabilities from 1 in 100 to greater
than 1 in 10,000 for a risk analysis.
These objectives are designed to address Safety of Dams (SOD) recommendation 2000-
SOD-E for Pueblo Dam: prepare a final hydrology report, suitable for use in a risk analysis
study. Studies conducted in 200 indicated that the hydrologic risk at Pueblo Dam exceeded
guidelines (Trojanowski, 2002). The objectives also include improving the theoretical basis and
data for making extreme flood probability statements; and reconciling results from some of the
previous flood studies that are described below. In order to meet the objectives, nine main tasks
are completed (Table 1.1). This report fulfills Task 8 of the Pueblo Flood Hazard Study, which
documets work completed in Tasks 1 through 7.
The Arkansas River watershed upstream of Pueblo Dam is large (4,669 mi
2
), and the
flood hydrology is relatively complex. There is a definite need to obtain historical and
paleoflood information within the Arkansas River basin. To estimate floods for dam safety as
remote as 1 in 10,000, additional paleoflood data are collected. A brief examination of regional
rainfall and streamflow data, combined with investigations by others, indicates a decrease in
extreme precipitation and floods with elevation. Preliminary data reviews indicate the most
extreme regional storms that produce extreme floods generally last less than about five days and
heavy precipitation typically lasts less than two days. In addition, it appears that snowmelt
runoff predominates at elevations greater than about 8,000 feet. This mixed-population rainfall
and flood data have a strong affect on scaling hydrograph peaks and durations.
The flood hazard study for Pueblo Dam is designed to address four issues for estimating
probabilistic flood inflow hydrographs to Pueblo Reservoir:
(1) What are peak discharge frequency curves and associated uncertainty estimates,
andprobabilistic flood hydrographs,for return periods from 1 in 100 to at least 1 in 10,000?
(2) What are probabilistic estimates of storm and flood runoff durations? The generalized PMP
assumptions are a 6-hour duration for the local storm and a 72-hour duration for the general
storm in this region (Hansen et al., 1988). How do these durations relate to the observed data
1
Table 1.1: Pueblo Flood Hazard Study Major Work Tasks
Task
No.
Task
1
Radar-Based Hydrometeorology Study (Princeton University). The hydrometeorology of extreme
rainfall in the upper Arkansas River basin (above Pueblo Dam) is examined through analysis of Doppler
radar from the Pueblo and Denver weather radars. The study centers on: 1) analyses of storm initiation,
structure and motion, with particular emphasis on their relation to terrain of the Colorado Front Range;
and 2) spatial and temporal distribution of extreme storm rainfall, with particular emphasis on their
relation to storm tracking properties and terrain.
2
Loss of Life Study. A consequence (loss of life) study for Pueblo Dam is conducted using Reclamations
A Procedure for Estimating Loss of Life Caused by Dam Failure, DSO-99-06.
3
Limited Paleoflood Study. Geochronology and river geometry are analyzed at several sites in the
Arkansas River basin to identify paleofloods and paleohydrologic bounds. The approximate frequency of
the June 1921 Arkansas River flood is estimated. 1-D hydraulic modeling is used to determine the
relationship of peak discharges to paleo bounds at these selected sites
4
Stochastic Storm Generation. Stochastic storm transposition with Depth-Area Duration data,
supplemented by radar data are used to generate extreme storms and storm probabilities for rainfall-
runoff modeling.
5
Rainfall-Runoff Modeling. A two-dimensional, distributed rainfall-runoff model is used to simulate
extreme flood hydrographs. The runoff model is coupled with stochastic storm transposition to estimate
flood frequency curves for comparison with paleoflood-based frequency curves.
6
Flood Frequency Analysis. Peak flow data at gages are combined with historical and paleoflood data to
estimate data-based flood frequency curves. These frequency curves are then used as a basis to adjust
rain-runoff model inputs and parameters.
7
Flood Routings. Perform routings of selected probabilistic hydrographs through the reservoir. Various
initial reservoir water surface elevations are assumed.
8 Report. A report that documents tasks 1 through 7 (this report).
9 Risk Analysis Support. Provide hydrologic inputs to subsequent risk analyses.
and can probability statements be made regarding each? Storm and flood durations that clearly
affect flood runoff volume estimates should be derived directly from the observed data and other
observations from the basin and region.
(3) What are the hydroclimatological mechanisms that cause large magnitude, extreme floods in
the Arkansas River basin and in the region? It appears that two broad storm classifications
(cyclonic and convective) (Hansen et al., 1988) apply for the region. It is also well known that
mixed populations (rainfall and snowmelt) contribute to flooding in the Colorado Front Range
(Follansbee and Sawyer, 1948; Elliott et al., 1982; Jarrett, 1987; Jarrett and Costa, 1988). The
effects of mixed population on flood frequency curves and probabilistic hydrographs are
unknown for the Arkansas River at Pueblo Dam.
2
(4) What are the spatial characteristics of observed extreme rainstorms and seasonal snowpack?
Are there any topographic barriers and/or elevation constraints to extreme rainfalls in the upper
Arkansas River basin upstream of Caon City? Preliminary analyses of the flood potential at the
Pueblo Dam site focused on rainfalls below Caon City. Both the peak and volume critical
hydrographs used for the Pueblo Dam final design utilized rainfalls over areas approximately
below El. 8,000. More data and analyses are needed to investigate and confirm extreme
precipitation and flood elevation trends in the Arkansas River basin.
1.2 Pueblo Dam and Reservoir
Pueblo Dam and Reservoir are located on the Arkansas River approximately six miles
west of the City of Pueblo, in southeastern Colorado (Figure 1.1). The dam was constructed
from 1970 to 1975 and is the largest reservoir in the Fryingpan-Arkansas Project. The dam
provides storage for irrigation, municipal and industrial use, and flood control. Pueblo Dam is a
concrete dam combined with two earthfill embankment wings on the right and left sides of the
dam. The concrete dam and massive-head buttress-type spillway is the principal control
structure for the reservoir. The dam is approximately 10,230 feet long at crest elevation 4,925 ft.
Pertinent data are listed in Table 1.2.
Figure 1.1. Location map of Pueblo Dam and Reservoir.
3
The spillway and flood control pool at Pueblo Dam were designed in cooperation with
the Corps of Engineers. The flood control pool was sized such that the City of Pueblo would
have 100-year flood protection with the flood storage in Pueblo Reservoir and the existing flood
levees within the city. The flood control pool was specifically designed to contain without
releases the first 57 hours of the 1921 flood that has a volume of 93,000 acre-ft (USACE, 1977
Plate 8-4). The spillway design flood was a volume-critical spring rainstorm with snowmelt
(described below). The spillway has never operated, as the maximum reservoir water surface to
date (4,888.4 ft.) is 10.3 feet below the spillway crest.
Table 1.2: Pueblo Dam Pertinent Data
Total Drainage Area (USGS): 4,669 mi
2
Contributing Drainage Area (Reclamation 1991 PMF): 4,560 mi
2
Spillway Crest El.: 4,898.7 ft.
Top of Conservation Pool El. 4,880.54 ft.
Top of Flood Control Pool El. 4,898.7 ft.
Top of Dam El.: 4,925 ft.
Maximum Observed Reservoir El.: 4,888.4 ft., 02/12/1996
Design Spillway Capacity: 191,500 ft
3
/s at El. 4,919 ft.
Spillway Type:
uncontrolled overflow and concrete flip
bucket
Low-level Outlet works capacity (all seven outlets): 5,767 ft
3
/s
Reservoir Capacity at Top of Conservation (El. 4,880.54): 234,347 acre-ft.
Flood Control Capacity (El. 4,893.79 to El. 4,898.7): 27,024 acre-ft.
Joint Use Capacity (El. 4,880.54 to El. 4,893.79): 65,952 acre-ft.
Reservoir Surface Area (El. 4,880.54): 4,641 acres
1.3 Flood Hydrology Background and Dam Safety Issues
One hydrologic loading condition used to assess the hydrologic adequacy of a
Reclamation dam is the Probable Maximum Flood (PMF). If a dam can successfully pass the
PMF, the hydrologic risk is considered adequate, even though the probability of the PMF is
unknown. The PMF is also considered an upper limit to the hydrologic risk (Reclamation, 2002;
Swain et al., 2004). The flood hydrographs that were utilized for Pueblo Dam spillway design
studies were completed in 1968 (Reclamation, 1968a, 1968b) and consisted of volume and peak
critical floods. The most recent Probable Maximum Precipitation/Probable Maximum Flood
(PMP/PMF) study for Pueblo Dam was completed in 1991 as part of the ongoing Safety of Dams
Program (Bullard and Leverson, 1991). A summary of design floods and current PMFs to
4
Pueblo Reservoir are shown in Table 1.3.
Table 1.3: Pueblo Dam Design Flood and Probable Maximum Flood Summary
Type (Study Year) Peak Volume (duration)
Maximum Reservoir
Level
(overtopping depth)
Volume Critical (1968)
[final design]
270,000 ft
3
/s
399,000 acre-ft. (2-day)
582,000 acre-ft. (14-day)
4,919 ft.
Peak Critical (1968) 550,000 ft
3
/s
329,000 acre-ft. (6-day)
438,000 acre-ft. (14-day)
General Storm PMF (1991) 835,000 ft
3
/s 1,390,000 acre-ft. (15-day)
4,931.27 ft.
(6.27 ft.)
Local Thunderstorm PMF (1991) 175,000 ft
3
/s 118,000 acre-ft. (52-hour) 4,899.89 ft.
The 1968 inflow design floods were based on limiting rainfall to an approximate 500 mi
2
area for the peak critical case, and developing storms from 500 to 2,500 mi
2
for the volume
critical case. A critical assumption in the studies was that storm rainfalls were limited to the
approximate 1,770 mi
2
area that is below elevation 8,000 ft. The 1968 study attempted to mimic
the rainstorms and flood conditions that had been observed in the region. The volume critical
flood was used for the Pueblo Dam final design.
The current General Storm PMF into Pueblo Reservoir is based on a 72-hour, June PMP
rainstorm over the 4,560 mi
2
watershed with a 100-year snowmelt as an antecedent flood
(Bullard and Leverson, 1991). The local storm PMF assumed snow free ground and that storm
rainfall occurred over a 500 mi
2
portion of the Arkansas Basin (Bullard and Leverson, 1991).
Hydrometeorological Report (HMR) No. 55A (Hansen et al., 1988) was used to estimate PMP
for the 1991 study.
The two main differences between the final design flood and the current PMFs are: (1)
the usage of generalized precipitation values (Hansen et al., 1988) in the 1991 PMF that are
substantially larger in terms of rainfall magnitude than site specific estimates used in 1968; and
(2) the significant increase in precipitation area coverage for the 1991 General Storm PMF as
compared to the 1968 volume critical final design flood. The inflow design flood and General
Storm PMF are shown in Figure 1.2.
Pueblo Dam cannot successfully pass the PMF without overtopping (Table 1.3). Because
Pueblo Dam cannot pass the PMF, the hydrologic risk for dam safety at Pueblo Dam needs to be
assessed for three potential failure modes (Stanton, 2000): overtopping and erosion of the
embankments; sliding along the shale seam beneath the concrete dam; and sliding along the
embankment foundation. Extreme flood probabilities are needed in order to assess the
hydrologic risk.
5
0 12 24 36 48 60 72 84 96 108 120 132 144 156 168
Time (hours)
0
100000
200000
300000
400000
500000
600000
700000
800000
D
i
s
c
h
a
r
g
e

(
f
t
3
/
s
)
Volume Critical Design Flood (Reclamation, 1968b)
PMF including snowmelt (Bullard and Leverson, 1991)
PMF peak 835,000 ft
3
/s
PMF volume 1,390,000 acre-ft
Design peak 270,000 ft
3
/s
Design volume 582,000 acre-ft
Figure 1.2. Pueblo volume-critical spillway design flood and current General Storm PMF hydrographs.
Several studies have been completed to estimate extreme flood probabilities for Pueblo
Dam. These include: assuming a probability of the PMF peak flow and scaling PMF
hydrographs (Reclamation, 1997); regional peak-flow frequency analysis using L-Moments
(Bullard, 1998a); regional volume frequency analysis using L-Moments (Bullard, 1998b); and a
reconnaissance paleoflood study (Levish, 1998). These studies are briefly summarized below.
General storm flood hydrographs for various return periods (1 in 100 to the 1991 PMF)
were prepared in 1997 (Reclamation, 1997) using peak discharge probability interpolation and
hydrograph ordinate interpolation techniques. A key assumption in this approach is estimating
the exceedance probability of the PMF. Exceedance probability (return period) estimates for the
PMF are very low and are subject to considerable uncertainties; the PMF return period was
estimated as 1 in one billion (1 x 10
-9
) for creating the probability hydrographs (Reclamation,
1997). This PMF peak discharge exceedance probability is three to five orders of magnitude
lower than the estimate from the 1991 study.
In contrast to the 1997 flood hydrographs, a new regional flood frequency method was
used to estimate return periods of peak discharge and streamflow volume at Pueblo Dam
(Bullard, 1998a,1998b). The most recent studies utilized regional index flood techniques with
probability distribution parameters estimated using L-Moments methods (Hosking and Wallis,
6
1997). The general conclusions from the 1998 studies were that for a 10,000-year event, the
one-day peak was less than 90,000 ft
3
/s, and the 15-day volume was consistent with previous
PMF estimates (Bullard, 1998b). In addition, the frequency of peaks and volumes for a fixed
magnitude increased for the 1998 study as compared to the 1997 study.
In addition to the regional flood study, a brief, reconnaissance-level paleoflood study was
completed for Pueblo Dam (Levish, 1998). Data presented as part of this brief study suggest that
a peak discharge in the range of about 125,000 to 191,000 ft
3
/s has not been exceeded in about
7,000 to 10,000 years.
The results from the above studies were used to route floods through Pueblo Reservoir,
compare flood routing results, and make recommendations concerning their use in risk analysis
studies (Ellingson, 1999). The 1998 flood hydrology studies were reviewed in detail by
individuals from the Pueblo Dam Modification Decision Analysis (MDA) risk analysis team.
The main reason for the review (Ellingson, 1999) was the 1998 flood frequency estimates were
prepared using new flood hydrology techniques for Reclamation. After routing hydrographs
based on the above methods (with some variants), it was determined that there is considerable
variability in the routing results and subsequent risk estimates. The investigations that are
completed as part of this flood hazard study, including utilization of site-specific paleoflood data
and regional extreme storm data, supersede these previous studies.
1.4 Loss of Life Estimates
Loss of life estimates were made for Pueblo Dam (Graham, 2003) using procedures in
Graham (1999). Results from the loss of life study are shown in Table 1.4. For flood-induced
dam failure, the estimated number of fatalities is 131 during the day and 376 at night. These
estimates are generally comparable with estimates made in 1992. The relatively large loss of life
estimates indicate that floods with exceedance probability estimates in the range from about 1 in
100,000 to 1 in 400,000 are needed for risk analysis. This is based on assuming an overtopping
failure probability of 0.9 and loss of life as shown in Table 1.4, in order to meet Reclamation
Annualized Loss of Life Criteria equal to 0.001 (Reclamation, 2003). The complete loss of life
report is attached as Appendix A.
1.5 Acknowledgments
This project was funded by the Bureau of Reclamation Dam Safety Office as part of dam
safety investigations for Pueblo Dam. The study was initiated by the Dam Safety Office, with
Bruce Muller, Chief of Dam Safety. The study scope was originally conceived by John England
and Dan Levish of the Technical Service Center (TSC), and supported by David Eubank (since
retired), Dam Safety Office Program Manager, Great Plains Region. The Dam Safety Office
also provided funding via contract 02CR810797 to Dr. Jim Smith and the Hydromet Group at
Princeton University for a radar hydrometeorology study.
7
Table 1.4*: Loss of Life Resulting from Various Causes of Dam Failure
Location
Dam Failure Cause
Static Seismic Flood Induced
Day Night Day Night Day Night
Pueblo Dam to western edge of
Pueblo (mile 0.0 to 5.0)
1.2 5.9 5.9 5.9 0.4 1.3
Western edge of Pueblo to
eastern edge of Pueblo (mile 5.0
to 10.4)
261 348 348 348 97 290
downstream from Baxter (mile
10.4 to 17.0)
47 47 47 47 18 53
Downstream from Baxter to
downstream from Boone (mile
17.0 to 32.0)
2 9.0 9.0 9.0 2.0 5.0
Downstream from Boone to
upstream from Fowler (mile 32.0
to 49.6)
0.4 0.9 0.4 0.9 0.2 0.4
Upstream from Fowler to Rocky
Ford (mile 49.6 to 73.4)
2.5 5.0 2.5 5.0 2.8 5.6
Rocky Ford to upstream from La
Junta (mile 73.4 to 84.1)
2.1 4.2 2.1 4.2 2.3 4.6
Upstream from La Junta to
downstream from La Junta (mile
84.1 to 86.5)
3.3 6.6 3.3 6.6 3.7 7.4
Downstream from La Junta to
upstream from Las Animas (mile
86.5 to 108.6)
0.3 0.7 0.3 0.7 0.4 0.7
Upstream from Las Animas to
head of John Martin Reservoir
(mile 108.6 to 117.2)
3.8 7.5 3.8 7.5 4.2 8.4
Total 324 435 422 435 131 376
*Table 9 from Graham (2003)
Dr. Robert Jarrett, U.S. Geological Survey, provided valuable review comments on
Sections 2 and 4 through 6, and made suggestions on future work. Drs. Pierre Julien, Jose Salas,
and Ellen Wohl of Colorado State University (CSU) provided general advice on the study and
gave constructive comments on these same sections. Portions of this report were included as
part of John England's Ph.D. dissertation at CSU (England, 2005).
Travis Bauer, John England, Wayne Graham, Jeanne Klawon and Ralph Klinger (TSC),
and Julie Javier and Drs. Jim Smith, Mary Lynn Baeck and Matthias Steiner (Princeton
University) contributed to the completion of this report. Jeanne Klawon, Ralph Klinger and
Travis Bauer contributed Section 3 - Paleoflood Hydrology, Appendix C - Field Descriptions of
Soils Properties, and Appendix E - HEC-RAS Model Cross Section Data. John England
8
contributed the remaining sections of the main body of the report and Appendices F and G.
Wayne Graham contributed Appendix A Loss of Life Study. Julie Javier, Jim Smith, Mary
Lynn Baeck and Matthias Steiner of Princeton University contributed Appendix B Radar Data
Study.
Several individuals, organizations and companies provided data and analyses for the
completion of the report. Dave Durovchin (Colorado Division of Water Resources) provided
streamflow data from selected state gages. Rick Crowfoot (U.S. Geological Survey) provided
streamflow data from USGS databases. Kathy Puseman at Paleo Research Institute (Golden,
Colorado) provided macrofloral analysis, identified detrital charcoal, and separated potentially
datable radiocarbon material from bulk sediment samples (Appendix D). Radiocarbon dating
was performed by Beta Analytic (Miami, Florida) for select samples (Appendix D).
This report has been peer reviewed by the following Reclamation TSC personnel: Robert
E. Swain (overall report), Louis C. Schreiner (Sections 2, 6, 7 and 8), David A. Raff (Sections 6
and 7), Jennifer Bountry (paleoflood hydraulics in Section 3), and Daniel R. Levish (Section 3).
This report reflects the comments and contributions of these peer reviewers.
9
10
2.0 Flood Hydrology and Hydrometeorology
This section presents flood hydrology and hydrometeorology data within the Arkansas
River basin and surrounding region. The main purpose is to gain a direct, data-based physical
understanding of extreme flood hydrology and hydrometeorology in order to develop predictive
models for extrapolation. Three main areas are explored: (1) review and documentation of
extreme floods, flood typology and seasonality within the Arkansas River and surrounding
region; (2) documentation of historical information that includes known floods prior to
establishment of gaging stations; and (3) review and documentation of an extreme storm catalog
and meteorology, that are supplemented by radar data. The focus is to describe the data and
flood mechanisms that will form the basis to estimate peak discharge probability relationships
and perform rainfall-runoff modeling of extreme floods at several scales within the Arkansas
River watershed. An extreme storm database is developed that is subsequently used for extreme
storm modeling (Section 5).
2.1 Arkansas River Basin Overview
The Arkansas River basin is a main tributary to the Mississippi River and covers
approximately 160,000 mi
2
in Colorado, Kansas, Oklahoma, and Arkansas (Iseri and Langbein,
1974). The study watershed within this basin is an approximately 4,660 mi
2
portion located in
the Arkansas Headwaters (11020001) and Upper Arkansas (11020002) hydrologic units (Seaber
et al., 1987) within Colorado. A general outline of the watershed, including major tributaries,
towns, and streamflow gage locations is shown in Figure 2.1.
The Arkansas River study watershed is located in the south-central and southeastern
portions of Colorado. The Arkansas River originates at the confluence of the East Fork
Arkansas River and Tennessee Creeks, high in the Colorado Rocky Mountains near Malta,
Colorado. The main river travels in a general southerly direction from Malta toward Salida. The
upper watershed in this section consists of narrow valleys and short, steep canyon reaches.
Numerous small tributary creeks flow into the main river in this section. Just upstream of
Salida, the Arkansas River flows in a southeasterly direction to Coaldale. Two majors tributaries
enter the Arkansas River within this section and include the South Arkansas River and Badger
Creek. The river then flows in a general northeasterly direction to Parkdale. From Salida to
Parkdale the river also flows through a canyon environment with relatively narrow valleys
connecting short, narrow canyon reaches. Main tributaries within this section include Bernard,
Texas, Currant, Tallahassee, and Cottonwood Creeks. The river then flows southeasterly
through the Royal Gorge (originally called the Grand Canyon of the Arkansas) to Caon City.
At Caon City, Grape and Fourmile Creeks enter the river. At Caon City, the topography and
river corridor change from steep canyons and narrow valleys to rolling terrain and an ever-
11
widening river valley. Main tributaries between Caon City and Pueblo include Eightmile,
Beaver, Oak, Hardscrabble, and Turkey Creeks.
Figure 2.1. Arkansas River Basin study watershed. Major towns and tributaries to the Arkansas River, and
streamflow gages within the watershed that are analyzed as part of this study (Table 2.1), are shown.
Elevations in the Arkansas River watershed range from 14,433 ft at Mt. Elbert to about
4,700 ft downstream of Pueblo Reservoir. The river cuts through the Colorado Front Range
between about Cotopaxi and Caon City. Upstream of Caon City, the mean elevation is 9,550
ft and the mean basin slope is 20.5%. Downstream of Caon City, the basin is generally lower
and flatter; the mean elevation is 6,152 ft and the mean slope is 9.9% within the watershed
between Caon City and Pueblo. For the entire watershed, the mean basin elevation is 8,655 ft
and the mean basin slope is 18.1%. The straight-line distance from Leadville to Pueblo is about
115 mi. Approximate distances along the main-stem Arkansas River are 35 mi from Leadville to
Buena Vista, 27 mi from Buena Vista to Salida, 17 mi from Salida to Coaldale, 23 mi from
Coaldale to Parkdale, 9 mi from Parkdale to Caon City, and 41 mi from Caon City to Pueblo.
The TREX rainfall-runoff model (Velleux et al., 2005) is used to estimate extreme flood
peaks and hydrographs. The available data within the watershed for TREX model parameter
estimation and calibration consists of three main types: GIS data, physical data, and
hydrographic and atmospheric measurement data from gages. The GIS data that are used
include: a Digital Elevation Model (DEM) of elevations in the watershed (Figure 2.2a); a map of
land use and land cover (Figure 2.2b); a surficial soils map (Figure 2.2c); and hydrography
12
(rivers and lakes) information. Physical data consists of river channel dimensions (thalweg
elevations, widths, bank heights, sideslopes, lengths, sinuosity).
(a)
(b)
(c)
Figure 2.2. Arkansas River study watershed main GIS data layers: (a) DEM; (b) landuse; and (c) soils.
13
The measurement data includes precipitation (rainfall rates and total accumulations), streamflow
(peaks, daily flows, unit values), and radar data from Pueblo and Denver. There are six main
land use classes present in the watershed based on the USGS National Land Cover Data
(NLCD): evergreen forest (35%), grasslands/herbaceous (29%), shrubland (23%), deciduous
forest (7%), bare rock/sand/clay (3%) and pasture/hay (2%). A description of each class is in
Anderson et al. (1976). Areas of the watershed with elevations greater than 9,840 ft are usually
snow-covered from November through mid-April. Snowmelt is the dominant runoff mechanism
in much of the watershed.
2.2 Extreme Floods in the Arkansas River Basin and Region
The Arkansas River basin study watershed falls within two flood peak discharge
hydrologic regions (Vaill, 2000): the Mountain region and the Plains region. The Arkansas
River watershed upstream of Wellsville (1,486 mi
2
) is in the Mountain region; the remaining
downstream portion of the watershed (3,174 mi
2
) is in the Plains region. The Mountain region
consists of the high topographic relief of the Rocky Mountains north of the continental Divide
and north of the Rio Grande drainage basin and is defined by the 7,500 ft elevation contour along
the eastern and western slopes of the Rocky Mountains. The western boundary of the Plains
region coincides with a line along an elevation of 7,500 ft in the South Platte River basin, south
to a transition zone near the Chaffee-Fremont County line, to a line along an elevation of about
9,000 ft in the Arkansas River basin (Vaill, 2000). These hydrologic region boundaries
described by Vaill were determined by McCain and Jarrett (1976) and Kircher et al. (1985). The
regions do not correspond to hydrologic units; rather, they are determined by examining
residuals from peak-flow regression equations and grouping similar areas. The spatial
distribution of elevations relative to 7,500 ft within the watershed is shown in Figure 2.3.
Streamflow data are used to understand and quantify floods and flood magnitude. Three
data sources from the U.S. Geological Survey and Colorado Division of Water Resources were
used to characterize streamflow in the Arkansas River basin: (1) annual peak discharge estimates
at gaging stations and miscellaneous sites; (2) daily mean discharge estimates and unit (hourly or
15-minute) values at gaging stations; and (3) qualitative and quantitative information from
USGS Water-Supply Papers and other reports.
The USGS and the Colorado Division of Water Resources have been measuring and
publishing streamflow records in Colorado since the late 1800s. The first regular streamflow-
gaging station was established by the State Engineer in 1883 on the Cache la Poudre River and
operated by L.G. Carpenter of the Colorado State Agricultural College (Colorado State Engineer,
1885; Fellows, 1902). The USGS established many stations in Colorado in 1888 (Fellows, 1902;
Frazier and Heckler, 1972). There are many long-term gaging stations located in the Arkansas
River basin upstream from the city of Pueblo, Colorado (Crowfoot et al., 2004). The stations
with the longest records are all located on the Arkansas River main stem and include: at Granite
(1895); at Salida (1895); at Canon City (1888); and at Pueblo (1885). Including tunnels, seepage
14
locations, and dams, there are approximately 25 active gaging stations in the Upper Arkansas
River basin upstream of Pueblo (Crowfoot et al., 2004). This study focuses on using data from
16 active and discontinued gaging stations located within the Arkansas River basin upstream
from Pueblo Dam (Figure 2.1). Streamflow data from these gaging stations were used for peak
discharge frequency and understanding flood hydrometeorology. The station names, location,
elevation and period of record at each site are summarized in Table 2.1.
Figure 2.3. Locations of elevations greater than (magenta) and less than or equal to (green) 7,500 ft within the
Arkansas River watershed, based on a DEM with 150 m cell size. Approximately 74% of the watershed area is
greater than 7,500 ft.
Streamflow data were obtained directly from the USGS National Water Information
System (NWIS) Database. These data were cross-checked with those published in USGS Water-
Data Reports and Water-Supply Papers, including: Fellows (1902), USGS (1923), Follansbee
and Jones (1922), Jarvis and others (1936), Follansbee and Sawyer (1948), USGS (1955), USGS
(1964), Patterson (1964), USGS (1969), and Crowfoot et al. (2004). The records were
supplemented by Colorado Division of Water Resources data published at selected locations.
These sources indicate that there are very long stream gaging records that are essentially
complete at certain scales, predominantly along the main stem. In addition to data at gaging
stations, historical information (discussed below) is used to supplement peak discharge estimates
and extend record lengths for peak flow frequency estimation.
15
Table 2.1: Streamflow Gaging Stations Analyzed
No.
(Fig.
2.1)
USGS
Gaging
Station No.
Station Name
Drainage
Area (mi
2
)
Lat. Long.
Gage
Elevation
(feet, NGVD)
Hydrologic
Unit Code
Period of
Record
(Water Years)
1 07083000 Halfmoon Creek near Malta 23.6 39.1722 -106.3886 9,830 11020001 1947-2003
2 07086000 Arkansas River at Granite 427 39.0428 -106.2653 8,914.86 11020001 1895-2003
3 07086500
Clear Creek above Clear Creek
Reservoir
67.1 39.0181 -106.2772 8,885 11020001 1946-1995
4 07089000
Cottonwood Cr. bl Hot Springs, near
Buena Vista
65 38.8128 -106.2217 8,532 11020001 1912-1986
5 07091500 Arkansas River at Salida 1,218 38.5458 -106.0100 7,050.45 11020001 1895-1979
6 07093700 Arkansas River near Wellsville, 1,485 38.5028 -105.9392 6,883.40 11020001 1961-2003
7 07093775
Badger Creek, lower station, near
Howard
211 38.4672 -105.8594 6,780 11020001 1981-2003
8 07094500 Arkansas River at Parkdale 2,548 38.4872 -105.3731 5,720 11020001 1946-2003
9 07095000 Grape Creek near Westcliffe 320 38.1861 -105.4831 7,690 11020001 1925-1995
10 07096000 Arkansas River at Canon City 3,117 38.4339 -105.2567 5,342.13 11020002 1889-2003
11 07096500 Fourmile Creek near Canon City 434 38.4364 -105.1908 5,254 11020002 1949-1997
12 07097000 Arkansas River at Portland 4,024 38.3883 -105.0156 5,021.59 11020002 1939-2003
13 07099050
Beaver Cr above Upper Beaver
cemetery near Penrose
122 38.5617 -105.0214 6,020 11020002 1991-2003
14 07099200 Arkansas River near Portland 4,280 38.3372 -104.9383 4,940 11020002 1965-1974
15 07099230
Turkey Creek above Teller Res near
Stone City
62.3 38.4650 -104.8258 5,520 11020002 1978-2003
16 07099500 Arkansas River near Pueblo 4,686 38.2672 -104.6572 4,689.74 11020002 1885-1975
2.2.1 Flood Process Typology
In order to understand and predict extreme floods within a particular watershed, it is
imperative to understand the causative mechanisms of floods. Meyer (1917) outlined three
mechanisms: floods due to rainfall on small and large basins; floods due primarily to snowfall;
and fall floods. Foster (1948) suggested three general classes for the causes of natural floods:
excessive rainfall; rapid melting of deep snow cover; and ground conditions (frozen or
saturated). Hoyt and Langbein (1955) describe five broad causative mechanisms of floods:
rainstorms; thunderstorms; hurricanes; snow; and ice jams. Jarrett and Costa (1988, p. 4)
summarized three types of floods that occur in the Colorado Front Range: snowmelt, rainfall;
and rain falling on snow or mixed-population. Cudworth (1989, p. 177-179) presents four
primary meteorologic classes of floods: thunderstorms; general rain type events; snowmelt
floods; and rain-on-snow events. He defined these classes as follows: (1) thunderstorm events
where the resulting flood is caused by high-intensity, short duration rainfall that produces high
peak discharges and relatively low volumes; (2) general rain type events where the resulting
flood is caused by moderate intensity, long duration rainfall; (3) snowmelt floods resulting from
16
the melting of an accumulated snowpack; and (4) floods resulting from a combination of rain
falling on a melting snowpack.
Doesken (1991) suggested that floods occur primarily from April through October in
Colorado from three principal causes: intense local thunderstorms; intense widespread rainfall;
and snowmelt. Loukas et al. (2000) classified peak flows in British Columbia into five types:
rainfall events; snowmelt events; winter rain-on-snow events; spring rain and snowmelt events;
and summer runoff events with contribution of glacier melt. Merz and Blschl (2003) recently
classified the causative mechanisms of floods for Austria into five major types: long-rain floods,
short-rain floods, flash floods, rain-on-snow floods, and snowmelt floods. Based on the past
work summarized above, the four class flood process typology from Cudworth (1989) are used
to describe floods within the Arkansas River basin and region. Prior work is reviewed on
extreme floods in the Colorado Front Range. Hydrometeorologic mechanisms for the observed
floods and storms are described in Section 2.3. The types of extreme floods in Colorado are
described by Follansbee and Hodges (1925, p. 105):
Floods in the Rocky Mountain region are of two types the floods in the larger streams due to
the general rains of several days' duration over large areas and the so-called cloudburst floods due
to intense rains of short duration covering well-defined small areas. Floods of the first type are
relatively infrequent, and, as they are well understood, their characteristics will not be discussed.
Only the severe floods of this type that occurred in 1923 are described in this report. Cloudburst
floods cause the streams to rise and fall suddenly.
Matthes (1922) discusses cloudburst mechanisms and major flood observations in the United
States; he suggests that cloudbursts in the foothills of the Rocky Mountains are common, at
higher elevations they are rare, and that the relation between topography and cloudburst
frequency needs to be studied.
The flood hydrology of the foothills and mountain streams in the Colorado Front Range
is characterized by mixed-population flooding from snowmelt and rainfall (Elliott et al., 1982;
Jarrett, 1987). Rain-on-snow floods do occur infrequently, but have been observed primarily in
the Rio Grande, Gunnison River, and Colorado River basins (Follansbee and Sawyer, 1948).
Short-duration, high-intensity local convective thunderstorms are the dominant mechanism for
causing flash floods and the largest instantaneous peak discharges on foothills streams (e.g,.
McCain et al., 1979). Follansbee and Sawyer (1948, p. 22) term these events cloudbursts, which
they define as follows.
A type of storm confined chiefly to the eastern foothills region below an altitude of about 7,500
feet and extending eastward from the mountains about 50 miles, is the so-called cloudburst,
which is a rainfall of great intensity confined to a very small area and lasting usually a very short
time. ... Cloudbursts occur only where there is a marked range in temperature within a relatively
small area. This condition exists chiefly within the foothills, where warm air from the plains
drifts toward the mountains, is deflected upward, and cools rapidly at the higher altitudes near the
heads of canyons. For this reason cloudbursts generally occur in the afternoon or early evening
17
of an unusually warm day. ... At the higher altitudes the differences in temperature are usually
insufficient and the mass of air too small to cause cloudbursts, although on rare occasions they
have occurred at high altitudes during unusually warm weather.
The Arkansas River watershed is subject to extreme flooding in the warm season (April
through August) from cloudburst rainfalls, snowmelt runoff, and spring general rainstorms.
Topography is the major influence on extreme precipitation. The largest recorded floods at
streamflow-gaging stations on the Arkansas River main stem upstream of Caon City have been
from snowmelt. At Caon City and downstream, the largest peak flows have been from rainfall-
runoff. Within the Arkansas River Basin, there are some mixed-population (rainfall-runoff and
snowmelt runoff) flooding on small tributaries at higher elevations. Follansbee and Hodges
(1925, p. 107) note the following, relevant to extreme floods at high elevations.
The east side of the Arkansas valley between Granite and Buena Vista, embracing the western
slope of the Park Range, is also subject to occasional cloudbursts, which, however, are not so
severe as those in the foothills region. Cloudbursts have also been recorded near the mouth of
Texas Creek above the Royal Gorge. ... Most cloudbursts occur at altitudes between 6,000 and
7,000 feet, although those near Ouray are between 8,000 and 9,000 feet, those near Granite about
9,500 feet, and the one series of cloudbursts recorded on the North Fork Shoshone River in
northern Wyoming at 10,000 feet. ... It is readily seen that they can seldom occur at higher
altitudes in the mountains, as there the differences in temperature are usually insufficient and the
mass of warm air in the high valleys is not great enough to cause any decided drift toward
adjacent mountains.
Twenty-three years later, Follansbee and Sawyer (1948, p. 73) clearly state their opinion on the
spatial distribution of extreme floods in the Arkansas River Basin:
Above the Royal Gorge, the Arkansas River is not subject to heavy floods. A few of its upper
tributaries are subject to cloudburst floods, but the volume of these floods is too small to affect
seriously the Arkansas River itself.
Hoyt and Langbein (1955, p. 272) also suggest that floods on the Arkansas River in Colorado
upstream of John Martin Reservoir are generally of the summer cloudburst type. Streamflow
and storm data are summarized below that show thunderstorms and general storms can cause
extreme floods within the Arkansas River basin east of Parkdale.
2.2.2 Largest Recorded Peak Discharges
Peak flow data are one of the most important measures of extreme floods (e.g.,
Dalrymple, 1964). The largest peak flows recorded within the Arkansas River watershed
upstream from Avondale are listed in Table 2.2; the flood season is typically from late April
through mid-September. Peak flows for rainfall-dominant portions of the watershed downstream
of Parkdale are from general storms in May and June, and local thunderstorms in July and
18
August. Snowmelt causes the largest flood peaks within the upper basin upstream of Parkdale,
and occurs from late May through mid July. The most important of the largest floods recorded
at these gaging stations were: June 3, 1921 at Pueblo; August 2, 1921 at Canon City; June 17-18,
1965 on the Arkansas River east of Pueblo; and April 30, 1999 on Fountain Creek. The largest
flood on the Arkansas River upstream of Pueblo, in terms of peak discharge and stage, occurred
on June 3, 1921 (Follansbee and Jones, 1922).
Table 2.2: Largest Peak Discharge Estimates in the Arkansas River Basin Upstream of Avondale
USGS
Gaging
Station
No.
Station Name
Drainage
Area
(mi
2
)*
Period of
Record
Date
Peak
Discharge
(ft
3
/s)
Flood Type
07086000
Arkansas River at Granite, CO 427 1897 - 2002 06/28/1957 5,360 snowmelt
07091200
Arkansas River near Nathrop, CO 1,060 1965 - 2002 07/14/1995 5,540 snowmelt
07091500
Arkansas River at Salida, CO 1,218 1895 - 1979 06/29/1957 9,220 snowmelt
07093700
Arkansas River near Wellsville, CO 1,485 1961 - 2002 06/12/1980 6,240 snowmelt
07094500
Arkansas River at Parkdale, CO 2,548 1946 - 2002 06/18/1995 6,830 snowmelt
07095000
Grape Creek near Westcliffe, CO 320 1925 - 1995 08/02/1966 7,460 thunderstorm
misc.
Grape Creek in Canyon above Caon City, CO (32) 1925 07/21/1925 14,500 thunderstorm
07096000
Arkansas River at Caon City, CO 3,117 1889 - 2002 08/02/1921 19,000 thunderstorm
misc.
Wilson Creek near Mouth 61.3 1941 07/04/1944 16,800 thunderstorm
07096500
Fourmile Creek near Caon City, CO 434 1949 - 1997 07/11/1951 4,260 thunderstorm
07097000
Arkansas River at Portland, CO 4,024 1939 - 2002 06/05/1949 21,100 thunderstorm
07099100
Beaver Creek near Portland, CO 214 1971 - 1981 09/09/1973 4,800 thunderstorm
07099200
Arkansas River near Portland, CO 4,280 1965 - 1974 08/21/1965 23,900 thunderstorm
07099400
Arkansas River Above Pueblo, CO 4,670 1966 - 2002 08/01/1966 10,100 thunderstorm
07099500
Arkansas River near Pueblo, CO 4,686 1895 - 1975 06/03/1921 103,000 general storm
misc.
Monument Creek at Colorado Springs 75 05/30/1935 50,000 general storm
07105500
Fountain Creek at Colorado Springs, CO 392 1976 - 2002 09/02/1994 10,100 thunderstorm
07105530 Fountain Cr Bl Janitell Rd Bl Colo. Springs,
CO
413 1990 - 2002 04/30/1999 13,800 general storm
07105800
Fountain Creek at Security, CO 495 1965 - 2002 07/24/1965 25,000 thunderstorm
misc.
Jimmy Camp Creek near Fountain, CO 54.3 06/17/1965 124,000 general storm
07106000
Fountain Creek near Fountain, CO 681 1939 - 2002 05/28/1940 22,100 general storm
07106300
Fountain Creek near Pinon, CO 849 1973 - 2002 04/30/1999 19,100 general storm
07106500
Fountain Creek at Pueblo, CO 926 1921 - 2002 06/17/1965 47,000 general storm
07109500
Arkansas River near Avondale, CO 6327 1939 - 2002 06/18/1965 50,000 general storm
07110500
Chico Creek near North Avondale, CO 864 1921 - 1946 06/04/1921 28,600 general storm
* values in parentheses (.) are contributing drainage area estimates from Follansbee and Sawyer (1948)
Regional flood peak discharge data from U.S. Geological Survey and Colorado Division
of Water Resources gaging stations and miscellaneous sites were examined to gain an
understanding of maximum flood experience to date in an area near and adjacent to the Arkansas
River basin. A regional envelope curve was prepared from these data, based on techniques
presented in Cudworth (1989). Regional peak discharge envelope curves are useful to: (1)
expand the flood data base for the watershed of interest with data from nearby streams; (2)
portray extreme flood potential in the area being studied; (3) gain an understanding of the
19
regional hydrometeorology based on the largest floods; and (4) compare probabilistic estimates
of peak discharge and/or design floods to the largest historic floods in the region. Envelope
curves do not typically have any probability or frequency associated with them (Crippen and
Bue, 1977; Crippen, 1982). Some investigators have proposed various methods to estimate
probabilities for the largest observed regional floods, such as those used to develop envelope
curves (Fuller, 1914; Carrigan, 1971; Wahl, 1982; Jarrett and Tomlinson, 2000 Figure 13; Vogel
et al. 2001; Troutman and Karlinger, 2003), but these techniques need further development and
testing prior to implementation.
The hydrologic region for the peak discharge envelope curve was selected based on the
eight Arkansas River hydrologic regions in Colorado (Patterson, 1964), the nine South Platte
hydrologic regions in Colorado (Matthai, 1968), the transition zone/South Platte region used by
Jarrett and Costa (1988), the Mountain region (east of Continental Divide) from Kircher et al.
(1985), and the Mountain (east of the Continental Divide) and Plains regions presented in Vaill
(2000). Peak discharge data from east of the Continental Divide within Colorado, excluding the
Rio Grande, were used. This mixes data from the Mountain and Plains regions. The area
encompasses eastern Colorado at the Continental Divide (headwaters of the Arkansas and South
Platte Rivers), north to the Wyoming border, east to the Kansas and Nebraska borders, and south
to the Rio Grande and New Mexico. Data from the Arkansas and South Platte drainage basins in
Colorado, that are subsets of the Missouri River basin, were utilized. This region is a subset of
Region 12 presented by Crippen and Bue (1977, Figure 1). Crippen and Bue included
southwestern Nebraska, eastern New Mexico and west Texas (including the Canadian, Devils
and Pecos Rivers).
The Mountain and Plains hydrologic regions from Vaill (2000) were combined in order
to obtain a larger sample of peak discharge data potentially applicable to the Arkansas River
basin. These regions were also combined in order to address one of the issues raised when
developing the current PMF using generalized procedures for Pueblo Dam (Bullard and
Leverson, 1991). This issue was the fact that record peak flows from June 1965 in the Colorado
Front Range were significantly larger than the 1968 inflow design flood peak for Pueblo Dam.
Peak-flow data within this region were segregated into three main groups: South Platte data,
upper Arkansas data (upstream from Pueblo), and lower Arkansas data. The peak flows were
further segregated by type: indirect measurement or estimate at a gaging station. Peak discharge
data from areas outside Colorado were not considered applicable to the Arkansas River basin.
Data from the Mountain and Plains regions include data from mixed populations. As
discussed below, and shown in Table 2.2, peak discharge estimates in the mountain region
(upper Arkansas and South Platte) are significantly lower than peak discharge estimates from the
Plains. The drainage area to Pueblo Dam spans both the Mountain and Plains Regions. McCain
and Ebling (1979), Jarrett (1987) and Jarrett and Costa (1988) describe a transition zone that
encompasses the area with elevations below about 7,500 to 8,000 feet in the canyons of the
Colorado Front Range, which is a transition between the two regions. There is clear evidence
20
that rainfalls have caused large magnitude flood peaks in the immediate area upstream and
downstream of Pueblo Reservoir. These floods are similar to those that have occurred on other
Front Range streams such as the South Platte River, but are distinctly different than the
snowmelt floods in upstream sections of the Arkansas basin.
A regional peak discharge envelope curve for Colorado, including areas east of the
Continental Divide within the Arkansas and South Platte basins, is shown in Figure 2.4. The
largest peak discharge estimates used to construct the curve are listed in Table 2.3 for this area.
The maximum, instantaneous peak discharge observed for a basin in eastern Colorado with a
comparably-sized drainage area to Pueblo Dam (4,560 mi
2
) is subject to large uncertainties. A
regional value is approximately 188,000 ft
3
/s for rainfall-dominated events, based on a relation
for drainage areas between 1,000 and 20,000 mi
2
(Figure 2.4). However, there is a distinct scale-
dependent feature to the data. It appears that rainfall over partial areas and/or watersheds less
than about 1,000 mi
2
follow a relation that has a higher slope and intercept than data from larger
watersheds. This relation would suggest a maximum observed peak flow of about 560,000 ft
3
/s
for a watershed with contributing drainage area of 1,000 mi
2
. Thus, the simple notion that flood
magnitude increases in some fashion with a concomitant increase in drainage basin scale is
rejected for this area in Colorado. Considering that the contributing rainfall area for the June
1921 Penrose storm (discussed below) is about 550 mi
2
(Follansbee and Jones, 1922), an
estimated peak for this storm and watershed scale is 428,000 ft
3
/s. These data indicate
approximate, maximum peak flood experience to date in the region, and are in general greater
than the existing design spillway capacity (191,500 ft
3
/s) at Pueblo Dam.
The regional relations shown in Figure 2.4 follow a typical Myers-type envelope curve
formula (Jarvis and others, 1936; Creagher et al., 1945):
QCA
n
(2.1)
where Q is peak discharge (ft
3
/s); C is a constant; A is the drainage area (mi
2
); and n is a slope.
This equation is empirical and has no theoretical basis. Based on data shown in Figure 2.4, it
appears that the parameters C and n are not constant and change with scale. Matthai (1969)
demonstrates a similar scale dependence with a change in parameters at 200 mi
2
. The curves are
considerably higher than Follansbee and Sawyer (1948), due to the addition of 1965 flood data.
The curves are similar in shape to that shown by Matthai (1969), but his equation was limited to
drainage basins less than 200 mi
2
. He used the same curve for larger areas as Hoyt and Langbein
(1955). However, the Hoyt and Langbein curve, as well as curves shown by most others (e.g.
O'Connor and Costa, 2004), includes data from most of the United States, such as Seco Creek
and the West Nueces River in Texas and the Eel River in California. The relation shown in
Figure 2.4 is considerably lower for drainage basins greater than 1,000 mi
2
than that depicted by
Crippen and Bue (1977) for their Region 12, as they also relied on data from Texas to define the
envelope curve for drainage basins greater than 1,500 mi
2
. The relation is also lower than the
Hoyt and Langbein (1955) and O'Connor and Costa (2004) curves because they also use data
from Texas and other locations within the United States. Those data are not considered
21
applicable to the Arkansas River basin in Colorado.
10 100 1000 10000
Drainage Area (mi
2
)
10
100
1000
10000
100000
1000000
P
e
a
k

D
i
s
c
h
a
r
g
e

(
f
t
3
/
s
)
S. Platte gages
upper Arkansas gages
lower Arkansas gages
S. Platte indirect meas.
upper Arkansas indirect meas.
lower Arkansas indirect meas.
approximate envelope medium basins
approximate envelope large basins
1
2
3
4
5
6
7
8
9
10
Q = 25,000 A
0.45
Q = 15,000 A
0.3
Figure 2.4. Maximum peak discharge data and drainage-area envelope curve for observations within the Arkansas
and South Platte River Basins in Colorado.
Four peak discharge estimates control the position of the envelope curve for scales less
than 1,000 mi
2
(Figure 2.4; Table 2.3): Bijou Creek near Wiggins (point 1), Rule Creek near
Toonerville (point 2), East Bijou Creek (point 3) at Deer Trail, and Jimmy Camp Creek near
Fountain (point 4). The June 1965 flood data (points 1, 3 and 4) were listed by Crippen and Bue
(1977) as the largest floods in Colorado for their respective drainage areas. The Jimmy Camp
Creek flood was also a point that defined the United States envelope curve (O'Connor and Costa,
2004). England (2004) suggested that the Jimmy Camp Creek peak may be overestimated and
that it be further reviewed. The USGS has agreed that the measurement rating be changed from
fair to poor.
Four peak discharge estimates control the position of the envelope curve for scales
greater than 1,000 mi
2
: South Fork Republican River (point 6), Purgatoire River (point 7),
Arkansas River at Nepesta (point 9), and the Arkansas River at La Junta (point 10). These points
are associated with the June 1921, May 1935 and June 1965 storms and floods. The largest peak
flow on the Arkansas River upstream of Fountain Creek (at Pueblo) is the June 1921 flood,
shown as point 8. It is slightly below the envelope curve relation.
22
Table 2.3: Largest Observed Peak Discharge Estimates that Define a Drainage Area-Peak Discharge Relation for
the Arkansas and South Platte River Basins
Point No.
(Fig. 2.4)
Station Name
Drainage
Area (mi
2
)
Date
Peak
Discharge
(ft
3
/s)
Measurement Type
Meas.
Rating
Flood Type
1 Bijou Creek near Wiggins, CO 1,314.0 06/18/1965 466,000 slope-area at misc. site poor general storm
2 Rule Creek nr Toonerville, CO 435.0 06/18/1965 276,000 slope-area at misc. site fair general storm
3
East Bijou Creek at Deer Trail,
CO
302.0 06/17/1965 274,000 slope-area at misc. site fair general storm
4
Jimmy Camp Creek near
Fountain, CO
54.3 06/17/1965 124,000 slope-area at misc. site fair general storm
5
Two Buttes Creek near Holly,
CO
817.0 06/17/1965 182,000 slope-area at misc. site good general storm
6
South Fork Republican River
near Idalia, CO
1,300.0 05/31/1935 103,000 slope-area at gage unknown general storm
7
Purgatoire River at Ninemile
Dam, near Higbee, CO
2,752.0 06/18/1965 105,000
estimated based on flow
over dam at gage
unknown general storm
8
Arkansas River near Pueblo,
CO
9
Arkansas River near Nepesta,
CO
10 Arkansas River at La Junta, CO 12,210.0 06/04/1921 200,000 slope-area at gage unknown general storm
There are three noteworthy features of the envelope curve. The first is that the nine
points that define the envelope curve all are observations that occurred east of the Front Range -
east of Pueblo, Colorado Springs and Denver. Orographics and topography appear to play a
significant role in extreme flood generation in the Colorado Front Range, and record flood peaks
have not been observed upstream of Pueblo other than the June 1921 flood. The second feature
is that three storms caused the extreme floods: June 1921 (three peaks), May 1935 (one peak),
and June 1965 (six peaks). The third feature is that all of the flood peaks that define the
envelope curve are classified as being caused by general storms. Storm classifications are
further discussed in Section 2.3.
There are four limitations to the current, regional envelope: (1) adequate sampling in
space and time for known, observed events; (2) scaling issues (changes in relationships at basins
greater than 1,000 mi
2
); (3) progressive regulation of flood peaks over time for these larger
basins; and (4) accuracy of the floods that define the envelope curve. Floods that are recorded at
multiple sources for the same rain-dominated flood (e.g. 1921, 1935, 1965, and 1976) limit
space-for-time substitutions and potentially determining an approximate exceedance probability
(return period) for the curve. Flood peaks for observed events are sometimes not well sampled
in space. There are extreme flood sampling and regulation problems for large basins. Crippen
and Bue (1977, p. 13) limit their analyses to basins less than 7,000 mi
2
because maximum
floods for larger basins in this region cannot be defined from data presently available. One
major problem is that the peak discharge estimates that define the envelope curves are uncertain;
23
their accuracy is either poor, fair or unknown (Table 2.3). Matthai (1990) describes several
other major problems with envelope curves, including data quality problems, partial area, and
that the curve may not be representative of the geologic and climatic conditions at one's point of
interest.
There is an additional, informative way to examine maximum peak discharge data in
orographic regions by including elevation. A relation combining peak flows, drainage area and
elevation, based on data in Jarrett (1987), England (1996), and updated through Water Year
2003, is shown in Figure 2.5. Data that control the location of the unit discharge-elevation
envelope are listed in Table 2.4. This relationship shows several distinct features. The most
important feature is the dramatic decrease in unit discharge with elevation. Jarrett (1987, 1990)
described this very same feature in terms of the South Platte River basin; here we expand the
focus to include the Arkansas River basin. The highest unit discharges are associated with very
high runoff from small watersheds less than about 50 mi
2
(Table 2.4). Most of the largest events
are associated with the July 1976 Big Thompson flood. The relation in Figure 2.5 supports
Follansbee and Sawyer's (1948) views on extreme floods within the Arkansas River basin
upstream from Parkdale. It is also nearly identical to that previously shown by Jarrett (1990).
Recent extreme floods (including those from July 1997) have not had higher unit discharges than
those from 1976. What one can also infer from this relationship is that extreme flooding on
large watersheds that include the foothills transition zone can be caused from partial-area runoff.
Thus, one needs to carefully consider the contributing drainage area above about 8,000 feet for
these watersheds (Jarrett and Costa, 1988). Two of the largest events that occurred within the
Arkansas watershed upstream from Pueblo are shown on Figure 2.5: Orman's Gulch (point 6)
and Arkansas tributary near Parkdale (point 8). However, these points, while relatively extreme
for the Colorado Front Range, do not define the unit discharge envelope curve. The peak-flow
and unit discharge data show that topography and physiography appear to play a dominant role
in extreme flood hydrology for the Arkansas River basin upstream of Pueblo. O'Connor and
Costa (2004) concluded that topography and basin physiography were important factors in
understanding and predicting high unit discharges.
2.2.3 Seasonality and Process Relationships
Understanding flood seasonality and rainfall/snowmelt processes (mixed populations)
are crucial to predicting extreme floods in Colorado Front Range watersheds. Here we focus on
annual peak flow and annual maximum mean daily flow data sets from selected gaging stations
within the Arkansas River basin to demonstrate flood seasonality and process relationships
within the watershed upstream from Pueblo.
24
4000 6000 8000 10000 12000
Elevation (ft)
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
12000
13000
U
n
i
t

P
e
a
k

D
i
s
c
h
a
r
g
e

(
f
t
3
/
s
/
m
i
2
)
S. Platte gages
upper Arkansas gages
lower Arkansas gages
S. Platte indirect meas.
upper/lower Arkansas indirect meas.
1
2
3
4
5
6
7
8
9
10
?
?
Figure 2.5. Maximum unit peak discharge data and elevation envelope curve for observations within the Arkansas
and South Platte River Basins in Colorado.
Table 2.4: Largest Observed Unit Discharge Estimates that Define an Elevation-Unit Discharge Relation
Point
No. (Fig.
2.5)
Station Name
Drainage
Area
(mi
2
)
Date
Peak
Discharge
(ft
3
/s)
Unit
Peak
(ft
3
/s/mi
2
)
Elev.
(ft)
Measurement
Type
Meas.
Rating
Flood Type
1
Big Thompson River trib. blw.
Glen Comfort
0.5 07/31/1976 6,950 13,113 7,400 SA (misc. site) poor thunderstorm
2 Dark Gulch at Glen Comfort 1.0 07/31/1976 7,210 7,210 7,280 SA (misc. site) poor thunderstorm
3
North Fork Big Thompson trib.
nr. Glen Haven
4
Big Thompson River trib. blw.
Loveland Heights
5 Devils Gulch nr. Glen Haven 0.9 07/31/1976 2,810 3,088 7,520 SA (misc. site) poor thunderstorm
6 Orman's Gulch nr. Swallows, 2.7 07/19/1965 7,000 2,632 5,187 SA (misc. site) fair thunderstorm
7
Jimmy Camp Creek near
Fountain
54.3 06/17/1965 124,000 2,284 5,546 SA (misc. site) fair general storm
8
Arkansas River trib. no. 2 at
Parkdale
0.2 07/27/1961 284 1,775 5,760 SA (misc. site) good thunderstorm
9 Molino Canyon nr. Weston 4.2 08/10/1981 5,100 1,206 6,730
culvert flow
(misc. site)
poor thunderstorm
10 Kiowa Creek at Elbert 28.6 05/30/1935 43,500 1,521 6,740 SA (gage) unknown general storm
25
Extreme floods in the Colorado Front Range typically occur from mid-April to late
August. Hoyt and Langbein (1955, p. 51) present a very generalized map of flood seasons for
the United States, and suggest that floods in Colorado and much of the Rocky Mountain west
occur in late spring. Hirschboeck (1991, p. 84) also suggested that late spring is the typical
season for which the largest flood peaks of the year occur in Colorado and much of the western
United States. Doesken (1991) simply states that floods occur primarily from April through
October in Colorado. The largest flood peaks in the Arkansas and South Platte River basins can
be used to determine flood seasonality. Based on the data shown in Tables 2.2 through 2.4, the
flood season for this region is late April through early September. General storms typically
occur in late April through mid-June, snowmelt floods occur in mid-June through mid-July, and
thunderstorms occur from early June through early September (Table 2.2). The most extreme
floods in terms of peak discharge (Table 2.3) and unit discharge (Table 2.4) confirm this
seasonality in the region.
Flood seasonality was explored using annual peak flow and maximum mean daily flow
data from six gaging stations on the Arkansas River main stem: Granite, Salida, Wellsville,
Parkdale, Canon City, and Pueblo (Figure 2.6). Runoff seasonality is very strong in the upper
watershed from Granite to Parkdale, and limited to May through August. Monthly distributions
of peak and maximum mean daily flows are unimodal, peak strongly in June, and daily and peak
frequencies are nearly identical. These data indicate that snowmelt runoff processes are the
dominant flood runoff mechanism in the upper watershed upstream of Parkdale. There is a fairly
consistent shape to the distributions from Granite to Parkdale (site a through d). At Canon City
and Pueblo, the distributions clearly change by having more spread, and indicate the influence of
the general storm rainfall-runoff process. Peaks at Canon become relatively more frequent in
July and August than at upstream sites. The peak distribution for Pueblo is nearly uniform for
June through August, and demonstrates that there are even differences between these two lower
elevation sites.
Snowmelt runoff within the basin shows a strong spatial coherence. The largest
snowmelt flood, in terms of peak discharge and volume, was the June 1957 runoff that lasted
approximately 21 days (Figure 2.7). Flow estimates are available at four locations for this
snowmelt flood: Granite, Salida, Canon City and Pueblo. The data from these locations show
that daily flows are remarkably similar at Salida, Canon City and Pueblo for the maximum
snowmelt runoff period. The data between the sites are highly correlated. However, there is
some variability and differences in runoff from Salida to Pueblo for several days. A rainfall-
runoff daily flow on May 17 at Pueblo was not observed at upstream locations. Likewise, there
are numerous small peaks on the falling limb of the snowmelt runoff hydrograph at Pueblo that
are not observed at upstream locations.
26
April May June July Aug. Sept.
0
0.2
0.4
0.6
0.8
peak
daily
0
0.2
0.4
0.6
0.8
peak
daily
0
0.2
0.4
0.6
0.8
peak
daily
0
0.2
0.4
0.6
0.8
peak
daily
0
0.2
0.4
0.6
0.8
peak
daily
0
0.2
0.4
0.6
0.8
peak
daily
(a)
(c)
(e)
(b)
(d)
(f)
F
r
e
q
u
e
n
c
y

(
n
o
.

e
v
e
n
t
s

e
a
c
h

m
o
n
t
h
/
N
)
Month
N=93 N=77
N=48 N=41
N=111
N=81
Figure 2.6. Histograms of annual peaks and 1-day maxima versus month for six locations within the Arkansas
River basin upstream of Pueblo. Sites are listed from upstream (snowmelt) to downstream (general storms): (a)
Granite; (b) Salida; (c) Wellsville; (d) Parkdale; (e) Canon City; and (f) Pueblo.
The maximum snowmelt runoff periods can also be substantially varying in shape,
volume and timing, but there is marked regularity within the basin. A recent very high snowmelt
runoff year was 1995; the record peak discharge at the Parkdale gage was observed for this
snowmelt runoff. Streamflow data for this runoff season were recorded at five locations:
Granite, Wellsville, Parkdale, Canon City, and Portland (Figure 2.8). The maximum runoff
during a 30-day period was nearly the same at locations within the basin from Wellsville to
Portland. As in 1957, there is some higher variability in runoff at the most downstream location.
In late May and early June 1995, there were several days from May 15 through about June 12
with slightly higher peaks and flows at Portland than at upstream locations. Overall, the data for
the highest snowmelt runoff periods demonstrate that there is similar response at various
27
locations within the watershed.
5/1 5/15 6/1 6/15 7/1 7/15 8/1 8/15 8/31
Day of Month (Water Year 1957)
0
2000
4000
6000
8000
10000
D
a
i
l
y

M
e
a
n

D
i
s
c
h
a
r
g
e

(
f
t
3
/
s
)
Pueblo
Canon
Salida
Granite
Figure 2.7. Record snowmelt flood within the Arkansas River basin during June 1957.
5/1 5/15 6/1 6/15 7/1 7/15 8/1 8/15 8/31
Day of Month (Water Year 1995)
0
1000
2000
3000
4000
5000
6000
7000
D
a
i
l
y

M
e
a
n

D
i
s
c
h
a
r
g
e

(
f
t
3
/
s
)
Portland
Canon
Parkdale
Wellsville
Granite
Figure 2.8. Recent snowmelt flood within the Arkansas River basin during June 1995.
In contrast to maximum snowmelt runoff, there are distinct differences in peak flow
behavior within the watershed. The location of the mixed-population transition zone between
28
snowmelt-dominant peak flows and rainfall-runoff dominant peak flows within the basin is
explored with peak flow and daily flow data. It is clear that the extreme flood mechanism at
Salida is snowmelt, and the mechanism at Pueblo is from spring general storms and summer
thunderstorms. The upper snowmelt and lower general storm flood hydrographs for these
locations are shown in Figure 2.9. The flood at Salida in 1957 was the largest peak and was due
to snowmelt. In contrast, the June 1921 flood at Pueblo was the largest recorded rainfall-
generated flood.
Figure 2.9. Largest recorded snowmelt (June 1957) and rainfall-generated flood (June 1921) hydrographs in the
Arkansas River Basin upstream of Pueblo.
Peak-flow relationships between the gaging stations at various locations in the watershed
clearly show that the transition between snowmelt-dominant runoff and rainfall runoff is
between Canon City and Parkdale (Figure 2.10). There is a relatively strong relationship
between Wellsville and Parkdale, indicating similar peak snowmelt runoff behavior. The
overlapping records between Salida and Canon City are the longest (77 years) for comparison
between snowmelt and snowmelt-rainfall flood peaks. These records show that there are clear
differences between snowmelt peak flows at Salida and the mixture of snowmelt and rainfall-
runoff dominant peaks at Canon City (Figure 2.10). Likewise, overlapping records at Canon and
Pueblo are relatively long (80 years). The peak-flow comparison shows essentially no
correlation between these two sites, indicating that the largest peak flows are caused by partial-
area rainstorms or storms with much higher rain rates closer to Pueblo. The concept of flood
processes can be further explored by comparison of peak flows and maximum mean daily flows
at a particular gaging station. Typically, for snowmelt-dominant systems, peak flows are not that
29
5/1 5/15 6/1 6/15 7/1 7/15 8/1 8/15 8/31
Day of Month
0
5000
10000
15000
20000
25000
30000
35000
D
a
i
l
y

M
e
a
n

D
i
s
c
h
a
r
g
e

(
f
t
3
/
s
)
Pueblo
Salida
much larger than the daily mean flows, and there is a strong relationship between annual peaks
and maximum mean daily flows. Relationships for Parkdale, Canon City, and Pueblo (Figure
2.11) clearly show that there is a strong snowmelt-dominant relation at Parkdale. The snowmelt
relationship is inferred when the data depict a near-linear function between peak and maximum
one day. Weak relationships at Canon and Pueblo indicate a mixture of snowmelt and rainfall-
runoff flood processes, where there are very large peak flows compared to maximum mean
flows. There is definitely snowmelt runoff at both locations; however, at Canon the near-linear
relationship is stronger than at Pueblo.
Figure 2.10. Peak-flow correlation relationships between upstream and downstream locations: Salida and Canon
City; Wellsville and Parkdale; Parkdale and Canon City, and Canon City and Pueblo.
30
0 5000 10000 15000 20000
Peak Flows at Canon City (ft
3
/s)
0
2000
4000
6000
8000
10000
P
e
a
k

F
l
o
w
s

a
t

S
a
l
i
d
a

(
f
t
3
/
s
)
0 2000 4000 6000
Peak Flows at Parkdale (ft
3
/s)
0
1000
2000
3000
4000
5000
6000
7000
P
e
a
k

F
l
o
w
s

a
t

W
e
l
l
s
v
i
l
l
e

(
f
t
3
/
s
)
0 2000 4000 6000 8000
Peak Flows at Canon City (ft
3
/s)
0
2000
4000
6000
8000
P
e
a
k

F
l
o
w
s

a
t

P
a
r
k
d
a
l
e

(
f
t
3
/
s
)
0 10000 20000 30000
Peak Flows at Pueblo (ft
3
/s)
0
5000
10000
15000
20000
P
e
a
k

F
l
o
w
s

a
t

C
a
n
o
n

C
i
t
y

(
f
t
3
/
s
)
N=77
r=0.33
N=38
r=0.91
N=45
r=0.83
N=80
r=0.13
0 1000 2000 3000 4000 5000 6000 7000
Parkdale Peak Discharge (ft
3
/s)
0
1000
2000
3000
4000
5000
6000
7000
P
a
r
k
d
a
l
e

M
a
x

1
D
a
y

F
l
o
w

(
f
t
3
/
s
)
0 5000 10000 15000 20000
Canon Peak Discharge (ft
3
/s)
0
2000
4000
6000
8000
10000
C
a
n
o
n

M
a
x

1
D
a
y

F
l
o
w

(
f
t
3
/
s
)
0 5000 10000 15000 20000 25000 30000
Pueblo Peak Discharge (ft
3
/s)
0
2000
4000
6000
8000
10000
P
u
e
b
l
o

M
a
x

1
D
a
y

F
l
o
w

(
f
t
3
/
s
)
N=47
r=0.94
N=110
r=0.47
N=79
r=0.56
Figure 2.11. Peak discharge-maximum mean daily flow relationships at Parkdale, Canon City, and Pueblo.
2.2.4 Historical Information and Floods
Historical information is valuable to extend peak flow estimates from gaging stations in
time. Historical information is defined, for the purpose of this study, as broad categories of data
collected by humans prior to establishing systematic protocols such as streamflow and
precipitation gaging stations. It generally consists of diaries, written accounts of settlements,
folklore, and descriptions that may document periods where no extreme weather and/or floods
31
have occurred. These accounts may also document historical floods. Historical floods are defined
as flood events which were directly observed by humans, generally in a non-systematic manner
by non-hydrologists (Baker, 1987). These events usually occurred and were described in some
qualitative and/or quantitative fashion prior to the peak flow gaging station (systematic) record.
Thomson et al. (1964), Gerard and Karpuk (1979) and England (1998) discuss historical data
that are useful for flood frequency analysis.
Historical information is typically utilized in flood frequency analysis for three main
purposes: (1) to extend the peak flow gaging station record length; (2) to provide estimates of
extreme storms and floods that may have occurred prior to the establishment of gaging stations;
and (3) to document potential limits on peak discharge magnitudes over time. The basic data
that are needed for flood frequency include: (1) a peak discharge time series; (2) some historical
period; (3) a discharge (stage) threshold for the historical period; and (4) knowledge of any
floods that exceeded the discharge threshold, or that no floods exceeded the threshold. Historical
information and historical floods are used to estimate these last three elements. The time of
people arriving in an area and establishing settlements is used as a base time to extend the
extreme peak flow observation record. Likewise, human observation and documentation of
large storms and floods, and development of floodplains during this period, allow one to estimate
discharge exceedance and/or nonexceedance thresholds and the number of historical floods
(possibly zero) that exceeded the threshold.
Four main sources were used to document the time of human settlement, travel routes,
population distribution, observations, and historical records along the upper Arkansas River in
Colorado: (1) a history of Colorado (Baker and Hafen, 1927); (2) a history of Colorado and
people (Hafen, 1948); (3) railroad history and development of the Denver and Rio Grande
(Campbell, 1922); and (4) railroad guide of the Royal Gorge (Osterwald, 2003). The goal of
reviewing information from these sources was to estimate the year to start the historical period at
each gaging station site. The focus was on five locations within the Arkansas River basin where
flood frequency estimates are made: Pueblo, Canon City, Parkdale, Wellsville, and Salida.
The earliest accounts of humans (other than native Americans) visiting the Arkansas
River basin was in November 1806 by Zebulon Pike, who camped near the confluence of
Fountain Creek and the Arkansas River (Campbell, 1922). Many people came to Colorado in
search of gold and silver. The town of Fountain City (site of Pueblo) was begun in November
1858 (Baker and Hafen, 1927; Hafen, 1948). Gold was soon discovered in South Park and on
the Blue River, and Canon City and Colorado City were formed in 1859 (Hafen, 1948). In June
of the same year (1859), Pueblo expanded, and the village consisted of some thirty cabins
(Hafen, 1948). Most of the cities and towns along the Arkansas River valley were settled and
subsequently grew in response to mining and railroads. The Canon City Times, the first
newspaper in Canon City, was established on September 8, 1860, and the town had a population
of 800 with 40 business houses (Hafen, 1948). By 1868 Canon City had achieved some
prominence, and the state penitentiary was located there (Campbell, 1922). The Denver and Rio
32
Grande Railroad completed its line to Leadville in 1880; the towns that grew up along the line to
the south of Leadville were Salida in 1880 and Buena Vista in 1879 (Baker and Hafen, 1927).
The largest town in the mountains west of Canon City is Salida; it was settled in 1880 at the time
the railroad was built up the Arkansas valley (Campbell, 1922).
Historical information within the Arkansas River basin, which includes large floods
outside the streamflow gaging station period of record, helps to extend the record length, and
place extreme floods within the record in their proper context. A longer record provides more
assurance for peak discharge probability model selection and reduced variance of estimated
extreme flood quantiles. In addition to the historical sources listed above, data and information
were gathered that documents the positive evidence of historical (pre-gaging station) flooding,
and periods of no flooding, in the upper Arkansas River basin upstream from Pueblo, Colorado.
The major sources of historical flood information and data used in this research were obtained
from Follansbee and Jones (1922), Munn and Savage (1922), Follansbee and Sawyer (1948),
Patterson (1964), and Crowfoot et al. (2004).
The historical information and data indicate large floods might have occurred in the
Upper Arkansas River basin in the vicinity of the city of Pueblo in water years 1826, 1864, 1884,
1889, 1893, 1894 and 1921 (Follansbee and Jones, 1922; Follansbee and Sawyer, 1948; Hafen,
1948). The most disastrous flood in the Upper Arkansas was the Pueblo flood of June 1921
(Hafen, 1948; Follansbee and Sawyer, 1948). However, there is only sufficient quantitative
information to determine approximate magnitudes for the 1864, 1893, 1894 and 1921 flood
peaks. Based on the information in these reports, the start of the historical period and the
historical flood years with quantitative estimates at each site are summarized in Table 2.5.
Estimates for these floods are discussed in Section 4.
Table 2.5: Historical Record Start Date and Floods at Select Locations Within the Arkansas River Basin
Flood Frequency Site
Start of Historical
Record
Historical Flood Years
(Outside Gage Record)
Source(s)
Arkansas River near Pueblo 1859
(1864), (1893), (1894)
June 1921
Campbell (1922); Follansbee and Jones (1922); Baker
and Hafen (1927); Hafen (1948)
Arkansas River at Canon City 1868 August 1921
Campbell (1922); Baker and Hafen (1927); Hafen
(1948)
Arkansas River at Parkdale 1868 ---
Campbell (1922); Baker and Hafen (1927); Hafen
(1948)
Arkansas River near Wellsville 1880 --- Campbell (1922); (Baker and Hafen, 1927)
Arkansas River at Salida 1880 1957
Campbell (1922); (Baker and Hafen, 1927); Crowfoot
et al. (2004)
2.3 Regional Extreme Storms and Hydrometeorology
The moisture source for precipitation forming in the Arkansas River basin and vicinity is
predominantly Gulf of Mexico and subtropical Atlantic moisture from the southeast, and some
Pacific moisture from the west for the headwaters (Doesken, 1991). The majority of
33
precipitation falls as snow during the winter months. The snowmelt period is typically during
the months of May and June, with little snowpack remaining in the basin in the summer months.
Two major precipitation patterns affect the Arkansas River watershed in spring and summer and
can result in significant rainfalls and large floods. From March through June, midlatitude
systems cross the region, strengthen on the leeward side of the Rockies and draw moisture into
eastern Colorado. This moisture and increased convective activity result in periodic, widespread
rainfall and occasionally severe thunderstorms east of the mountains (Doesken, 1991).
Subtropical moisture from the Atlantic drifts northward to eastern Colorado starting in early
July. This monsoon moisture peaks near the beginning of August, then gradually weakens and
moves out of the region in late summer. The monsoon is responsible for the frequent summer
thunderstorms in the southern Rocky Mountains.
2.3.1 Extreme Storm Database
An electronic database of extreme storms and pertinent characteristics was developed
from existing data sets. The database was developed to provide quantitative estimates of
extreme storm rainfall in space and time for estimating extreme storm probabilities (Section 5)
and subsequent runoff modeling (Section 6). Two main sources of extreme storm data were used
to develop the database. The first is depth-area-duration (DAD) data from the Corps of
Engineers storm catalog (USACE, 1945- ) and Bureau of Reclamation cooperative storm studies.
The DAD data are used nearly exclusively in developing regionalized hydrometeorological
reports that provide Probable Maximum Precipitation (PMP) estimates (e.g., Hansen et al.,
1988). The second source is a recently developed extreme storm catalog for Colorado (McKee
and Doesken, 1997).
An initial geographic region was used to select storms from the DAD catalog for
consideration in developing extreme storm probability estimates. This region covers the United
States from the Canada to Mexico borders, and between about longitude 99
o
and the Continental
Divide. This region is nearly the same as that used by Hansen et al. (1988). About 110 storms
with some DAD data (not all complete) are located in the region. Most of these storms
correspond to the major and supplemental storms listed in Hansen et al. (1988). An
electronic database was developed using these 110 storms and includes the following
components: DAD data; storm start and end dates; assignment number; total duration; maximum
center location (nearest town and state); latitude and longitude of storm center; storm period; and
storm start and end times. The DAD data are given for specific durations (typically six-hour
increments) and area sizes. An example of a portion of DAD data for a storm is shown in Table
2.6. The storm orientation and major-to-minor axis ratio were estimated from the DAD
summary for each of the 110 storms and included in the database.
34
Table 2.6: Example Depth-Area Duration Data Table
Rainfall Depth (in) for Duration (hours)
Area (mi
2
) 6 12 18 24 30
10 10.4 11.3 12.0 12.0 12.0
100 8.8 10.4 11.0 11.1 11.1
200 7.9 9.7 10.3 10.4 10.4
500 6.5 8.4 9.0 9.1 9.1
1,000 5.4 7.1 7.8 7.8 7.8
2,000 4.2 5.4 6.1 6.2 6.2
The Colorado Climate Center (CCC) storm catalog (McKee and Doesken, 1997) is a
listing of 328 storms in the Rocky Mountain region, including 14 states in the intermountain
west and Great Plains. It consists of a simple table with the following elements: storm number;
storm name; state; storm date; region; storm type; latitude and longitude of storm center;
maximum precipitation; remarks; and notes if there are Reclamation storm files and DAD
studies available. The CCC catalog is a comprehensive index to storms in Colorado and parts of
the Rocky Mountain region. It is used to define the most important extreme storms that should
be considered in the Colorado Front Range. However, it has limited utility as there are no
detailed, quantitative storm data with depths, durations, areal distributions, etc. available for each
storm. Summaries do exist for all the storms, and the information in the CCC files for each
storm is of varying quality. The catalog is most useful for extending the DAD storm catalog by
determining locations of extreme storms, seasonality, approximate frequency of storm center
locations, and approximate duration. Additionally, the July 1997 Fort Collins (Doesken and
McKee, 1998) storm was included in the electronic database after manual processing of isohyetal
maps to develop DAD tables.
A review of the combined USACE/Reclamation DAD and CCC extreme storm catalog
for the Colorado and Rocky Mountain region indicates that there are several observed, extreme
flood-producing rainstorms in the lower elevation portions of the study watershed. There are 23
storms with DAD data and 154 storms from the CCC catalog that have been observed within
Colorado east of the Continental Divide (Figure 2.12). The majority of observed Colorado Front
Range flood-producing storms are shorter duration (generally less than 24 hours), high-intensity
convective (cloudburst) events. In some cases these are embedded in longer-duration storms that
produce somewhat heavy rainfall over several days. Large magnitude rainfall from these storms
is typically limited in areal extent, from tens to several hundreds of square kilometers. There is a
lack of evidence of longer duration, cyclonic storms that cause flooding in the Colorado Front
Range, especially at higher elevations (above about El. 7,500 ft). Based on an examination of
regional rainfall and streamflow records, Jarrett (1987, 1993) hypothesized that an elevation
limit exists for extreme floods caused predominately by rainfall in this region. The elevation
limits are approximately between 7,500 ft (2,290 m) and 8,000 ft (2,440 m) in the South Platte
and Arkansas River basins. However, Jarrett (1987) recommended that further work was needed
35
to document physical evidence of flooding (or lack of flooding) in the Arkansas River basin.
There are few observed flood-producing storms in the Upper Arkansas headwaters region
upstream of Wellsville (Figure 2.12); storm and streamflow data indicate that these storms are
very localized and cover small areas (several kilometers). The largest flood-producing storm
that occurred within the watershed is the June 1921 Penrose storm. The second-largest storm
and flood was the May 1894 Ward District storm. Other storms within the region are also
considered for transposition.
Figure 2.12. Locations of extreme storms near the Arkansas River watershed from the CCC catalog (green dots)
and those from the DAD database (labeled red dots).
In order to estimate extreme storm and flood probabilities for the Arkansas River basin
above Pueblo, storm transposition is used to increase the data base forprobability estimation.
From the DAD catalog, 77 storms are considered for analysis. These storms were chosen using
the Continental Divide to 103
rd
Meridian (CD-103) geographic region (Hansen et al., 1988)
within the United States, and also included the April 1900 Springfield storm, the October 1908
May Valley storm, and the May 1935 Hale storm centered in eastern Colorado. In addition to
these 77 storms, 33 supplemental storms located between about the 103
rd
and 98
th
meridian are
also examined. McKee and Doesken (1997) describe a recommended final list of thirty-six
storms for consideration in investigating extreme rainfall potential in the Rocky Mountain region
of Colorado. Six of these storms occurred in the Arkansas and South Platte River basins and
have available DAD data. A final subset of 40 storms was closely examined and considered for
36
stochastic storm transposition (Table 2.7).
Table 2.7: Extreme Storms from DAD Catalog Considered for Transposition to Pueblo Watershed
Date Assignment No. Location State
Duration
(hours)
Orientation
(degrees)
total 10mi
2
depth
max 24hr
10mi
2
depth
areal extent mi
2
05/30/1935 MR 3-28A Cherry Creek CO 24 47 22.20 22.20 6,300
09/20/1941 GM 5-19 McColleum Ranch NM 78 16 21.20 12.10 38,000
05/30/1935 MR 3-28AZoneA Hale CO 24 32 21.20 21.20 1,291
05/04/1969 19690504bemCO Big Elk Meadow CO 96 14 18.21 11.83 5,000
06/13/1965 SW 3-23 Plum Creek CO 181 5 18.10 13.20 39,266
09/27/1923 MR 4-23 Savageton WY 108 46 16.90 9.50 95,000
06/06/1964 NP 2-23 Gibson Dam MT 36 141 16.40 14.90 12,096
06/17/1921 MR 4-21 Springbrook MT 108 38 15.10 13.30 52,600
06/09/1972 MR 10-12 Rapid City SD 12 14.90 2,000
07/29/1997 19970729pawCO Pawnee Creek CO 24 65 14.00 12.00 1,070
06/06/1906 MR 5-13 Warrick MT 54 90 13.30 10.20 40,000
07/21/1905 GM 3-13 Elk NM 108 80 13.10 5.70 44,000
06/12/1949 R7-2-5 Prospect Valley CO 36 63 13.00 9.10 360
06/02/1921 SW 1-23 Penrose CO 24* 0 12.00 12.00 1,000*
07/27/1997 19970727ftcCO Fort Collins CO 32 0 12.00 10.00 1,000
07/31/1976 19760731bgtCO Big Thompson CO 4 25 11.70 50
08/30/1938 MR 5-8 Loveland CO 126 42 10.60 7.00 21,500
06/01/1953 19530601bltMT Belt MT 48 10.40 8.60 14,000
05/20/1941 GM 5-18 Prairieview NM 108 5 10.00 6.50 44,000
10/09/1930 SW 2-6 Porter NM 60 26 9.90 9.90 27,700
07/19/1915 SW 1-18 Tajique NM 240 26 9.90 5.20 95,000
04/29/1914 SW 1-16 Clayton NM 66 22 9.60 9.00 36,500
09/15/1919 GM 5-15B Meek NM 54 25 9.50 7.40 75,000
06/17/1947 MR 7-16 Gering NE 10 156 9.40 220
08/30/1938 R4-1-23 Waterdale CO 108 35 9.20 8.50 57,000
06/19/1916 R6-1-8 Sun River Canyon MT 66 106 8.90 6.20 10,000
06/16/1948 19480616dprMT Dupuyer MT 48 8.80 2,275
05/29/1894 MR 6-14 Ward District CO 60 165 8.50 5.60 25,300
08/29/1942 SW 2-29 Rancho Grande NM 84 12 8.00 7.90 35,600
06/03/1908 MR 5-15 Evans MT 72 42 8.00 6.50 20,000
06/06/1913 SW 1-14 Fort Union NM 132 161 7.90 5.10 23,000
09/26/1904 SW 1-6 Rociada NM 90 29 7.90 6.60 70,000
05/26/1937 GM 5-17 Ragland NM 84 69 7.80 4.40 37,000
05/30/1948 MR 7-18 Fort Collins CO 8 73 7.80 83
09/03/1911 MR 5-18 Knobles Ranch MT 72 36 7.60 3.70 37,000
09/27/1941 SW 3-1 Tularosa NM 48 49 7.50 5.70 66,700
04/14/1921 MR 4-19 Fry's Ranch CO 42 167 7.50 7.30 9,200
06/01/1915 MR 5-21 Adel MT 108 33 6.70 3.60 12,800
08/06/1929 SW 2-27 Valmora NM 144 37 6.60 4.60 49,000
04/22/1900 MR 5-10 Big Timber MT 60 0 6.60 3.80 30,000
* restricted storm area and duration to the core rainfall area directly over Penrose and the Arkansas watershed above Pueblo.
37
2.3.2 Radar Data and Flood Zones
Radar data are a good source for understanding and modeling precipitation over large
watersheds in space and time. The NWS WSR 88D radar data from Pueblo (KPUX) were used
by Javier et al. (2004, 2005) to examine space-time storm rainfall properties in the Arkansas
River basin. Results of their work were used to document storm rainfall with elevation. There
are several quality issues with ground rainfall estimates in orographic regions within the western
United States that result from using WSR 88D radar data (e.g. NRC, 2005). These typically
include beam blockage, degradation of the signal with distance from the radar, and calibration of
the rainfall-reflectivity (Z-R) relationship. However, the radar data are extremely valuable for
two purposes: estimation of storm rainfall in space (areal extent, coverage, coherence, spatial
correlation); and estimates of rainfall depth with time at all locations in space. From these data,
one can determine nondimensional depth-duration curves, and the storm spatial structure,
regardless of magnitude. It is very difficult to obtain this information based on raingage
networks as the network coverage in the western United States is fairly sparse, and does not
capture orographic effects particularly well. Radar data from the NWS KPUX radar in Pueblo
were processed by Princeton University. Javier et al. (2004, 2005) described the methods used
in developing a radar-based catalog of sixty-six storms based on data from 1995 through 2003
from the KPUX radar. The radar data were used to derive 5-minute rain rates at a spatial
resolution of 1 km
2
for each event, over a 200 km by 200 km area. The storm events from the
radar data are for the warm season during the months June, July and August. Storm activity for
a radar grid cell is computed as the number of hours of storms over a grid divided by the total
number of stormy hours over the entire domain. Storm activity highlights regions with larger
storms as they account for more time over each grid point. Based on aggregate properties of the
sixty-six events, storm activity is high along the Front Range, especially at the base of Pikes
Peak, and the area between Pueblo and Canon City (Figure 2.13). The larger values of storm
activity over the Plains illustrate the development of larger thunderstorm systems as storm
elements move eastward from the Front Range. There is very little extreme storm activity in the
watershed upstream of Wellsville, and at higher elevations (Figure 2.14). Accumulated
precipitation from the 66 events also shows a strong decrease with increasing elevation (Figure
2.15). These results suggest that restrictions should be placed on storm center location and on
storm areal extent in modeling. Stochastic storm transposition and storm modeling (Section 5)
incorporate these observations.
Based on streamflow data analyses, the storm database, and radar data, an estimate of the
extreme flood zones for rainfall, transition (rainfall to snowmelt) and snowmelt for the Arkansas
River basin is shown in Figure 2.16. The zones integrate the effects of drainage area, river
network, elevation, streamflow, storm rainfall and radar information.
38
Figure 2.13. Geographical distribution of storm activity (percent) of the sixty-six storms events from 1995-2003
based on KPUX radar (Javier et al., 2005).
1000-
1500
1500-
2000
2000-
2500
2500-
3000
3000-
3500
3500-
4000
4000-
4500
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6
E
l
e
v
a
t
i
o
n

[
m
]
Avg storm activity [%]
Figure 2.14. Storm activity (percent) as a function of elevation of the sixty-six storms events from 1995-2003
based on KPUX radar (Javier et al., 2005).
39
1000-
1500
1500-
2000
2000-
2500
2500-
3000
3000-
3500
3500-
4000
4000-
4500
0 50 100 150 200 250 300 350 400 450
E
l
e
v
a
t
i
o
n

[
m
]
Avg rainfall accumulation [mm]
Figure 2.15. Storm accumulations as a function of elevation of the sixty-six storms events from 1995-2003 based
on KPUX radar (Javier et al., 2005).
Figure 2.16. Estimated extreme flood zones in the Arkansas River basin.
40
2.4 Summary
This section presented streamflow and storm data within the Arkansas River basin and
surrounding region. Analysis of streamflow records revealed several important features that
reflect the physical hydrology of this watershed. A database of storm rainfall was developed for
modeling extreme storms; it included DAD data and storm properties to implement the elliptical
storm model presented in Section 5.
The largest peak flows in the region resulted from three extreme storms: June 1921, June
1965 and July 1976. There appears to be a very clear break in scaling relationships between
peak flows and drainage areas at scales larger than about 1,000 mi
2
(Figure 2.4). It is inferred
that this change in relationship is due to partial-area storms in orographic areas, including
elevation effects (Figure 2.5). Extreme storms and radar data suggest the partial-area storm
concept is valid. Peak flow, mean daily flow and hydrograph analyses clearly demonstrate
mixed-population rainfall-runoff and snowmelt runoff areas within the Arkansas River
watershed. Peak flow magnitudes within the basin are dramatically reduced with elevation
(Figure 2.5), supporting Jarrett's (1987, 1993) elevation limit hypothesis. This hypothesis
suggested that above approximately elevation 7,500 ft there is a limit to substantial rainfall-
produced flooding.
Streamflow records were analyzed for seasonality and correlations between locations in
the basin. Snowmelt runoff shows a strong spatial coherence, as demonstrated by the largest
snowmelt runoff hydrographs in June 1957 and June 1995. The upper watershed upstream of
Parkdale has maximum peaks and daily flows in May and June. A transition in monthly peak
distributions is observed between Parkdale and Canon City. Upstream of Parkdale, snowmelt
runoff is the dominant streamflow mechanism and the cause of the largest peaks. Downstream
of Parkdale, peaks and maximum mean daily flows are caused by large rainfall-runoff events,
and sometimes snowmelt runoff. The maximum flows usually occur in June through August
(Figure 2.6). Peak-flow and mean daily flow correlations support the transition between
snowmelt runoff upstream of Canon City and rainfall-caused large floods in the lower portion of
the watershed. Peak flows in the upper snowmelt-dominant basin are highly correlated; peaks in
the lower watershed between Canon City and Pueblo are essentially uncorrelated.
41
42
3.0 Paleoflood Hydrology
3.1 Introduction
Paleoflood data is a critical component for assessing hydrologic hazards at Reclamation
dams. By providing information beyond the historical gaging record, it typically serves as a
data-based means to extend the flood frequency curve beyond the 100-year return period. It also
provides an important check for the validity of flood frequency and rainfall-runoff modeling
results because it is based on direct observations of the physical system. Through field-based
studies, hydraulic modeling, and laboratory analysis, a paleoflood hydrology study can provide
data on the magnitude and frequency of floods and/or can provide a limit to flood magnitude
over a specified time interval. When used in flood frequency analysis, paleoflood data can
greatly improve the flood hazard estimation for return periods beyond the period of record.
When extreme floods or high outliers are present in the record, such as the June 1921 flood on
the Arkansas River, their return period may be much greater than the gage period of record.
Placing a large historical flood into frequency context with geological deposits provides a more
accurate estimate of the true return period of the extreme flood.
3.2 Objective
The paleoflood analysis fulfills Task 3 of the detailed probabilistic flood hazard study for
Pueblo Dam. The objective of this task is to collect paleoflood data in the Arkansas River basin
upstream of Pueblo Dam in order to extend the temporal and spatial distribution of existing
observations. The task is limited in scope, and therefore focuses on describing the magnitude
and frequency of floods in short reaches of the upper Arkansas River. The analysis, however,
should be adequate to provide physical information that will complement modeling components
of the overall study and will be used specifically as input to flood frequency analysis (Section 4)
and for rainfall-runoff modeling comparisons (Section 6).
3.3 Geologic Setting
The Upper Arkansas River Basin lies in the Southern Rocky Mountain Physiographic
Province, a subdivision of the Rocky Mountains, which separate the Interior Plains from the
Intermontane Plateaus and Pacific Mountain System (Hunt, 1967). The Rocky Mountains were
formed by folding, faulting, and uplift during the Laramide orogeny of Late Cretaceous and
Tertiary time. Much of the current physiography as well as geology reflects these events.
The Southern Rocky Mountain Physiographic Province extends from southern Wyoming
to northern New Mexico; from east to west the province is generally composed of two parallel
43
mountain ridges separated by intermontane basins. The eastern slope includes the Laramie
Mountains, Front Range, Wet Mountains, and Sangre de Cristo Mountains while the western
slope includes the Park Range, Gore Range, Sawatch Range, Elk Mountains, San Juan
Mountains, and Jemez Mountains. Basins that separate the two mountain ranges are the North
Platte River Basin, North Park, Middle Park, South Park, Arkansas River Valley, San Luis Basin,
and Rio Grande trough. The mountain ranges are anticlinal folds formed during uplift, in which
overlying sedimentary units were stripped off, exposing the PreCambrian crystalline core. Uplift
occurred throughout the Tertiary along with volcanism (such as expressed in the San Juan
Mountains) and a series of intrusions into rocks in the central part of the province. Near the
mountain front (at the eastern slope), resistant sedimentary layers were turned up steeply,
forming hogbacks (Loomis, 1938; Hunt, 1967; Madole et al., 1987).
Although many competing hypotheses have been proposed in the last century, it appears
that at least one erosional surface or pediment (termed peneplain by early workers) formed
during or following the uplift phase (late Eocene?) and is preserved in several locations as a
planar bedrock surface at high elevations. Many workers have named this surface the Rocky
Mountain erosion surface, but other names, such as Sherman and late Eocene surface have been
given in different areas to what is believed by some to be the same feature (Bradley, 1987).
During the Quaternary, Pleistocene alpine glaciations have sculpted the upper elevations
of the range into erosional features such as cirques, and U-shaped valleys and shaped lower
elevations with thick and extensive gravelly outwash deposits that have terrace forms to which
piedmont alluvium is graded. During the Pleistocene, five glaciations and interstades shaped the
physiography of the Rocky Mountains. The glaciations can be grouped into pre-Bull Lake
(oldest three), Bull Lake, and Pinedale. Several stages within the latter two glaciations have also
been identified by the position of moraines, indicating advance and retreat within each major
glaciation. Two minor Holocene advances are also recognized in the Rocky Mountain region
(Richmond, 1965). Along the Upper Arkansas River, pre-Bull Lake, Bull Lake, and Pinedale
outwash terraces have been identified and are the dominant landforms along the river upstream
of the Royal Gorge. Fluvial processes in the Holocene are for the most part incapable of
reshaping the Pleistocene landforms in this area, leaving the majority of valleys and canyons
filled with glacial outwash. Downstream of Canon City, glacial outwash deposits disappear and
in their place, Tertiary and Pleistocene alluvial deposits associated with the erosion of the Rocky
Mountains overlie Cretaceous sedimentary bedrock (e.g., Taylor et al., 1975a,b; Scott, 1972a,b).
3.4 Previous Work
This review of previous work is a brief summary of research that is specifically
applicable to the paleoflood component of the detailed flood hazard study in the Upper Arkansas
River Basin. Early research outlines the formation of Tertiary and Quaternary landforms that
dominate the physiography of the Upper Arkansas River basin. Researchers such as Behre
44
(1933) and Powers (1933; 1934; 1935) provide a geologic history of Tertiary uplift and erosion
and of Quaternary glaciation and glaciofluvial processes. Behre (1933) discusses a 12,000-ft
erosional surface cut on bedrock that forms rock benches on the sides of wider valleys or flat
surfaces in mountainous areas. He hypothesizes that this surface is associated with the Tertiary
uplift of the Rocky Mountains. Behre (1933) also describes four terraces in the headwaters
between the Mosquito and Sawatch Ranges that postdate the formation of the 12,000-ft erosional
surface. The first and oldest (No. 1 terrace) of the four terraces is believed to be associated with
rapid uplift and erosion of the mountain front during the late Tertiary. The younger three
terraces are interpreted as glacial outwash deposits associated with the most recent major
glaciations in the region. Powers (1935) refers to work of Behre (1933), but expands on the
number of terraces while focusing on a longer reach of the Upper Arkansas River from Leadville
to Canon City. Powers finds evidence for two late Tertiary (?) erosional surfaces. The oldest
surface is the same as Behres 12,000-ft erosional surface, which they both correlate to the
Rocky Mountain peneplain of eastern Colorado (Lee, 1923). Powers believes the surface to be
late Pliocene in age, since it is not deformed like Pliocene sediments in the Upper Arkansas
River Basin. Powers refers to the second surface as the broad valley stage, in which high
erosional surfaces or dissected rock pediments ranging from ~10,000 ft at Leadville to 6,800 ft at
Royal Gorge were eroded in response to a second late Pliocene uplift of 2,000 to 3,000 ft.
Powers also describes seven fluvial terraces, which range in age from pre-glacial to post-glacial.
The oldest terrace (Terrace No. 7) is correlated to Behres No. 1 terrace, which is approximately
390 ft above the modern Arkansas River and was formed during or following the uplift. Terrace
No. 6 is also pre-glacial in age at ~300 ft above the Arkansas River as it cannot be traced to any
glacial moraines near the mountain front. Terrace No.s 5 to 3 at 200-80 ft above the modern
river are pre-Wisconsin, or Bull Lake, in age although not all can be traced to glacial moraines.
Terrace No. 2 (60 ft) can be traced to moraines associated with the Wisconsin (Pinedale)
glaciation. Terrace No. 1 is post-glacial in age and is approximately 20 ft above the modern
Arkansas River.
While no paleoflood studies and very few Quaternary-specific studies have been
conducted within the upper Arkansas River Valley, there are several geologic maps that are
applicable to the current study. Several researchers have mapped geologic deposits along the
upper Arkansas River. Mapping scale ranges from 1:24,000 to 1:250,000. Typically, the small-
scale maps are compilations of more detailed mapping or include reconnaissance-level mapping
in areas that have not been mapped in detail. The most useful small scale maps for this study
were compiled by Scott et al. (1978) and Tweto et al. (1976). These maps show geologic
relationships throughout the basin and are useful for locating more detailed mapping. The large-
scale maps, such as the 1:24,000 scale, typically delineate bedrock and Pleistocene units in
detail, but are still somewhat lacking in their differentiation of Holocene alluvium. Due to the
limited scope of this study, we were able to gain important information from previous mapping
45
on the extent and age of units that are pre-Holocene in age. Publications used in mapping the
detailed study sites include Taylor et al. (1975a; 1975b), Scott (1972a; 1972b) and Van Alstine
(1974). Several of these maps are still reconnaissance-level; despite their lack of detail, they
provide the only known geologic mapping in the areas of interest.
3.5 Site Selection
An initial paleoflood site reconnaissance of the Arkansas River basin upstream of Pueblo
Reservoir was conducted from July 26-30, 2004. During the field reconnaissance four major
observations were made: (1) the lack of preserved Holocene terraces ( 1,000 years) along the
river; (2) strath terraces and bedrock control upstream of Canon City; (3) major anthropogenic
influences along entire system consisting of roads, railroads, and mining; and (4) locations of
promising sites to conduct detailed stratigraphy and hydraulic modeling. Six potential locations
for detailed study were initially identified: (1) upstream of Pueblo Dam and Reservoir; (2) Canon
City/Parkdale near Parkdale gage; (3) Loma Linda; (4) Swissvale (between Salida and Howard);
(5) Adobe Park; and (6) the Riverside Narrows BLM campground upstream of Buena Vista.
Figure 3.1. Map of the upper Arkansas River basin showing the location of stream gages and paleoflood study sites.
The rainfall-dominated and snowmelt-dominated areas of the basin are colored light green and dark green,
respectively.
46
Four of these sites, Pueblo State Park, Parkdale, Loma Linda and Adobe Park, were
ultimately selected for detailed study (Figure 3.1). These four reaches were surveyed and
modeled using HEC-RAS in order to make an estimate of the discharge required to overtop
specific terrace surfaces. At each of the four sites, soils were described and charcoal samples
were collected to estimate the age for the stabilization of the terrace surfaces. Peak discharge
estimates and radiocarbon dates were used to develop paleoflood estimates and nonexceedance
bounds for the upper Arkansas River. Site selection was based on several important factors:
geographic location, preservation of Pleistocene and Holocene deposits, hydraulic
characteristics, and magnitude of human impact. The geographic location of sites was important
in order to quantify the magnitude of floods in different parts of the basin. It is believed that
paleoflood magnitudes should reflect behavior similar to gaging records in that floods in various
parts of the basin are rainfall-dominated (lower elevations downstream of Canon City) or
snowmelt-dominated (higher elevations upstream of Canon City) (Figure 3.1). The Pueblo State
Park and Parkdale sites were selected to represent rainfall-dominant locations; snowmelt runoff
is reflected at the Loma Linda and Adobe Park sites.
Preserved Holocene and Pleistocene terraces provide information for paleoflood bounds
and are therefore critical components in each study reach. Downstream of Canon City, the most
extensive surfaces near the Arkansas River are late Holocene terraces, while upstream of Canon
City, Pleistocene glacial outwash terraces adjacent to the main channel are extensive. Holocene
alluvium in the upper basin is generally limited, principally as narrow floodplain surfaces.
The hydraulic geometry of each site was important because the one-dimensional
hydraulic model generally requires a relatively straight or gradually curved, single main channel
reach. Study sites should be located near the upstream end or in the middle of the modeling
reach in order to reduce downstream boundary effects in the model and potentially reduce
uncertainty of peak discharges at the stratigraphic sites. The reach must also be sufficient in
length to model stages that inundate terraces. Reaches that show minimal human disturbance or
interference by infrastructure such as railroads, highways, and bridges are also preferable. In
canyon reaches, which are typically optimal sites for paleoflood analysis, highways and bridges
can occupy a significant portion of the cross section and may introduce modifications to the
landscape that are large enough to rule out some reaches as suitable for paleoflood analysis.
3.6 Methods
3.6.1 Paleoflood Hydrology and Nonexceedance Bounds
Most conventional estimates for the frequency of large floods are based on extrapolations
from stream gaging records, commonly with record lengths shorter than 100 years. While these
estimates can be vastly improved by including historical information, which in the United States
may extend the record up to several hundred years, estimating the frequency of lower probability
floods (>100 years) becomes problematic. Rainfall-runoff models can also be used to make
47
flood frequency estimates, but commonly the results are based on an estimate of the precipitation
frequency. By depending on the precipitation record length and the geomorphic characteristics
of the basin in question, this latter method can also incorporate significant uncertainty.
Sensitivity analyses in flood frequency calculations have shown that the addition of paleoflood
information that spans a range of hundreds to thousands of years are valuable and can have a
significant impact on the shape of the flood frequency curve (Stedinger and Cohn, 1986; Blainey
et al., 2002; OConnell et al., 2002; England et al., 2003).
Paleoflood hydrology has a long history in a wide variety of settings throughout the
world (Costa, 1986; Patton, 1987; Baker et al., 1988). Early studies by Mansfield (1938) on the
Ohio River and Jahns (1947) on the Connecticut River recognized that historical floods on those
rivers overtopped stream terraces that had not been inundated for thousands of years. The
stratigraphic record present along streams in the form of terrace and floodplain deposits represent
direct indicators of the magnitude of large floods on a river and may be 10 to 100 times longer
than conventional stream gaging records of large floods (e.g., Patton, 1987; Baker, 1989; Jarrett,
1991; House et al., 2002). Thus, the study of paleofloods not only offers a means to reduce the
uncertainty in flood frequency analyses and verify the results of rainfall-runoff models, but also
helps to extend the record many times longer than the length of the stream and precipitation
gages or historical records (Costa, 1978).
Paleoflood studies can also provide a long-term perspective that can place historical
large-magnitude floods into temporal context and assist in reconciliation of conflicting
information (Boner and Stermitz, 1967; Helley and LaMarche, 1973; Klinger, 2006). If a gaging
record contains an exceptionally large flood, in a traditional flood frequency analysis it may be
assigned an unrealistically short return period, omitted from the frequency analysis as an outlier,
or recognized as a relatively rare event and given an arbitrary return period. Any estimate of a
flood with a recurrence greater than several hundred years that is based only on short-term gage
or even a long historical record of a few hundred years can incorporate large uncertainties.
Additionally, there are often large uncertainties associated with the reported peak discharges as
well. These uncertainties can similarly be reduced or clarified with further evaluation such as
hydraulic modeling.
One widely used technique in the study of paleofloods uses the fine-grained
sedimentological record that accumulates in backwater areas (slackwater) to construct a detailed
history of past floods (e.g., Patton et al., 1979; Kochel and Baker, 1988). This technique can be
extremely useful in characterizing the frequency of large floods, but can fall victim to the
inherent assumption that a sequence of slackwater sediments represents a complete and
continuous record of floods at a particular site. In addition, the physical setting of a backwater
site may not be ideally suited for reconstructing or accurately estimating the peak discharge for
the flood that is associated with a particular sequence of slackwater deposits in hydraulic models.
48
This can often be accomplished with more elaborate hydraulic modeling (e.g., Denlinger et al.,
2002).
Another methodology uses the age of a terrace surface that lacks evidence of recent
inundation, or alternatively displays evidence for long-term stability, to establish an upper limit
to flooding. This non-inundation approach can be extremely useful in flood hazard assessment
because geomorphic and stratigraphic information derived from the terrace surface can provide
an upper limit or bound on the age and magnitude of extreme floods (Levish, 2002). Rather than
constructing a detailed record of past floods, the non-inundation approach focuses on defining a
nonexceedance bound. Defining a nonexceedance bound is accomplished by identifying terrace
surfaces that serve as limits for the paleostage of large floods and estimating ages for those
terraces (Figure 3.2). These bounds do not represent actual floods, but instead limits on flood
magnitude over a measured time interval. Simply stated, a nonexceedance bound is a maximum
stage (and associated peak discharge) that has not been exceeded in the time period since the
terrace surface stabilized. It is not necessary to develop evidence of specific paleofloods using
this methodology. What is important is determining the discharge for a flood that has not been
exceeded over the time interval represented by the preserved geologic record. The
nonexceedance bound does not imply that the estimated peak discharge has ever occurred or that
such a flood is even physically possible.
Figure 3.2. Idealized channel cross-section illustrating the fluvial landforms important to the concept of a
nonexceedance bound.
49
Stable terrace surfaces are simply floodplains that have been abandoned due to either
stream incision and/or channel migration. Once abandoned, recognizable changes in the surface
characteristics of the terrace will follow a predictable path with time. An abandoned terrace
surface will tend to lose all evidence of having been inundated, becoming more planar and
smooth. Once stabilized, soil will begin to form on the sediment underlying the terrace surface.
Thus, stable terrace surfaces represent the field expression for the level of non-inundation or a
nonexceedance bound. They are a direct indication for the physical upper limit of floods along a
stream through the time period represented by the soil. The geometric characteristics of the
channel and terrace surfaces define the channel conveyance. The minimum overtopping depth
required for the initiation of large-scale erosion of and/or deposition on the stable terrace surface
can be evaluated formally in terms of shear stress or stream power (e.g., Parker, 1978; Andrews,
1984; Baker and Costa, 1987) or directly compared to empirical data from historical floods. On
high gradient streams, relatively shallow flow depths may produce higher values of stream
power or shear stress. The depth of flow associated with a nonexceedance bound is that which is
sufficient to cause modification of the terrace surface that is overtopped. Through step-
backwater modeling or other one-dimensional techniques (e.g., Webb and Jarrett, 2002), a peak
discharge for a nonexceedance bound can be easily derived from stage. In flood frequency
analyses, a nonexceedance bound is set as the time interval (age of the terrace surface) during
which the modeled peak discharge has not been exceeded.
The ages associated with the terrace surfaces that form the nonexceedance bounds should
almost always be considered minimum ages because of the problems related to dating the precise
time when a particular surface was abandoned and soil began forming. Soils contain a wide
variety of organic components of different ages due to the continual input of new carbon that is
mixed into the profile by biotic processes (e.g., Johnson, 1990; Hseih, 1992, 1993; Birkeland,
1999; Puseman and Klinger, 2001). Therefore, an estimate for the maximum discharge over the
minimum time interval since the terrace surface stabilized makes the bound a conservative
estimate for the flood hazard. These estimates are made even more conservative because through
time, channels may down cut and erode laterally resulting in apparently larger cross sections and
peak discharges than those that may have actually been present or occurred, so underestimation
of a nonexceedance bound in most cases is typically not a problem.
3.6.2 Data Collection
Soil and stratigraphic data were gathered at each site along stream bank exposures or in
hand dug pits approximately 2 ft (65 cm) wide and 3 ft (100 cm) deep (Appendix C). Soil
properties and sedimentary structures were described using terminology from Birkeland (1999)
and Boggs (1995). To estimate the age of terraces, individual charcoal pieces and bulk sediment
were collected. These samples were floated for macrobotanical materials, which were then
identified to a species level by Paleo Research Institute (Appendix D). Selected samples
50
underwent radiocarbon analysis, which determines the age of a sample based on the principles of
the radioactive decay of the carbon atom, at Beta Analytic, Inc. These ages can then be used in
conjunction with soil data and used to interpret the age of the stratigraphy of the stream terrace
and any individual flood deposits preserved in the stratigraphic record.
Each detailed study area was mapped based on aerial photo interpretation, using field
observations, and existing geologic data. Geomorphic mapping is very useful for placing study
sites into their proper geologic context. The geomorphic maps compiled for this study show the
distribution of Pleistocene and Holocene alluvium, the location of bedrock, HEC-RAS hydraulic
model cross sections, and soil pit locations where stratigraphic and age data were collected.
The topographic data for this project was collected between October 5 and October 7,
2004 using a Trimble 5800 GPS system. A Real Time Kinematic (RTK) survey was done at
each of the four sites, and data points were collected in the NAD 83 (feet) State Plane coordinate
system using the Colorado Central horizontal datum and the NAVD 88 vertical datum. A survey
control network was not used in this survey. The base station was set up over an autonomous
position at each site, therefore the coordinates of collected data points differ slightly from real
world coordinates. Relative to each other, all the points at a site are spatially correct.
At each site between seven and nine cross sections were surveyed. Cross sections were
extended as far as possible onto the adjacent terraces so that modeled high flows would be
contained within the measured cross section. If parts of the river cross section were too deep or
fast to survey, they were skipped. In these cases notes were made so that discharge information
at the time of the survey could be included in the HEC-RAS model. The cross section locations
at each site are included in Appendix E.
3.6.3 Hydraulic Modeling
After each site survey was completed, the data points were imported into Arc Map. The
data points were used to develop a surface representation or Triangular Irregular Network (TIN)
of each site. The HEC-GeoRAS extension for Arc Map was used to export cross sections from
the TIN for use in HEC-RAS modeling. Cross sections were edited in HEC-RAS to remove
redundant points. Roughness values (Mannings n) were applied separately to each cross section
(Table 3.1). Uniform values were applied to the channel and left and right floodplains based on
field observations and published values (Barnes, 1967; Jarrett, 1984; Jarrett, 1985). Normal
depth was used as the downstream boundary condition for each reach. The normal depth water
surface slope was estimated from a best-fit line through water surface elevations in each reach
using flows similar to those during the survey (300 ft
3
/s). It was assumed that the water surface
slope from the low survey flow would be similar to the water surface slope for high flows.
The vertical accuracy of the modeled water surface is highly dependent on channel
geometry, roughness, and discharge. Channel and floodplain roughness can change with
discharge because of complex channel geometry and vegetation. The highest level of model
51
Table 3.1: Study Reach Characteristics
Site
Site
Name
Cross
Sections
Reach
Length
(ft)
Reach Bed
Slope
(ft/ft)
Mannings n
Left
Overbank
Main
Channel
Right
Overbank
1 Pueblo 7 7,496 0.0030 0.050 0.040 0.040
2 Parkdale 8 2,331 0.0045 0.060 0.045 0.055
3
Loma
Linda
9 2,342 0.0019 0.050 0.045 0.050
4
Adobe
Park
8 1,869 0.0052 0.045 0.040 0.045
accuracy is achieved when the modeled water surface is matched to observed water surface
elevations. The study reach models were not calibrated because observed water surface
elevations were only available for the survey flow (approximately 300 ft
3
/s). Modeled
discharges were up to several orders of magnitude larger than the survey flow. These flows also
exceed the largest current meter and slope-area estimates by factors of 1.5 to 10.
A range of roughness values were used because the models were not calibrated and high
flow observations are not available in the study reaches. The channel and floodplain roughness
was increased and decreased by one hundredth from the roughness initially assumed. When the
roughness is increased the same surface will be inundated at a lower discharge. If the roughness
is decreased, the surface will be inundated at a higher discharge. The discharge required to
overtop a surface can therefore be given as a range of flows rather than a single flow. Using a
range of flows helps account for modeling errors related to complex channel hydraulics, lack of
data, and uncertainty in other input parameters.
The model for each reach was run with a range of flows to determine the flow required to
overtop each of the study sites by three feet. Three feet of inundation was used rather than the
wetting point because it was assumed that it would take up to three feet of water to produce the
shear stress required to erode an otherwise stable surface. A three foot depth also ensures that
the inundation depth is greater that the assumed vertical accuracy of the model, which is perhaps
plus or minus 0.5 feet. Initial model runs used 5,000 to 10,000 ft
3
/s flow increments to bracket
the overtopping discharge at each site. Once the coarse overtopping discharge was determined,
the flow increment was decreased to 1,000 ft
3
/s. Water surface elevations from the nearest cross
section were compared to ground elevations from the survey to determine the overtopping flow
to the nearest 1,000 ft
3
/s.
3.7 Geomorphic Data and Peak Discharge Modeling
This section contains the results of the paleoflood study for Pueblo Dam. Data including
soil/stratigraphic descriptions, radiocarbon analysis, geomorphic mapping, and peak discharge
modeling with HEC-RAS are presented for each of the four detailed study sites.
52
3.7.1 Pueblo State Park
The Pueblo State Park reach is 1.5 miles long and located upstream of Pueblo Reservoir
(Figure 3.1). The high water pool for the dam, if ever reached, would inundate the downstream
portion of this reach. In this reach, the Arkansas River flows in a ~-mile wide canyon incised
into Pleistocene gravels and Cretaceous limestones and shales. Pleistocene gravels are
differentiated based mostly on their height above the modern Arkansas River and are named
using terminology developed originally by Scott (1960). Pleistocene gravels mapped in the
study reach by Scott (1972a) include the Rocky Flats alluvium, Slocum alluvium, and Louviers
alluvium, which are 300, 120, and 70-80 ft above the modern Arkansas River, respectively.
These deposits are generally about 20 ft thick and form rounded gravelly hills in the study reach.
They lie unconformably on upper Cretaceous rocks of the Greenhorn Limestone, Carlile Shale,
and Niobrara Formation, which includes the Fort Hays Limestone member and Smokey Hill
Shale member. Holocene alluvial surfaces along the study reach are exclusively late Holocene in
age (labeled Qy2 in Figure 3.3). Previous research along the Front Range in Colorado identified
these deposits as Post Piney Creek in age (Malde, 1955), which is approximately 1500 years old
(Scott, 1963). Older Holocene deposits of Piney Creek age (>1500 years) do occur downstream
of this reach, and were mapped prior to the construction of Pueblo dam (Scott, 1972b). These
terraces are now inundated by the reservoir and therefore cannot be used in the current study to
develop nonexceedance bounds.
Figure 3.3. Geomorphic map of the Arkansas River in Pueblo State Park reach upstream of Pueblo dam and
reservoir. River flow is from upper left to lower right. Soil pits are denoted by white dots.
53
Three soil pits denoted AR1, AR2 and AR9 were dug in the study reach (Figure 3.3) The
stratigraphy and properties of the soils developed were described (Appendix C) to establish the
flooding history at each site. Site AR1 is a low terrace at the downstream end of the Pueblo State
Park reach and is composed of silty sand, sandy clay and gravel with weakly developed soil
properties (Figure 3.4). The surface horizon contains coal fragments and railroad slag suggesting
some relationship to the railroad, which may explain why the A horizon is over-thickened.
Three C horizons described from 39 to 100 cm are generally sandy with clay-rich lenses. The
lowermost horizon has a high gravel content with subrounded to rounded clasts of white quartz
to pink feldspar typically less than 4 cm in diameter.
Sites AR2 and AR9 are located at the upstream end of the study reach at about 12 and 19
ft above the active channel on an extensive late Holocene right-bank terrace (Qy1; Figure 3.3)
that grades to an alluvial fan near the valley margin. Site AR2 is near the lower end of the Qy1b
terrace and exhibits a continuous soil with a highly bioturbated A horizon and weak calcium
carbonate development at a ~60 cm depth in the Bk horizon (Figure 3.5). Parent material is fine-
grained alluvium, suggesting that this sediment was deposited exclusively by the Arkansas River.
Radiocarbon analysis of Juniper (Juniperius) charcoal from the Bk horizon yielded an age of
1740-1550 cal B.P. This suggests that the AR2 soil is approximately 1600 to 1800 years old.
Figure 3.4. Schematic of soil profile at site AR1.
Site AR9 is located upslope from AR2 on the north side of the old railroad grade (see
Figure 3.3). The upper 22 cm has characteristics typical of alluvial fan sediments with coarse
angular gravel on the surface and a 7-cm thick gravelly layer of locally derived angular
limestone fragments (Figure 3.5). The lower 48 cm are composed of river sediments with sandy
loam and silt loam textures and massive to moderate subangular blocky structure. Carbonate
morphology was not observed; however, the entire profile strongly to violently effervesces with
54
Figure 3.5. Schematic profiles of soils at sites AR2 and AR9. Sample AR2-3JU collected from the interval 63-78
cm yielded a radiocarbon age of 175040 yrs B.P. (1740-1550 cal yrs B.P.). Sample AR9-1JU collected from 56
cm yielded a radiocarbon age of 83040 yrs B.P. (790-680 cal yrs B.P.).
the application of 1N HCl, which may simply be due to the limestone parent material.
Radiocarbon analysis of charcoal recovered from a depth of 56 cm yielded an age of 790-680 cal
B.P. It is difficult to compare the radiocarbon ages from AR2 and AR9 because it is difficult to
correlate individual units in each deposit. Assuming equivalent fluvial deposition at both sites
and ignoring the alluvial fan deposit at the top of the AR9 profile (Figure 3.5; 0-22 cm) would
place the younger radiocarbon age at 34 cm, which would be higher in the stratigraphic section
than the dated sample at AR2. This suggests that these ages are perhaps in stratigraphic order,
which is supported by the extent of soil development at both sites. Assuming that these deposits
are similar in age, we would assign an age of 700 to 1800 years for a nonexceedance bound at
AR9. This age is similar to ages of about 1500 years previously developed for the Post Piney
Creek alluvium and reported by Malde (1955) and Scott (1963).
The Pueblo State Park reach is a relatively wide reach for developing paleoflood
nonexceedance bounds because of the potential for shifts in channel position, therefore, the
channel morphology within the reach was investigated to ensure that it would be adequate for
developing the necessary paleoflood information. On the basis of historical aerial photography,
the Arkansas River channel appears to have been in relatively the same position for at least the
last 68 years (ca. 1937; Figure 3.6). In addition, the first railroads were completed through this
reach in 1872 (Campbell, 1922) and much of the original roadbed is still intact on the landscape,
so the hydraulic modeling with the current topography should produce valid estimates for the
1921 flood peak discharge and stage. It is also unlikely that the channel has shifted significantly
55
Figure 3.6. Aerial photograph comparison of the Pueblo State Park reach between 1937 and 1999. A) the location
of the meander cutoff presumably by railroad construction; B) sediment deposited by Red Creek that forms the
Holocene terrace and the nonexceedance bound.
56
near AR2 and AR9 during the time frame of the nonexceedance bound. It appears that sediment
from Red Creek (Figure 3.1) has forced the river to the north side of the valley, and probably has
controlled the position of the river for at least the last 1000-2000 years. This idea is supported
by the radiocarbon ages (175040 14C yrs B.P.) and the formation of a calcic B horizon, as well
as the lack of any evidence that the river flowed south of this site on the landscape.
The overtopping discharges at each of the three sites near Pueblo Reservoir vary
dramatically. Site AR1 is located on the lowest and youngest alluvium (Qy2; Figure 3.3) in the
reach and is overtopped by a discharge of approximately 10,000 ft
3
/s. Site AR2, while being
immediately adjacent to the active channel, is on a slightly higher and older terrace (Qy1b;
Figure 3.3). The site is also located just upstream of cross section 5 and a large expansion in
valley width associated with the abandoned meander (Figure 3.3). This site is overtopped by
discharges ranging from about 30,000 to 40,000 ft
3
/s (Figure 3.7). Site AR9 is located just
downslope of an abandoned railroad grade and similarly just upstream of cross section 5 (Figure
3.3). The stratigraphy at AR9 indicates that the site is situated at the distal margin of a small
alluvial fan that has been shed across the back edge of the Qy1a terrace, but may have been
truncated or reworked by flow associated with the mainstem Arkansas River. The minimum
peak discharge needed to overtop this site is approximately 91,000 ft
3
/s. Peak discharges that
overtop the site by about 1-3 ft (estimation of nonexceedance bound) range from approximately
130,000 to 160,000 ft
3
/s. The overtopping discharge at this site did not vary dramatically with
changes in roughness because all the modeled flows defaulted to critical depth at this location.
This in part is the result of the rapid expansion in valley width just downstream from the site.
The abandoned railroad grade was present in the cross section geometry, but it was not modeled
as a levee so it does not significantly affect water surface elevations. Computed water-surface
profiles of flows for the preferred range of discharges are illustrated in Figure 3.8.
3.7.2 Parkdale
The Parkdale reach is located upstream of Canon City and just downstream of the
Parkdale gage and the U.S. Highway 50 bridge. Canyon walls in the reach are composed of
Precambrian granodiorite (Xgd on Figure 3.9); however, just upstream, Jurassic and Cretaceous
sedimentary rocks are juxtaposed against these Precambrian igneous rocks along the Mikesell
Gulch Fault. Glacial outwash terrace deposits of Pinedale age (>10,000 years) are also quite
extensive upstream of the study reach and Highway 50 (labeled Qpo in the northwest corner of
Figure 3.9). In the study reach, glacial outwash of presumed Pinedale age is exposed in a bank
exposure buried underneath Holocene alluvium and colluvium (Qy1/Qpo between cross sections
3-5; Figure 3.9). The youngest deposits mapped in the study reach include the active channel
(Qyc) and the floodplain (Qy2) sediments.
57
Figure 3.7. Schematic cross section showing the stage of the preferred nonexceedance bound peak discharge of
modeled flows and its relationship to geomorphic surfaces and stratigraphic sites in the Pueblo State Park reach.
Arkansas River at Pueblo State Park
4875
4880
4885
4890
4895
4900
4905
4910
4915
4920
4925
4930
0 1000 2000 3000 4000 5000 6000 7000 8000
Station, ft
E
l
e
v
a
t
i
o
n
,

f
t
Thalweg
300 cfs
130,000 cfs
150,000 cfs
160,000 cfs
Critical WSE
Site AR9
Figure 3.8. Computed water-surface profiles for estimated nonexceedance bound peak discharge in the Pueblo State
Park reach.
58
Figure 3.9. Surficial photogeologic map showing stratigraphic sites and the approximate location HEC-RAS cross
sections 3 through 8 in the Parkdale reach. Cross sections 1-2 at the downstream end of the reach are not shown.
Stream flow is from the top to bottom of the photograph.
Soil/stratigraphic descriptions at sites AR3, AR4 and AR5 demonstrate the variability in
deposits along the length of the Qy1/Qpo exposure. Sites AR3 and AR4 are on the Qy1/Qpo
map unit and represent the same soil with different parent material. Both soils exhibit a 20- to
30-cm thick loamy sand A horizon with less than 10% gravel (Figure 3.10). The underlying Bw
horizon is about 30 cm thick and has a medium loamy sand texture and subangular blocky
structure. Parent material was observed to vary between profiles such that potassium feldspar-
rich colluvium is described at AR4 and fine-grained alluvium is described at AR3. The Bk
horizon, which underlies the Bw horizon, is a loamy sand with subangular blocky structure.
Carbonate accumulation in the Bk horizon is in the form of a whitened matrix or filamentous
coatings on the undersides of gravel. Two radiocarbon samples from the overlying Bw horizon
were dated at 1230 40 B.P. and 1210 40 B.P. Calibration of these ages indicate that this soil
59
is about 1100 to 1300 years old. The contact at the change in parent material from fine-grained
sandy alluvium and the coarse-grained gravelly alluvium at each of the sites (Appendix C) marks
the unconformity or break between Holocene and Pleistocene deposits.
Figure 3.10. Soil exposure at site AR3. Sample AR3-1PI was collected from 48 cm and yielded a radiocarbon age
of 123040 yrs B.P. (1260-1060 cal yrs B.P.). Sample AR3-2PI was collected from the interval 30-54 cm and
yielded a radiocarbon age of 121040 yrs B.P. (1250-1050 cal yrs B.P.).
At the downstream end of the Qy1/Qpo deposit, a historical flood deposit is inset against
this stratigraphy on the lee side of a large boulder which created a suitable setting for the
deposition of flood sediment. Below an irregular slope of wind-blown sand lies a 5-cm thick
dark brown to black bed, which appears to be a concentration of mafic minerals deposited either
by the flood or placer mining activity (Figure 3.11). Below this horizon are three sedimentary
beds: a 9-cm thick pebble lens with well sorted and well rounded clasts less than 3 cm in
diameter, a 6-cm thick silt bed and a 45-cm thick cross-bedded and loosely consolidated medium
sand. A piece of mesh wire was recovered from the cross-bedded sand unit at a depth of 50 cm
below the top of the black bed, indicating that the entire sequence was deposited historically.
The wire mesh was woven and not welded suggesting some antiquity, but no definitive age
estimate could be made as woven wire is still quite common. The lack of any recognizable soil
60
development on the flood deposit is also consistent with the conclusion that the deposit is
historical. The contact between historical and Pleistocene alluvium is 65 cm below the top of the
black bed. The Pleistocene alluvium is composed of poorly sorted and well rounded boulders,
cobbles, gravel and sand with continuous carbonate coatings that appear very similar to the lower
Bk horizon described at sites AR3 and AR4 (Appendix C).
Figure 3.11. Photograph at site AR5 showing the historical flood deposit. Mesh wire was recovered from the cross-
bedded sand, indicating that the deposit to a depth of 60 cm is composed of sediment that is historical. Colored
increments on the measuring tape are 10-cm long.
The modeled peak discharges needed to inundate sites AR3, AR4 and AR5 are very
similar, primarily because these sites are very close together and all of the sites are on the same
terrace (Qy1/Qpo; Figure 3.9). The Qy1 terrace is overtopped by peak discharges between
24,000 and 34,000 ft
3
/s, when inundated by the approximately 1-3 ft of water that forms the
nonexceedance bound for this surface (Figure 3.12). The historical flood deposit at site AR5 is
overtopped by flows between 18,000 and 22,000 ft
3
/s. The computed water-surface profiles for
these preferred ranges of peak discharges are illustrated in Figure 3.13. There have been some
anthropogenic modifications in the Parkdale reach, principally placer mining of the Qpo terrace
gravels and construction of the railroad through the canyon. While the mining activity may have
61
enlarged the cross sectional area due to the excavation of an unknown amount of the terrace, the
placement of the railroad grade along the margin of the river channel would have decreased the
cross sectional area. In the study reach, it is apparent that the construction of the railroad
involved some widening of the canyon by blasting and excavation of the bedrock. Both
modifications add an unknown degree of uncertainty to the nonexceedance bound peak discharge
estimates; but these modifications would not have had an effect on estimates of peak discharges
for the historical deposit at AR5. Thus, the peak discharge estimate for the historical flood does
not include this additional uncertainty.
modeled flows and its relationship to geomorphic surfaces and stratigraphic sites in the Parkdale reach.
3.7.3 Loma Linda
The Loma Linda reach is located approximately two miles downstream of Coaldale
(Figure 3.1) in a narrow canyon reach of the Arkansas River. This site was chosen for its
intermediate location in the basin and to verify that the hydrology of the upper basin is
snowmelt-dominated. In a narrow canyon setting such as the Arkansas River between Parkdale
and Loma Linda sites, highway and railroad construction can have major effects on channel
geometry. The Loma Linda site was selected because the Santa Fe Railroad and U.S. highway
50 would have minimal effect on the model cross sections. This reach was also ideal because it
preserves a late Holocene flood deposit (site AR7; Figure 3.14), which occurs at the downstream
end of the model reach.
62
Arkansas River at Parkdale
5690
5695
5700
5705
5710
5715
5720
5725
5730
0 500 1000 1500 2000 2500
Station, ft
E
l
e
v
a
t
i
o
n
,

f
t
Thalweg
300 cfs
18,000 cfs
20,000 cfs
22,000 cfs
24,000 cfs
30,000 cfs
34,000 cfs
Sites AR3, AR4, AR5
Figure 3.13. Computed water-surface profiles for the range of preferred nonexceedance peak discharges in the
Parkdale reach.
The canyon at the Loma Linda site trends northeast to southwest and is cut into the
Precambrian granodiorite bedrock mapped by Taylor et al. (1975a, 1975b) (Xgd in Figure 3.14).
Few Holocene deposits are preserved in this reach, which was typically observed in other narrow
reaches in the upper Arkansas River basin such as this one. Glacial outwash deposits, primarily
of Pinedale age (Qpo in Figure 3.14), are preserved along the reach and are composed of
metamorphic and igneous boulders, cobbles, pebbles, and sand (Taylor et al., 1975a, 1975b).
Bull Lake glacial outwash deposits (Qbo in Figure 3.14) as well as alluvial fan deposits at the
mouths of small tributaries (Qf in Figure 3.14) form a minor component of the surficial deposits
preserved in the Loma Linda reach.
Through the modeled reach, the Arkansas River flows in a narrow channel with deep
pools and large fluvially-sculpted boulders. The Pinedale outwash terrace on the right bank is
about 30 feet high, and bedrock on the left bank has for the most part constrained the Arkansas
River to its present position, limiting any significant lateral movement. The Pinedale outwash
terrace forms a nonexceedance bound at this site that is greater than 10,000 years old (~10,000 to
14,000 years). A soil profile described at site AR6 exhibits a 10-cm thick A horizon formed on
sandy parent material (Appendix C). A 15-cm weakly developed cambic horizon (Bw) is
reddened and contains minor amounts of rounded gravel compared to the overlying A horizon.
A juvenile argillic horizon (Btj) at the base of the soil pit consists of poorly sorted, well rounded
gravel and moderately developed distinct clay coatings on the gravel. The Btj horizon marks a
63
distinct change in parent material and depositional processes. While the Btj horizon is distinctly
fluvial in origin, the lack of gravel and predominance of sand in the uppermost horizons suggest
that they are composed mostly of loess, formed as wind blown sediment was deposited on the
surface and incorporated into the soil.
Figure 3.14. Surficial photogeologic map of the Loma Linda reach showing stratigraphic sites and the approximate
location of HEC-RAS cross sections. Stream flow is from the lower left to the upper right of the photograph.
A slackwater deposit preserved at site AR7 is one of the few Holocene deposits preserved
in this reach and is inset at the base of the Pinedale outwash terrace on the downstream end of the
modeled reach (Figure 3.15). Its preservation is afforded by several large fluvially-sculpted
boulders at a channel expansion, which during flood flows creates a back eddy and a prime
setting for slow velocities and sediment deposition.
64
Figure 3.15. Vegetated slackwater deposit at site AR7 (note person in the vegetation for scale). Stream flow is from
right to left.
The soil formed on the slackwater deposit at site AR7 consists of a 20-cm thick A
horizon with single grain structure that grades to granular structure with depth (Figure 3.16;
Appendix C). This horizon overlies an AB horizon that is transitional having properties of both
A and B horizons. Its color is similar to that of the A horizon, which is a dark brown, but its
structure resembles that of the underlying Bk horizons. Filamentous carbonate occurs on the
undersides of clasts and in pores at the base of the AB horizon. The Bk and Bk2 horizons exhibit
an increase in soil structure development as well as a whitened matrix and stronger
effervescence, indicating greater carbonate accumulation. Soil color in the Bk horizon is also
lighter and more yellow than the darker and browner A or AB horizons.
Radiocarbon dates from Pinus charcoal in the AB and Bk horizons are 830 40 yrs B.P.
and 2100 40 yrs B.P., respectively (Appendix D). The calibrated age range for this soil
incorporating a 2 uncertainty is 700 to 2200 cal yrs B.P. The soil at site AR7 appears to be one
continuous soil, with no apparent breaks in soil formation or sediment deposition. This supports
the interpretation that this stratigraphy represents a single flood or series of floods deposited in
rapid succession and at a rate greater than the rate of soil formation. Since the formation of this
soil, floods have not deposited any significant amount of sediment that could clearly be
identified as a flood deposit. It is possible that floods have inundated this surface shallowly, but
not at a depth great enough to leave a recognizable record. Continuous soil development
indicates overall stability, and the lack of evidence for individual floods within the sequence
65
establishes the AR7 geomorphic surface as a nonexceedance bound over the last 700 to 2200
years.
Figure 3.16. Soil profile at site AR7. Sample AR7-1PI collected from 20-52 cm yielded a radiocarbon age of
83040 yrs B.P. (790-680 cal yrs B.P.). Sample AR7-2PI collected from 52-80 cm yielded a radiocarbon age of
210040 yrs B.P. (2150-1980 cal yrs B.P.).
Peak discharges required to inundate sites AR6 (on the Qpo outwash) and AR7
(correlative to Qy1 deposits at Parkdale and Qy1a deposits at Pueblo State Park; Appendix D)
expectedly have very different nonexceedance bound peak discharges. Site AR6 (in cross
section 5; Figure 3.14) is overtopped by discharges between 50,000 and 60,000 ft
3
/s (Figure
3.17). Site AR7 (in cross section 4; Figure 3.14) is overtopped by discharges ranging from 9,000
to 13,000 ft
3
/s. A nonexceedance bound estimate made by modeling approximately 1-3 ft of
water over the AR7 geomorphic surface ranges from 13,000 to 18,000 ft
3
/s (Figure 3.17). The
computed water-surface profiles for this preferred range of peak discharges are illustrated in
Figure 3.18.
3.7.4 Adobe Park
The Adobe Park reach is located upstream of Salida along County Road 63 (Figure 3.1).
This reach was chosen as the uppermost modeling reach in the upper Arkansas River basin for
paleoflood analysis based on the preservation of older and higher Holocene and Pleistocene
terraces, its suitability for hydraulic modeling, and access to the site. Nine different aged
Pleistocene units (Van Alstine, 1974) are comprised of alluvium deposited during various
glaciation and interglaciation periods (Table 3.2). The four highest and oldest units include
66
pediments, or erosional surfaces, that overlie the Tertiary Dry Union Formation in some areas.
Pinedale and Bull Lake alluvium form the five youngest outwash terraces.
modeled flows and its relationship to geomorphic surfaces and stratigraphic sites in the Loma Linda reach.
Arkansas River at Loma Linda
6300
6305
6310
6315
6320
6325
6330
6335
6340
6345
6350
0 500 1000 1500 2000 2500
Station, ft
E
l
e
v
a
t
i
o
n
,

f
t
Thalweg
300 cfs
13,000 cfs
14,000 cfs
18,000 cfs
50,000 cfs
60,000cfs
Site AR7
Site AR6
Figure 3.18. Computed water-surface profiles for nonexceedance bound peak discharges in the Loma Linda reach.
67
Table 3.2: Pleistocene Map Units of Van Alstine (1974)
Gravel unit (Qg)
Height
(feet above modern channel)
Probable age
9 10 Pinedale III
8 40 Pinedale II
7 60 Pinedale I
6 110 Bull Lake II
5 150 Bull Lake I
4 220 Illinoian
3 300 Kansan
2 370 Nebraskan
1 470 Nebraskan
taken from Van Alstine, 1974; table 1 on p. 9.
In the study reach, the oldest Pinedale (Qg7) and youngest Bull Lake (Qg6) outwash
terraces form extensive planar surfaces at about 60 and 110 ft above the modern Arkansas River
(Figure 3.19). A younger Pinedale age terrace (labeled Qg8? in Figure 3.19) also appears to be
present along the left bank although it is not mapped by Van Alstine (1974). He mentions,
however, that small remnants of the two youngest Pinedale terraces (Qg8 and Qg9) were
included with Holocene alluvium map unit (Qal). This is probably the case in the Adobe Park
reach.
Two Holocene deposits (labeled Qy2 and Qy1) that form the modern floodplain and low
terraces as well as the active channel (Qyc) of the Arkansas River were also mapped in the
Adobe Park reach (Figure 3.19). Typically, the terrace surfaces formed on these deposits have
irregular topography and are approximately 5 to 6 ft above the active channel. Exposures of
soils on the Qy2 deposits show that they are weakly developed, exhibit primary depositional
features, and distinct beds, the result of overbank flows. The Qy1 deposits are composed of
sandy river alluvium and the soil formed on these deposits include a 10-cm thick A horizon and
several underlying B horizons (Figure 3.20). The B horizons at site AR8 (Appendix C) were
distinguished from each other primarily on the basis of parent material changes (sand to sandy
gravel).
Below a depth of 42 cm, the soil becomes clay rich and exhibits filaments and nodules of
an unidentified salt. The salts appear much whiter than what is typical for soil carbonate and do
not effervesce. A single radiocarbon sample from the base of the B2 horizon yielded an
radiocarbon age of 400 40 B.P. Calibration and adjustment of the age to the 1950 datum for
the radiocarbon methodology indicates that the soil may be 400 to 600 years old (Appendix D).
Within the study area, Van Alstine (1974) shows that the terraces about 10 ft above the
modern river form the youngest late Pleistocene outwash deposits (Pinedale III; Qg9, Table 3.2).
While the Qy1 terrace mapped for this study is about 10 ft above the modern channel, it is not
considered to be correlative to the Pinedale III terrace since it exhibits characteristics more
68
Figure 3.19. Surficial photogeologic map of the Adobe Park reach showing the approximate location of HEC-RAS
cross sections and stratigraphic site AR8. Stream flow is from left to right. Nomenclature for Pleistocene map units
is derived from Taylor et al. (1975a; Qpo and Qbo), and Van Alstine (1974; Qg6, Qg7, and Qg8?).
common on Holocene alluvium with its soil development and sediment grain size (sand versus
sandy gravel). The terrace surface heights are generalizations made by Van Alstine over a large
map area and may include Holocene deposits, as described previously. It is also possible that the
estimates were made for wider reaches where the younger Holocene deposits might well be
lower than in the more narrow Adobe Park study reach, where the youngest Pinedale gravel
appears to be more than 15 ft above the modern channel.
Peak discharges required to inundate site AR8 (cross section 7; Figure 3.19) by
approximately 1-3 ft range from about 17,000 to 27,000 ft
3
/s (Figure 3.21; Appendix D). Trees
along the channel and other surface irregularities on the Qy1 terrace are likely to create higher
roughness values when flood flows exceed the channel capacity than was modeled. Therefore,
the nonexceedance bound peak discharge is most likely to be at the low end of the above range.
The computed water-surface profiles for this preferred range of peak discharges are illustrated in
Figure 3.22.
69
Figure 3.20. Soil profile at site AR8. Sample AR8-1PI collected from 42 cm yielded a radiocarbon age of 40040
yrs B.P. (520-420; 390-320 cal yrs B.P.).
modeled flows and its relationship to geomorphic surfaces and stratigraphic sites in the Adobe Park reach.
70
Arkansas River at Adobe Park
7130
7135
7140
7145
7150
7155
7160
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Station, ft
E
l
e
v
a
t
i
o
n
,

f
t
Thalweg
300 cfs
17,000 cfs
20,000 cfs
27,000 cfs
Site AR8
Figure 3.22. Computed water-surface profiles for nonexceedance bound peak discharges in the Adobe Park reach.
3.8 Discussion
3.8.1 General Observations
Quaternary deposits along the Upper Arkansas River consist of predominantly
Pleistocene deposits upstream of Parkdale and predominantly Holocene deposits downstream of
Canon City. Upstream of Parkdale, the floodplain occupies a narrow space between Pleistocene
gravelly outwash terraces, leaving little room for the preservation of Holocene deposits.
Downstream of Canon City, Holocene alluvium is common and the floodplain is generally wide.
In addition to the obvious physiographic and geologic differences, it also appears that a change
also occurs in the size of paleofloods based on information from the Loma Linda site to the
Pueblo State Park site. For instance, for a similar age surface and accounting for differences in
basin area, floods at Pueblo State Park are substantially larger than floods at Loma Linda. Based
strictly on the paleoflood data, this suggests that there is a transition in flood hydrology to
snowmelt-dominant floods within the basin upstream of Canon City.
3.8.2 Development of Paleoflood Nonexceedance Bounds for Pueblo Dam
Paleoflood nonexceedance bounds were developed for each of four study sites in the
Upper Arkansas River basin. The nonexceedance bounds were established using age estimates
derived from radiocarbon analysis and peak discharge estimates from HEC-RAS hydraulic
modeling. Peak discharges corresponding to 1-3 ft of inundation over each site were used to
71
develop the nonexceedance bounds (Table 3.3). For paleoflood deposits, the minimal depth
required to overtop the surface was used.
Table 3.3: Range of Nonexceedance Bound Peak Discharge Estimates
Location Stratigraphic Site Minimum Flow, ft
3
/s Maximum Flow, ft
3
/s
Pueblo State Park AR1 8,000 12,000
AR2 30,000 40,000
AR9 130,000 160,000
Parkdale AR3 24,000 34,000
AR4 24,000 34,000
AR5 18,000 22,000
Loma Linda AR6 50,000 60,000
AR7 13,000 18,000
Adobe Park AR8 17,000 27,000
Range of peak discharges that overtop bounding surface by 1-3 feet.
At Pueblo State Park, the estimated peak discharge for the June 1921 flood (about
103,000 ft
3
/s) marginally inundates site AR9. Little evidence of historical flooding was
observed. Angular rock fragments on the ground surface at site AR9 and approximately 20 cm
of alluvium in the upper part of the soil profile were interpreted as being alluvial fan deposits
derived from the alluvial fan upslope. Since the railroad grade immediately south of site AR9
(Figure 3.3) was built in the 1870s, this site has been effectively isolated from alluvial fan
deposition historically. It is thus logical that the alluvial fan sediment observed at site AR9 must
pre-date the railroad. In the soil profile, the parent material on which the 3B1 horizon is formed,
interpreted to be flood sediment, would represent the last time a large flood inundated the site. A
radiocarbon age on charcoal recovered from these sediments (3B3 horizon; Figure 3.5) indicates
that the overlying soil is at least 730-840 years old. Based on these data, it appears that there has
not been a flood larger than 150,000 ft
3
/s, the peak discharge needed to inundate site AR9 by 3 ft,
in the last 730 to 840 years (Table 3.3).
In the Parkdale reach, a peak discharge for both a nonexceedance bound and a paleoflood
were developed. The paleoflood represented by the flood deposit at site AR5 is considered to be
historical because of the incorporated wire mesh recovered from the deposit. The age of the
flood can be bracketed by documentation of the historical development and placer mining in the
area since the late 1860s and by the gage record at Parkdale, which began in 1946. While the
age of this paleoflood cannot be stated with certainty, it seems logical to conclude that the flood
deposit resulted from the August 1921 flood that was recorded at the Canon City gage. Peak
discharge estimates at the Parkdale site from the hydraulic modeling and that reported at the
Canon City gage are very similar, 18,000-22,000 ft
3
/s versus 19,000 ft
3
/s, respectively. This
conclusion is further supported by an analysis of the post-1946 gaging record at Parkdale to the
Canon City gage indicate that peak discharges at the two sites historically are highly correlated
72
(Section 2). A nonexceedance bound developed for sites AR3 and AR4 indicates that 24,000-
34,000 ft
3
/s has not been exceeded in 1100-1300 years (Table 3.3). This estimate is derived from
modeled peak discharge estimates required to inundate the Qy1 terrace surface by 3 ft and on
two radiocarbon ages from the soil profile at site AR3 (Appendix D).
Estimates of nonexceedance bounds were developed in the Loma Linda reach on Late
Holocene and Late Pleistocene deposits. The Pleistocene deposit is mapped as a Pinedale
outwash terrace by Taylor et al. (1975a) and forms the basis for a nonexceedance bound peak
discharge estimate of 50,000-60,000 ft
3
/s with an age range of 10,000-14,000 years.
Radiocarbon ages derived from a stable soil formed on a small Holocene terrace inset into and
below the Pleistocene outwash terrace indicates that this soil is between approximately 700 and
2200 years old. Modeled peak discharges of 13,000-18,000 ft
3
/s required to inundate the terrace
surface by 3 ft forms the stage-discharge portion of the nonexceedance bound (Table 3.3).
In the Adobe Park reach, a late Holocene terrace (Qy1) at site AR8 was determined to be
between 400-600 years old on the basis of radiocarbon analysis of charcoal recovered from a soil
formed on the terrace surface. This soil is the basis for a nonexceedance bound, in which
modeled peak discharges of 17,000-27,000 ft
3
/s required to inundate the terrace surface by 1-3
feet have not been exceeded in 400-600 years (Table 3.3).
A quick comparison of the nonexceedance bound peak discharge data between the upper
basin sites (values in Table 3.3 exclusive of Pueblo State Park) suggests that the Adobe Park
nonexceedance estimate is anomalous. The basin area at the Adobe Park site is about half of that
at the Parkdale site, the age for the nonexceedance bound is also about half as old, and the peak
discharge estimate is roughly equal. This potentially indicates that paleofloods in the upper part
of the basin have been somewhat larger than at downstream sites. However, closer analysis of
the paleoflood data and the geomorphic setting of each of the upper basin sites suggest that the
apparent anomaly falls well within expected data uncertainty (described in the following
section). First, the basin area at the Loma Linda site downstream is perhaps 20% larger and the
lower limit of the age range is about 100 years older than the same properties at the Adobe Park
site while the peak discharge estimates for nonexceedance bounds overlap, admittedly only
slightly. Relative to the Parkdale site, while the basin area and the age range is about double that
of the Adobe Park site, the minimum estimate for the nonexceedance bound peak discharge is
also half of the maximum estimate at the Parkdale site.
The most important factor to consider in making comparisons between the
nonexceedance bounds between sites is believed to be the geomorphology between the sites.
Both the Parkdale and Loma Linda sites are located in narrow reaches where the preservation of
Holocene deposits is relatively rare compared to much more extensive late Pleistocene deposits
present at each site. In addition, the channel geometry at both of these sites is relatively
confined, so the estimates of peak discharge in the hydraulic model are probably more precise.
The geomorphic setting at both of these sites is considered near ideal for making paleoflood
73
nonexceedance bound estimates for this reason. The Adobe Park site is situated in a wide, lower
relief valley than the other downstream sites. As a result, the channel geometry is not as well
constrained so given the same increase in stage (1-3 ft) across a wider channel, an estimate of
peak discharge will be significantly larger. The channel at the Parkdale and Loma Linda sites is
between about 150 and 250 feet wide; at Adobe Park the channel width is about 750 feet.
3.8.3 Data Uncertainty
Several sources of discharge uncertainty are encompassed in each paleoflood bound or
paleoflood estimate. In the HEC-RAS model, channel roughness (Mannings n) as well as
expansion/contraction coefficients can be varied and may therefore substantially affect the
resulting peak discharge. While values of 0.1 and 0.3 for the expansion/contraction coefficients
are considered standard, the roughness coefficient depends on the vegetation and sediment
characteristics of a particular reach. The final range of peak discharge estimates are based on a
range of roughness values in order to account for some of the uncertainty associated with
channel roughness during flood flows. The selected roughness values for each reach are within
accepted ranges and on the low side resulting in a more conservative discharge estimate. If
higher roughness values were used, discharge estimates would be lower at most of the study
sites. The nonexceedance method assumes that 3 ft, or one meter of water, is sufficient to
generate enough stream power to leave a clearly recognizable record of erosion and/or
deposition. There is some uncertainty in this assumption, since no data were collected on the
terrace surfaces where nonexceedance bounds were established to characterize the sediment or
evaluate the stream power/shear stress relationship in this system. It is recognized that in
overbank areas of the floodplain and on higher terraces stream power can be highly erratic.
Radiocarbon analysis used to establish age constraints for paleofloods and
nonexceedance bounds also has its own set of uncertainties. Among the problems associated
with determining the radiocarbon age of soil and young deposits is the possibility of
incorporating older material that has been reworked and deposited with younger sediments. This
scenario would suggest that the radiocarbon age is a maximum age for the deposit. Younger
material can also be incorporated into the deposit by bioturbation. Rodents and insects use
organic matter as foodstuffs and bedding and may thereby translocate materials to a greater depth
as they burrow through the soil. Dating these types of materials gives a minimum age for the
deposit. Because of these issues, bulk samples can be problematic and may introduce greater
uncertainty in age estimates for soils and young deposits. These problems can be minimized by
identifying specific material to be dated. Charcoal and shell material that has been specifically
identified provides important information about nature of the material, the paleoecology of the
site, and any possible contamination. This information can be very helpful in interpreting the
depositional context of each sample, should help the researcher determine which samples are
74
most suitable for radiocarbon analysis, and thus provide a more defensible age estimate of the
deposit.
3.9 Summary
The paleoflood study of the Upper Arkansas River Basin utilizes detailed information
from four widely spaced sites to characterize paleofloods and nonexceedance bounds for the
Pueblo Dam flood hazard study. The sites were strategically selected to provide information
about hydrologic conditions at various locations in the basin. Of most importance was to
characterize flood hazards both upstream and downstream of the Royal Gorge, where there
appears to be a transition from snowmelt-dominated to rainfall-dominated floods. Suitable sites
were found and studied in Adobe Park, the Loma Linda recreation area, Parkdale, and Pueblo
State Park. Nine soil/stratigraphic descriptions (Appendix C) and seven radiocarbon dates
(Appendix D) of key deposits were used in conjunction with geomorphic mapping and HEC-
RAS flow modeling to determine age estimates of each soil and peak discharges required to
inundate the surfaces. Estimates are summarized in Table 3.4. With the exception of Adobe
Park, nonexceedance bounds for similar age surfaces appear to increase in the downstream
direction and change markedly between Loma Linda and Pueblo State Park from approximately
14,000 ft
3
/s to 150,000 ft
3
/s for Holocene alluvium between the ages of 700 and 2000 years.
Table 3.4: Summary of Paleoflood Nonexceedance Bounds, Upper Arkansas River
Location
Stratigraphic
Site
Type of Estimate
Age Range
(years)
Peak
Discharge
Range (ft
3
/s)
Preferred
Peak
Discharge
(ft
3
/s)
Pueblo
State Park
AR9
nonexceedance
bound
730-840
130,000-
160,000
150,000
Parkdale
AR5 paleoflood
historical
(post 1870
ca. 1921)
18,000-
22,000
20,000
AR3, AR4
nonexceedance
bound
1100-1300
24,000-
34,000
30,000
Loma
Linda
AR7
nonexceedance
bound
700-2200
13,000-
18,000
14,000
AR6
nonexceedance
bound
10,000-
14,000
50,000-
60,000
50,000
Adobe Park AR8
nonexceedance
bound
400-600
17,000-
27,000
20,000
75
76
4.0 Flood Frequency Analysis with Paleoflood Data
This section describes peak discharge frequency analysis at sites within the Arkansas
River Basin. A moments-based 3-parameter flood frequency model (Cohn et al., 1997) was used
to conduct at-site peak discharge frequency analysis. This model and methods for comparing
frequency curves in a regional frequency context are briefly presented. Peak discharge,
historical and paleoflood data used in the frequency analysis at each site are summarized. The
frequency analysis results for each site are discussed. Regional frequency analysis results are
then presented. These frequency curves are later used as a basis to compare frequency curves
from the TREX rainfall-runoff model.
Paleoflood investigations and peak discharge probability estimates were made at four
main sites within the Arkansas River basin: the Arkansas River at Pueblo State Park near Pueblo;
the Arkansas River at Parkdale; the Arkansas River at Wellsville (Loma Linda); and the
Arkansas River at Salida (Figure 4.1). Paleoflood peak discharge estimates, and estimates of
paleohydrologic bounds, were combined with available historical information and peak
discharge estimates from gaging stations at these sites. Peak discharge estimates from several
gaging stations were combined, where appropriate, to obtain longer records. At selected
locations, peak discharge estimates from gaging stations were deliberately censored. The peak-
flow data and uncertainties used in peak discharge frequency analysis at each site are
summarized below. Soils stratigraphy, age estimates, and hydraulic modeling associated with
paleofloods and paleohydrologic bounds are described in detail in Section 3.
The flood frequency results for each site are presented immediately following the data at
each site. The major goal of the flood frequency analysis is to develop probability relationships
directly from historical, streamflow gage and paleoflood data. These relationships are then used
as a basis for comparing TREX rainfall-runoff model flood frequency curves. In this way, the
paleoflood data are used as an independent check on the rainfall-runoff model.
4.1 Flood Frequency Methods
Two flood frequency methods are used in this work: at-site flood frequency and regional
flood frequency.
Peak-flow frequency estimates were made for annual instantaneous peak discharge
estimates. Peak discharge probabilities are estimated directly from the data using Cunnane's
plotting position with the threshold-exceedance formula (Stedinger et al., 1993) that includes
historical and paleoflood data. The data were assumed to follow a log-Pearson Type III (LP-III)
distribution. The method of moments was used to estimate the LP-III parameters for peak
discharge estimates using Expected Moments Algorithm (EMA) techniques (Cohn et al., 1997;
England, 1999). EMA (Cohn et al., 1997, 2001; England et al., 2003a) is a new moments-based
77
Figure 4.1. Locations of four paleoflood/flood frequency study sites within the Arkansas River basin.
parameter estimation procedure that was designed to incorporate many different types of
systematic, historical, and paleoflood data into flood frequency analysis. EMA assumes the LP-
III distribution is the true distribution for floods. EMA was designed to handle the four different
classes of historical and paleoflood data beyond the applicability of the Bulletin 17B historical
weighting procedure (IAWCD, 1982). As noted by Cohn et al. (1997, 2001) and England
(1998), EMA is philosophically consistent with, and is an improvement to, the Bulletin 17B
method of moments procedure when one has historical or paleoflood information. EMA is
specifically designed to use historical and paleoflood data, in addition to annual peak flows from
gaging stations, in a manner similar to Maximum Likelihood Estimators (Lane and Cohn, 1996).
It is a more logical and efficient way to use historical and paleoflood data than the current
Bulletin 17B historical method, and it is a natural extension to the moments-based framework of
Bulletin 17B. Confidence intervals were estimated using the approach in Cohn et al. (2001).
Because the record length was long, no regional skew weighting was performed. An at-site
estimate of the station skewness coefficient was used in the analysis. EMA has been rigorously
peer reviewed in the literature (Cohn et al., 1997, 2001; England et al., 2003a, 2003b) and
provides a suitable flood frequency model. EMA has been applied at many sites for peak-flow
frequency (England et al., 2003b).
A simplified regional frequency analysis was conducted for the four sites with paleoflood
data, using the index flood method (Stedinger et al., 1993; Hosking and Wallis, 1997). The
78
regional frequency analysis was conducted to compare the distributions from each site for
consistency in a qualitative sense. The goal was to determine if the estimated frequency curve
for the Arkansas River immediately upstream of Pueblo was similar to frequency curves from
the other three sites. Similarity in the frequency curve at Pueblo, to those from other sites,
would provide additional confidence in estimating extreme flood probabilities at Pueblo Dam.
Differences between the frequency curves could clearly highlight mixed-population flood effects
within the basin. A regional distribution, based on the frequency curves and data from the four
sites, was not estimated. Following Smith (1989), the preferred peak discharge frequency curve
at each site was non-dimensionalized using the at-site 0.10 exceedance probability (10-year
flood). This estimate was made from the LP-III model frequency curve at each site that included
gage data, historical and paleoflood peak flow estimates.
4.2 Arkansas River at Pueblo State Park
Peak discharge estimates on the Arkansas River at Pueblo State Park are combined from
three gaging stations (Table 4.1). Records were chosen in order to gain a complete record of all
large floods that exceeded approximately 10,000 ft
3
/s for the period of record. It was assumed
that peak discharge estimates from these gages were unaffected by upstream regulation.
Information in Patterson (1964) indicates that this assumption is approximately valid. The total
combined gage record length, excluding historical data, is 110 years (1895-2004). There are no
significant tributaries between the stations. Although there are slight differences in drainage
areas between the three sites, no adjustments were made to transfer the records to upstream of
Pueblo Reservoir.
Table 4.1: Peak discharge records from the Arkansas River near Pueblo State Park that are used for frequency
analysis
USGS
Gaging
Station
No.
Station Name
Drainage
Area (mi
2
)
Period of
Record
(Water Year)
Records Used
for Frequency
Analysis (Water
Year)
Maximum Discharge (ft
3
/s)
and Date
07097000
Arkansas River at Portland, CO 4,024 1939 - 2002 1975 - 1976 21,100; 06/05/1949
07099200
Arkansas River near Portland, CO 4,280 1965 - 1974 1974 23,900; 08/21/1965
07099500
Arkansas River near Pueblo, CO 4,686 1895 - 1975 1895 - 1973 103,000; 06/03/1921
Based on reviews of available historical information (Section 2), including Follansbee
and Jones (1922), Munn and Savage (1922), and Follansbee and Sawyer (1948) (among others),
there is a substantial amount of historical flood information on the Arkansas River in Pueblo that
was usable for frequency analysis. The historical record was estimated to begin in 1859,
resulting in a 146-year period (1859-2004). The criteria for inclusion of historical floods in
frequency analysis were: the ability to rank the floods relative to the June 1921 event; and to
estimate magnitudes for the individual floods. Three historical floods were included: June 1864,
79
July 1893, and May 1894. The magnitudes of these floods were large relative to the floods in the
gaging record; estimates within a range were based on Follansbee and Sawyer (1948) and
included in the flood frequency analysis. These estimates in general all have relatively large
uncertainties as compared to the smaller floods in the gage record (Figure 4.2).
A paleohydrologic bound of about 840 years was estimated at this site (Table 3.4) for
inclusion in the flood frequency curve. The estimate is based on three soils pits, two radiocarbon
ages, and hydraulic modeling of a 7,500 foot reach (Section 3). No estimates of individual
paleofloods were made at this site, due to the relatively wide channel geometry and the lack of
apparent stratigraphic evidence of large paleofloods during the limited field study. A time series
plot of the peak discharge, historical flood and paleohydrologic bound data is shown in Figure
4.2.
Figure 4.2. Approximate unregulated peak discharge, historical and paleoflood estimates, Arkansas River at
Pueblo Dam. A scale break is used to separate the gage and historical data from the longer paleoflood record.
Arrows on the 1864, 1893, 1894 and 1921 floods indicate floods in a range.
The flood frequency results are shown in Figure 4.3 and Table 4.2. Peak discharge
estimates from the gage are shown as open squares with estimated data uncertainty for some of
the largest floods that were described in a range. One can observe the large positive skew (0.8
log space) and relatively steep transition between snowmelt-dominant floods to rainfall dominant
floods greater than about 10,000 ft
3
/s. These large rainfall-caused floods are responsible for the
shape of the upper portion of the frequency curve. The return period of the largest flood on
record (June 1921) is about 270 years from the exceedance-based plotting position, and about
1,600 years from the LP-III model.
80
Figure 4.3. Approximate peak discharge frequency curve, Arkansas River at Pueblo State Park, including gage,
historical and paleoflood data.
Table 4.2: Arkansas River at Pueblo State Park Peak Discharge Frequency Results
Annual
Exceedance
Probability (%)
Return Period
(years)
Peak Discharge (ft
3
/s)
Model Estimate*
5% Confidence
Limit*
95% Confidence
Limit*
10 10 16400 14,100 19,400
4 25 23,800 19,400 29,700
2 50 31,000 23,900 41,000
1 100 39,800 29,000 57,100
0.5 200 50,600 34,800 79,900
0.2 500 68,700 43,700 125,000
0.1 1,000 86,100 51,400 176,000
0.05 2,000 107,000 60,100 249,000
0.02 5,000 143,000 73,400 393,000
0.01 10,000 177,000 85,000 556,000
0.005 20,000 218,000 98,100 787,000
*Results shown in italics are extrapolated beyond available data
The results indicate that the LP-III model provides an adequate fit to the gage, historical
and paleohydrologic bound data. The model fits the bulk of the data well, including most of the
large floods, but undershoots the largest flood (June 1921) due to the addition of paleoflood data.
81
The paleohydrologic bound data at Pueblo State Park increases the peak discharge record length
substantially to about 840 years, and has an effect on the upper end of the extrapolated frequency
curve principally by reducing the skewness coefficient.
One can estimate design flood probabilities based on significant extrapolation of the LP-
III model and 90 percent confidence interval (Figure 4.3). The spillway design outflow capacity
for Pueblo Dam (191,000 ft
3
/s) has an estimated return period of 13,000 years, and a return
period estimate for the volume-critical spillway design inflow peak (270,000 ft
3
/s) is about
42,000 years.
4.3 Arkansas River at Parkdale
Peak discharge estimates on the Arkansas River at Parkdale are based on the Parkdale
gage with historical information, the largest flood from the Canon City gage, and the basin
snowmelt flood of record in June 1957 (Section 2). It was assumed that peak discharge estimates
from these gages were unaffected by upstream regulation. The gage record length at Parkdale,
excluding historical data and gaps in the gage record, is 48 years (1946-2004).
Based on reviews of available historical information (Section 2), including Follansbee
and Jones (1922) and Follansbee and Sawyer (1948) (among others), there is some amount of
historical flood information on the Arkansas River in Canon City that was usable for frequency
analysis. The historical record was estimated to begin in 1868, resulting in a 137-year period
(1868-2004). It was also clear from the very long streamflow record at Canon City (Figure 4.4),
that the storms causing the largest floods at Pueblo (June 1921 and May 1894) did not cause
significantly large peaks at Parkdale or Canon City. The May 1894 flood at Pueblo was only
about 4,400 ft
3
/s at Canon City. The rainstorm that caused the June 1921 flood at Pueblo
centered downstream of Canon City. The largest peak of record at Canon was the August 2,
1921 flood (Figure 4.4).
A paleohydrologic bound of about 1,250 years was estimated at Parkdale (Table 3.4) for
inclusion in the flood frequency curve. The estimate is based on two soils pits, two radiocarbon
ages, and hydraulic modeling of a 2,330 foot reach (Section 3). A historical flood estimate was
made based on stratigraphy in a third soils pit. The flood age was estimated to be historical (post
1870). Given the streamflow data at Canon (Figure 4.4) and the correlations between Parkdale
and Canon (Section 2), it is estimated that this deposit is from the August 1921 flood. A time
series plot of the peak discharge, historical flood and paleohydrologic bound data is shown in
Figure 4.5. The peak discharge estimates at Parkdale include the 1957 snowmelt flood, based on
estimates at Salida and Canon City, and a censoring threshold based on this 1957 flood to
account for the missing or unmeasured floods from 1956-1964 and 2002-2004. The use of this
threshold indicates that the unmeasured floods have unknown magnitudes that are less than
about 9,000 ft
3
/s.
82
Figure 4.4. Approximate unregulated peak discharge and historical flood estimates, Arkansas River at Canon City.
Note the distinct differences in magnitude for the largest floods versus those at Pueblo.
Figure 4.5. Approximate unregulated peak discharge, historical and paleoflood estimates, Arkansas River at
Parkdale. A scale break is used to separate the gage and historical data from the longer paleoflood record. Arrows
on the 1921 and 1957 floods indicate floods in a range.
83
estimates from the gage are shown as open squares with estimated data uncertainty for the two
largest floods that were described in a range. One can observe that the skewness coefficient (0.3
log space) is positive, but much reduced from downstream at Pueblo. Peak-flow magnitudes are
also dramatically less. There is only one observation that exceeds 10,000 ft
3
/s (August 1921 rain
flood); this flood is partly responsible for the shape of the upper portion of the frequency curve.
The majority of the observations, including the second-largest and third-largest floods, are from
snowmelt. The return period of the largest flood on record (August 1921) is about 270 years
from the exceedance-based plotting position, and about 1,300 years from the LP-III model.
The results indicate that the LP-III model provides an adequate fit to the gage, historical
and paleohydrologic bound data. The model fits the bulk of the data well, including all of the
large floods, except for undershooting the largest flood (August 1921) due to the addition of
paleoflood data. The paleohydrologic bound data at Parkdale increase the peak discharge record
length substantially to about 1,250 years, and have an effect on the upper end of the extrapolated
frequency curve principally by reducing the skewness coefficient for a very large flood in a
shorter record. The 90 percent confidence interval encompasses most of the data, including
paleohydrologic bounds.
Figure 4.6. Approximate peak discharge frequency curve, Arkansas River at Parkdale, including gage, historical
and paleoflood data.
84
Table 4.3: Arkansas River at Parkdale Peak Discharge Frequency Results
Annual
Exceedance
Probability (%)
Return Period
(years)
Peak Discharge (ft
3
/s)
Model Estimate*
5% Confidence
Limit*
95% Confidence
Limit*
10 10 6,910 6,030 7,950
4 25 8,730 7,280 10,300
2 50 10,200 8,150 12,300
1 100 11,800 9,000 14,800
0.5 200 13,500 9,840 17,600
0.2 500 16,000 11,000 22,300
0.1 1,000 18,100 11,900 26,600
0.05 2,000 20,400 12,800 31,800
0.02 5,000 23,700 14,000 40,100
0.01 10,000 26,400 14,900 47,800
0.005 20,000 29,400 15,900 56,900
4.4 Arkansas River at Loma Linda
The Loma Linda site was selected for its intermediate location in the watershed and
reflects the upper basin snowmelt hydrology. Peak discharge estimates on the Arkansas River at
Loma Linda are based on the Wellsville gage with historical information and the basin snowmelt
flood of record in June 1957 (Section 2). It was assumed that peak discharge estimates from this
gage were unaffected by upstream regulation. The gage record length at Wellsville, excluding
historical data and gaps in the gage record, is 41 years (1961-2004).
Based on reviews of available historical information (Section 2), including Campbell
(1922), Baker and Hafen (1927) and Crowfoot et al. (2004) (among others), there is some limited
amount of historical flood information on the Arkansas River at Salida that was usable for
frequency analysis. The historical record was estimated to begin in 1880, resulting in a 125-year
period (1880-2004). It was also clear from the relatively long streamflow record at Salida
(shown below) in conjunction with this site and Parkdale, that the largest floods are caused by
snowmelt. The largest peak of record at Salida and in the upper Arkansas River watershed was
the June 1957 snowmelt flood (Section 2).
Two paleohydrologic bounds (Table 3.4) for inclusion in the flood frequency curve were
estimated at Loma Linda: a Holocene bound with an age of about 1,500 years; and a Pleistocene
bound with an age of about 10,000 years. The estimates are based on two soils pits, two
radiocarbon ages, and hydraulic modeling of a 2,340 foot reach (Section 3). No estimates of
individual paleofloods were made at this site, as there are few deposits of Holocene alluvium
preserved in the reach. A time series plot of the peak discharge, historical flood and
paleohydrologic bound data is shown in Figure 4.7. These data include a censoring level for the
historical period, including unobserved floods, that was chosen based on the June 1957
snowmelt peak. This threshold also includes missing observations from 2002-2004.
85
estimates from the gage are shown as open squares with estimated data uncertainty for the
largest flood that was described in a range. One can observe that the skewness coefficient (-0.2
log space) is negative, and reduced from downstream at Parkdale. Peak-flow magnitudes are
also dramatically less. There is only one observation that exceeds 7,000 ft
3
/s (June 1957
snowmelt flood); this flood fits well with the other snowmelt flood observations.
Figure 4.7. Approximate unregulated peak discharge, historical and paleoflood estimates, Arkansas River at Loma
Linda. A scale break is used to separate the gage and historical data from the longer paleoflood record. A second
scale break separates the gage/historical and lower paleohydrologic bound from the larger paleohydrologic bound.
Arrows indicate the 1957 flood in a range.
Nearly all the observations, including the largest floods, are from snowmelt. The return period
of the largest flood on record (June 1957) is about 250 years from the exceedance-based plotting
position, and about 450 years from the LP-III model.
The results indicate that the LP-III model provides a very good fit to the gage, historical
and paleohydrologic bound data. The model fits the bulk of the data well, including all of the
large floods and the Holocene paleohydrologic bound. The paleohydrologic bound data at Loma
Linda significantly increase the peak discharge record length to about 10,000 years. The
Pleistocene paleohydrologic bound has little to no influence on the frequency curve because of
the negative skew and the very large difference in flow magnitude between the Pleistocene and
Holocene bounds. The paleohydrologic bounds are consistent with the observed streamflow
86
data; there have been no floods in the geologic record substantially larger than the snowmelt
floods observed from the gage record. The 90 percent confidence interval encompasses the
observed data.
Figure 4.8. Approximate peak discharge frequency curve, Arkansas River at Loma Linda, including gage,
Table 4.4: Arkansas River at Loma Linda Peak Discharge Frequency Results
Annual
Exceedance
Probability (%)
Return Period
(years)
Peak Discharge (ft
3
/s)
Model Estimate*
5% Confidence
Limit*
95% Confidence
Limit*
10 10 5,420 4,850 6,050
4 25 6,380 5,640 7,220
2 50 7,080 6,170 8,130
1 100 7,760 6,630 9,090
0.5 200 8,440 7,040 10,100
0.2 500 9,320 7,520 11,500
0.1 1,000 9,980 7,830 12,700
0.05 2,000 10,600 8,110 13,900
0.02 5,000 11,500 8,440 15,700
0.01 10,000 12,200 8,660 17,100
0.005 20,000 12,800 8,850 18,600
87
4.5 Arkansas River at Adobe Park
The Adobe Park site was selected for its relatively higher location in the watershed to
document snowmelt hydrology and potentially limit areal extents of extreme storms. Peak
discharge estimates on the Arkansas River at Adobe Park are based on the Salida gage with
historical information and the basin snowmelt flood of record in June 1957 (Section 2). It was
assumed that peak discharge estimates from this gage were unaffected by upstream regulation.
The gage record length at Salida, excluding historical data and gaps in the gage record, is 77
years (1895-1979).
Based on reviews of available historical information (Section 2), including Campbell
(1922), Baker and Hafen (1927) and Crowfoot et al. (2004) (among others), there is some limited
amount of historical flood information on the Arkansas River at Salida that was usable for
frequency analysis. The historical record was estimated to begin in 1880, resulting in a 125-year
period (1880-2004). It was also clear from the relatively long streamflow record at this site that
the largest floods are caused by snowmelt. The largest peak of record at Salida and in the upper
Arkansas River watershed was the June 1957 snowmelt flood (Section 2).
One paleohydrologic bound (Table 3.4) for inclusion in the flood frequency curve was
estimated at Adobe Park, a Holocene bound about 550 years. The estimate is based on mapping
two Holocene surfaces, one soils pit, one radiocarbon age, and hydraulic modeling of a 1,870
foot reach (Section 3). No estimates of individual paleofloods were made at this site, as there are
few deposits of younger Holocene alluvium preserved within Pinedale and Bull Lake outwash
terraces. A time series plot of the peak discharge, historical flood and paleohydrologic bound
data is shown in Figure 4.9. These data include a censoring level for the historical period,
including unobserved floods, that was chosen based on the June 1957 snowmelt peak. This
threshold also includes unmeasured floods since the gage was discontinued after 1979.
The flood frequency results are shown in Figure 4.10 and Table 4.5. One can observe
that the skewness coefficient (-0.2 log space) is negative, and identical to that at Loma Linda.
Peak-flow magnitudes are also about the same (slightly less) as Loma Linda, and dramatically
less than at Pueblo. There is only one observation that exceeds 6,000 ft
3
/s (June 1957 snowmelt
flood); this flood is a bit larger than the other snowmelt flood observations. Nearly all the
observations, including the largest floods, are from snowmelt. The return period of the largest
flood on record (June 1957) is about 250 years from the exceedance-based plotting position, and
about 4,000 years from the LP-III model.
The results indicate that the LP-III model provides a very good fit to the gage, historical
and paleohydrologic bound data. The model fits the bulk of the data well, but departs slightly
from the larger floods in order to attempt to fit the largest flood. The paleohydrologic bound
data at Adobe Park increase the peak discharge record length to about 550 years. This Holocene
paleohydrologic bound has a small influence on the frequency curve because of the negative
skew and the very large difference in flow magnitude between the Holocene bound and the
largest flood. The paleohydrologic bound is consistent with the observed streamflow data and
88
suggests that there have been no floods in the geologic record substantially larger than the
snowmelt floods observed from the gage record. The 90 percent confidence interval
encompasses the observed data.
Figure 4.9. Approximate unregulated peak discharge, historical and paleoflood estimates, Arkansas River at Adobe
Park. A scale break is used to separate the gage and historical data from the longer paleoflood record.
Table 4.5: Arkansas River at Adobe Park Peak Discharge Frequency Results
Annual
Exceedance
Probability (%)
Return Period
(years)
Peak Discharge (ft
3
/s)
Model Estimate*
5% Confidence
Limit*
95% Confidence
Limit*
10 10 4,910 4,570 5,360
4 25 5,660 5,180 6,350
2 50 6,180 5,570 7,120
1 100 6,690 5,910 7,920
0.5 200 7,180 6,210 8,750
0.2 500 7,820 6,560 9,900
0.1 1,000 8,290 6,790 10,800
0.05 2,000 8,750 7,000 11,800
0.02 5,000 9,360 7,240 13,200
0.01 10,000 9,820 7,410 14,300
0.005 20,000 10,300 7,560 15,400
89
Figure 4.10. Approximate peak discharge frequency curve, Arkansas River at Adobe Park, including gage,
4.6 Regional Frequency Analysis
The main purpose of the regional frequency analysis was to compare peak discharge
frequency curves from the four sites, and show distinct differences between the locations. In
concept, the frequency curves should show clear differences between downstream, large rainfall-
runoff events and upstream snowmelt-dominant events. The goal is to demonstrate these mixed-
population differences, and document upper and lower basin frequency curve changes with scale,
process and elevation. If the frequency curves from the snowmelt-dominant upstream sites were
similar, and substantially different from downstream sites, one can use these results to limit areal
extent of extreme storms.
Non-dimensional peak discharge frequency curves, based on LP-III models derived from
the data set at each site, are shown in Figure 4.11. The at-site 10-year peak discharge was used
to non-dimensionalize the flood frequency model results. The lower-basin frequency curves
(Pueblo and Parkdale) clearly reflect rainstorms (mixed-population snowmelt and large
rainstorms), whereas the upper basin sites (Loma Linda and Adobe Park) are from snowmelt.
The frequency curves for the downstream locations are clearly different for the most extreme
floods; they have a much steeper shape. The curvature is primarily determined by the LP-III
90
model skewness coefficient. The log-space skew is positive at Pueblo (0.8) and Parkdale (0.3),
and negative at Loma Linda (-0.2) and Adobe Park (-0.2). The extreme peak-flow magnitudes at
Pueblo are substantially larger than at upstream locations within the same time period, giving a
much steeper frequency curve. The shapes and slopes of the Adobe Park and Loma Linda
frequency curves are very similar, and are clearly different than the two downstream locations.
The streamflow and paleoflood data do not show any evidence of extreme floods substantially
larger than that recorded in the gage record for the upper basin snowmelt locations. The
Parkdale frequency curve is similar in shape to the upstream curves for flows less than about the
10-year peak. This suggests a separation in flood process in the record; the upper end of the
Parkdale frequency curve behaves similarly to Pueblo, but is not as steep. One can infer from
this that there is a transition in peak-flow frequency behavior between Pueblo and Loma Linda;
storms that affect Pueblo and cause extreme floods do not cause as large peaks at Parkdale. The
results show that the use of a partial-area storm concept on the lower part of the basin is
warranted.
Figure 4.11. Approximate unregulated non-dimensional peak discharge frequency curves for the four sites within
the Arkansas River basin. Each curve is non-dimensionalized by its respective at-site 10-year model peak flow.
91
4.7 Summary
Flood frequency analysis was conducted using peak-flow (gage), historical and
paleoflood data at four locations along the main stem of the Arkansas River upstream from
Pueblo. The Expected Moments Algorithm was used with the LP-III distribution to estimate
flood frequency curves at Pueblo, Parkdale, Loma Linda (Wellsville) and Adobe Park (Salida).
Confidence intervals for each frequency curve were also estimated. Paleohydrologic bounds
spanned 550 to 10,000 years within the watershed and provide substantially longer record
lengths for frequency analysis. Peak flows in the lower watershed at Pueblo and Parkdale
reflected extreme floods from rainfall, and were relatively much larger than at the two upstream
locations. The lower sites had positive log-space skews. The largest floods at upstream sites
were from snowmelt; these locations had relatively flat frequency curves and negative log-space
skews. Using a regional index-flood approach, it was shown that there is a different population
of floods between upstream snowmelt and downstream rainfall-runoff sites. The frequency
curves are later used in Section 6 to contrast flood frequency curves estimated with a rainfall-
runoff model.
92
5.0 Extreme Storm Modeling
This section presents an extreme storm model that predicts basin-average extreme rainfall
depths and probabilities using stochastic storm transposition. The model is applied for the first
time in a mountainous region for the Arkansas River watershed; predictions are then used as
input to the TREX rainfall-runoff model (Section 6). A brief sensitivity analysis demonstrates
storm location and areal distribution effects on basin-average rainfall depth predictions.
5.1 Storm Transposition Overview
Extreme storms are considered in two dimensions (x,y) in space, in order to describe and
numerically model them. Cumulative rainfall totals for an extreme storm are shown in Figure
5.1. Two-dimensional models are able to quantitatively describe the extreme storm rainfall
magnitude at each specified coordinate (x,y) location. The vertical (third) dimension (z) is
ignored in the present work, as we focus on the magnitude of rainfall on the ground in space (x
and y). Operational storm models used for Probable Maximum Precipitation estimation typically
are developed in two dimensions (e.g., Wiesner, 1970). In addition to the spatial distribution of a
storm, the rainfall magnitudes are also described in time t at each (x,y) location by either a mass
curve (cumulative rainfall depth with time), or a hyetograph (rainfall intensity versus time).
Typical mass curves and hyetographs are described in standard hydrology textbooks (e.g. Chow
et al., 1988). Reclamation has used two hyetograph patterns for distributing PMP estimates in
time: a standard alternating block pattern arrangement with the maximum at 2/3t (Cudworth,
1989); and an exponential-like front-end loaded pattern with the maximum at t=0. The spatial
and temporal storm model used in this project is described in Section 5.2.
An important concept in hydrology is regionalization, which means studying the
hydrologic properties of a large and homogeneous region with the objective of applying the
results to a watershed within the region (Laurenson and Kuczera, 1998). Within a specific river
valley, there are often inadequate records of extreme storms (Wiesner, 1970). Frequency
analysis is a problem in hydrology because sufficient information is seldom available at a site to
adequately determine the frequency of rare events (Stedinger et al., 1993). The at-site storm
record within a watershed can be extended significantly by using data from the surrounding
region. Using the regional storm data, the hydrologist is substituting space for time. Space for
time substitution is one of the three principles advocated by NRC (1988) to improve estimates of
extreme flood probabilities. The regional data of given length are effectively equivalent to a
much longer record at the watershed of interest.
93
Figure 5.1. Typical extreme storm total rainfall accumulation estimates in space (x,y) for the June 2-6, 1921
Penrose, CO storm (114 hours). Isohyets (solid lines) represent total storm rainfall in inches over the area
encompassed by the line.
Storm transposition is a regionalization concept that involves moving (transposing)
storms within an area to the watershed of interest. Transposition involves relocating individual
storm precipitation within a region considered homogeneous relative to topographic and
meteorologic characteristics deemed significant to that storm (Cudworth, 1989). Transposition
greatly increases the available data for evaluating the rainfall potential for a drainage (Schreiner
and Riedel, 1978). Transposition concepts are illustrated in Figure 5.2. Stochastic storm
transposition is a generalization of the concept of storm transposition, which is the basis for
estimating PMP in the United States (e.g., Hansen et al., 1988). In the PMP application, storm
transposition is based on the assumption that there exist meteorologically homogeneous regions
such that a major storm occurring somewhere in the region could occur anywhere else in the
region, with the provision that there may be differences in the averaged depth of rainfall
94
produced based upon differences in moisture potential (NRC, 1988; Cudworth, 1989).
Stochastic storm transposition (SST) extends this concept by incorporating the probability of
occurrence (Fontaine and Potter, 1989).
Figure 5.2. Storm transposition concepts, showing transposition area A
tr
, watershed area A
c
, and storm area A
s
(modified from Foufoula-Georgiou, 1989).
The storm transposition area A
tr
is the area within which all the occurred storms can be
transposed anywhere in the region either with the same depths and an adjustment to their
probability of occurrence, or with the same probability of occurrence but with an adjustment to
their depths (Foufoula-Georgiou, 1989).
5.2 Stochastic Storm Transposition Model
5.2.1 Overview of Probability Concepts
When estimating the frequency of rare floods for a given catchment using storm rainfall
data, a pertinent question is: What is the probability of a rainfall averaging more than d inches
(in a specified duration) occurring over the catchment in question within a long period, such as
the life of a dam (Alexander, 1963)? The general approach for stochastic storm transposition
95
consists of two main parts: (1) estimating a basin-average depth of storm rainfall

d
c
over a
particular watershed with area A
c
during a time period At ; and (2) estimating the cumulative
probability distribution function (cdf) of the basin-average depth
F
d
c
( At )
during that time
period. The cdf of
d
c
depends on the joint distribution of the storm properties (described
below) and the storm position (x,y). For conceptual understanding, one can consider this joint
distribution in two parts: a transposition probability P
t
that represents the probability of a
storm center (x,y) falling within or near the watershed of interest; and a probability of occurrence
P
r
of a storm with depth that exceeds some minimum depth or rank in one year (Alexander,
1963). Alexander (1963) suggested that the probability of occurrence P
c
of a transposable storm
occurring over the watershed in any given year is equal to the product P
r
P
t
. Gupta (1972),
YAEC (1984) and Fontaine and Potter (1989) utilized other alternative definitions of SST
probabilities; these are summarized by Wilson (1989). The SST method described below uses
the theoretical analysis framework of Foufoula-Georgiou (1989) to estimate probabilities.
5.2.2 Stochastic Storm Transposition Theory
The formal theoretical framework for the stochastic storm transposition (SST) model that
is used here is adopted from Foufoula-Georgiou (1989) and Wilson and Foufoula-Georgiou
(1990). The SST model describes the annual exceedance probability of the maximum basin-
average depth over a catchment. The theory and equations to estimate

d
c
and its annual
exceedance probability are described below and follow that of Wilson and Foufoula-Georgiou
(1990). The SST model is later coupled with a rainfall-runoff model in order to estimate extreme
flood probabilities.
The maximum areally averaged depth that can occur over a catchment of area A
c
during a
time period t is estimated via:
d
c
(At )=
1
A
c
||
Ac
[
d( x , y , t
s
+At )-d ( x , y , t
s
)
]
dxdy
(5.1)
where
d
c
is the maximum areally-averaged depth, (x,y) are spatial coordinates and t is a
critical duration of rainfall in terms of flood production. This critical duration may be a fixed
time period such as 6, 12, 24 or 72 hours, up to the total storm duration t
r
, and t
s
is defined as:
t
s
=
||
Ac
[
d ( x , y , t
s
+At )-d( x , y ,t
s
)
]
dxdy
>
||
Ac
[ d( x , y , t +At )-d( x , y , t )] dxdy \t t
r
-At
(5.2)
and we interpret the random variable

d
c
(At )
as the maximum average depth from a storm
over a catchment with area A
c
during a time period equal to t.
An important step in SST is the storm selection. A storm severity criterion E is used to
select storms (Wilson, 1989):
Criterion E: {
d[ At , A] >d
min
}
(5.3)
96
where

d [ At , A] denotes the average storm depth over an area A accumulated over a period of
time t, and d
min
is some minimum depth value based on t and A, such that all storms having an
effect on the exceedance probabilities are included in the sample set.
Let
s
represent the random vector of storm characteristics that describe a storm. These
characteristics may be random variables, such as the average depth over an area (e.g., 10 mi
2
depth), or random functions, such as a function that describes the storm depth at a position (x,y)
and at time t (a stochastic model of the rainfall field). Let
p
represent a two-dimensional vector
which describes the position of the storm based on the x,y-coordinates of the location of its
center. The center of the storm can be defined as the location of the maximum observed total
depth, the location of the maximum accumulated depth over a period of time, or as the center of
mass of the storm. It is assumed here that the storm center is defined as the location of the
maximum observed total depth, and is estimated from the depth-area duration (DAD) catalog.
The cumulative distribution function F of the maximum average total storm depth
d
c
(At )
can be written as:
F
d
c
(At )
( d)=pr [
d
c
(At ) d]
(5.4)
This cumulative distribution F is now defined in terms of the joint distribution of
s
and
p
(Wilson and Foufoula-Georgiou, 1990), noting that (
s
,
p
) are the lower case variates:
F
d
c
( At )
( d )=
|
[ A
s
]
|
[ A
p
]
pr [

d
c
(At ) d\
s
, \
p
] dF
A
s
, A
p
(\
s
, \
p
)
(5.5)
and
F
A
s
,A
p
(\
s
,\
p
)
is the cumulative joint distribution function of the random vectors
s
and
p
. Wilson (1989) notes that [
s
] and [
p
] represent the functional spaces within which the
vectors
s
and
p
vary. One challenge is to estimate this joint distribution of storm
characteristics and storm position. Using Bayes' theorem, these distributions can be analyzed as
f (\
s
,\
p
)= f (\
s
\
p
) f (\
p
)
(5.6)
where
f (\
s
\
p
)
denotes the probability that a storm that occurred at a position given by
p
had
the storm properties
s
.
One is most interested in estimating the exceedance probability. Using the superscript a
for the annual probability of a variable, the annual exceedance probability G
a
of the maximum
average total storm depth

d
c
(At )
is:
G
d
c
(At )
a
(d)=1-F
d
c
(At )
a
(d)
(5.7)
To determine annual probabilities, let Z(t) represent a counting process of the number of extreme
storms in an interval of t years. One can eliminate

d
c
(At )
subscripts on F and G and write
(Wilson and Foufoula-Georgiou, 1990):
G
a
(d)=1-
_
+=0
pr
[
d
c
(At ) dZ (1)
]
pr [ Z (1)=+] (5.8)
This equation may be simplified by using a discrete summation over actual storms (Foufoula-
97
Georgiou, 1989):
G
a
(d)=p
a
(
d
c
>d)= p
s
_
j =1
N
s
p
j
(
d
c
>d
)
(
A
eff , j
A
tr
)
(5.9)
The stochastic storm transposition model is defined by coupling equations 5.5, 5.6, and
5.9. In order to develop a complete SST model, one needs to define the contents of the position
vector
p
and the storm vector
s
. In addition to these vectors, one needs to define a model that
describes the storm rainfall distribution in space, and a model that describes the storm center
occurrence in space. The maximum storm center depth needs to be described with a model. In
order to conduct rainfall-runoff simulations, a temporal model that describes the time-varying
rainfall depth or rates over the storm duration is also needed (e.g., Franchini et al., 1996).
5.2.3 Storm Spatial Distribution
There are several approaches to develop a spatial distribution of an extreme storm. The
approach that is implemented here is a simple, parsimonious model based on DAD data. It is
assumed that the storm is single-centered, and isohyets are geometrically similar in the form of
an ellipse (e.g., Hansen et al., 1982). The equation for the ellipse geometry is (Grossman, 1984):
(
x -x
s
)
2
a
2
+
(
y-y
s
)
2
b
2
=1
(5.10)
where (x
s
,y
s
) is the storm center and a and b are major and minor axes, respectively. This
assumed storm shape describes both within-storm amounts and storm totals, and can be an
adequate spatial representation based on the DAD data (Foufoula-Georgiou and Wilson, 1990).
Although storm shapes are generally very complex, Hansen et al. (1982) recommended using a
standardized elliptical pattern to represent the storm isohyets. The storm spatial pattern that is
adopted has geometrically similar ellipses with a major (a) to minor (b) axis ratio c, where c =
a/b (Figure 5.3). The storm orientation is defined clockwise in degrees from North (y) in the
half-plane (0, 180), following Hansen et al. (1982). The (x',y') coordinates of the ellipse after
rotation are (Grossman, 1984):
x ' =
(
x-x
s
)
cos0+
(
y-y
s
)
sin0
(5.11a)
y' =-
(
x-x
s
)
sin0+
(
y-y
s
)
cos 0
(5.11b)
Using this geometry to represent a complete storm, the parameters are c and , and user-defined
storm center (x
s
,y
s
). The ellipse parameter c is limited to [1.0, 8.0] where 1.0 represents a circle.
Typical c values for extreme storms range between 1.0 and 3.0 (Hansen et al., 1982; Foufoula-
Georgiou and Wilson, 1990). The storm orientation angle is limited to 180
o
in the half plane.
98
Figure 5.3. Storm spatial representation with storm center location (x
s
, y
s
), orientation , and cartesian coordinate
system.
5.2.4 Storm Center Distribution
The two general options for SST storm center locations are that storms may occur
anywhere within a homogeneous region with differences in average depth, or the depth may be
the same throughout the region with differences in probabilities. The approach that is considered
in this work is to vary the average depth in space for a fixed exceedance probability, duration,
and area, using the nonhomogeneous spatial point process model of Wilson and Foufoula-
Georgiou (1990) for depths greater than d
min
. This model considers the storm center location
(x,y) to be dependent on the storm center depth (d
o
) by determining the conditional pdf
f
XYD
o
( x , yd
o
)
and then the marginal pdf
f
D
o
(d
o
)
. The distribution function shape is
estimated empirically based on the available data. Wilson and Foufoula-Georgiou (1990)
modeled storm spatial occurrences in the Midwest using two functions: (1) a uniform distribution
for d < d
min
; and (2) a transformed bivariate distribution for
d>d
min
. In this work, we adopt a
uniform probability distribution of storm centers in space (Fontaine and Potter, 1989; Franchini
et al., 1996). The uniform distribution for (x,y) denoted f
XY
is:
f
XY
= f
XYD
o
>d
min
( x , yd
o
>d
min
)=
1
A
tr
I ( x , y)
(5.12)
99
where I(x,y) is an indicator function defined over A
tr
as
I ( x , y)=1 if ( x , y)-A
tr
I ( x , y)=0 otherwise
(5.13)
This relationship applies to a storm conceptualized with a single storm center (England, 2005).
It is later modified for application to the Arkansas River watershed, by limiting storm center
locations for larger-area storms within the basin. The storm center may be located at lower
elevations, with rainfall rates reduced at higher elevations.
5.2.5 Storm Temporal Distribution
The major theoretical developments of the SST approach by Foufoula-Georgiou (1989)
and Fontaine and Potter (1989) focused on estimating basin-average rainfall depths for a
specified duration. They did not include procedures to include the temporal distribution of
rainfall within the storm. This step is needed for rainfall-runoff modeling and hydrograph
generation. The input to a rainfall-runoff model such as TREX is in the form of rainfall
hyetographs or mass curves. There are several ways of distributing basin-average storm rainfall
depths over time. The method that is used as part of this project is normalized mass curves
(Huff, 1967; Koutsoyiannis and Foufoula-Georgiou, 1993) using DAD storm data and radar data.
Franchini et al. (1996) used mass curves from Huff (1967) to distribute basin-average depths
from SST in the Midwest.
Normalized mass curves are obtained using DAD data by: (1) dividing the cumulative
storm depth d at time t by the total storm depth; and (2) dividing the time t by the total storm
time (e.g., Koutsoyiannis and Foufoula-Georgiou, 1993). Examples of these curves for two
Colorado Front Range storms (June 1921 and May 1894) are shown in Figure 5.4. Rain rates r
[L/T] are determined by simple difference from each successive cumulative depth d:
r (t )=
d (t )
i+1
-d(t )
i
t
i+1
-t
i
(5.14).
5.3 Extreme Storm Modeling Results and Discussion
Extreme storm DAD data (Section 2) were used to implement the SST model. Several
simplifying assumptions were made in order to demonstrate the main concepts and estimate
extreme storm probabilities for application to the Arkansas River basin. The criterion for storms
to be transposed, using equation (5.3), is:
Criterion E: {
d[ At =t
r
, A=10 mi
2
] > 11.0in.} (5.15)
which means that the storm set is composed of storms that had a maximum observed 10 mi
2
total
depth over the whole storm duration (t
r
) exceeding 11 inches. Wilson (1989) and Foufoula-
Georgiou (1989) used similar criteria for selecting and transposing storms in the Midwest.
Based on this criterion, fifteen storms are identified and selected from the 77 extreme storms
100
0 0.2 0.4 0.6 0.8 1
t/t
r
0
0.2
0.4
0.6
0.8
1
d
/
d
t
r
10 mi
2
100 mi
2
200 mi
2
500 mi
2
1,000 mi
2
2,000 mi
2
5,000 mi
2
10,000 mi
2
20,000 mi
2
(a) June 2-6, 1921
0 0.2 0.4 0.6 0.8 1
t/t
r
0
0.2
0.4
0.6
0.8
1
d
/
d
t
r
10 mi
2
100 mi
2
200 mi
2
500 mi
2
1,000 mi
2
2,000 mi
2
5,000 mi
2
10,000 mi
2
20,000 mi
2
(b) May 29-31, 1894
Figure 5.4. Normalized mass curves from DAD data for: (a) June 2-6, 1921 (SW1-23) storm; and (b) May 29-31,
1894 (MR6-3) storm.
101
(Section 2); these storms are listed with pertinent properties in Table 5.1. Sensitivity of
estimated basin-average depth probabilities to this criterion is discussed below.
A storm transposition region A
tr
for the Arkansas River basin was selected based the
DAD data (Section 2) and hydrometeorological analyses in Hansen et al. (1988), and
encompasses the area between the Continental Divide (CD) and the 103
rd
meridian (CD-103).
The 397,200 mi
2
region includes parts of eight states, and also includes the Arkansas River study
watershed (Figure 5.5). One important feature of this region is the CD boundary and the location
of the study watershed on the western edge of the region. The basin orography and CD boundary
play a major role in estimating the spatial distributions of extreme storm centers and storm areas
within the region and the watershed. The region also has an irregular shape, being long and
relatively narrow. This geometry can affect the selection of a storm center location distribution.
Table 5.1: Fifteen Extreme Storms from DAD Catalog Transposed to Arkansas Watershed
Date Assignment No. Location State
Duration
(hours)
Orientation
(degrees)
total 10mi
2
depth
max 24hr 10mi
2
depth
areal extent
mi
2
Ellipse c
05/30/1935 MR 3-28A Cherry Creek CO 24 47 22.20 22.20 6,300 4.0
09/20/1941 GM 5-19 McColleum Ranch NM 78 16 21.20 12.10 38,000 3.5
05/30/1935 MR 3-28AZoneA Hale CO 24 32 21.20 21.20 1,291 2.0
05/04/1969 19690504bemCO Big Elk Meadow CO 96 14 18.21 11.83 5,000 2.0
06/13/1965 SW 3-23 Plum Creek CO 181 5 18.10 13.20 39,266 1.5
09/27/1923 MR 4-23 Savageton WY 108 46 16.90 9.50 95,000 2.5
06/06/1964 NP 2-23 Gibson Dam MT 36 141 16.40 14.90 12,096 2.0
06/17/1921 MR 4-21 Springbrook MT 108 38 15.10 13.30 52,600 1.0
06/09/1972 MR 10-12 Rapid City SD 12 172 14.90 14.90 2,000 2.0
06/06/1906 MR 5-13 Warrick MT 54 90 13.30 10.20 40,000 1.2
07/21/1905 GM 3-13 Elk NM 108 80 13.10 5.70 44,000 1.5
06/12/1949 R7-2-5 Prospect Valley CO 36 63 13.00 9.10 360 2.0
06/02/1921 SW 1-23 Penrose CO 114 0 12.00 12.00 1,000 2.5
07/27/1997 19970727ftcCO Fort Collins CO 32 0 12.00 10.00 1,000 2.0
07/31/1976 19760731bgtCO Big Thompson CO 4 25 11.70 11.70 50 3.5
The spatial occurrence of storm centers was assumed to be a homogeneous Poisson
process. Given the limited number of storms used, and their geographic centers (Figure 5.5), this
assumption means that storm transposition probabilities are equal within A
tr
, and are
independent of storm properties. The spatial model for all depths in space is based on equation
(5.12), and
p
consists of (x,y). This simplification was also made by Foufoula-Georgiou (1989)
and Franchini et al. (1996), and is sufficient for providing initial estimates of basin-average
rainfall depth probabilities for subsequent runoff modeling.
The temporal occurrence model of extreme storms was simplified to a Bernoulli process
with a success probability p
s
and estimated by:
102
p
s
=
N
s
N
(5.16)
where N
s
is the number of extreme storms, and N is the number of years of record. Based on the
DAD storm data presented in Section 2,
p
s
is 0.144 with N
s
=15 and N=104 years (1894-1997).
Figure 5.5. Storm transposition region and spatial distribution of the 15 selected extreme storms.
The joint estimation of storm properties
s
was simplified to a summation over all storms
sampled, rather than integration of the pdf over each property. In this way, the stochastic storm
transposition was employed using the actual storm properties for duration, orientation, areal
extent, and ellipse parameters (Table 5.1). Spatial and temporal properties from each observed
storm were used directly in the summation. This assumption results in equation (5.9) being used
for annual exceedance probabilities, following Foufoula-Georgiou (1989):
G
a
(d)=p
a
(
d
c
>d)= p
s
_
j =1
N
s
p
j
(
d
c
>d
)
(
A
eff , j
A
tr
)
(5.9)
103
where A
eff,j
is the effective area of the j
th
storm, A
tr
is the transposition region, and
p
j
(
d
c
>d)
is
the probability for the j
th
storm that the average depth over the catchment is greater than some
value d. This probability value is determined by simulating each observed storm j 1,000 times
uniformly within the effective area, using a 960m grid cell resolution over the A
eff,j
domain.
The major factors in (5.9) are the number of storms N
s
,
p
s
and A
eff,j
. The effective area
is defined as the area within which the storm must be centered and still cover at least one point
within the catchment (Wilson and Foufoula-Georgiou, 1990). This is a crucial definition that
allows the probability estimation to include extreme rainfall depths for storms centered outside
the catchment that partially cover or fully cover the watershed. The storm effective area changes
every time a new storm is simulated over the watershed. Previous investigators (e.g., Foufoula-
Georgiou, 1989; Wilson and Foufoula-Georgiou, 1990) used simple shapes for storms and
catchments, such as circles, ellipses, triangles and rectangles, and were able to use analytical
methods to estimate the catchment-storm interaction. Here, a new approach was developed using
numerical methods to account for the actual basin geometry and interaction with each storm
elliptical pattern. Using the June 1921 storm as an example, the orientation, areal extent and
ellipse parameter c dictate the intersection between the watershed and the storm (Figure 5.6). By
placing the storm footprint at every watershed cell, one obtains the storm-dependent effective
area. This area is a combined image of the watershed and storm geometries.
Figure 5.6. Example effective storm area determined from the intersection of the June 1921 storm and the Arkansas
watershed.
104
Based on the assumptions stated above, basin-average depth probabilities for the 15
storms are estimated and combined using (5.9). The results of the simulation are shown in
Figure 5.7. The shape of the basin-average depth distribution is generally similar to that
obtained for 1,000 mi
2
(Foufoula-Georgiou, 1989) and 2,000 mi
2
(Wilson, 1989) hypothetical
circular watersheds in the Midwest. The upper tail of the distribution is also similar in shape to
Fontaine and Potter (1989) and Wilson and Foufoula-Georgiou (1990). This upper tail shape is
due primarily to the limited storm sample, the use of fixed storm parameters, and interactions of
the storms with the watershed. The lower part of the distribution appears to give relatively large
average depths for this watershed. For a 0.01 annual probability, the basin-average depth is 7.0
inches; this value appears high compared to NOAA Atlas II published point rainfall information.
This is a result of the extreme storms used in the analysis. Their extreme space-time
characteristics result in large depth estimates for more common probabilities. For subsequent
runoff modeling, the rainfall distribution is truncated for AEPs greater than 0.01. The depth
distribution does encompass the design basin average rainfall estimate for Pueblo Dam (10.52
inches in 27 hours over 2,000 mi
2
), but falls short of the PMP general storm basin average depth
estimate (13.71 inches in 72 hours over 4,686 mi
2
). An alternative shape for the upper tail is
considered below as part of the sensitivity analysis.
3 4 5 6 7 8 9 10 11 12
Basin Average Depth (inches)
1e-06
1e-05
0.0001
0.001
0.01
0.1
1
A
n
n
u
a
l

E
x
c
e
e
d
a
n
c
e

P
r
o
b
a
b
i
l
i
t
y

p
a
15 storms; 1,000 samples per storm
Figure 5.7. Annual exceedance probability of the average rainfall depth d
c
over the Arkansas River watershed.
A sensitivity analysis was performed to demonstrate the effects of the number of storms,
number of simulations, storm locations, and storm area sizes. The procedure to estimate A
eff
was
105
modified to enable limiting storm center locations based on location. The flood runoff
characteristics within the Arkansas River basin indicate that the transition zone between rainfall-
runoff and snowmelt runoff is near Parkdale (Section 2). Based on the flood hydrology,
distribution of extreme storm center locations with elevation, and storm radar investigations
(Javier et al., 2005), it is hypothesized that extreme storm centers are restricted to locations east
of Parkdale. This assumption reduces the initial effective area (Figure 5.6) to that shown in
Figure 5.8. It is also hypothesized that there are restrictions to the size of extreme storms that
can occur in the Arkansas watershed. The most extreme storms that have occurred immediately
adjacent to the foothills, such as the June 1921 Penrose storm, the July 1976 Big Thompson
storm and the July 1997 Fort Collins storm, occurred over areal extents less than 5,000 mi
2
.
Figure 5.8. Example restricted effective storm area determined from intersecting the June 1921 storm and the
Arkansas watershed, and limiting storm centers west of Parkdale; 1,000 simulated centers shown.
The sensitivity of the probability distribution of storm depths was determined for four
cases. The first was to increase the number of extreme storms included in the analysis from 15
to 20, and to increase the number of simulations from 1,000 to 10,000. The second case was to
restrict the storm centers to locations east of Parkdale. The third and fourth cases were to restrict
the storm center locations and to restrict the maximum storm area to 5,000 mi
2
and 2,000 mi
2
,
respectively. Results of these simulations are shown in Figure 5.9. Increasing the number of
storms from 15 to 20 shifts the distribution to more frequent probabilities for the lower
magnitude depths. The results indicate that increasing the number of simulations has little effect;
the shape is nearly identical to the base case with 15 storms and 1,000 simulations. The
106
restriction of storm centers results in a shifted frequency distribution to smaller exceedance
probabilities, due to a reduction of A
eff
by about a factor of two or more (Table 5.2). The extreme
tail of the distribution also drops more quickly, but predicted maximum depths are about the
same. A restriction on storm areas has the largest effect on the probability distribution. The
distributions are shifted much lower, by over an order of magnitude for the 2,000 mi
2
case, and
maximum depths are dramatically reduced. The shape of the upper tail for all the simulated
cases is approximately similar. This suggests that the simulated distributions are a function of
the total depth, duration, orientation, and estimated storm shape characteristics from the actual
storms used in the analysis.
3 4 5 6 7 8 9 10 11 12
Basin Average Depth (inches)
1e-06
1e-05
0.0001
0.001
0.01
0.1
1
A
n
n
u
a
l

E
x
c
e
e
d
a
n
c
e

P
r
o
b
a
b
i
l
i
t
y

p
a
restricted storm centers east of Parkdale;
restricted storm area 2,000 mi
2
;
restricted storm centers
2
;
Figure 5.9. Sensitivity of the annual exceedance probability of the average rainfall depth d
c
over the Arkansas River
watershed due to number of extreme storms, restricted storm center locations, and restricted storm areas.
The SST sensitivity indicates that the critical factors are storm center locations and the
distribution of storm areas. Basin average depth and probability estimates for the base case and
restricted cases are subsequently used in runoff modeling to explore runoff prediction sensitivity
to the depth and probability estimates. Because a limited sample of storms were used, and can
potentially affect the upper tail of the rainfall distributions, we briefly explore alternative shapes.
The main portions of the rainfall frequency curves appear to be similar to a Normal distribution
(Figure 5.10) based on plotting the functions on probability paper. Alternative tail distributions
are postulated as straight lines (Figure 5.10); these give much larger AEPs for a specific depth.
The sensitivity of the flood peak distributions to these different tails are subsequently examined
107
in Section 6.
Table 5.2: Sensitivity of Effective Area Estimates for Restricted Storm Centers and Storm Areas
Assignment No.
Simulated Storms
Restricted Storm Centers
West of Parkdale
West of Parkdale and
Storm Areas Limited to
5,000 mi
2
West of Parkdale and
Storm Areas Limited to
2,000 mi
2
Aeff (km
2
) Aeff/Atr Aeff (km
2
) Aeff/Atr Aeff (km
2
) Aeff/Atr Aeff (km
2
) Aeff/Atr
MR 3-28A 105,591.4 0.103 56,668.3 0.055 47,833.8 0.046 27,371.5 0.027
MR 3-28AZoneA 40,456.4 0.039 18,221.9 0.018 18,221.9 0.018 18,221.9 0.018
GM 5-19 286,011.2 0.278 191,908.5 0.187 44,387.9 0.043 25,844.4 0.025
19690504bemCO 73,805.4 0.072 39,851.8 0.039 39,851.8 0.039 23,086.1 0.022
SW 3-23 219,303.0 0.213 150,522.2 0.146 34,109.3 0.033 34,109.3 0.033
MR 4-23 537,095.6 0.522 305,152.8 0.297 40,058.3 0.039 23,367.2 0.023
NP 2-23 106,270.6 0.103 64,369.2 0.063 37,176.4 0.036 22,024.4 0.021
MR 4-21 210,591.1 0.205 103,666.2 0.101 25,880.4 0.025 16,810.0 0.016
MR 10-12 45,050.6 0.044 22,843.7 0.022 22,843.7 0.022 22,843.7 0.022
MR 5-13 189,742.7 0.184 116,066.3 0.113 26,818.6 0.026 17,272.6 0.017
GM 3-13 240,497.0 0.234 143,526.3 0.140 30,258.9 0.029 18,677.1 0.018
R7-2-5 25,682.2 0.025 10,281.4 0.010 10,281.4 0.010 10,281.4 0.010
SW 1-23 36,035.5 0.035 16,528.0 0.016 16,528.0 0.016 16,528.0 0.016
19970727ftcCO 36,035.5 0.035 16,528.0 0.016 16,528.0 0.016 16,528.0 0.016
19760731bgtCO 16,624.7 0.016 6,388.5 0.006 6,388.5 0.006 6,388.5 0.006
5.4 Summary
A stochastic storm transposition model was presented to estimate basin-average rainfall
depths and probabilities. The spatial distributions of storms were described as an elliptical
pattern with major to minor axis ratio c, orientation and storm center (x
s
,y
s
). Temporal
distributions were estimated using mass curves. Two new features were developed and
implemented to extend the SST method: a numerical procedure to determine the effective area
for an arbitrary-shaped watershed rather than simple geometric shapes (circles); and a restriction
on storm center locations to account for basin orography. The SST method was applied for the
first time to a large, orographic region and to an actual watershed, rather than a hypothetical one.
Actual storms were used in the simulation to estimate extreme basin average rainfall depth
probabilities. Fifteen storms were selected and used in the stochastic storm generation based on
a criterion that the storm had a maximum observed 10 mi
2
depth exceeding 11 inches in the
region. A limited sensitivity analysis demonstrated that the effects of restricting storm centers
and storm areas are important on the basin-average depth frequency curve. These factors had not
been previously examined by others. Further research efforts in stochastic storm transposition
are needed to investigate impacts of storm center, orientation, duration and ellipse parameters of
storms, and their interactions with watersheds. The current SST model that was implemented
108
and applied to the Arkansas watershed should eventually be improved by completing further
DAD data analysis and simulating storms using Monte Carlo techniques.
Figure 5.10. Postulated alternative tail (Normal) distributions for basin-average rainfall depth probability curves.
Dashed lines represent more frequent, less accurate estimates that are not used in runoff modeling.
109
110
6.0 Flood Frequency with the TREX Model
This section presents an application of the TREX model to the Arkansas River basin.
TREX is the Two-dimensional Runoff, Erosion, and eXport model, and is a generalized
watershed rainfall-runoff, sediment transport and contaminant transport model. The TREX
model is described. The focus is a new application of the model to a large watershed and
demonstration that the model can be used to provide flood frequency estimates. The input data
for the Arkansas watershed are presented. Model calibration and validation results for the two
largest flood events are discussed, and the model ability to simulate extreme floods on a large
watershed is shown. Performance of the model for flood frequency estimation and comparisons
to a streamflow and paleoflood data-based peak-flow frequency curve are made. A sensitivity
analysis of selected hydrologic and hydraulic factors is conducted.
6.1 TREX Model and Inputs
The rainfall-runoff model that is utilized in this project is TREX, the Two-dimensional
Runoff, Erosion, and eXport model (Velleux et al., 2005; England, 2005; Velleux, 2005). The
TREX model is a physically-based, two dimensional hydrodynamic watershed model. TREX is
based on the CASC2D model that was originally developed by Pierre Julien and students at
Colorado State University. CASC2D details and theory are summarized in Julien and Saghafian
(1991), Julien et al. (1995) and Ogden and Julien (2002). It is classified as an event model as it
simulates the Hortonian (overland flow) surface watershed response from a single storm with no
soil infiltration capacity recovery between events. The major components of the model include:
rainfall interception, Green-Ampt infiltration, surface and channel runoff routing using the
diffusive wave method, soil erosion and sediment transport, and chemical transport. Surface
runoff is routed on the overland planes (in two dimensions) and through the channel network (in
one dimension) using an explicit finite difference formulation of the diffusive wave equation.
The model is described in a user's manual (Velleux et al., 2005) and in England (2005).
As part of this project, TREX has been applied to the Arkansas River above Pueblo,
Colorado. The available data within the Arkansas River watershed were presented in Section 2.
Here, the focus is on TREX model parameter estimation and calibration. An elevation grid,
obtained and processed from the USGS National Elevation Data set (NED), was used as the base
for the watershed hydraulic routing. A 960 m grid cell size was used in the modeling as a good
compromise between capturing detailed spatial variability faster model run times (Figure 6.1).
The number of active grid cells in the 11,869 km
2
watershed is 12,879. Based on this elevation
grid, channels were derived using a 100 cell area threshold for initiation; this resulted in 764
defined channel cells (shown as black squares in Figure 6.1) and 69 links.
111
Figure 6.1. Elevation grid (960 m) and channel cells (black squares) for modeling the Arkansas River basin
upstream of Pueblo.
A field reconnaissance was conducted within the Arkansas River basin on July 26-30,
2004 to locate paleoflood study sites and estimate river channel dimensions at select locations.
Channel base width and bank height properties were measured at 20 locations within the
watershed (Table 6.1). As in Orlandini and Rosso (1998), and many others, power functions
were fit based on drainage area to estimate channel base width (Figure 6.2) and bank height
(Figure 6.3) in space. Widths ranged from 11 to 57 m; a spatial map is shown in England
(2005). Bank heights ranged from 0.9 to 3.1m. A rectangular channel shape (sideslope z =0)
and sinuosity equal to 1.0 were assumed. Manning n was assumed constant throughout the
network.
The remaining model parameters to be estimated included Manning n for overland flow,
and Green-Ampt parameters for effective porosity
e
, effective suction head , saturated
hydraulic conductivity K
s
and effective soil saturation S
e
. Overland Manning n roughness values
were estimated by correlation with USGS National Land Cover Data set (NLCD) land classes
based on Engman (1986). Nine land use classes were used (Figure 6.4). Green-Ampt
parameters were estimated based on the STATSGO soils database and the equations presented in
England (2005). Eighteen soils classes were used (Figure 6.5).
112
Table 6.1: Channel Width and Bank Height Measurements in the Arkansas River Basin
Location
Drainage
Area (km
2
)
Channel
Width (m)
Channel Bank
Height (m)
Cottonwood Creek at Mouth 168.35 12.19 0.82
Chalk Creek near Nathrop (CODWR gage) 251.23 19.81 1.22
Texas Creek at Mouth 372.96 15.54 0.91
South Arkansas River near Salida (CODWR gage at mouth) 538.72 21.03 2.07
Badger Creek at Mouth 546.49 17.37 1.52
Beaver Creek at Mouth 554.26 38.1 1.52
Grape Creek near Westcliffe 828.80 24.38 1.37
Arkansas River at Granite (CODWR gage) 1,105.93 34.44 1.52
Fourmile Creek at Mouth 1,124.06 16.76 1.83
Arkansas River near Howard Lakes 1,437.45 25.6 1.52
Arkansas River at Buena Vista (near former USGS gage) 1,582.49 24.38 3.05
Arkansas River near Nathrop (USGS gage) 2,745.40 27.43 3.66
Arkansas River at Adobe Park 2,991.45 50.29 1.65
Arkansas River at Salida (CODWR gage) 3,154.62 44.5 1.22
Arkansas River at Swissvale (near Wellsville, u/s Badger Creek) 3,846.15 27.25 3.26
Arkansas River at Loma Linda 4,817.40 38.1 1.83
Arkansas River at Texas Creek 5,439.00 70.71 1.68
Arkansas River at Parkdale gage 6,599.32 41.45 3.05
Arkansas River in Royal Gorge upstream from Grape Creek 6,734.00 41.76 2.53
Arkansas River at Pueblo State Park 11,564.35 60.96 3.66
The remaining model inputs consisted of time step selection and rainfall. Constant time
steps equal to 2.5 and 5 seconds, depending on rainfall inputs, were used to ensure computational
stability for model simulations. Storm rainfall based on the DAD data for extreme storms
(Section 2) was used as rainfall input. These storms usually have rainfall depths specified in 6-
hour increments for various fixed area sizes. The spatial distribution of each storm was
represented with an ellipse (Hansen et al., 1982).
113
100 1000 10000
Drainage Area (km
2
)
1
10
100
F
l
o
o
d
p
l
a
i
n

W
i
d
t
h

(
m
)
width = 2.696 * area
0.32428
R
2
=0.65, n=20
Figure 6.2. Spatial channel width estimation from Arkansas River basin data.
100 1000 10000
Drainage Area (km
2
)
0.1
1
10
F
l
o
o
d
p
l
a
i
n

D
e
p
t
h

(
m
)
y = 0.26998 *area
0.25941
R
2
=0.48, n=20
Figure 6.3. Spatial channel bank height estimation from Arkansas River basin data.
114
Figure 6.4. Spatial Manning n overland flow index map. Classes are listed in Table 6.3.
Figure 6.5. Spatial Green-Ampt parameter index map. Classes are listed in Table 6.4.
115
6.2 Calibration and Validation to Largest Observed Floods
The criteria used for calibration and validation were peak discharge, runoff volume, and
time to peak of flood hydrographs. The parameters that were used to calibrate the model were
Manning n for overland cells and channel segments, saturated hydraulic conductivity, and initial
soil moisture.
The June 3-4, 1921 storm was selected for TREX model calibration. This flood is the
largest on record in the Arkansas River basin near Pueblo, and resulted in at least 78 deaths
(Follansbee and Jones, 1922). The rainfall amounts, hydrometeorology and flood runoff of this
extreme event are principally described in Munn and Savage (1922) and Follansbee and Jones
(1922), and exceeded 12 inches. Rainfall estimates are used from USACE (1945-) and Hansen
et al. (1988). Mass curves for the storm are shown in Figure 6.6; storm total isohyets are shown
in Figure 5.1 of Section 5. The peak flow from this flood is estimated to be between 83,500 ft
3
/s
and 103,000 ft
3
/s, depending on the timing of tributary flow from Dry Creek (Follansbee and
Jones, 1922). It was shown that the peak flow of this flood has a return period between about
150 and 1,000 years (Section 4). The calibration runoff data for this event are based on a table
and descriptions by Munn and Savage (1922, p. 10):
The main part of the flood started at 5:00 pm on June 3
rd
, and totaled more than 78,000 acre-feet
in the following 18 hours. At 10:30 am June 4
th
, the river was still flowing 40,000 second-feet, as
estimated by Mr. Hosea. There is, however, no information on which to base an accurate
estimate of the volume added after this time. It is probable that fully 20,000 acre-ft was added
and that the main part of the flood, starting at 5:00 pm June 3, and ending some time during the
night of June 4
th
, totaled about 100,000 acre-ft.
Their estimated data for the June 3-4 hydrograph are shown in Figure 6.7.
0 3 6 9 12 15 18 21 24
Time (hours) from 15:00 June 3, 1921
0
2
4
6
8
10
12
R
a
i
n
f
a
l
l

D
e
p
t
h

(
i
n
c
h
e
s
)
10 mi
2
100 mi
2
200 mi
2
500 mi
2
1,000 mi
2
Figure 6.6. Mass curves for the June 1921 storm.
116
The results of the calibration are shown in Figure 6.7. The model does a good job at
matching the peak, volume and time to peak (Table 6.2). The shape of the model hydrograph is
considered good given the uncertainty in the data. Calibrated estimates of spatially varying
Manning n (Table 6.3) and spatially varying infiltration parameters (Table 6.4) are within
published ranges (Engman, 1986; Rawls et al., 1993). A calibrated Manning n for channels was
set equal to 0.050. This value is within the range of published estimates in Colorado (Jarrett,
1985). Model results, consisting of rainfall rate, water depth, and cumulative infiltration, are
shown for selected time steps in Figures 6.8 through 6.10. Based on this calibration, the model
can successfully be used to simulate extreme floods on the Arkansas River watershed.
0 6 12 18 24 30 36 42 48
Time (hours) from 15:00 June 3, 1921
0
20000
40000
60000
80000
100000
D
i
s
c
h
a
r
g
e

(
f
t
3
/
s
)
June 1921 estimated hydrograph (Munn and Savage)
TREX model hydrograph
Figure 6.7. June 1921 extreme flood hydrograph and TREX model calibration.
Table 6.2: Calibration Results for the June 1921 Flood
Hydrograph Peak Discharge (ft
3
/s) Runoff Volume (acre-ft) Time to Peak (hours)
Observed 100,000 100,000 11.5
TREX Model 100,200 105,600 10.3
Percent Difference 0.2 5.6 -10.5
117
Table 6.3: Calibrated Manning n Estimates for Overland Flow Grid Cells
Map
No.
Land Use
Class No.
USGS NLCD Land Use Class Name
Calibrated
Manning n
Percent of
Watershed
1 11 Open Water; Perennial Ice/Snow 0.07 0.71
2 21 Low and High Intensity Residential; Commercial/Ind/Transp. 0.02 0.60
3 31 Bare Rock/Sand/Clay; Quarries/Strip Mines/Gravel Pits 0.03 3.14
4 41 Deciduous Forest; Evergreen Forest; Mixed Forest 0.52 42.07
5 51 Shrubland 0.59 22.85
6 71 Grasslands/Herbaceous 0.20 28.36
7 81 Pasture/Hay 0.46 2.09
8 82 Row Crops; Small Grains; Fallow 0.21 0.15
9 85 Urban/Recreational Grasses 0.33 0.04
Table 6.4: Calibrated Green-Ampt Infiltration Parameters for Overland Flow Grid Cells
Soils
No.
USDA Texture Class
(STATSGO Database)
Porosity
(cm
3
/cm
3
)
Effect.
Poros.
e
(cm
3
/cm
3
)
Effective
Saturation
S
e
Suction
Head
(cm)
Sat. Hydraul.
Conductivity
K
s
(cm/hr)
Percent of
Watershed
1 very bouldery sandy loam 0.363 0.455 0.1 27.72 0.3 6.22
2 cobbly loam 0.437 0.450 0.1 20.76 0.48 7.89
3 very cobbly sandy loam 0.321 0.407 0.1 19.03 0.57 1.36
4 clay loam 0.528 0.426 0.1 27.42 0.2 2.61
5 channery loam 0.464 0.418 0.1 22.63 0.4 4.99
6 fine sandy loam 0.465 0.411 0.1 12.58 1.2 9.73
7 gravelly coarse sandy loam 0.377 0.352 0.1 23.75 0.3 2.03
8 gravelly sandy loam 0.446 0.415 0.1 20.24 0.5 12.83
9 very gravelly loam 0.498 0.463 0.1 30.54 0.22 3.40
10 very gravelly sandy loam 0.431 0.400 0.1 29.81 0.21 28.52
11 loam 0.473 0.408 0.1 26.21 0.25 1.98
12 loamy sand 0.472 0.422 0.1 7.44 4.38 0.13
13 silt loam 0.491 0.413 0.1 34.97 0.13 6.97
14 sandy loam 0.528 0.460 0.1 7.75 3.93 1.84
15 stony sandy loam 0.448 0.399 0.1 10.74 1.57 0.75
16 very stony loam 0.165 0.470 0.1 20.64 0.52 0.71
17 very stony sandy loam 0.257 0.418 0.1 16.17 0.91 0.31
18 extremely stony loam 0.050 0.408 0.1 31.41 0.18 7.73
118
Figure 6.8. Cumulative rainfall, surface depth, and cumulative infiltration results at 3.5 hours.
Figure 6.9. Cumulative rainfall, surface depth, and cumulative infiltration results at 10.3 hours (at peak).
Figure 6.10. Cumulative rainfall, surface depth, and cumulative infiltration results at 24 hours.
119
The model was validated with rainfall and runoff from the May 31, 1894 flood. This
flood caused the third largest estimated peak flow on the Arkansas River at Pueblo (Section 4),
and was the most destructive flood in the history of the Arkansas valley prior to June 1921
(Follansbee and Jones, 1922 p. 38). Rainfall data for this event were obtained from Reclamation
files and the U.S. Army Corps of Engineers catalog for storm MR6-14 (Section 2). Mass curves
of rainfall for the storm are shown in Figure 6.11. This storm was of lower intensity and longer
duration than June 1921. Flood flow for this event is based on estimates published by the USGS
at Canon City (staff gage read twice daily), and a peak flow estimate (39,100 ft
3
/s) made by the
Pueblo city engineer using slope-conveyance with a cross section just upstream of Pueblo.
0 6 12 18 24 30 36 42 48 54 60
Time (hours) from 12:00 May 30, 1894
0
1
2
3
4
5
6
7
8
9
R
a
i
n
f
a
l
l

D
e
p
t
h

(
i
n
c
h
e
s
)
10 mi
2
100 mi
2
200 mi
2
500 mi
2
1,000 mi
2
Figure 6.11. Mass curves for the May 1894 storm.
The validation results are shown in Table 6.5 and Figure 6.12. The initial soil moisture
was increased so that the model matched the observed runoff data. All other parameters retained
their calibrated values. The model does a fairly good job at matching the peak flow,
approximate time to peak and total runoff volume (Table 6.5), within the uncertainty of the data.
Given the storm rainfall, the validation run balances matching the peak flow at Pueblo and the
runoff volume at Canon. If the modeled peak flow at Pueblo were to increase, this would also
increase the runoff volume at Canon City. The other factor in the validation run is the potential
errors in rainfall inputs. Fontaine (1995) shows that errors in precipitation data are the primary
source of uncertainty in calibrating rainfall-runoff models for extreme floods. As the rainfall
input data are inexact for this storm, changes in the rainfall forcing can have an effect on the
validation.
120
Table 6.5: Validation Results for the May 1894 Flood
Variable
Canon City Pueblo
Observed Model
Percent
Difference
Observed Model
Percent
Difference
Peak Discharge (ft
3
/s) 39,100 35,100 -10
60-hour Volume (acre-ft) 18,460 20,300 9.9
0 6 12 18 24 30 36 42 48 54 60 66 72
Time (hours) from 12:00 May 30, 1894
0
5000
10000
15000
20000
25000
30000
35000
40000
D
i
s
c
h
a
r
g
e

(
f
t
3
/
s
)
TREX Pueblo model hydrograph
TREX Canon City model hydrograph
Observed Canon City hydrograph
Estimated Pueblo Peak
Figure 6.12. May 1894 extreme flood hydrographs and TREX model validation.
6.3 Peak-Flow Frequency Estimation with Extreme Storms
The calibrated and validated model was then used to estimate a flood frequency curve.
The inputs to the model were storms used in stochastic storm transposition (Section 5). Basin-
average depths were selected for specified annual exceedance probabilities in the general range
of 0.01 to 0.0001. The lower limit (0.01) was chosen because the SST results suggested that
depths at more frequent probabilities were not reliable (Section 5). Storms were then
constructed to distribute the basin-average rainfall depths in space and time over the watershed.
Because a simplified SST methodology was implemented to estimate a rainfall depth frequency
curve based on actual storms, an actual storm is used as input to TREX. The following storm
properties are considered fixed: storm center location, orientation, storm area, ellipse parameter,
spatial and temporal distributions. The storm that contributed the largest depths in the SST
(NP2-23, June 1964) was selected as the base pattern. This storm covers 12,096 mi
2
, has an
orientation of 141
o
, c = 2.0 and the duration is 36 hours. The center was transposed to just west
121
of Parkdale, and the storm spatial distribution covers the entire Arkansas River watershed
(Figure 6.13). The orientation most likely optimizes flood runoff production, as the major axis is
nearly in line with the river in the lower watershed.
Figure 6.13. Spatial storm pattern for TREX model runs and flood frequency.
The input to the TREX model for flood frequency consisted of the storm with area
amounts adjusted to meet the target basin-average depth, and the calibrated model parameters.
The initial soil saturation was set to 0.5 for each run. The flood frequency curve was then
determined by selecting rainfall depths from the basin-average rainfall frequency curve and
running the TREX model for each specified depth. In this way, the peak-flow probabilities are
conditioned on the rainfall depth probabilities, as the inputs are held constant except for the
rainfall depth, so the rainfall term contributes to the peak flow distribution (England, 2005). The
complete rainfall depth frequency curve is discretized into sixteen points for subsequent runoff
modeling. The computer time to complete 16 TREX model runs for the rainfall depths ranged
from five to 16 hours, depending on model inputs.
A flood frequency curve is estimated based on the model inputs, storm and rainfall
depths. The SST basin-average depth rainfall probability distribution with no storm area or
location restrictions (Figure 5.7) is used as the input. The rainfall curve and resulting TREX
flood frequency curve are shown in Figure 6.14. The flood frequency curve has the same overall
shape as the rainfall frequency curve. The main portion of each distribution from 1 to 0.1
percent is nearly Normal; the tails of both distributions flatten. The rainfall tail behavior was
discussed in Section 5; clearly the rainfall distribution upper tail affects the shape of the peak-
flow frequency curve. Flood frequency curves at Salida, Wellsville, Parkdale and Pueblo are
122
shown in Figure 6.15. The reduction in peak flow magnitude from downstream to upstream is
readily apparent.
Figure 6.14. Flood frequency curve at Pueblo from TREX with corresponding SST basin-average depth curve with
no area or location restrictions.
2 1 0.5 0.2 0.1 0.05 0.02 0.01 0.005
Annual Exceedance Probability (%) | Rainfall Depth [p
a
(q) | p
a
(d)]
0
50000
100000
150000
200000
250000
300000
P
e
a
k

D
i
s
c
h
a
r
g
e

(
f
t
3
/
s
)
Salida
Wellsville
Parkdale
Pueblo
Figure 6.15. Flood frequency curves at Pueblo, Parkdale, Wellsville and Salida from TREX.
123
The spatial rainfall distribution and storm center location are the principal factors in the
relatively dramatic reduction of peak flows from downstream to upstream sites. Runoff
hydrographs for the largest simulated rainfall depth (basin average 10.9 inches) that has an
estimated AEP of 0.00007 are shown in Figure 6.16. The hydrograph shapes indicate that the
model is stable and can simulate extreme floods on a watershed of this scale. It has been
demonstrated that the model can be used to estimate a flood frequency curve. Model predictions
can now be compared to data-based flood frequency curves. Sensitivity analysis is conducted to
examine the effects of some factors on model predictions.
0 6 12 18 24 30 36 42 48 54 60 66 72
Time (hours)
0
50000
100000
150000
200000
250000
300000
D
i
s
c
h
a
r
g
e

(
f
t
3
/
s
)
Salida
Wellsville
Parkdale
Pueblo
Figure 6.16. Runoff hydrographs for the largest simulated rainfall depth at four locations in the watershed.
6.4 Model and Paleoflood Data-Based Peak-Flow Frequency
Comparisons
The TREX model flood frequency curves at four locations (downstream to upstream),
Pueblo, Parkdale, Wellsville and Salida, are compared with LP-III peak-flow frequency curves
based on streamflow, historical and paleoflood data (Section 4). Here we are interested in the
differences in flood frequency shapes, if the TREX model-generated frequency curve matches
the LP-III curves, if it falls within the confidence intervals, and if it matches magnitudes and
plotting position estimates of the largest floods and paleofloods. The basic concept for making
the comparisons is that the TREX model flood frequency curve should be consistent with the
gage, historical and paleoflood peak flow estimates.
124
The frequency curves for the four sites are shown in Figure 6.17. LogNormal probability
paper is used for the graphical comparison; the color and symbol for each TREX frequency
curve corresponds to that shown in Figure 6.15. Recall that from Sections 2 and 6, runoff
mechanisms were distinguished between snowmelt (Salida and Wellsville) and rainfall-runoff
(Parkdale and Pueblo). The focus of these comparisons is thus in the lower watershed at
Parkdale and Pueblo as we are simulating runoff from extreme rainfalls. Nevertheless, TREX
flood frequency curves at the upper sites are compared to data-based snowmelt-dominant
frequency curves to see if the rainfall-runoff generated curves are much larger than the snowmelt
curves.
The comparisons indicate that the TREX flood frequency curves are larger than the data-
based frequency curves at the four locations. At Pueblo, the upper tail of the TREX frequency
curve is within the data-based confidence interval. This upper tail is directly related to the
rainfall distribution. The impact of changing the rainfall frequency curve on the resulting flood
frequency is conducted below as part of sensitivity analysis. The lower portion of the curve at
Pueblo, as well as the entire curve at Parkdale, appears too high compared to the data. This
suggests that the rainfall magnitudes and potentially the initial soil moisture might be too high.
The shapes of the TREX flood frequency curves are also distinctly different than the positive-
skewed data-based LP-III distributions; they have flatter slopes in the main portion of each curve
and relatively flat tails. The curves at upstream locations (Wellsville and Salida) also have
dramatically different shapes compared to the data-based frequency curves. The lower portions
are much steeper, reflecting rainfall-runoff, and the upper portions clearly exceed the streamflow
observations, but are consistent with the paleohydrologic bounds. Recall that paleohydrologic
bounds are defined as a time interval where a given discharge has not been exceeded (Levish,
2002). At Wellsville, the TREX curve is close to the lower paleohydrologic bound, and is less
than the 10,000-year paleohydrologic bound. At Salida, the curve is less than the
paleohydrologic bound within the time frame of the bound. However, the curves at the upper
sites are still much larger than any observation. This suggests that rainfall amounts in this part
of the watershed are too large.
6.5 Sensitivity Analysis
A sensitivity analysis was conducted to investigate the impacts of three main factors on
the flood frequency curves: initial conditions, spatial distributions of storms, and storm temporal
distributions. Frequency curve comparisons to data-based flood frequency curves are then
revisited.
6.5.1 Initial Conditions
As discussed in England (2005), initial soil moisture conditions play an important role in
runoff predictions, and can affect infiltration rate, peak and volume. The role of initial soil
moisture in peak flow predictions is explored by changing the initial soil saturation S
e
.
125
Figure 6.17. TREX model flood frequency and streamflow/paleoflood frequency curves at (a) Pueblo, (b)
Parkdale, (c) Wellsville and (d) Salida. Based on SST curve with no area or location restrictions.
126
Figure 6.17 (continued). TREX model flood frequency and streamflow/paleoflood frequency curves at (a) Pueblo,
(b) Parkdale, (c) Wellsville and (d) Salida. Based on SST curve with no area or location restrictions.
127
The value used for the results shown above is 0.5, and represents a 50% saturation level across
the watershed. Three saturation levels are chosen for sensitivity: 0.05 (dry), 0.2 (slight
saturation) and 0.8 (near saturation). The initial soil moisture amounts were assumed to be
uniform in space across the watershed domain.
The soil moisture sensitivity results are shown in Figure 6.18 for the three sensitivity
cases as compared to the base case. In all runs, the main SST depth-average rainfall curve was
used, and all other parameters were held constant. It is apparent that the soil saturation has a
moderate to large effect on the predictions. The curves are shifted from the base run (shown in
blue), but the frequency curves do not change shape substantially. The peak discharges and
volumes (not shown) can increase by a factor of 1.24 to 1.41 for S
e
= 0.5 to 0.8, or decrease by
about 1.18 to 1.34 for S
e
= 0.5 to 0.2. As rainfall (peak discharge) increases, the change in initial
soil moisture has less of an effect; larger percent differences are observed for the lower peaks.
The slope of the curves change slightly. The amount of change for the largest floods with S
e
=
0.8 to 0.05 is 1.58, and for the more frequent floods the ratio with S
e
= 0.8 to 0.05 is 2.15. Thus,
the initial soil moisture does play a major role in flood peak predictions from TREX.
2 1 0.5 0.2 0.1 0.05 0.02 0.01 0.005
a
(q) | p
a
(d)]
0
100000
200000
300000
400000
P
e
a
k

D
i
s
c
h
a
r
g
e

(
f
t
3
/
s
)
Se=0.05
Se=0.2
Se=0.5
Se=0.8
Figure 6.18. TREX model flood frequency curves at Pueblo with varying initial soil saturation Se.
6.5.2 Spatial Distributions of Storm Rainfall
The sensitivity of runoff peak flows to the spatial distribution of storm rainfall was
explored by placing restrictions on the storm location, storm area, and storm pattern. Three
128
experiments were performed. The first was to restrict the storm center (x,y) locations such that
storm centers were allowed east of Parkdale. This restriction mimics what has been seen in the
historical record, where storms have centered near Penrose in the lower watershed. The second
experiment was to restrict storm centers to east of Parkdale and restrict storm areas to 5,000 mi
2
or less. The third experiment was to restrict storm centers to east of Parkdale, restrict storm
areas to 2,000 mi
2
or less, and use a different extreme storm with different properties (location,
duration, orientation, and ellipse ratio). These experiments correspond to the variations in SST
rainfall depth frequency curves shown in Section 5, and reflect conditions for some of the
extreme storms in the database and the radar data (Section 2).
The results of the spatial distribution changes are shown in Figure 6.19. Each frequency
curve corresponds to an experiment; the case with no area or location restrictions is shown in
black diamonds and the curve colors and symbols match those in Figure 5.9 (Section 5). It is
evident that a restriction in the storm spatial distribution can significantly affect the peak flow
frequency curve shape. As the storm location within the watershed is restricted, and storm areas
reduced, the peak flow frequency curves shift to the right so that a given peak discharge is less
frequent. There is variability in frequency curve shape due to the shifts in basin-average depth
probabilities. The effective storm area reduction (described in Section 5) causes a reduction in
flood probabilities; the tails of the distributions are extended. The frequency curve for restricting
storm areas to less than 5,000 mi
2
steepens (blue curve with squares) because for a fixed basin-
average rainfall, the depth is distributed over a smaller area. The outer two ellipses of the storm
pattern (Figure 6.13) are eliminated. The largest reduction in flood peaks is due to restricting the
storm area to 2,000 mi
2
and utilizing a different storm pattern, as shown by the green line with
inverted triangles (Figure 6.19). In this case, the SW3-23 June 1965 storm was used as the
rainfall pattern. The longer storm duration (181 hours versus 36 hours for the main case) and
areal reduction cause the dramatic shift and reduction in flood peak discharge. These flood
frequency curves mimic the shapes of the rainfall frequency curves (Figure 5.9 in Section 5).
6.5.3 Storm Duration and Temporal Distribution
In order to show effects of storm duration on the flood frequency curve, two experiments
are conducted with the 36-hour NP2-23 storm. The first experiment is to stretch the existing
storm from 36 hours to 48 hours, a factor of 1.33. In this way, each of the 6-hour rainfall
amounts in the mass curves are extended by this factor. The second experiment is to change the
within-storm temporal distribution. The existing storm temporal distributions for the nine area
sizes are shown in Figure 6.20. All curves are very similar, and follow a first quartile
distribution (Huff, 1967). This curve is then flipped to follow a fourth quartile storm. These
two quartiles span the range of temporal distributions of large storms.
129
2 1 0.5 0.2 0.1 0.05 0.02 0.01 0.005 0.002
a
(q) | p
a
(d)]
0
50000
100000
150000
200000
250000
300000
P
e
a
k

D
i
s
c
h
a
r
g
e

(
f
t
3
/
s
)
2
;
restricted storm centers; sw3-23 storm
2
;
restrict storm centers east of Parkdale;
15 storms, 1,000 samples per storm
Figure 6.19. TREX model flood frequency curve at Pueblo with varying basin-average depth rainfall frequency.
0 0.2 0.4 0.6 0.8 1
Dimensionless Time t/t
r
0.2
0.4
0.6
0.8
1
D
i
m
e
n
s
i
o
n
l
e
s
s

D
e
p
t
h

d
/
d
t
r
50 mi
2
100 mi
2
200 mi
2
500 mi
2
1,000 mi
2
2,000 mi
2
5,000 mi
2
10,000 mi
2
12,096 mi
2
temporal
rearranged
Figure 6.20. Rearranged temporal distribution of the 36-hour NP2-23 storm of June 1964.
130
The results of the two temporal distribution experiments are shown in Figure 6.21 with
the base run (black line with diamonds). Given this storm, the rearrangement of the temporal
distribution results in a minor decrease (7 to 9 percent) in peak flows; the curve shifts slightly
lower. The increase in storm duration has a much larger effect. Peak flows decrease by factors
between 1.46 and 1.66, and the slope of the frequency curve flattens, especially at the upper tail.
2 1 0.5 0.2 0.1 0.05 0.02 0.01 0.005
a
(q) | p
a
(d)]
0
50000
100000
150000
200000
250000
300000
P
e
a
k

D
i
s
c
h
a
r
g
e

(
f
t
3
/
s
)
duration stretch 36 to 48 hours
rearrange temporal distribution
15 storms, 1,000 samples per storm
Figure 6.21. TREX model flood frequency curves at Pueblo with rearranged temporal distributions and stretched
durations.
6.5.4 Model and Peak Flow Frequency Revisited
Based on the sensitivity analysis, we now revisit the comparisons between TREX and
data-based flood frequency curves. It was shown earlier that model flood frequency curves are
larger than the data and data-based flood frequency curves. The storm spatial distribution (via
changes in basin-average depth frequency), initial soil moisture, and storm duration all can affect
the model frequency curve shape and peak-flow magnitudes. Considering only the restriction on
the storm location to east of about Parkdale and storm area less than 5,000 mi
2
(Figure 6.22), we
compare model frequency curves with data-based ones at the four locations. The SST basin-
average rainfall curve that is used is shown in Figure 5.9 (blue with filled squares). A Normal
distribution upper tail for the rainfall distribution is also considered. By limiting the
comparisons to the rainfall curve, we can isolate the effect on the resulting flood frequency
curves.
131
Figure 6.22. Restricted spatial storm pattern for TREX model runs and flood frequency.
The model flood frequency curves at the four sites, using the restricted storm center and
5,000 mi
2
restricted area, are shown in Figure 6.23. The TREX model frequency curves at the
four locations changed fairly substantially from the base case results (Figure 6.17). The lower
basin curves at Pueblo and Parkdale have shifted down and have steepened slopes. The curve at
Pueblo is well within the LP-III confidence intervals, but the shape has a strong negative skew
with a very flat upper tail. The Normal distribution tail approximation on the rainfall curve
results in a curve that is closer to the data slope. The curve at Parkdale is close to the data in the
lower tail, but still appears to be much higher than the data in the upper tail and inconsistent with
the paleohydrologic bound. These two curves have different shapes than the data-based curves
clearly because of the rainfall frequency distribution. As this rainfall distribution is based on
storm transposition, there is no reason why the model and data-based flood distributions should
necessarily be the same, as storms transposed into the watershed are not part of the observed
flood records. The data-based flood frequency curve is based in-part on the June 1921 and May
1894 storms; these events were used in calibration/validation as well as in SST. The
comparisons suggest that storms should be further limited in areal distribution, location, and/or
intensity to reduce the peaks at Parkdale.
In the upper watershed, the curves have shifted substantially downward and steepened
due to the limited rainfall extent (partial area) for the storm (Figure 6.22). The upper tails are
consistent with the confidence intervals from the snowmelt-dominant frequency curves.
132
Figure 6.23. TREX model flood frequency and streamflow/paleoflood frequency curves at (a) Pueblo, (b)
Parkdale, (c) Wellsville and (d) Salida. Based on SST curve with restricted storm centers and maximum 5,000 mi
2
storm area. Short solid line represents peak probabilities based on alternative Normal distribution rainfall
extrapolation.
133
Figure 6.23 (continued). TREX model flood frequency and streamflow/paleoflood frequency curves at (a) Pueblo,
(b) Parkdale, (c) Wellsville and (d) Salida. Based on SST curve with restricted storm centers and maximum 5,000
mi
2
storm area. Short solid line represents peak probabilities based on alternative Normal distribution rainfall
extrapolation.
134
Normal distribution rainfall extensions appear to be too steep in this area of the watershed.
Because the data-based frequency curves represent snowmelt, further reductions in rainfall
amounts may be warranted in this part of the watershed. In order to effectively match the peak-
flow and paleoflood data at Pueblo, restricted storm rainfall areas and storm center locations
need to be used. The peak-flow frequency curve based on storm centers restricted to near east of
Parkdale and storm areas less than or equal to 5,000 mi
2
is used as the best estimate for
subsequent reservoir elevation frequency (Section 7). An upper estimate for reservoir
frequency is based on the SST curve with no restrictions.
6.6 SUMMARY
The TREX model was applied to the Arkansas River basin upstream of Pueblo, Colorado.
The inputs for estimating model parameters were described. The model was successfully
calibrated to the June 1921 flood, the largest flood on record in the basin. The model was
validated with the May 1894 flood, the third largest flood on record. It was shown that the
model can successfully be applied to a large watershed of this scale (12,000 km
2
) and simulate
extreme floods.
The calibrated model was used to estimate flood frequency curves. Basin-average
rainfall depths and probabilities were obtained from a stochastic storm transposition model that
was applied to the watershed (Section 5). It was demonstrated that the model can be used to
develop a flood frequency curve. TREX model frequency curves at four sites within the
watershed were compared with data-based peak-flow frequency curves with paleoflood data.
The comparison showed that peak flows from the rainfall-runoff model were generally larger
than the data-based flood frequency curves. The shapes of the rainfall frequency curves dictated
the shapes of the flood frequency curves. Rainfall frequency and flood frequency curve extreme
upper tails were relatively flat compared to LP-III curves.
A series of sensitivity analysis experiments were performed to illustrate the effects of
initial soil moisture, rainfall distribution, storm location and area, and storm duration and
temporal distribution on flood frequency predictions. A particular factor was changed and others
were held constant. Initial soil moisture was found to be important and could affect peak flows
by a factor of 1.18 to 2.15 for the four cases considered. The effect of storm spatial distribution
was simulated by restricting storm locations and areas for three cases, and was an important
factor. Frequency curves generally became stretched out (less frequent probabilities for a fixed
discharge), and in some cases steeper as storm areas were reduced and centers restricted to the
lower watershed. A change in rainfall temporal distribution had less of an effect than extending
the storm duration. Based on the sensitivity analysis, a better match between model frequency
curves and peak-flow and paleoflood data-based frequency curves was obtained at some
locations by restricting storm area and storm center locations. This frequency curve is used for
reservoir elevation estimation.
135
136
7.0 Reservoir Elevation Frequency and Risk Analysis
Inputs
This section presents an analysis of reservoir elevation data for Pueblo Dam, reservoir
routings of extreme flood hydrographs, reservoir elevation probability estimates, and inputs for
subsequent hydrologic risk analysis studies. Daily reservoir elevations for the flood season are
combined with extreme flood hydrographs, and probabilities are estimated from stochastic storm
transposition (Section 5) and the TREX runoff model (Section 6). The goal is to estimate
reservoir elevations and probabilities, and develop event trees for later risk analysis studies.
7.1 Initial Reservoir Levels
An initial reservoir level is needed in order to perform level-pool routing of extreme
flood hydrographs through the reservoir, and determine if the dam can successfully withstand the
inflow hydrograph without overtopping. The typical Reclamation practice for routing PMF
hydrographs is to use a maximum initial reservoir level as the starting condition; it is usually
set equal to the top of active conservation storage. We examine actual reservoir performance
since Pueblo Dam was built (1975 2005) to consider alternative initial reservoir levels.
Daily and monthly reservoir elevation data for Pueblo Dam were obtained from the
Bureau of Reclamation Hydromet system and the U.S. Geological Survey. Daily reservoir
storage and elevation data are available from the Hydromet system from October 1, 1981 to
present. Data prior to October 1981 were obtained from USGS water data reports. The flood
season at Pueblo Dam is considered to be from April 15 through October 31 to operate the flood
control pool (USACE, 1977). As expected, this flood season corresponds well with the extreme
storm and flood season (Section 2). Within the Arkansas River upstream of Pueblo Dam, the
largest floods have typically occurred in late May through August.
Seasonal frequency analysis and daily reservoir elevation percentiles were estimated from
available data. Annual probabilities of maximum reservoir levels were estimated for the April
through October flood control season (Figure 7.1), using Cunnane's plotting position (Stedinger
et al., 1993). The frequency distribution shows that the top of active conservation (elevation
4,880.5) is reached about every 4.5 years. The upper tail of the distribution is very flat, and
reflects the highest reservoir level (4,882.16) over the last 25 years. These data and probability
estimates suggest that the reservoir is operated to maximize storage, and it is quite likely during a
major flood that the reservoir would be at or near elevation 4880.5. Daily reservoir levels for
each month (Figure 7.2) suggest that 50 percent of the time the reservoir ranges from about
4,868 to 4,862 feet during April through July, and is at 4,880 feet about 10 percent of the time.
137
99 90 80 70 60 50 40 30 20 10 5 2 1
Annual Exceedance Probability (%)
4800
4810
4820
4830
4840
4850
4860
4870
4880
4890
R
e
s
e
r
v
o
i
r

F
o
r
e
b
a
y

E
l
e
v
a
t
i
o
n

(
f
t
.
)
Figure 7.1. Annual exceedance probability of maximum reservoir levels at Pueblo Dam for the flood season (April
15-October 31), based on data from 1975 through 2005.
Apr-Oct Apr May June July Aug Sept Oct
Season or Month
4810
4820
4830
4840
4850
4860
4870
4880
4890
4900
4910
4920
D
a
i
l
y

R
e
s
e
r
v
o
i
r

F
o
r
e
b
a
y

E
l
e
v
a
t
i
o
n

(
f
t
.
)
Figure 7.2. Daily reservoir elevation percentiles for the flood season (April 15-October 31) based on data from
1981-2005.
138
Based on annual maximum reservoir elevation probabilities, the initial reservoir level
selected for routing extreme flood hydrographs is the top of active conservation (4,880.5 ft).
Depending on results of the routings, an initial level equal to 4,865 feet (approximately a May
through July median level) may also be used to investigate the sensitivity of the initial level.
There is approximately 61,450 acre-ft. of storage between reservoir elevations 4,865 and 4,880.
7.2 Hydrograph Routings and Elevation Probabilities
Hydrographs for level-pool reservoir routing were selected for two cases (Sections 5 and
6): a best estimate that is based on storms restricted to east of Parkdale and 5,000 mi
2
area; and
an upper estimate with no restrictions on storm locations and areas. The best estimate
considered basin-wide initial soil saturation S
e
= 0.5, a 36-hour storm, and optimally-centered
storm location for basin-average rainfall depths. A 15-day snowmelt volume (Bullard and
Leverson, 1991) was considered as a base flow addition to the rainfall-runoff hydrographs.
Fifteen hydrographs, spanning the range of basin-average rainfall depth exceedance probabilities,
are summarized in Table 7.1 and Figure 7.3. The 100-year, 15-day snowmelt hydrograph was
added to these TREX hydrographs. These combined hydrographs were routed using the
Reclamation FLROUT program, with an initial reservoir level equal to 4,880.5 ft. Results of the
routings are shown in Table 7.2 and Figure 7.4. Annual exceedance probabilities for these
reservoir levels are conditional probabilities; they are obtained directly from basin-average
rainfall frequency (Section 5). The results indicate that the maximum reservoir levels are less
than the dam crest, and are comparable to the design maximum water surface (El. 4,919).
0 6 12 18 24 30 36 42 48 54 60 66 72
Time (hours)
0
50000
100000
150000
200000
250000
300000
D
i
s
c
h
a
r
g
e

(
f
t
3
/
s
)
Figure 7.3. Fifteen extreme flood hydrographs from the TREX model, based on the best estimate rainfall
frequency curve.
139
Table 7.1: Summary of Best Estimate Extreme Flood Hydrographs from the TREX Model
Basin-Avg Rainfall
Depth (in)
Exceedance
Probability
Return Period
(years)
Peak Discharge
(ft
3
/s)
Runoff Volume
(acre-ft)
4.00 0.006628714 151 28,400 57,100
5.10 0.003267515 306 61,300 119,700
5.60 0.002194952 456 82,100 162,000
6.10 0.001336010 748 105,000 209,500
7.10 0.000524851 1,910 157,000 317,900
7.60 0.000321104 3,110 187,000 377,400
7.80 0.000234466 4,270 201,000 402,200
8.00 0.000176619 5,660 214,000 426,500
8.20 0.000126012 7,940 227,000 451,900
8.30 0.000089489 11,200 235,000 464,700
8.40 0.000064634 15,500 241,000 477,600
8.50 0.000037919 26,400 248,000 490,400
8.60 0.000028730 34,800 255,000 503,300
8.70 0.000016441 60,800 262,000 516,700
8.80 0.000010356 96,600 269,000 529,800
Table 7.2: Reservoir Elevation Frequency Summary
Exceedance
Probability
Return
Period
Maximum Reservoir Elev. (ft)*
Best Estimate Upper Estimate
0.005000 200 4,901.22 4,910.39
0.002000 500 4,905.12 4,913.90
0.001000 1,000 4,908.89 4,916.97
0.000500 2,000 4,912.44 4,918.49
0.000200 5,000 4,915.81 4,919.50
0.000100 10,000 4,917.33 4,920.38
0.000050 20,000 4,918.10 4,921.08
0.000020 50,000 4,919.01 4,922.02
0.000010 100,000 4,919.63 4,922.70
0.000005 200,000 4,920.27 4,923.35
0.000002 500,000 4,921.09 4,924.19
0.000001 1,000,000 4,921.69 4,924.80
* Results shown in italics are extrapolated beyond available data
140
Figure 7.4. Reservoir elevation frequency curves based on the best and upper estimate rainfall curves.
In addition to the best estimate, hydrographs for an upper estimate were also routed;
results are shown in Table 7.2 and Figure 7.4. This upper estimate attempts to incorporate the
input uncertainty of the rainfall data. The rainfall input is from the SST rainfall frequency curve
that includes no restrictions on storm center location or storm size (Figure 5.7). Based on the
streamflow and radar analyses (Section 2, Appendix A), as well as paleoflood data (Section 3), it
is extremely unlikely that storms could center anywhere in the watershed as well as cover the
entire basin. The routing results demonstrate that the maximum reservoir level is at or slightly
higher than the design maximum reservoir level for this rainfall case. Maximum reservoir levels
predicted by the TREX model using stochastic storm transposition did not exceed the top of the
dam (El. 4,925), and did not reach the maximum level (El. 4,931.27) used in risk analysis event
trees (Steers and Trojanowski, 2000). Results shown in Figure 7.4 are subsequently extrapolated
to estimate higher reservoir levels and probabilities.
7.3 Hydrologic Risk Inputs
There have been several risk analysis meetings and investigations for Pueblo Dam (Steers
and Trojanowski, 2000). Four hydrologic failure modes have been identified: overtopping
erosion of the embankment, sliding in the spillway buttress foundation, sliding in the
141
1 0.5 0.2 0.1 0.05 0.02 0.01 0.001 0.0001
a
(elev)|p
a
(d)]
4895
4900
4905
4910
4915
4920
4925
4930
R
e
s
e
r
v
o
i
r

E
l
e
v
a
t
i
o
n

(
f
t
)
Best Estimate
Upper Estimate
embankment foundation, and concrete dam stress. The event trees for these failure modes, as
well as static failure modes, included eight hydrologic elevation classes and one static case. In
order to be useful in a risk analysis, we estimate reservoir elevation probabilities for specific
elevations that correspond to the event tree classes. These estimates are made via linear
interpolation of reservoir levels and standard Normal probabilities within the range of modeled
extreme floods (Table 7.3). Because reservoir elevations are also needed beyond model
estimates, we extrapolate the reservoir elevation frequency curves assuming a Normal
distribution (Figure 7.5).
1 0.5 0.2 0.1 0.05 0.02 0.01 0.001 0.0001 0.00001 0.000001 0.0000001 0.00000001
a
(elev)|p
a
(d)]
4895
4900
4905
4910
4915
4920
4925
4930
4935
R
e
s
e
r
v
o
i
r

E
l
e
v
a
t
i
o
n

(
f
t
)
Extrapolation Best Estimate
Best Estimate
Extrapolation Upper Estimate
Upper Estimate
Figure 7.5. Extrapolated TREX reservoir elevation frequency curves assuming a Normal distribution.
The reservoir elevation frequency curves were extrapolated from El. 4,920 to El. 4,931.27
(Figure 7.5). The results show that the probability of exceeding elevations greater than 4,924 is
less than 1 in 18,000,000 years for the best estimate and less than 1 in 400,000 years for the
upper estimate. Probability estimates for risk analysis event trees are listed in Table 7.4. It is
recommended that the upper estimate be assigned a weight equal to 0.05 for use in a tree as an
approximate upper confidence limit. The minimum predicted reservoir elevation based on the
TREX model for the upper estimate was 4,905 feet (Figure 7.4); thus the probability estimates
start at 4,906 feet. These results were based on using a fixed initial reservoir level equal to the
142
top of active conservation. If lower initial levels were chosen, the reservoir elevation
probabilities would decrease. Probabilities for other reservoir elevations than those listed in
Tables 7.3 and 7.4 can be provided on an as-needed basis to the risk analysis team.
Table 7.3: Exceedance Probability Estimates for Specified Reservoir Elevations*
Reservoir
Elevation
Best Estimate
Exceedance Probability
Best Estimate
Return Period
Upper Estimate
Exceedance Probability
Upper Estimate
Return Period
El. 4931.27 0.0000000000010100 990,129,000,000 0.0000000002018287 4,955,000,000
El. 4926 0.0000000037426667 267,000,000 0.0000002441915478 4,100,000
El. 4925 0.0000000149681636 67,000,000 0.0000007973050634 1,250,000
El. 4924 0.0000000566586061 18,000,000 0.0000024718503103 400,000
El. 4912.8 0.0004625480138271 2,200 0.0027949049564111 360
El. 4906 0.0017071982324810 590 0.0104167358732433 96
El. 4902.34 0.0031805430709404 310 --- ---
El. 4898.7 0.0074247812663645 130 --- ---
El. 4893.8 0.0099676826857708 100 --- ---
* Results shown in italics are extrapolated beyond available data
Table 7.4: Reservoir Elevation Probability Estimates for Risk Analysis Event Tree
Failure Mode
Best Estimate Exceedance
Probability
Upper Estimate Exceedance
Probability
H1 El. 4926 to El. 4931.27 0.0000000037416568 0.0000002439897191
H2 El. 4925 to El. 4926 0.0000000112254969 0.0000005531135157
H3 El. 4924 to El. 4925 0.0000000416904425 0.0000016745452469
H4 El. 4912.8 to El. 4924 0.0004624913552209 0.0027924331061008
H5 El. 4906 to El. 4912.8 0.0012446502186539 0.0076218309168322
H6 El. 4902.34 to El. 4906* 0.0014733448384594 0.9895832641267570
H7 El. 4898.7 to El. 4902.34 0.0042442381954241 ---
H8 El. 4893.8 to El. 4898.7 0.0025429014194063 ---
Static Below El. 4893.8 0.9900323173142290 ---
* Below El. 4906 for Upper Estimate
7.4 Summary
This section presented a frequency analysis of reservoir elevation data for Pueblo Dam,
reservoir routings of extreme flood hydrographs, reservoir elevation probability estimates, and
inputs for subsequent hydrologic risk analysis studies. From a 25-year reservoir record, the
maximum elevation observed during the April 15 October 31 flood season was 4,882.16 feet.
143
The top of active conservation (4,880.5 feet) was reached on average every 4.5 years. It is quite
likely that the reservoir would be full during a major flood. Based on flood season daily
elevations, the reservoir ranges from about 4,868 to 4,862 feet during April through July, and is
at 4,880 feet about 10 percent of the time. Based on routing of extreme flood hydrographs, a
best estimate return period for the design maximum reservoir elevation (4,919 feet) is 50,000
years, with an upper estimate equal to 5,000 years. The maximum reservoir elevations were in
the range of 4,919 to 4,920 feet based on routing the hydrographs, and return periods were about
100,000 years (best estimate) to 20,000 years (upper estimate). Reservoir elevations and
probabilities were estimated based on extrapolation with a Normal distribution for eight event
tree classes that were defined in prior risk analysis studies.
144
8.0 Conclusions
This section presents major conclusions of the Pueblo Flood Hazard Study. The purpose
of this study was to assess the flood hazard for Pueblo Dam near Pueblo, Colorado. The three
main objectives of the flood hazard study were:
1. significantly extend the time base and spatial distribution of the existing peak-flow
observations by collecting historical and paleoflood data in the Arkansas River basin;
2. estimate constraints on (a) storm intensity, duration and spatial distribution and (b) flood
durations and volumes for the Arkansas River basin upstream from Pueblo Dam; and
3. develop probabilistic flood hydrographs (peak, volume, and duration) for exceedance
probabilities from 1 in 100 to greater than 1 in 10,000 for a risk analysis.
These objectives were designed to address Safety of Dams (SOD) recommendation 2000-SOD-E
for Pueblo Dam: prepare a final hydrology report, suitable for use in a risk analysis study.
In order to meet the objectives, nine main tasks were completed: (1) radar-based
hydrometeorology study (Princeton University); (2) loss of life study; (3) limited paleoflood
study; (4) stochastic storm generation; (5) rainfall-runoff modeling; (6) flood frequency analysis;
(7) flood routings; (8) a report; and (9) risk analysis support. Main conclusions from each part of
this work are summarized below. A longer summary of each topic is included at the end of each
major section in the report. The loss of life study focused on estimating fatalities from flooding
and dam failures. Loss of life estimates ranged from 131 to 376 for a flood-induced failure and
324 to 435 for a static-induced failure. The radar study focused on rainfall relations with
elevation and storm movement. Additional details of the loss of life and radar hydrometeorology
studies are given in Appendices A and B, respectively.
Investigations into the flood hydrology and hydrometeorology within the Arkansas River
basin and surrounding region were based on streamflow and radar data. The largest peak flows
in the region resulted from three extreme storms: June 1921, June 1965 and July 1976. There
appears to be a very clear break in scaling relationships between peak flows and drainage areas at
scales larger than about 1,000 mi
2
. It is inferred that this change in relationship is due to partial-
area storms in orographic areas. Extreme storms and radar data suggest the partial-area storm
concept is valid. Peak flow, mean daily flow and hydrograph analyses clearly demonstrate
mixed-population rainfall-runoff and snowmelt runoff areas within the Arkansas River
watershed. Peak flow magnitudes within the basin are dramatically reduced with elevation,
supporting Jarrett's (1987, 1993) elevation limit hypothesis. Upstream of Parkdale, snowmelt
runoff is the dominant streamflow mechanism and the cause of the largest peaks. Downstream
of Parkdale, peaks and maximum mean daily flows are caused by large rainfall-runoff events,
and sometimes snowmelt runoff. Peak-flow and mean daily flow correlations support the
145
transition between snowmelt runoff upstream of Canon City and rainfall-caused large floods in
the lower portion of the watershed. A database of storm rainfall was developed for modeling
extreme storms; it included DAD data and storm properties to implement an elliptical storm
model for extreme rainfall and runoff modeling.
Warm-season (June through August) radar data from the period 1995 through 2003
suggested the following. The geographical distribution of storm total rainfall accumulations
shows an increasing gradient of heavy rainfall with decreasing terrain elevation. An area of
maximum rain is observed on the plains east of Pikes Peak. As shown by the geographical
distribution of storm speed and direction, storm propagation is from west to east. Because of the
storms' dynamics and evolution, heavy rainfall is distributed mainly on the lower basin. The
upper basin (above Wellsville) rarely experiences extreme rainfall as evidenced by the
geographical distribution of storm total rainfall and storm activity. Heavy rains and extreme
storm occurrence are concentrated on the plains east of the mountains.
The paleoflood study of the Upper Arkansas River Basin utilized detailed information
from four sites to characterize paleofloods and nonexceedance bounds for Pueblo Dam at Adobe
Park, the Loma Linda recreation area, Parkdale, and Pueblo State Park. Nine soil/stratigraphic
descriptions and seven radiocarbon dates of key deposits were used in conjunction with
geomorphic mapping and HEC-RAS flow modeling to determine age estimates of each soil and
peak discharges required to inundate the surfaces. With the exception of Adobe Park,
nonexceedance bounds for similar age surfaces appear to increase in the downstream direction
and change markedly between Loma Linda and Pueblo State Park from approximately 14,000
ft
3
/s to 150,000 ft
3
/s for Holocene alluvium between the ages of 700 and 2,000 years. The
paleoflood data confirm that the upper basin is characterized by snowmelt runoff, and the lower
basin (near Parkdale and downstream) is dominated by rainfall-runoff. At Pueblo State Park, the
paleohydrologic bound suggests that floods slightly larger in magnitude than the June 1921 flood
have not been experienced in the past 800 years.
Flood frequency analysis was conducted using peak-flow (gage), historical and
paleoflood data at four locations along the main stem of the Arkansas River upstream from
Pueblo. The Expected Moments Algorithm was used with the LP-III distribution to estimate
flood frequency curves at Pueblo, Parkdale, Loma Linda (Wellsville) and Adobe Park (Salida).
Confidence intervals for each frequency curve were also estimated. Paleohydrologic bounds
spanned 550 to 10,000 years within the watershed and provided substantially longer record
lengths for frequency analysis. Peak flows in the lower watershed at Pueblo and Parkdale
reflected extreme floods from rainfall, and were relatively much larger than at the two upstream
locations. The largest floods at upstream sites were from snowmelt; these locations had
relatively flat frequency curves. Using a regional index-flood approach, it was shown that there
is a different population of floods between upstream snowmelt and downstream rainfall-runoff
sites. The frequency curves were later used to constrain flood frequency curves estimated with
the TREX rainfall-runoff model.
146
A stochastic storm transposition (SST) model was implemented to estimate basin-average
rainfall depths and probabilities based on depth-area-duration (DAD) data. The spatial
distributions of storms were described as an elliptical pattern with major to minor axis ratio,
orientation and storm center. Temporal distributions were estimated using mass curves. Two
new features were developed and implemented to extend the SST method: a numerical procedure
to determine the effective area for an arbitrary-shaped watershed rather than simple geometric
shapes (circles); and a restriction on storm center locations to account for basin orography.
Fifteen storms were used in the simulation to estimate extreme basin average rainfall depth
probabilities, based on a criterion that the storm had a maximum observed 10 mi
2
depth
exceeding 11 inches in the region. Basin-average rainfall depths for a 36-hour storm ranged
from 4.0 to 10.9 inches.
The TREX model was applied to the Arkansas River basin upstream of Pueblo, Colorado.
The model was successfully calibrated to the June 1921 flood, the largest flood on record in the
basin. The model was validated with the May 1894 flood, the third largest flood on record. The
calibrated model was used to estimate flood frequency curves. Basin-average rainfall depths and
probabilities were obtained from the SST model. TREX model frequency curves at four sites
within the watershed were compared with data-based peak-flow frequency curves with
paleoflood data. The comparison showed that peak flows from the rainfall-runoff model were
generally larger than the data-based flood frequency curves. The shapes of the rainfall frequency
curves dictated the shapes of the flood frequency curves. Rainfall frequency and flood frequency
curve extreme upper tails were relatively flat compared to LP-III curves. A better match between
TREX model frequency curves and data-based frequency curves was obtained at some locations
by restricting storm area and storm center locations.
TREX model hydrographs were used in reservoir routing to determine maximum
elevations for extreme floods. From a 25-year reservoir record, the top of active conservation
(4,880.5 feet) was reached on average every 4.5 years. It is quite likely that the reservoir would
be full during a major flood. Based on flood season daily elevations, the reservoir ranges from
about 4,868 to 4,862 feet during April through July, and is at 4,880 feet about 10 percent of the
time. Based on routing of extreme flood hydrographs, a best estimate return period for the
design maximum reservoir elevation (4,919 feet) is 50,000 years, with an upper estimate equal
to 5,000 years. The maximum reservoir elevations were in the range of 4,919 and 4,920 feet
based on routing the hydrographs, and return periods were about 100,000 years (best estimate)
to 20,000 years (upper estimate). These results suggested that the overtopping probability is
remote.
Based on the the data and results presented in this report, it is recommended that the risk
analysis results for Pueblo Dam be revisited by a Reclamation Senior Engineer or risk analysis
team. Reservoir elevation frequency curves presented in this report are comparable to those
previously published.
147
148
9.0 References
Alexander, G.N. (1963) Using the probability of storm transposition for estimating the frequency
of rare floods. J. Hydrol., 1, p. 46-57.
Anderson, J.R., Hardy, E.E., Roach, J.T. and Witmer, R.E (1976) A land use and land cover
classification system for use with remote sensor data. U.S. Geological Survey Professional
Paper 964, 28 p.
Andrews, E.D. (1984) Bed-material entrainment and hydraulic geometry of gravel-bed rivers in
Colorado: Geological Society of America Bulletin, v. 95, p. 371-378.
Baker, J.H. and Hafen, L.R. (1927) History of Colorado, Volume II. Linderman Co., Denver, pp.
429-867.
Baker, V.R. (1987) Paleoflood hydrology and extraordinary flood events. J. Hydrology, 96, nos.
1-4, pp. 79-99.
Baker, V.R. (1989) Magnitude and frequency of palaeofloods, in Beven, K., and Carling, P.,
(eds.), Floods: Hydrological, sedimentological, and geomorphological implications: John
Wiley and Sons, New York, p. 171-183.
Baker, V.R. and Costa, J.E. (1987) Flood power, in Mayer, L., and Nash, D., eds., Catastrophic
Flooding: Allen & Unwin, Boston, MA, p. 1-21.
Baker, V.R., Kochel, R.C., and Patton, P.C., eds. (1988) Flood geomorphology: John Wiley and
Sons, New York, 503 p.
Barnes, H.H. (1967) Roughness characteristics of natural channels. U.S. Geological Survey
Water-Supply Paper 1849, 213 p.
Behre, C.H., Jr. (1933) Physiographic history of the Upper Arkansas and Eagle rivers, Colorado:
Journal of Geology, v. 41, no. 8, p. 785-814.
Birkeland, P.W. (1999) Soils and geomorphology: New York, Oxford University Press, 412 p.
Blainey, J.B., Webb, R.H., Moss, M.E., and Baker, V.R. (2002) Bias and information content of
paleoflood data in flood-frequency analysis: in House, P.K., Webb, R.H., Baker, V.R., and
Levish, D.R., (eds), Ancient Floods, Modern Hazards: Principles and Applications of
Paleoflood Hydrology, Water Sciences and Application Volume 5, American Geophysical
Union, Washington D.C., p. 161-174.
Boggs, S., Jr. (1995) Principles of Sedimentology and Stratigraphy: Prentice Hall, Englewood
Cliffs, New Jersey, 774 p.
Boner, F.C., and Stermitz, F. (1967) Floods of June 1964 in Northwestern Montana: U.S.
Geological Survey Water-Supply Paper 1840-B, B1-B114.
Boyer, M.C. (1957) A correlation of the characteristics of great storms. Trans. Amer. Geophys.
149
Union, 38(2), pp. 233-238.
Bradley, W.C. (1987) Erosion surfaces of the Colorado Front Range: A review in Madole, R.F.,
and others, 1987, Rocky Mountains, Ch. 7, in Graf, W.L., ed., Geomorphic Systems of North
America: Geological Society of America, Centennial Special Volume 2, Boulder, Colorado p.
215-220.
Bullard, K.L. (1998a) Pueblo Dam, Colorado - Peak Flow Frequency Analysis by Index Flood
with L-Moments - An Approach Proposed by Hosking and Wallis 1997, Bureau of
Reclamation, Flood Hydrology Group, January, 5 p., 1 table, and 3 appendices.
Bullard, K.L. (1998b) Pueblo Dam, Colorado - Regional Analysis of Duration Frequency Curves
with Hydrographs Based on Regional Volume-Frequency Curves, Regional Analysis with
Index Flood and L-Moments - An Approach Proposed by Hosking and Wallis - 1997, Bureau
of Reclamation, Flood Hydrology Group April, 10 p., 8 tables, and 11 figs.
Bullard, K.L. and Leverson, V. (1991) Pueblo Dam, Fryingpan-Arkansas Project Probable
Maximum Flood (PMF) Study. Flood Section, Bureau of Reclamation, Denver, CO, dated
June, 1991, 23 p. and enclosures.
Bureau of Reclamation (Reclamation) (1968a) Supplement to Final Inflow Design Flood Study
of January 16, 1967 by Region 7, Pueblo Damsite, Fryingpan - Arkansas Project, Memorandum
from Donald L. Miller to Head, Flood Hydrology Section, dated January 24, 1968.
Bureau of Reclamation (Reclamation) (1968b) Volume Inflow Design Flood, Pueblo Damsite,
Fryingpan - Arkansas Project, Memorandum from Donald L. Miller to Head, Flood Hydrology
Section, dated March 18, 1968.
Bureau of Reclamation (Reclamation) (1997) Pueblo Dam - Return Period Hydrographs, email
message from Ken Bullard, Flood Hydrology Group, D-8530 to John Trojanowski, Waterways
and Concrete Dams Group, D-8130, dated February 4, 1997, 1 p., 1 fig., 1 table.
Bureau of Reclamation (Reclamation) (2002) Interim Guidance for Addressing the Risk of
Extreme Hydrologic Events. Dam Safety Office and Technical Service Center, Bureau of
Reclamation, Denver, CO, dated August 19, 2002, 3 p.
Campbell, M.R. (1922) Guidebook of the Western United States, Part E, The Denver and Rio
Grande Western Route: U.S. Geological Survey Bulletin 707, 266 p.
Carrigan, P.H. (1971) A flood-frequency relation based on regional record maxima. U.S.
Geological Survey Professional Paper 434-F, 22 p.
Chow, V.T., Maidment, D.R. and Mays, L.W. (1988) Applied Hydrology: McGraw-Hill, New
York, 572 p.
Colorado State Engineer (1885) Report of the State Engineer to the Governor of Colorado for the
years 1883 and 1884, Denver, Colorado.
Cohn, T.A., Lane, W.L, and Baier, W.G. (1997) An algorithm for computing moments-based
flood quantile estimates when historical information is available. Water Resour. Res. 33(9), pp.
2089-2096.
Cohn, T.A., Lane, W.L., and Stedinger, J.R. (2001) Confidence intervals for EMA flood quantile
150
estimates. Water Resour. Res. 37(6), pp. 1695-1706.
Costa, J.E. (1978) Holocene stratigraphy in flood frequency analysis: Water Resources Research,
v. 14, p. 626-632.
Costa, J.E. (1986) A history of paleoflood hydrology in the United States: Eos, Transactions of
the American Geophysical Union, v. 67, no. 17, p. 425, 428-430.
Creagher, W.P., Justin, J.D. and Hinds, J. (1945) Engineering for Dams. Volume I. General
Design. John Wiley and Sons, New York, pp. 99-140.
Crippen, J.R. (1982) Envelope curves for extreme flood events. J. Hydraul. Engr., ASCE, 108
(HY10), pp. 1208-1212.
Crippen, J.R. and Bue, C.D. (1977) Maximum floodflows in the Conterminous United States.
U.S. Geological Survey Water-Supply Paper 1887, 52 p.
Crowfoot, R.M., Payne, W.F. and O'Neill, G.B. (2004) Water Resources Data, Colorado, Water
Year 2003, Volume 1. Missouri River Basin, Arkansas River Basin, and Rio Grande Basin.
U.S. Geological Survey Water-Data Report CO 03 1, 577 p.
Cudworth, A.G. Jr. (1989) Flood Hydrology Manual. A Water Resources Technical Publication,
U.S. Department of Interior, Bureau of Reclamation, Denver, Colorado, 243 p.
Dalrymple, T. (1964) Hydrology of Flood Control, Part I. Flood Characteristics and Flow
Determination. In Handbook of Hydrology, Chow, V.T. (ed.), McGraw-Hill, New York,
Section 25-I, pp. 25-1 25-33.
Denlinger, R.P., OConnell, D.R.H., and House, P.K. (2002) Robust determination of stage and
discharge: An example from an extreme flood on the Verde River, Arizona: in House, P.K.,
Webb, R.H., Baker, V.R., and Levish, D.R., (eds), Ancient Floods, Modern Hazards: Principles
and Applications of Paleoflood Hydrology, Water Sciences and Application Volume 5,
American Geophysical Union, Washington D.C., p. 127-146.
Dixon, M. and Weiner, G. (1993) TITAN: Thunderstorm Identification tracking, analysis and
nowcasting - A radar-based methodology. J. Atmos. Ocean Technol., 10(6), pp. 785-797.
Doesken, N.J. (1991) General Climatology in Colorado, in National Water Summary 1988-89-
Hydrologic Events and Floods and Droughts, R.W. Paulson, E.B. Chase, R.S. Roberts, and
D.W. Moody (compilers), U.S. Geological Survey Water-Supply Paper 2375, pp. 207-208.
Doesken, N.J. (1998) A post-evaluation of rainfall reports associated with the Pawnee Creek
flood of July 29-30, 1997 in eastern Weld and western Logan counties in northeast Colorado,
Climatology Report #98-3, Department of Atmospheric Science, Colorado State University,
Fort Collins, Colorado, 33 p.
Doesken, N.J. and McKee, T.B. (1998) An Analysis of Rainfall for the July 28, 1997 Flood in
Fort Collins, Colorado, Climatology Report #98-1, Department of Atmospheric Science,
Colorado State University, Fort Collins, Colorado, 55 p.
Durrans, S.R., Julian, L.T., and Yekta, M. (2002) Estimation of depth-area relationships using
radar-rainfall data. J. Hydrol. Eng., ASCE, 7(5), pp. 356-367.
Ellingson, J. (1999) Hydrologic Methods Investigations and Flood Routing Results, Pueblo Dam,
151
Fryingpan-Arkansas Project, Colorado, Technical Memorandum No. PUE-8130-MOD-TM-99-
2, Bureau of Reclamation, Denver, CO, 17 p. and 5 appendices.
Elliott, J.G., Jarrett, R.D., and Ebling, J.L. (1982) Annual snowmelt and rainfall peak-flow data
on selected foothills region streams, South Platte River, Arkansas River, and Colorado River
Basins, Colorado. U.S. Geological Survey Open-File Report 82-426, 86 p.
England, J.F. Jr. (1996) Compilation of an Extreme Streamflow Data Base as Part of an Extreme
Precipitation Data Study. Prepared for the Colorado Climate Center, Colorado State University,
dated May 1, 1996, 32 p., 1 figure, 28 tables, and 2 appendices.
England, J.F. Jr. (1998) Assessment of Historical and Paleohydrologic Information in Flood
Frequency Analysis. Unpublished M.S. Thesis, Department of Civil Engineering, Colorado
State Univ., Fort Collins, Colorado, 292 p.
England, J.F. Jr. (1999) Draft Users manual for program EMA, At-Site Flood Frequency
Analysis with Historical/Paleohydrologic Data. Flood Hydrology Group, Bureau of
Reclamation, Denver, CO, 52 p.
England, J.F. Jr. (2004) Review of Selected Large Flood Estimates in the United States for the
U.S. Geological Survey. Bureau of Reclamation, Denver, CO, August, 14 p. and appendices.
England, J.F. Jr. (2005) Frequency Analysis and Two-Dimensional Simulations of Extreme
Floods on a Large Watershed. Ph.D. Dissertation, Department of Civil Engineering, Colorado
State University, Fort Collins, CO, 237 p.
England, J.F. Jr., Salas, J.D., and Jarrett, R.D. (2003a) Comparisons of two moments-based
estimators that utilize historical and paleoflood data for the log-Pearson Type III distribution,
Water Resour. Res. 39(9), pp. 5-1 5-16, doi:10.1029/2002WR001791.
England, J.F. Jr., Jarrett, R.D., and Salas, J.D. (2003b) Data-based comparisons of moments
estimators that use historical and paleoflood data. J. Hydrol., 278(1-4), pp. 170-194.
Engman, E.T. (1986) Roughness coefficients for routing surface runoff. J. Irrig. Drain. Engr.,
ASCE, 112(1), pp. 39-53.
Fellows, A.L. (1902) Water Resources of the State of Colorado. U.S. Geological Survey Water-
Supply and Irrigation Paper No. 74, 151 p.
Follansbee, R. and Hodges, P.V. (1925) Some floods in the Rocky Mountain region. U.S.
Geological Survey Water-Supply Paper 520-G, pp. 105-125.
Follansbee, R. and Jones, E.E. (1922) Arkansas River flood of June 3-5, 1921. U.S. Geological
Survey Water-Supply Paper 487, 44 p.
Follansbee, R. and Sawyer, L.R. (1948) Floods in Colorado. U.S. Geological Survey Water-
Supply Paper 997, 151 p.
Fontaine, T.A. (1995) Rainfall-runoff model accuracy for an extreme flood. J. Hydraul. Engr.,
ASCE, 121(4), pp. 365-374.
Fontaine, T.A. and Potter, K.W. (1989) Estimating probabilities of extreme rainfalls. J. Hydraul.
Engr., ASCE, 115(11), pp. 1562-1575.
Foster, E.E. (1948) Rainfall and Runoff. Macmillan, New York, 487 p.
152
Foufoula-Georgiou, E. (1989) A probabilistic storm transposition approach for estimating
exceedance probabilities of extreme precipitation depths. Water Resour. Res., 25(5), pp. 799-
815.
Foufoula-Georgiou, E. and Wilson, L.L. (1990) In search of regularities in extreme rainstorms. J.
Geophys. Res., 95(D3), pp. 2061-2072.
Franchini, M., Helmlinger, K.R., Foufoula-Georgiou, E., and Todini, E. (1996) Stochastic storm
transposition coupled with rainfall-runoff modeling for estimation of exceedance probabilities
of design floods. J. Hydrol., (175)1-4, pp. 511-532.
Frazier, A.H and Heckler, W. (1972) Embudo, New Mexico, Birthplace of Systematic Stream
Gaging. U.S. Geological Survey Professional Paper 778, 23 p.
Fuller, W.E. (1914) Flood flows. Trans. Am. Soc. Civ. Eng., Paper No. 1293, vol. 77, pp. 564-
617, with discussion pp. 618-694.
Gerard, R. and Karpuk, E.W. (1979) Probability analysis of historical flood data. J. Hydraulics
Div. ASCE, 105(HY9), pp. 1153-1165.
Graham, W.J. (1999) A Procedure for Estimating Loss of Life Caused by Dam Failure. Dam
Safety Research Program Research Report DSO-99-06, U.S. Department of Interior, Bureau of
Graham, W.J. (2003) Pueblo Dam, Colorado, Loss of Life Assessment. Bureau of Reclamation,
Denver, CO, 21 p.
Gupta, V.K. (1972) Transposition of storms for estimating flood probability distributions.
Hydrology Paper 59, Colorado State Univ., Fort Collins, Colorado, 35 p.
Gupta, V.K. and Waymire, E. (1993) A statistical analysis of mesoscale rainfall as a random
cascade. J. Appl. Meteorol., 32, pp. 251267.
Hafen, L.R. (1948) Colorado and Its People. A Narrative and Topical History of the Centennial
State, Volume I. Lewis Historical Publishing, New York, 644 p.
Hansen, E.M., Schreiner, L.C., and Miller, J.F. (1982) Application of Probable Maximum
Precipitation Estimates-United States East of the 105th Meridian. Hydrometeorological Report
No. 52, National Weather Service, National Oceanic and Atmospheric Administration, U.S.
Department of Commerce, Washington, DC, 168 p.
Hansen, E.M., Fenn, D.D., Schreiner, L.C., Stodt, R.W., and Miller, J.F. (1988) Probable
Maximum Precipitation Estimates-United States between the Continental Divide and the 103rd
Meridian. Hydrometeorological Report No. 55A, National Weather Service, National Oceanic
and Atmospheric Administration, U.S. Department of Commerce, Silver Spring, MD, 242 p.
Harris, D., Menabde, M., Seed, A. and Austin, G. (1996) Multifractal characterization of rain
fields with a strong orographic influence. J. Geophys. Res. (Atmos.), 101(D21, pp. 26,405-
26,414.
Helley, E.J., and LaMarche, V.C. (1973) Historic flood information for northern California
streams from geological and botanical evidence: U.S. Geological Survey Professional Paper
485-E, 16 p.
153
Hirschboeck, K.K. (1991) Climate and Floods, in National Water Summary 1988-89-Hydrologic
Events and Floods and Droughts, R.W. Paulson, E.B. Chase, R.S. Roberts, and D.W. Moody
(compilers), U.S. Geological Survey Water-Supply Paper 2375, pp. 67-88.
Horton, R.E. (1924) Discussion of The distribution of intense rainfall and some other factors in
the design of storm-water drains flows. Proc. Am. Soc. Civ. Eng., 50, pp. 564-617, with
discussion pp. 660-667.
Hosking, J.R.M. and Wallis, J.R. (1997) Regional Frequency Analysis - An Approach based on
L-Moments. Cambridge University Press, 224 p.
House, P.K., Webb, R.H., Baker, V.R., Levish, D.R. (2002) Ancient floods, modern hazards:
principles and applications of paleoflood hydrology. Am. Geophys. Union Water Sci. Appl.
Ser. 5, 385 p.
Hoyt, W.G. and Langbein, W.B. (1955) Floods. Princeton University Press, Princeton, NJ, 469
p.
Hsieh, Y.P. (1992) Pool size and mean age of stable soils organic carbon in cropland: Soil
Science Society of America Journal, v. 56, p. 460-464.
Hsieh, Y.P. (1993) Radiocarbon signatures of turnover rates in active soil organic carbon pools:
Soil Science Society of America Journal, v. 57, p. 1020-1022.
Huff, F.A. (1967) Time distribution of rainfall in heavy storms. Water Resour. Res. 3(4), pp.
1007-1019.
Hunt, C.B. (1967) Physiography of the United States: W.H. Freeman and Company, San
Francisco, 480 p.
Interagency Committee on Water Data (IACWD) (1982) Guidelines for determining flood flow
frequency: Bulletin 17-B. Hydrology Subcommittee, March 1982 (revised and corrected), 28 p.
and appendices.
Iseri, K.T. and Langbein, W.B (1974) Large Rivers of the United States. U.S. Geological Survey
Circular 686, 10 p.
Jahns, R.E. (1947) Geologic features of the Connecticut Valley, Massachusetts, as related to
recent floods: U.S. Geological Survey Water-Supply Paper 996, 158 p.
Jarrett, R.D. (1984) Hydraulics of high-gradient streams. J. Hydraul. Engrg., ASCE, HY11, 110,
pp. 1519-1539.
Jarrett, R.D. (1985) Determination of roughness coefficients for streams in Colorado. U.S.
Geological Survey Water-Resources Investigations Report 85-4004, 54 p.
Jarrett, R.D. (1987) Flood Hydrology of Foothill and Mountain Streams in Colorado. Ph.D.
Dissertation, Department of Civil Engineering, Colorado State Univ., Fort Collins, Colorado,
239 p.
Jarrett, R.D. (1990) Paleohydrologic techniques used to define the spatial occurrence of floods.
Geomorphology, 3, pp. 181-195.
Jarrett, R.D. (1991) Paleohydrology and its value in analyzing floods and droughts, in Paulson,
R.W., Chase, E.B., Roberts, R.S., and Moody, D.W., compilers, U.S. Geological Survey Water-
154
Supply Paper 2375, p. 105-116.
Jarrett, R.D. (1993) Flood elevation limits in the Rocky Mountains. In Kuo, C.Y. (ed.)
Engineering Hydrology-Proceedings of the symposium sponsored by the Hydraulics Division,
ASCE, July 25-30, 1993, San Francisco, CA, pp. 180-185.
Jarrett, R.D. and Costa, J.E. (1988) Evaluation of the flood hydrology in the Colorado Front
Range using precipitation, streamflow, and paleoflood data for the Big Thompson River Basin.
U.S. Geological Survey Water-Resources Investigations Report 87-4117, 37 p.
Jarrett, R.D. and Tomlinson, E.M. (2000) Regional interdisciplinary paleoflood approach to
assess extreme flood potential. Water Resour. Res. 36(10), pp. 2957-2984.
Jarvis, C.S. and others (1936) Floods in the United States, Magnitude and Frequency. U.S.
Geological Survey Water-Supply Paper 771, 497 p.
Javier, J.R., Smith, J.A., England, J.F., and Baeck, M.L. (2004) A Radar Climatology of Extreme
Rainfall in the Front Range of the Rocky Mountains. Eos Trans. AGU, 85(47), Fall Meet.
Suppl., Abstract H11F-0376.
Javier, J.R., Smith, J.A., Baeck, M.L., and Steiner, M. (2005) A Radar Climatology of Extreme
Rainfall in the Colorado Front Range of the Rocky Mountains. Draft report submitted to
Bureau of Reclamation. Department of Civil and Environmental Engineering, Princeton
University, Princeton, NJ, 48 p.
Johnson, D.L. (1990) Biomantle evolution and the redistribution of earth materials and artifacts:
Soil Science, v. 149, no. 2, p. 84-102.
Julien, P. Y. and Saghafian, B. (1991) CASC2D users manual - A two dimensional watershed
rainfall-runoff model. Civil Eng. Report CER90-91PYJ-BS-12, Colorado State University, Fort
Collins, Fort Collins, CO, 66 p.
Julien, P. Y., Saghafian, B., and Ogden, F. L. (1995) Raster-Based hydrologic modeling of
spatially-varied surface runoff. Water Resources Bulletin, AWRA, 31(3), pp. 523-536.
Kircher, J.E., Choquette, A.F. and Richter, B.D. (1985) Estimation of Natural Streamflow
Characteristics in Western Colorado. U.S. Geological Survey Water-Resources Investigations
Report 85-4086, 28 p.
Klinger, R.E. (2006) Evaluation of Extreme Floods in the Western United States: [draft technical
report], Bureau of Reclamation, Technical Service Center, Denver, CO, 41 p.
Kochel, R.C., and Baker, V.R. (1988) Paleoflood analysis using slackwater deposits, in Baker,
V.R., Kochel, R.C., and Patton, P.C., (eds.), Flood geomorphology: John Wiley and Sons, New
York, p. 357-376.
Koutsoyiannis, D. and Foufoula-Georgiou, E. (1993) A scaling model of a storm hyetograph.
Water Resour. Res., 29(7), pp. 2345-2361.
Lane, W.L., and Cohn, T.A. (1996) Expected moments algorithm for flood frequency analysis, in
N. Am. Water and Environ. Cong. 1996, edited by C.T. Bathala, Amer. Soc. Civ. Eng.,
Anaheim, California.
Laurenson, E.M. and Kuczera, G.A. (1998) Annual Exceedance Probability (AEP) of the
155
Probable Maximum Precipitation (PMP) Report on a review and recommendations for
practice. Prepared for NSW Department of Land and Water Conservation and the Snowy
Mountains Hydro-Electric Authority, 63 p.
Lee, W.T. (1923) Peneplains of the Front Range and Rocky Mountain National Park, Colorado:
U.S. Geological Survey Bulletin 730, p. 4-5, 15.
Levish, D.R. (1998) Preliminary paleoflood information for Pueblo Dam. Bureau of
Levish, D.R. (2002) Paleohydrologic Bounds: Nonexceedance information for flood hazard
assessment: in House, P.K., Webb, R.H., Baker, V.R., and Levish, D.R., (eds), Ancient Floods,
Modern Hazards: Principles and Applications of Paleoflood Hydrology, Water Sciences and
Application Volume 5, American Geophysical Union, Washington D.C., p. 175-190.
Loomis, F.B. (1938) Physiography of the United States: Doubleday, Doran & Company, Inc.,
New York, 350 p.
Loukas, A., Vasiliades, L., and Dalezios, N.R. (2000) Flood producing mechanisms
identification in southern British Columbia, Canada. J. Hydrology, 227, pp. 218-235.
Lovejoy, S, Schertzer, D. (1995) Multifractals and rain. In New Uncertainty Concepts in
Hydrology and Hydrological Modeling, Kundzewicz, Z.W. (ed.), Cambridge University Press,
New York, pp. 61103.
Madole, R.F., Bradley, W.C., Loewenherz, D.S., Ritter, D.F., Rutter, N.W., and C.E. Thorn
(1987) Rocky Mountains, Ch. 7, in Graf, W.L., ed., Geomorphic Systems of North America:
Geological Society of America, Centennial Special Volume 2, Boulder, Colorado p. 211-257.
Malde, H.E. (1955) Surficial geology of the Louisville quadrangle, Colorado: U.S. Geological
Survey Bulletin 996E, p. 217-259.
Mansfield, G.R. (1938) Flood deposits of the Ohio River, January-February, 1937, a study of
sedimentation, in Grover, N.C., Floods of Ohio and Mississippi Rivers, January-February,
1937: U.S. Geological Survey Water Supply Paper 838, p. 693-736.
Matthai, H.F. (1968) Magnitude and frequency of floods in the United States Part 6-B,
Missouri River Basin below Sioux City, Iowa. U.S. Geological Survey Water-Supply Paper
1680, 491 p.
Matthai, H.F. (1969) Floods of June 1965 in South Platte River Basin, Colorado. U.S. Geological
Survey Water-Supply Paper 1850-B, 64 p.
Matthai, H.F. (1990) Floods. In M.G. Wolman and H.C. Riggs (eds.) Surface Water Hydrology,
Geol. Soc. America, the Geology of North America, 0-1, pp. 97-120.
Matthes, G.H. (1922) Floods on small streams caused by rainfall of the cloudburst type. Trans.
Am. Soc. Civ. Eng., Paper No. 1505 (symposium), vol. 85, pp. 1388-1399.
McCain, J.F. and Ebling, J.L. (1979) A plan for the study of flood hydrology of foothill streams
in Colorado. U.S. Geological Survey Open-File Report 79-1276, 29 p.
McCain, J.F. and Jarrett, R.D. (1976) Manual for estimating flood characteristics of natural-flow
streams in Colorado. Colorado Water Conservation Board Technical Manual 1, 68 p.
156
McCain, J.F., Hoxit, L.R., Maddox, R.A., Chappel, C.F. and Caracena, F. (1979) Storm and
flood of July 31-August 1, 1976 in the Big Thompson and Cache la Poudre River Basins,
Larimer and Weld Counties, Colorado, Part A. Meteorology and Hydrology in the Big
Thompson River and Cache la Poudre River Basins. U.S. Geological Survey Professional Paper
1115, pp. 1-85.
McKee, T.B. and Doesken, N.J. (1997) Colorado Extreme Storm Precipitation Data Study,
Climatology Report #97-1, Department of Atmospheric Science, Colorado State University,
Fort Collins, Colorado, 34 p. and appendices.
Merz, R. and Blschl, G. (2003) A process typology of regional floods. Water Resour. Res. 39
(12), 1340, doi:10.1029/2002WR001952, pp. SWC 5-1 SWC 5-20.
Meyer, A.F. (1917) Elements of Hydrology. John Wiley and Sons, New York, 487 p.
Munn, J. and Savage, J.L. (1922) The Flood of June 1921 in the Arkansas River, at Pueblo,
Colorado. Trans. Am. Soc. Civ. Eng., Paper No. 1480, vol. 85, pp. 1-35, with discussion pp.
36-65.
National Research Council (NRC) (1988) Estimating Probabilities of Extreme Floods: Methods
and recommended research. National Academy Press, Washington, D.C., 141 p.
National Research Council (NRC) (2005) Flash Flood Forecasting over Complex Terrain, With
an Assessment of the Sulphur Mountain NEXRAD in Southern California. National Academy
Press, Washington, D.C., 191 p.
OConnell, D.R.H., Ostenaa, D.A., Levish, D.R., and Klinger, R.E. (2002) Bayesian flood
frequency analysis with paleohydrologic bound data: Water Resources Research, v. 38, no. 5,
pp. 16-1-16-14.
O'Connor, J.E. and Costa, J.E. (2004) Spatial distribution of the largest rainfall-runoff floods
from basins between 2.6 and 26,000 km
2
in the United States and Puerto Rico. Water Resour.
Res. 40, W01107, doi:10.1029/2003WR002247, 11 p.
Ogden, F.L. and Julien, P.Y. (2002) CASC2D: A Two-Dimensional, Physically-Based,
Hortonian Hydrologic Model. In Singh, V.P. and Frevert, D. (eds.) Mathematical Models of
Small Watershed Hydrology and Applications, Ch. 4, Water Resources Publications, Littleton,
CO, pp. 69-112.
Osterwald, D.B. (2003) Rails Thru the Gorge. A mile-by-mile guide for the Royal Gorge Route.
Western Guideways, Ltd., Hugo, CO, 167 p.
Over, T.M. and Gupta, V.K. (1996) A space-time theory of mesoscale rainfall using random
cascades. J. Geophys. Res. (Atmos.), 101(D21, pp. 26,319-26,331.
Parker, G. (1978) Self-formed straight rivers with equilibrium banks and mobile bed. Part 2. The
gravel river: Journal of Fluid Mechanics, v. 89, part 1, p. 127-146.
Patterson, J.L. (1964) Magnitude and frequency of floods in the United States Part 7, Lower
Mississippi River Basin. U.S. Geological Survey Water-Supply Paper 1681, 636 p.
Patton, P.C. (1987) Measuring rivers of the past: A history of fluvial paleohydrology, in Landa,
E.R., and Ince, S., (eds.), The History of Hydrology: History of Geophysics: Volume 3,
157
American Geophysical Union, Washington, D.C., p. 55-67.
Patton, P.C., Baker, V.R., and Kochel, R.C. (1979) Slack water deposits: A geomorphic
technique for the interpretation of fluvial paleohydrology, in Adjustments of the fluvial system:
Rhodes, D.D., and Williams, G.P., (eds.), Kendall/Hunt Publishing Company, Dubuque, Iowa,
p. 225-253.
Powers, W.E. (1933) Physiography of the Royal Gorge of the Arkansas River: Geological
Society of America Proceedings, p. 102.
Powers, W.E. and Behre, C.H., Jr. (1934) Physiographic history of the Upper Arkansas River
Valley and the Royal Gorge, Colorado: Annals of the Association of American Geographers, v.
24, p. 64.
Powers, W.E. (1935) Physiographic history of the Upper Arkansas River Valley and the Royal
Gorge, Colorado: Journal of Geology, v. 43, no. 2, p. 184-199.
Puseman, K. and Klinger, R.E. (2001) Dating bulk soil versus identified organics at
archaeological sites: [abstract] 66th Annual Meeting of the Society for American Archaeology,
New Orleans, LA.
Rawls, W.J., Ahuja, L.R., Brakensiek, D.L., and Shirmohammadi, A. (1993) Infiltration and soil
water movement. In Handbook of Hydrology, Maidment, D.R. (ed.), McGraw-Hill, New York,
Ch. 5, pp. 5.1-5.51.
Richmond, G.M. (1965) Glaciation of the Rocky Mountains, in H.E. Wright, Jr., and D.G. Frey,
eds., The Quaternary of the United States: Princeton University Press, Princeton, New Jersey,
922 p.
Schreiner, L.C. and Riedel, J.T. (1978) Probable Maximum Precipitation Estimates, United
States East of the 105th Meridian. Hydrometeorological Report No. 51, National Weather
Service, National Oceanic and Atmospheric Administration, U.S. Department of Commerce,
Silver Spring, MD, 87 p.
Scott, G.R. (1960) Subdivision of the Quaternary alluvium east of the Front Range near Denver,
Colorado: Geological Society of America Bulletin, v. 71, no. 10, p. 1541-1543.
Scott, G.R. (1963) Quaternary geology and geomorphic history of the Kassler Quadrangle
Colorado: U.S. Geological Survey Professional Paper 421A, 70 p., 1 sheet, scale 1:24,000.
Scott, G.R. (1972a) Reconnaissance geologic map of the Hobson Quadrangle, Pueblo and
Fremont Counties, Colorado: USGS Miscellaneous Field Studies Map MF-353, 1 sheet, scale
1:24,000.
Scott, G.R. (1972b) Reconnaissance geologic map of the Swallows Quadrangle, Pueblo County,
Colorado: USGS Miscellaneous Field Studies Map MF-354, 1 sheet, scale 1:24,000.
Scott, G.R., Taylor, R.B., Epis, R.C., and Wobus, R.A. (1978) Geologic map of the Pueblo 1 X 2
Quadrangle, south-central Colorado: USGS Miscellaneous Investigations Series I-1022, 2
sheets, scale 1:250,000.
Seaber, P.R., Kapinos, F.P., and Knapp, G.L. (1987) Hydrologic Unit Maps. U.S. Geological
158
Smith, J.A. (1989) Regional flood frequency analysis using extreme order statistics of the annual
peak record. Water Resour. Res., 25(2), pp. 311-317.
Stedinger, J.R., and Cohn, T.A. (1986) Flood frequency analysis with historical and paleoflood
information: Water Resources Research, v. 22, p. 785-793.
Stanton, D. (2000) Comprehensive Facility Review, Pueblo Dam, Fryingpan-Arkansas Project,
Colorado. Bureau of Reclamation, Denver, CO.
Steers, M. and Trojanowski, J. (2000) Pueblo Dam Risk Analysis Modified Dam Including
Contraction Joint Leakage and Concrete Dam Tension Issues. Bureau of Reclamation,
Technical Service Center, Denver, CO, 29 p. and appendices.
Swain, R.E., England, J.F. Jr., Bullard, K.L. and Raff, D.A. (2004) Hydrologic hazard curve
estimating procedures. Dam Safety Research Program Research Report DSO-04-08, U.S.
Department of Interior, Bureau of Reclamation, Denver, Colorado, 79 p.
Taylor, R.B., Scott, G.R., Wobus, R.A., and Epis, R.C. (1975a) Reconnaissance geologic map of
the Cotopaxi 15-minute Quadrangle, Fremont and Custer counties, Colorado: USGS
Miscellaneous Field Investigations Map I-900, 1 sheet, scale 1:62,500.
Taylor, R.B., Scott, G.R., Wobus, R.A., and Epis, R.C. (1975b) Reconnaissance geologic map of
the Royal Gorge Quadrangle, Fremont and Custer counties, Colorado: USGS Miscellaneous
Investigations Map I-869, 1 sheet, scale 1:62,500.
Thomson, M.T., Gannon, W.B., Thomas, M.P., and Hayes, G.S. (1964) Historical floods in New
England. U.S. Geological Survey Water-Supply Paper 1779-M, 105 p.
Trojanowski, J. (2002) Report of Findings, Pueblo Dam Modification, Fryingpan-Arkansas
Project, Colorado. Bureau of Reclamation, Denver, CO, 13 p.
Troutman, B.M. and Karlinger, M.R. (2003) Regional flood probabilities. Water Resour. Res. 39
(4), 1095, doi:10.1029/2001WR001140, pp. SWC 4-1 SWC 4-15.
Tweto, O., Steven, T.A., Hail, W.J., Jr., and Moench, R.H. (1976) Preliminary geologic map of
the Montrose 1 X 2 Quadrangle, southwestern Colorado: USGS Miscellaneous Field Studies
Map MF-761, 1 sheet, scale 1:250,000.
U.S. Army Corps of Engineers (USACE) (1945 - ) Storm Rainfall in the United States (ongoing
publication). Washington, D.C.
U.S. Army Corps of Engineers (USACE) (1977) Master Water Control Manual, Appendix C,
Pueblo Dam and Reservoir, Arkansas River Watershed, Arkansas River above mouth of
Walnut Creek, Kansas, Colorado and Kansas, Arkansas River Basin. U.S. Army Engineer
District, Albuquerque, NM.
U.S. Geological Survey (USGS) (1923) Surface water supply of the United States, 1921, Part 7,
Lower Mississippi River Basin. U.S. Geological Survey Water-Supply Paper 527, 39 p.
U.S. Geological Survey (USGS) (1955) Compilation of records of surface waters of the United
States through September 1950, Part 7, Lower Mississippi River Basin. U.S. Geological
U.S. Geological Survey (USGS) (1964) Compilation of records of surface waters of the United
159
States, October 1950 to September 1960, Part 7, Lower Mississippi River Basin. U.S.
Geological Survey Water-Supply Paper 1731, 552 p.
U.S. Geological Survey (USGS) (1969) Surface water supply of the United States, 1961-65, Part
7, Lower Mississippi River Basin, Volume 2, Arkansas River Basin. U.S. Geological Survey
Water-Supply Paper 1921, 878 p.
Vaill, J.E. (2000) Analysis of the Magnitude and Frequency of Floods in Colorado. U.S.
Geological Survey Water-Resources Investigations Report 99-4190, 35 p.
Van Alstine, R.E. (1974) Geology and mineral deposits of the Poncha Springs SE Quadrangle,
Chaffee County, Colorado: USGS Professional Paper 829, 19 p., ? sheets, scale 1:24,000.
Velleux, M. L. (2005) Spatially distributed model to assess watershed contaminant transport and
fate. Ph.D. Dissertation, Department of Civil Engineering, Colorado State Univ., Fort Collins,
Colorado, 261 p.
Velleux, M.L., England, J.F. Jr., and Julien, P.Y. (2005) TREX Watershed Modeling Framework
User's Manual: Model Theory and Description. Department of Civil Engineering, Colorado
State University, Fort Collins, Colorado. Last Revised August 2005, 83 p.
Vogel, R.M., Zafirakou-Koulouris, A. and Matalas, N.C. (2001) Frequency of record-breaking
floods in the United States. Water Resour. Res. 37(6), pp. 1723-1731.
Wahl, K.L. (1982) Simulation of regional flood-frequency curves based on peaks of record. In:
Biswas, M.R. and Biswas, A.K., eds., Alternative strategies for desert development and
management, vol.3, Pergamon Press, New York, pp. 760-769.
Webb, R.H., and Jarrett, R.D. (2002) One-dimensional estimation techniques for discharges of
paleofloods and historical floods: in House, P.K., Webb, R.H., Baker, V.R., and Levish, D.R.,
(eds), Ancient Floods, Modern Hazards: Principles and Applications of Paleoflood Hydrology,
Water Sciences and Application Volume 5, American Geophysical Union, Washington D.C., p.
111-126.
Wiesner, C.J. (1970) Hydrometeorology. Chapman and Hall, London, 232 p.
Wilson, L.L. (1989) Assessment of the exceedance probability of extreme rainfalls in the
Midwest by stochastic storm transposition. M.S. Thesis, Iowa State University, Ames, IA, 127
p.
Wilson, L.L. and Foufoula-Georgiou, E. (1990) Regional rainfall frequency analysis via
stochastic storm transposition. J. Hydraul. Engr., ASCE, 116(7), pp. 859-880.
Yankee Atomic Energy Company (YAEC) (1984) Probability of extreme rainfalls and the effect
on the Harriman Dam. Yankee Atomic Energy Company, Framingham, MA, 16 p. and four
appendices.
160
APPENDIX A
LOSS OF LIFE STUDY
Pueblo Dam Colorado
Loss of Life Assessment
September 10, 2003
Introduction
Mr. John England requested that an evaluation be made to determine the loss of life that
would result from the failure of Pueblo Dam. This information will be used as part of a
detailed probabilistic flood hazard study for Pueblo Dam.
Pueblo Dam is located on the Arkansas River in Pueblo County, about 6 miles upstream and
west of the city of Pueblo. The dam, constructed during the period 1970-1975, is the
terminal storage feature for the Fryingpan-Arkansas Project. The dam is earthfill with a
massive concrete buttress with an overflow section. The height of the dam above the
streambed is 191 feet. The dam has a crest length of 10,200 feet. The reservoir has a total
storage capacity of 357,700 acre-feet at elevation 4898.7 feet, 26.3 feet below the 4925.0 feet
dam crest elevation. The reservoir has a capacity of 535,300 acre-feet with the reservoir at
the dam crest. The reservoir has a surface area of 4,646 acres (7.25 square miles) at elevation
4898.7 feet and 8,027 acres (12.5 square miles) at the dam crest. The uncontrolled, overflow
type, spillway has a crest elevation of 4898.7 feet. The spillway has a capacity of 191,500
cubic feet per second (ft
3
/s) at a reservoir elevation of 4919 feet, 6 feet below the dam crest.
Procedure Used for Estimating Life Loss
Reclamations A Procedure for Estimating Loss of Life Caused by Dam Failure was used
in conducting this study. The procedure is useful, robust, easy to apply and produces
plausible results. The procedure was developed using data from every U.S. dam failure that
resulted in more than 50 fatalities and every post-1960 U.S. dam failure that resulted in 1 or
more fatalities. This dam failure data provided information on warning, population at risk,
flood severity, and fatality rates. It was concluded that loss of life resulting from dam failure
is highly influenced by three factors: 1) The number of people occupying the flood plain
before any evacuation takes place, 2) The amount of warning that is provided to people
exposed to dangerous flooding, and 3) The severity of the flooding.
The procedure for estimating loss of life from dam failure is comprised of 7 steps:
1) Determine dam failure scenarios to evaluate.
2) Determine time categories for which loss of life estimates are needed.
1
3) Determine area flooded for each dam failure scenario.
4) Estimate the number of people at risk for each dam failure scenario and time
category.
5) Determine when dam failure warnings would be initiated.
6) Select appropriate fatality rate for estimating life loss.
7) Evaluate uncertainty in various parameters which lead to uncertainties in the life loss
estimates.
These steps, as applied to Pueblo Dam, will be described in detail:
1) Determine dam failure scenarios to evaluate.
The failure of a dam can occur from various causes and at different times of the day, week or
year. The consequences of dam failure or sudden releases from the dam would be different
depending upon the cause of failure or sudden release and when the event occurs.
The following scenarios will be evaluated:
a) Dam fails from piping with the reservoir water surface at elevation 4893.7 feet (top of
Joint Use Capacity) during normal weather conditions. (Up through the year 2000,
the reservoir has never reached this elevation, however, it has been a few feet below
this level). The reservoir water surface would be 31.3 feet below the dam crest at the
initiation of dam failure. Approximately 330,700 acre-feet of water would be
released in the failure.
b) Dam fails as a result of an earthquake with the reservoir water surface elevation at
4893.7 feet during normal weather conditions. The reservoir water surface would be
31.3 feet below the dam crest at the initiation of dam failure. Approximately 330,700
acre-feet of water would be released in the failure.
c) Dam fails during a flood that approaches the probable maximum flood during
extreme weather conditions. The reservoir water surface elevation is near the dam
crest at the initiation of dam failure. Approximately 535,300 acre-feet of water would
be released in the failure. Major releases from the spillway would occur prior to dam
failure.
2) Determine time categories for which loss of life estimates are needed.
Two different time categories were chosen for inclusion in the study. Loss of life will be
developed for a failure that occurs during the day as well as for a failure that occurs during
2
the night. The time of day (or night) at which failure occurs will likely influence when
warnings are initiated, how people respond to the warnings, and the resultant fatality rates.
3) Determine the area flooded for each dam failure scenario.
The failure of Pueblo Dam would cause flooding along an extended reach of the Arkansas
River. Pueblo, Colorado would be the hardest hit community because it is close to the dam
and is also the largest community downstream from the dam. There are many other
communities between Pueblo and John Martin Reservoir, and these communities would also
be seriously impacted by dam failure flooding. The failure of John Martin Dam could occur,
depending upon several factors, if Pueblo Dam failed from a major flood.
Dam Failure Inundation studies for Pueblo Dam have been prepared. A Dam Failure
Inundation Study for Pueblo Dam was prepared in June 1981 by what was then the Lower
Missouri Regional Office. The 1981 inundation study used an inflow design flood having a
peak inflow to the reservoir of 270,000 ft
3
/s. The breach was assumed to form with the
reservoir water surface at 4918.88 feet, the maximum water surface elevation reached during
the inflow design flood. The breach was assumed to fully form in 12 minutes and the breach
was assumed to be rectangular with a width of 120 feet. These breach parameters, when used
in the National Weather Services Dam Break Flood Forecasting Model, produced a peak
discharge of 998,759 ft
3
/s. The peak discharge was 771,500 ft
3
/s at Pueblo and 528,700 ft
3
/s
at Fowler, a community located about 50 river miles downstream from Pueblo Dam. The
study extended 117 miles from Pueblo Dam to John Martin Reservoir.
A new Inundation Study was prepared in May 1992 as part of a Threat to Life Assessment
for Pueblo Dam. The new inundation study was prompted by the preparation of a new
Probable Maximum Flood Study for Pueblo Dam in June 1991. The general storm PMF
developed in 1991 had a peak inflow of 834,700 ft
3
/s and an inflow volume of 1,390,000
acre-feet. Another major change between the 1992 study and the 1981 study is the size and
timing of the breach used in the model. The 1981 study used a beach width that was less
than the dam height. The use of such a small breach width typically does not yield either
conservative or realistic results. The breach used in the 1992 study had a base width of 330
feet and a top width of 660 feet, yielding an average breach width of 495 feet. The breach
width used in the 1992 was about 400 percent of the average breach width used in the 1981
study. The full breach in the 1992 study was assumed to form in one hour, commencing
when the reservoir water surface elevation was one foot above the dam crest. The peak dam
failure outflow was 3,159,000 ft
3
/s. The peak discharge was 2,694,000 ft
3
/s at Pueblo and
1,297,000 ft
3
/s at Fowler.
3
In addition to dam failure during the PMF, the 1992 inundation study evaluated other flood
scenarios. These scenarios included:
a) 1991 PMF without dam (peak flow at Pueblo Dam site of 834,700 ft
3
/s).
b) 43% of the 1991 PMF without dam (peak flow at Pueblo Dam site of 359,000 ft
3
/s).
c) 43% of the 1991 PMF with releases from an intact existing dam (peak flow from
Pueblo Dam of 198,000 ft
3
/s).
Forty-three percent of the 1991 PMF was chosen in these scenarios because this flood
reaches elevation 4919.06 feet, the approximate original design maximum water surface
elevation. Flood boundaries were delineated on a set of USGS topographic maps for the dam
failure event, PMF without dam, and 43% of the PMF with releases from intact existing dam
from Pueblo Dam to Fowler. In August 1999, using the results from the 1992 study, flood
boundaries for dam failure and 43% of the PMF with releases from intact existing dam were
continued downstream from Fowler to John Martin Reservoir.
A dam failure inundation study does not exist for the failure of Pueblo Dam with the
reservoir at or near the top of joint use elevation. The failure of the dam with the reservoir
water surface elevation about 32 feet below the dam crest would result in a peak dam failure
outflow less than that obtained for an overtopping failure.

Several methods and procedures were used to verify the validity of the peak dam failure
outflow used in the existing dam failure inundation study and to determine the dam failure
outflow that would occur from dam failure with the reservoir at elevation 4898.7 feet.
The following procedures were used for estimating the peak breach outflow for Pueblo Dam:
The Bureau of Reclamation issued inundation mapping guidelines in the early 1980s
(reference 1). The guideline contains an equation for estimating peak outflow from a dam
failure. The magnitude of the peak outflow is solely the function of dam height.
In 1995, David Froehlich prepared, Peak Outflow from Breached Embankment Dam
(reference 2). Tony Wahl, the author of Reclamations Prediction of Embankment Dam
Breach Parameters, DSO-98-004, 1998, describes Froehlichs procedure as one of the better
available methods for direct prediction of peak breach discharge. In Froelichs procedure,
the peak outflow from a dam failure is a function of the reservoir storage at the time of
4
failure and reservoir depth at the time of failure, measured from the bottom of the final
breach to the reservoir surface.
In 1997, Walder and OConnor prepared, Methods for predicting peak discharge of floods
caused by failure of natural and constructed earthen dams (reference 3). Tony Wahl
describes this procedure as a relatively simple, physically-based model of dam-breach
formation. Mr. Wahl stated that the procedure offers the advantage of a theoretical
foundation that accounts for the differences and limitations imposed by large- and small-
reservoir cases. This procedure uses total volume of water drained from the reservoir, drop
in reservoir level during the failure and a breach erosion rate which is chosen by the user of
the procedure.
Empirical dam failure data indicates that most dams are empty within a few hours of the
breach initiation. Peak outflow estimates were made assuming that the reservoir was empty
in either 2 hours or 4 hours. The peak was assumed to be 200% of the average outflow
during the time to empty.
The peak dam failure outflows for Pueblo Dam obtained with these various methods and
procedures is summarized in Table 1.
5
Table 1
Pueblo Dam
Maximum Dam Failure Outflow (cubic feet per second)
Source, Method or
Procedure
Failure with Reservoir at
Elevation 4893.7 feet
Failure with Reservoir at or near
Elevation 4925 feet (Dam Crest)
1981 Inundation Study Not evaluated 998,600
(reservoir at elevation 4918.9)
1992 Inundation Study Not evaluated 3,159,000
Equation Q=75D
1.85
992,300 1,355,100
Froehlichs Procedure 986,200 1,400,800
Walder and OConnor
Procedure
2,902,000 4,142,000
Reservoir empties in 2
hours
4,002,000 6,477,000
Reservoir empties in 4
hours
2,000,100 3,239,000
Table 1 indicates that there is significant uncertainty regarding peak dam failure outflows.
Failure of the dam with the reservoir at elevation 4893.7 feet results in about 60 to 75%
(depending upon procedure used) of the dam failure outflow with the reservoir at the dam
crest. Applying this percentage to the peak obtained in the 1992 Inundation Study, the
maximum outflow from failure of the dam with the reservoir at elevation 4893.7 feet would
be somewhere in the range of 1,900,000 ft
3
/s to 2,400,000 ft
3
/s. The area flooded from a
failure with the reservoir at elevation 4893.7 feet would be slightly smaller than from a
failure with the reservoir at the dam crest. Flooding would still be catastrophic.
4) Estimate the number of people at risk for each dam failure scenario.
The failure of Pueblo Dam would place people along the Arkansas River at risk.
In the Pueblo area, the flood boundaries for the 1991 PMF without the dam in place is only
slightly smaller than the flood boundary for the 1991 PMF with dam failure. The peak dam
failure outflow for a sunny day or piping failure would be more than double the discharge
associated with the 1991 PMF without the dam in place. The flood boundary for a sunny day
or piping failure (neither was studied) would be slightly smaller than the boundary for failure
of the dam during a flood that causes dam overtopping. The number of people at risk for a
6
sunny day or piping failure was assumed to be 90% of the number of people at risk for a
flood induced dam failure.
Major flooding would occur downstream from Pueblo Dam prior to a flood induced failure of
the dam. A Risk Assessment of Pueblo Dam was prepared August 26, 1997. Flood routings
used in the risk assessment indicate that the dam can be overtopped by as much as 6.27 feet
during the 1991 PMF. With or without dam overtopping, major outflows would occur from
the dam prior to a flood induced dam failure. If the dam was overtopped by one foot prior to
the initiation of dam failure, the total outflow from the dam would be about 325,000 ft
3
/s. If
the dam was overtopped by two feet prior to failure, the total outflow prior to failure would
be about 380,000 ft
3
/s. These discharges are more than three times larger than the discharge
that went through Pueblo in 1921. The flooding that would precede dam failure would place
many people at risk. The number of people at risk for a flood induced dam failure will
include the total number of people at risk, both before failure from flows prior to dam failure,
and after failure, from the increase in outflow resulting from the failure.
The number of people at risk is summarized in Table 2:
Table 2
Pueblo Dam
Number of People at Risk
Location
Dam Failure Cause
*
39 39 39 39 43 43
38 38
eastern edge of Pueblo (mile
5.0 to 10.4)
8690 8690 8690 8690 9652 9652
2037 2037
10.4 to 17.0)
1580 1580 1580 1580 1756 1756
285 285
17.0 to 32.0)
900 900 900 900 1000 1000
847 847
upstream from Fowler (mile
32.0 to 49.6)
185 185 185 185 204 204
88 88
Upstream from Fowler to
Rocky Ford (mile 49.6 to 73.4)
2520 2520 2520 2520 2800 2800
224 224
7
Rocky Ford to upstream from
La Junta (mile 73.4 to 84.1)
2085 2085 2085 2085 2317 2317
326 326
downstream from La Junta
(mile 84.1 to 86.5)
3320 3320 3320 3320 3688 3688
1209 1209
upstream from Las Animas
(mile 86.5 to 108.6)
325 325 325 325 362 362
228 228
(mile 108.6 to 117.2)
3770 3770 3770 3770 4186 4186
3681 3681
Total 23414 23414 23414 23414 26008 26008
8963 8963
*
For the Flood Induced Failure, the bottom number in each row represents the number of
people at risk from the flooding that precedes dam failure, and the top number in each row is
the number of people at risk from dam failure, including the number of people at risk from
the flooding that precedes dam failure.
5) Determine when dam failure warnings would be initiated.
The most important factor that determines the loss of life from dam failure is when dam
failure flood warnings are initiated. Guidance regarding when dam failure warnings would
be initiated for Pueblo Dam was obtained from A Procedure for Estimating Loss of Life
Caused by Dam Failure, DSO-99-06.
An Emergency Action Plan (EAP) for Pueblo Dam has been prepared. The latest revision is
dated April 14, 1998. The EAP states that Pueblo Dam is not manned on a permanent basis,
but personnel are assigned to the local area at the Operations and Maintenance shop at the
Pueblo Field Office which is located at the toe of the dam near the right embankment section.
The Pueblo Dam operator visits the dam daily and may be contacted by telephone or by
radio. Around-the-clock attendance at the dam is required when the reservoir water surface
is in the flood pool (which is defined seasonally, i.e., part of the flood pool is designated as a
joint use pool and is used for conservation storage from November 1 through April 15.
Initial notification of a problem affecting Pueblo Dam would be made by the dam operating
and maintenance personnel who attend Pueblo Dam on a daily basis and who perform
damtending duties as well as maintenance on the dam and appurtenant features. Water
Scheduling/Safety of Dams staff and/or their division chief(s) and/or Area Manager, Eastern
Colorado Area Office is responsible for providing the initial notifications to Pueblo County
8
public safety officials. The Division Chief and/or Area Manager, Eastern Colorado Area
Office is responsible for providing subsequent notification and protective action
recommendations to the Pueblo County public safety officials.
Table 3 summarizes the assumptions made regarding the initiation of dam failure warnings:
Table 3
Pueblo Dam
Assumptions Regarding When Dam Failure Warnings Would be Initiated
1 hour before
failure
0.5 hour after
dam failure
0.25 hour after
dam failure
0.5 hour after
dam failure
4 hours before
failure
2 hours before
failure
The amount of warning that is issued in any particular locations would depend not only on
when a dam failure warning is initiated but also on how long it takes floodwater to travel
from the dam to the location of interest. Flood wave travel times to downstream areas were
obtained from the 1992 inundation study.
9
Table 4 summarizes the flood wave travel times to downstream locations:
Table 4
Pueblo Dam
Flood Wave Travel Times (Measured from the start of the breach)
Location Distance Downstream
from Pueblo Dam
(miles)
Travel Time for Leading
Edge of Flood (hours)
Pueblo Dam 0 0
Pueblo 5 0.5
Baxter 13 1.5
Devine 17 2.0
Avondale 25 3.5
Boone 32 4.5
Fowler 50 8
Elder 54 9
Manzanola 60 10
Rocky Ford 74 13
Swink 81 14
La Junta 87 15
Riverdale 101 18
Las Animas 112 21
Flood severity understanding is a factor that has an impact on fatality rates. The terms vague
and precise are used to gage the quality of the warning that is issued in areas downstream
from the dam. Vague understanding means that warning issuers have not seen an actual dam
failure or do not comprehend the true magnitude of the flooding. Precise understanding
means that the warning issuers have an excellent understanding of the flood magnitude due to
observations of the flooding made by themselves or others.
For daytime static or seismic failures, vague warning was assumed for areas that would be
flooded during the first two hours after the initiation of dam failure. For nighttime static or
seismic failures, vague warning was assumed for areas that would be flooded during the first
3 hours after the initiation of dam failure.
For a daytime overtopping failure, precise warning was assumed for all areas. For a
nighttime overtopping failure, vague warning was assumed for areas that would be flooded
during the first 2 hours after the initiation of dam failure.
10
Table 5 summarizes the amount of warning time in hours and assumption regarding flood
severity understanding for the different locations and dam failure causes.
Table 5
Pueblo Dam
Flood Wave Travel Time and Warning Time for Each Location (hours)
Location Flood Wave
Travel Time
(average)
Dam Failure Cause
0.25 1.25
Vague
0
Vague
0
Vague
0
Vague
4.25
Precise
2.25
Vague
Western edge of Pueblo to eastern
edge of Pueblo (mile 5.0 to 10.4)
0.8 1.8
Vague
0.3
Vague
0.5
Vague
0.3
Vague
4.8
Precise
2.8
Vague
downstream from Baxter (mile 10.4
to 17.0)
1.7 2.7
Vague
1.2
Vague
1.4
Vague
1.2
Vague
5.7
Precise
3.7
Vague
downstream from Boone (mile 17.0
to 32.0)
3.3 4.3
Precise
2.8
Precise
3.0
Precise
2.8
Precise
7.3
Precise
5.3
Precise
upstream from Fowler (mile 32.0 to
49.6)
6.2 7.2
Precise
5.7
Precise
5.9
Precise
5.7
Precise
10.2
Precise
8.2
Precise
Upstream from Fowler to Rocky
Ford (mile 49.6 to 73.4)
10.5 11.5
Precise
10.0
Precise
10.2
Precise
10.0
Precise
14.5
Precise
12.5
Precise
Rocky Ford to upstream from La
Junta (mile 73.4 to 84.1)
14 15.0
Precise
13.5
Precise
13.7
Precise
13.5
Precise
18.0
Precise
16.0
Precise
downstream from La Junta (mile
84.1 to 86.5)
15 16.0
Precise
14.5
Precise
14.7
Precise
14.5
Precise
19.0
Precise
17.0
Precise
upstream from Las Animas (mile
86.5 to 108.6)
18 19.0
Precise
17.5
Precise
17.7
Precise
17.5
Precise
22.0
Precise
20.0
Precise
Upstream from Las Animas to head
of John Martin Res. (108.6 to 117.2)
21 22.0
Precise
20.5
Precise
20.7
Precise
20.5
Precise
25.0
Precise
23.0
Precise
6) Select appropriate fatality rate for estimating life loss.
A Procedure for Estimating Loss of Life Caused by Dam Failure, DSO-99-06, contains
recommended fatality rates for estimating loss of life resulting from dam failure. The
recommended rates were developed using data from approximately 40 floods, many of which
were caused by dam failure. The 40 floods included nearly all U.S. dam failures causing
more than 50 fatalities. Other flood events were selected in an attempt to cover a full range
11
of flood severity and warning combinations. Events occurring outside of the U.S. were
included in the data set.
At least two major floods have occurred on the Arkansas River and information associated
with these floods provides some useful insights that may help determine appropriate fatality
rates to use for dam failure. A major flood occurred in the vicinity of Pueblo in June 1921
and a major flood occurred in scattered sections of Eastern Colorado in June 1965. The 1965
flood was most severe in the area downstream from John Martin Dam.
Table 6 summarizes the peak discharges on the main stem of the Arkansas River that
occurred during the 1921 and 1965 floods.
Table 6
Arkansas River
Summary Information for the 1921 and 1965 Floods
Location
(listed from upstream
to
downstream)
USGS
Gauging
Station
Number
1921 1965
Peak
Discharge
(ft
3
/s)
Date Peak Discharge
(ft
3
/s)
Time/Date
Near Pueblo 07099500 103,000 June 3 9,260 2130 hours, June 17
Near Nepesta 07117000 180,000 June 4 43,100 1130 hours, June 18
At La Junta 07123000 200,000 June 4 31,700 0530 hours, June 19
At Las Animas 07124000 - - 22,100 1630 hours, June 19
Below John Martin
Dam
07130500 - - 900 (daily
discharge
June 16
At Lamar 07133000 130,000 June 5 73,800 0800 hours, June 18
At Holly 07135500 90,000 June 5 - -
Near Coolidge, KS 07137500 - - 158,000 June 17
At Syracuse, KS 07138000 62,000 June 6 174,000 2000 hours, June 17
At Garden City, KS 07139000 - - 130,000 0300 hours, June 19
At Dodge City, KS 07139000 - June 7 82,000 2400 hours, June 19
At Great Bend, KS 07141300 - June
10
27,800 1200 hours, June 23
1921 Flood
The Weather Bureau described the June 1921 flood as the most disastrous flood of record in
Colorado (ref. 4). There were three distinct flood peaks, but the second flood peak was
clearly the most devastating. Heavy rains fell upstream from Pueblo with as much as 14
inches occurring during the afternoon and night of June 3. At about 8:45 p.m., levees in
Pueblo overtopped at a river stage of 18.1 feet. At about midnight the maximum stage of
24.66 feet was reached. In the 45 minutes ending at 11:55 p.m., the water level rose 5.36 feet
12
(ref. 5). The peak discharge on the Arkansas River at Pueblo was estimated at 103,000 ft
3
/s
(ref. 6).
When the levees overtopped at about 8:45 p.m., an immense volume of water flowed across
the old flood plain and through the heart of the business district, which in 1921 was on both
sides of the river. At the time of the 1921 flood, the Arkansas River was located north of its
present location. The river was relocated to the southerly edge of the floodplain in the
downtown area after the 1921 flood. Flood depths in some areas were as much as 15 feet
above street level. The area inundated was 3 square miles (ref. 6).
The first warning of the approaching flood reached the city about 6 p.m. on the 3
rd
, stating
that a wall of water was rushing down the river. Messengers were sent out at once to warn
the people living in the lowlands. Hundreds of people rushed to the levees to witness the
approach of the great wall of water, not thinking that the city could be inundated, as
the levees were believed high enough to protect it. The sudden breaking of the levees cut
off the people from higher land, and in endeavoring to escape many were drowned, as were
many others in the houses in the lowlands who had refused to heed the flood warning (ref. 6).
Estimates of property damage were very precise whereas estimates of the loss of life varied
widely. Flood damage totaled $19,080,000 and the City of Pueblo suffered slightly more
than half of this (ref. 5). A report to the Pueblo city council stated that 510 dwellings were
washed away, 98 buildings wrecked, and 61 buildings washed from their foundations (ref. 6).
The 1920 census reported a population of 42,908 people in Pueblo. Only a small fraction of
the community was flooded by the 1921 flood. A report by Reclamations Savage and
Munn, published before July 16, 1921, stated that there were 120 known deaths, 49 known to
be missing and 93 not accounted for (from Pueblo Star-Journal and Chieftain, June 3, 1971).
The U.S.G.S. (ref. 6) stated that the official list places the number of bodies recovered at 78,
but many bodies that were washed downstream were never recovered. In 1971, the Pueblo
Star Journal reported: But death certificates at the Pueblo City-County Health Department
were issued as late as February 1922. According to the health department, 88 bodies were
recovered here. The total at the end of October 1921 was 78. The paper went on to talk
about bodies of victims that were found in debris at Fowler for several months after the
flood. The Weather Bureau (ref. 4) stated that 120 persons are known to have lost their
lives because of the flood, 70 of whom were drowned at Pueblo, 12 at points above Pueblo
and the remainder at places below that city. An article in the Pueblo Star-Journal on March
14, 1946, the 25
th
anniversary year of the flood, stated that the toll, a year after the flood,
stood at 200 dead and missing.
13
Subsequent to the 1921 flood, the Pueblo Conservancy District constructed a floodway (river
confined to a concrete lined channel) for the Arkansas River through Pueblo.
1965 Flood
The 1965 peak discharges along the main stem of the Arkansas River in Colorado in the
vicinity of Pueblo were less than those in 1921, but tributary peaks were probably greater at
many sites (ref. 7). In the three state area of Colorado, Kansas and New Mexico, property
damage exceeded $60 million (ref. 7).
The 1965 flood caused fatalities in Colorado, Kansas and New Mexico. The Weather
Bureaus Storm Data publication reported the death of 5 people within the Arkansas River
Basin in Colorado (ref. 8). The Weather Bureaus Climatological Data publication
reported that the number of deaths along the Colorado portion of the Arkansas River and
tributaries will probably total between 20 and 25(ref. 9). The Corps of Engineers prepared a
Flood Report on the 1965 flood. The Corps reported that 4 deaths occurred upstream from
John Martin Dam, 10 deaths occurred in Colorado and Kansas downstream from John Martin
Dam, and 2 deaths occurred in the Canadian River Basin in New Mexico (ref. 10). This
indicates a 3 state total of 16 deaths. The U.S.G.S. reported that 16 people died in the floods,
perhaps taking the number from the Corps of Engineers report (ref. 7).
Newspapers in some of the major towns and cities in the Arkansas River Basin were searched
to learn the details associated with the flood related deaths in Colorado. (These newspapers
were available on microfilm at the Colorado Historical Society Library in downtown
Denver). Newspapers published during the period of about June 15 to June 30, 1965 in
Colorado Springs, Pueblo, La Junta and Lamar provided valuable information. The deaths
associated with the flooding are described starting in the upstream (western) area and
proceed downstream.
David Wilson, age 21, died while trying to flee from a cabin. He and a cabin mate were
awakened by the sound of rocks hitting the side of the cabin. The flooding was caused by the
failure of 2 or 3 dams in series on the West Fork of West Beaver Creek. The dams were
located a few miles from Cripple Creek, Colorado (ref. 11).
Robert Reutter, age 15, died of electrocution while helping a feedlot owner in Pueblo clean
up after the flood (ref. 12).
14
Glenn Leeman, age 53, died when he attempted to drive across a washed out bridge. The
bridge was located about 23 miles east of Colorado Springs and about 0.5 miles west of
Ellicott (ref. 11).
Mrs. James Blundell, age 56, died near Ninaview which is located about 30 miles south of
Las Animas. The woman, in ill health, was being carried out of her trailer home by her
husband when he stepped into the entranceway to a storm cellar and lost hold of her. The
husband washed ashore about a quarter of a mile downstream and survived (ref. 13).
Two children, ages 3 and 7, of Mr. and Mrs. Bill Piper, died about 40 miles south south-east
of Lamar, or about 2 miles east of the Deora Community. The two children died, and the
parents survived, when the adults were attempting to cross a draw by driving on a concrete
slab. The vehicle engine stalled, and in the process of attempting to pull the stalled vehicle
out with a chain attached to another vehicle, the stalled vehicle slipped off the concrete and
the man and two children went under the vehicle. The man was washed a few hundred feet
downstream and was able to get out of the rushing water (ref. 14).
Mrs. Manuel Guajardo, and three of her children, died when their car overturned in flood
water. They were from the West Farm area east of Lamar. It is uncertain where the car was
located when it encountered flooding (ref. 15 and ref. 16).
None of the 10 deaths that occurred within the Arkansas River Basin in Colorado occurred
along the main stem of the Arkansas River (ref. 17). There were fatalities in New Mexico
and Kansas. Some of the fatalities in Kansas could have been on the Arkansas River but
records were not obtained or analyzed.
15
Fatality Rate Selection for Failure of Pueblo Dam
To estimate the loss of life from the failure of Pueblo Dam, a fatality rate is applied to each
area or community impacted by dam failure flooding. Fatality rates are based on the flood
severity, warning time and warning quality. The fatality rates were selected based on
guidance contained in DSO-99-06 as well as community specific information.
Flood severity is an indicator of the destructive capability of the flooding. There are three
categories of flood severity: Low severity flooding occurs when no buildings are washed off
of their foundations; medium severity flooding occurs when homes are destroyed but trees or
mangled homes remain for people to seek refuge in or on; and high severity flooding occurs
when the flood sweeps the area clean and nothing remains. High severity flooding is
generally reserved for the instantaneous failure of concrete dams.
In determining whether flooding is low severity or medium severity, low severity is generally
selected if most of the structures would be exposed to flood depths of less than 10 feet and
medium severity is generally selected if most of the structures would be exposed to flood
depths of 10 feet or more. Another method used for differentiating low from medium
severity is to divide the flood discharge at a particular location by the flood width at that
location. When this parameter is less than 50 ft
2
/s, low severity flooding is generally selected
and when this parameter is 50 ft
2
/s or more, medium severity flooding is generally selected.
The failure of Pueblo Dam would cause the most severe flood ever recorded in Colorado
measured in terms of flood depth and area covered. The width of the flooding would be
similar for both a static/seismic failure and for an overtopping failure. Flood widths were
available for an overtopping failure and are as follows: Pueblo, 7,000 feet; Boone, 9,000
feet; Fowler, 11,600 feet; Rocky Ford, 10,500 feet; La Junta, 7,000 feet; and 16,500 feet at
Las Animas. Dam failure discharges (for an overtopping failure) at these locations are as
follows: Pueblo, 2,694,000 ft
3
/s; Boone, 1,706,000 ft
3
/s; and Fowler, 1,297,000 ft
3
/s.
Discharges were not estimated for the area downstream from Fowler. Medium severity
flooding will be assumed to occur for the entire reach between Pueblo Dam and John Martin
Reservoir.
16
Table 7 summarizes the flood severity assumptions used in this study.
Table 7
Pueblo Dam
Flood Severity Category
Location
Dam Failure Cause
Pueblo Dam to western edge of Pueblo
(mile 0.0 to 5.0)
Western edge of Pueblo to eastern edge of
Western edge of Pueblo to downstream
from Baxter (mile 10.4 to 17.0)
Downstream from Baxter to downstream
from Boone (mile 17.0 to 32.0)
Downstream from Boone to upstream
from Fowler (mile 32.0 to 49.6)
Upstream from Fowler to Rocky Ford
(mile 49.6 to 73.4)
Rocky Ford to upstream from La Junta
(mile 73.4 to 84.1)
Upstream from La Junta to downstream
from La Junta (mile 84.1 to 86.5)
Downstream from La Junta to upstream
from Las Animas (mile 86.5 to 108.6)
Upstream from Las Animas to head of
John Martin Reservoir (mile 108.6 to
117.2)
Medium Flood Severity is assumed for
all failure causes and all locations
Pueblo Dam is located only a few miles upstream from Pueblo. It may be difficult to provide
sufficient and forceful warning to residents of Pueblo. It is reasonable to assume that a large
percentage of the loss of life resulting from the failure of Pueblo Dam would occur in the
community of Pueblo. There are no major population centers for many miles downstream
from Pueblo Dam; and the communities that are downstream from Pueblo Dam would
benefit from more warning time and warnings that would be enhanced with the knowledge of
the flood magnitude as it went through the community of Pueblo. The people exposed to
life-threatening flooding in the downstream communities would likely receive warnings that
clearly describe the seriousness of the situation. Warnings in Pueblo, which may be issued
before dam failure, may be more timidly issued as officials responsible for issuing and
disseminating the warning may be afraid of crying wolf.
17
Dam failure would cause flooding through Pueblo that would be more than a mile wide. This
flooding would cause havoc and massive destruction. However, the majority of Pueblo, and
its residents, would not be directly impacted by dam failure. With most of Pueblo unaffected
by dam failure, resources for warning, evacuation, rescue and reconstruction, would more
widely available than if the entire community was impacted by dam failure.
The failure of Pueblo Dam during a major flood would place thousands of people at risk from
the flooding that would precede dam failure. Many of the people exposed to this pre-failure
flooding would evacuate from their residences or places of business in response to this
flooding. An undetermined number of people would remain in the flooded area, even upon
being warned. Such a situation presents an extreme danger during a dam failure because dam
failure would cause flood levels to rise by 20 to 40 feet and it may not be possible for these
stragglers to evacuate due to impassable roads. Most structures would not be able to
withstand this additional flooding (or would be totally submerged). The largest loss of life
from a U.S. dam failure occurred in 1889 when the South Fork Dam, located near Johnstown,
Pennsylvania, failed during a time when floods were occurring both upstream and
downstream from the dam. Flooding in Johnstown, which preceded the flooding caused by
dam failure, prevented people from evacuating from the community (the warnings were also
nonexistent or poor). Loss of life for the failure of Pueblo Dam will be determined using the
population at risk for the entire dam failure flood plain which includes that portion of the
dam failure floodplain that would be flooded prior to dam failure.
Table 8 summarizes the fatality rates which will be used to estimate loss of life. The fatality
rates were obtained from DSO-99-06 and are based on the flood severity, warning time and
warning quality. The fatality rates were obtained from Table 7 of DSO-99-06. Between
Pueblo Dam and mile 7.0, the rates suggested in Table 7 were used without modification.
Downstream from mile 17, the rates were reduced from the suggested values to account for
the more accurate and forceful warnings that would likely be issued in these downstream
areas. For daytime failures, fatality rates used were 20% of suggested values for areas
receiving 4 to 8 hours of warning and 10% of suggested values for areas receiving more than
8 hours of warning. For nighttime failures, fatality rates used were 50% of suggested values
for areas receiving 4 to 8 hours of warning and 20% of suggested values for areas receiving
more than 8 hours of warning.
18
Table 8
Pueblo Dam
Fatality Rates Used for Estimating Loss of Life
Location
Dam Failure Cause
.03 .15 .15 .15 .01 .03
5.0 to 10.4)
.03 .04 .04 .04 .01 .03
10.4 to 17.0)
.03 .03 .03 .03 .01 .03
17.0 to 32.0)
.002 .01 .01 .01 .002 .005
32.0 to 49.6)
.002 .005 .002 .005 .001 .002
.001 .002 .001 .002 .001 .002
.001 .002 .001 .002 .001 .002
(mile 84.1 to 86.5)
.001 .002 .001 .002 .001 .002
(mile 86.5 to 108.6)
.001 .002 .001 .002 .001 .002
(mile 108.6 to 117.2)
.001 .002 .001 .002 .001 .002
19
Based on the fatality rates shown above, estimates of loss of life were made which are
included in Table 9.
Table 9
Pueblo Dam
Loss of Life Resulting from Various Causes of Dam Failure
Location
Dam Failure Cause
1.2 5.9 5.9 5.9 0.4 1.3
5.0 to 10.4)
261 348 348 348 97 290
10.4 to 17.0)
47 47 47 47 18 53
17.0 to 32.0)
2 9.0 9.0 9.0 2.0 5.0
32.0 to 49.6)
0.4 0.9 0.4 0.9 0.2 0.4
2.5 5.0 2.5 5.0 2.8 5.6
2.1 4.2 2.1 4.2 2.3 4.6
(mile 84.1 to 86.5)
3.3 6.6 3.3 6.6 3.7 7.4
(mile 86.5 to 108.6)
0.3 0.7 0.3 0.7 0.4 0.7
(mile 108.6 to 117.2)
3.8 7.5 3.8 7.5 4.2 8.4
Total 324 435 422 435 131 376
The loss of life estimates above can be compared to loss of life estimates that were prepared
several years ago. A Threat to Life Assessment for Pueblo Dam was prepared June 24, 1992.
Loss of life was estimated for various scenarios using the PMF that was developed in 1991.
Loss of life was estimated using the estimating equations that are contained in Reclamations
1989 Policy and Procedures for Dam Safety Modification Decisionmaking.
20
The maximum operational release is approximately 198,000 ft
3
/s. The assessment indicated
that there are approximately 5 hours between the beginning of discharge through the spillway
and the water surface reaching the crest of the dam. The majority of the people at risk should
have adequate time to notice operational releases causing rising water or to receive
evacuation warnings, and to safely move to higher ground.
The 1992 Threat to Life Assessment also estimated loss of life for the full PMF occurring
without the dam in place. The 1992 study estimated that, without Pueblo Dam in place, the
full 1991 PMF would place about 3,700 people at risk and cause 138 fatalities. The loss of
life was determined assuming that all of the people at risk get at least 15 minutes of warning.
(The number of people at risk is only for those areas in which people receive 0 to 90 minutes
of warning).
The 1992 Threat to Life Assessment also provided estimates of loss of life for two different
warning scenarios. It was estimated that 13,947 people would be at risk from dam failure of
the dam during the 1991 PMF. This population at risk estimate includes areas that would be
flooded from operational releases. Loss of life was estimated for two different warning
assumptions.
With a dam failure warning initiated when the reservoir water level reaches the design water
surface elevation, 41 fatalities were estimated. This was determined based on 0 fatalities
for the 0 people who have 0 to 15 minutes of warning, 38 fatalities for the 430 people who
have 15 to 90 minutes of warning, and 3 fatalities for the 13,544 people who have more than
90 minutes of warning.
With a dam failure warning initiated when the reservoir water level reaches the dam crest,
310 fatalities were estimated. This was determined based on 24 fatalities for the 47 people
who have 0 to 15 minutes of warning, 286 fatalities for the 12,400 people who have 15 to 90
minutes of warning, and 0 fatalities for the 1,500 people who have more than 90 minutes of
warning.
The issue of incremental loss of life is important in the case of a flood induced failure of
Pueblo Dam. As shown in the 1992 Threat to Life Assessment, loss of life from dam failure
during a PMF can be less than if the same flood occurs without the dam in place. Much
depends on warning assumptions and the responsiveness of people at risk to warnings that
are issued.
21
7) Evaluate uncertainty in various parameters which lead to uncertainties in the life loss
estimates.
Various types of uncertainty can influence loss life estimates. Quantifying uncertainty is
difficult. There is uncertainty associated with various inputs used in this analysis. Factors
contributing to uncertainty include the following:
The rate at which the failures occur and the ultimate size of the failures.
The breach characteristics including breach time and geometry.
The depth and velocity of flooding within the flooded area is uncertain; therefore it is
not known if the flooding would be severe enough to wash buildings off of their
foundations.
The number of people at risk varies as people depart and arrive their residences
throughout the day.
When warnings are initiated
The quanity and quality of warnings that are issued.
Flood wave travel times are subject to considerable variation. Existing dam failure
computer models can yield varying travel times as input parameters are changed.
The proper fatality rates to use.
The loss of life values presented in this study are only estimates. They should be used
cautiously in making decisions regarding risk reduction at Pueblo Dam.
References
1) U.S.B.R., Guidelines for Defining Inundated Areas Downstream from Bureau of
Reclamation Dams, June 1982.
2) Froehlich, David C., Peak Outflow From Breached Embankment Dam, Journal of Water
Resources Planning and Management, Vol. 121, No. 1, January/February, 1995.
3) Walder, Joseph S. and Jim E. OConnor, Methods for predicting peak discharge of floods
caused by failure of natural and constructed earthen dams, Water Resources Research, Vol.
33, No. 10, October 1997.
4) U.S. Department of Agriculture, Weather Bureau, Climatological Data, Colorado Section,
Vol. XXVI, No. 6, June 1921.
22
5) American Society of Civil Engineers, Munn and Savage with discussion by Ridgway,
Hosea, Anderson , Follansbee and Jones, Transactions, Paper No. 1480, The Flood of June
1921, in the Arkansas River, at Pueblo, Colorado, presented at the meeting of October 5,
1921.
6) U.S.G.S., Follansbee and Jones, The Arkansas River Flood of June 3-5, 1921, Water
Supply Paper 487, Washington, D.C., 1922.
7) U.S.G.S., Snipes and others, Floods of June 1965 in Arkansas River Basin, Colorado,
Kansas, and New Mexico, Water Supply Paper 1850-D, Washington, D.C., 1974.
8) U.S. Department of Commerce, Weather Bureau, Storm Data, Volume 7, No. 6, June
1965, Asheville, North Carolina, 1965.
9) U.S. Department of Commerce, Weather Bureau, Climatological Data National
Summary, Volume 16, No. 6, Asheville, North Carolina, 1965.
10) U.S. Army Engineer District, Albuquerque Corps of Engineers, Flood Report, Arkansas
River Basin, Flood of June 1965, Colorado, Kansas & New Mexico, April 1966.
11) Colorado Springs Gazette (newspaper), David Wilson Perishes in Beaver Creek, and
Dave Was Trapped When I Came Out, He Was Gone, Sunday, June 20, 1965.
12) Pueblo Chieftan (newspaper), Pueblo Boy Dies in Flood Clean-Up, Monday, June 21,
1965.
13) La Junta Tribune-Democrat (newspaper), Stories about the flood from other
newspapers, Tuesday, June 29, 1965. (This article apparently first appeared in the Las
Animas Democrat).
14) La Junta Tribune-Democrat (newspaper), Stories about he flood from other
newspapers, Monday, June 28, 1965. (This article apparently first appeared in the Bent
County Democrat).
15) Pueblo Star Journal and Sunday Chieftan (newspaper), Five Persons Drown in Lamar
Area, Sunday, June 20, 1965.
23
16) Pueblo Chieftan (newspaper), 17 Prowers County Persons Still Missing After Flood,
Tuesday, June 22, 1965.
17) La Junta Tribune-Democrat (newspaper), Seven Lives Lost in Valley, Tuesday, June
22, 1965.
24
Pueblo Dam Colorado
Loss of Life Assessment
Study prepared by: ________________________ _________________
Wayne J. Graham Date
Peer reviewed by: ________________________ _________________
James T. Melena Date
Note: Messrs. Graham and Melena signed this signature page on September 10, 2003. The
signature page will be retained in D-8540 files.
25
APPENDIX B
RADAR DATA STUDY

A Radar Climatology oI Extreme RainIall in the Colorado Front

Range oI the Rocky Mountains

Julie Rose Javier
James A.Smith
Mary Lynn Baeck
Matthias Steiner

Department oI Civil and Environmental Engineering
Princeton University
Princeton, New Jersey

Report Submitted to

John England
Bureau oI Reclamation
Denver, Colorado

September 13, 2005

ii
Abstract
Analyses oI the spatial and temporal distribution oI extreme rainIall in the Arkansas River
basin above Pueblo, Colorado are based on analyses oI volume scan reIlectivity observations Irom
the Pueblo and Denver WSR-88D radars Ior the period oI 1995-2003. A storm catalog oI 66
rainIall events during the 9-year study period has been developed. For each oI the 66 events
rainIall Iields have been developed Ior the Arkansas River basin at a spatial scale oI 1 km and time
increment oI 5 minutes. These rainIall Iields are designed Ior use in hydrologic modeling studies
oI extreme Ilood response oI the Arkansas River basin and to directly examine the climatology oI
extreme rainIall in the Arkansas River basin. Climatological analyses oI extreme rainIall are carried
out both Irom an Eulerian perspective, in which distributional aspects oI rainIall at Iixed locations
are examined, and a Lagrangian perspective in which distributional aspects oI rainIall are based on
storm tracking algorithms. OI particular interest is the spatial heterogeneity oI extreme rainIall in
the complex terrain oI the upper Arkansas River basin. Lagrangian analyses are used to
characterize the spatially varying distribution oI storm initiation, storm motion, and storm structure.
Analyses are motivated by problems oI dam saIety in which distributional properties oI extreme
rainIall are oI most interest. Climatological analyses based on the radar rainIall data set indicate a
lack oI spatial coherence in extreme events over the basin, with the upper basin rainIall climatology
exhibiting pronounced contrasts with that oI the lower basin The analyses support the inIerence that
convective rainIall in the upper Arkansas River (above Wellsville) does not contribute to extreme
Ilood response oI the Arkansas River at Pueblo. These conclusions are consistent with Jarrett`s
hypothesis that extreme, Ilood-producing rainIall does not occur in the high elevation regions oI the
Rocky Mountains. The diurnal variation in extreme rainIall is a key element oI Ilood response in
the Arkansas River basin. There is pronounced diurnal variation in warm season rainIall in the
Arkansas River basin, with the pattern oI temporal variation exhibiting pronounced topographic
controls. The temporal maximum in basin-averaged rainIall rate occurs at approximately 2300
UTC. Climatological analyses oI extreme rainIall in the upper Arkansas River basin are examined
relative to the spatial and temporal properties oI two extreme events that occurred in June 1921 and
June 1965. For 10 oI the 66 storm events in the Pubelo storm catalog, a simple transIormation oI
rainIall rate and storm duration is used to provide alternative representations oI the space and time
distribution oI rainIall events emulating the June 1921 storm.
iii

Table oI Contents

Abstract -------------------------------------------------------------------

List oI Figures------------------------------------------------------------

List oI Tables ------------------------------------------------------------

1. Introduction ---------------------------------------------------------

2. 3-4 June 1921 Pueblo Storm -------------------------------------

3. Data and Methodology ---------------------------------------------

4. Storm Catalog -------------------------------------------------------

5. Discussion and Analyses
5.1.Diurnal cycle oI Convective RainIall -----------------------
5.2.Geographical Distribution oI Storm Properties -----------
5.3.Storm Environment -------------------------------------------
5.4.3-4 June 1921 Storm Reconstruction ----------------------
5.5.Depth-Area-Duration Analyses oI the Ten Storms -------
5.6.Hydrometeorological Synthesis -----------------------------

6. Summary and Conclusions -----------------------------------------

7. ReIerences ------------------------------------------------------------

ii

iv

vi

1

2

3

4

5
5
6
7
7
8

8

10

iv
List of Figures
page

1. Topography oI the Arkansas River drainage area above Pueblo, Colorado.
2. Number oI extreme storms per day oI the year Irom 1860-2000. Data Irom the
Colorado Climate Center Database.
3. Storm total rainIall map |in inches| Ior June 3-4, 1921 Ior 12 hours ending at
3:00 A.M. MST oI June 4. Map is Irom HMR 10.
4. Shaded areas are locations oI heavy rainIall Ior the June 3-4, 1921 storm event
at Pueblo, Colorado.
5. Area-averaged rainIall rate time series Ior the July 15, 2003 storm event
6. Storm total accumulation |mm| Ior the July 15, 2003 storm event
7. TITAN algorithm Ior a) storm identiIication; b) storm tracking. The black
solid lines are possible storm paths while the red dash lines are paths which are
too long`. Storm area projections are approximated as ellipses.
8. An example oI how TITAN identiIies and tracks a storm. Shown is an image
oI the July 13, 2000 storm at 20:51 UTC.
9. Diurnal cycle oI rainIall accumulation and lightning Irequency based on the
sixty-six storm events.
10.Geographical distribution oI the aggregate storm properties oI the sixty-six
storm events: a) storm initiation sites; b) storm cell speed and direction; c)
storm activity ||; d) storm total rainIall |mm|.
11.Average number oI storm initiation sites per km
2
as a Iunction oI terrain
elevation. Based on the sixty-six storm events.
12.Average storm activity |expressed as a percent| as a Iunction oI terrain
elevation. Based on the sixty-six storm events.
13.Average rainIall accumulation as a Iunction oI terrain elevation. Based on the
14.Pre-storm wind speed and direction at the a) surIace (taken at the Iirst 500
meters Irom the ground level); b) steering-level (taken to be at the 500 mb
pressure height).
15.RainIall maximization Ior June 26, 1997: a) duration adjustment; b) magnitude
adjustment.
16.RainIall maximization Ior July 13, 2001: a) duration adjustment. In this case, it
was unnecessary as the storm lasted long enough; b) magnitude adjustment.
17.RainIall maximization Ior July 14, 2001: a) duration adjustment. In this case, it
was unnecessary as the storm lasted long enough; b) magnitude adjustment
18.RainIall maximization Ior July 26, 2001: a) duration adjustment; b) magnitude
adjustment.
adjustment.

12

13

14

15
16
16

17

18

19

20-21

22

23

24

25

26

27

28

29

30

v

20.RainIall maximization Ior August 28, 2002: a) duration adjustment; b)
magnitude adjustment
21.RainIall maximization Ior August 29, 2002: a) duration adjustment; b)
magnitude adjustment
adjustment
adjustment
24.RainIall maximization Ior July 15, 2003: a) duration adjustment; b) magnitude
adjustment
25.Depth-Area-Duration curves Ior a) June 26, 1997; b) July 13, 2001; c) June 14,
2001; d) July 26, 2001; e) June 25, 2002; I) August 28, 2001; g) August 29,
2002; h) June 10, 2003; i) June 13, 2003; j) July 15, 2003.

page

31

32

33

34

35

36-40

vi

List of Tables page

1. Summary table oI available MDV scans Ior Denver, Colorado (KFTG).
A day is Irom 0:00 UTC to 23:59 UTC.
2. Summary table oI available MDV scans Ior Pueblo, Colorado (KPUX).
A day is Irom 0:00 UTC to 23:59 UTC.
3. Summary table oI available tracking Ior Denver, Colorado (KFTG). A
day is Irom 16:00 UTC to 15:59 UTC oI the Iollowing day.
4. Summary table oI available tracing Ior Pueblo, Colorado (KPUX). A
day is Irom 16:00 UTC to 15:59 UTC oI the Iollowing day.
5. Summary oI the sixty-six events chosen Ior Iurther analyses. Area 25
(area 50) is the Iraction oI the area (200 km x 200 km domain) with
rainIall accumulations greater than 25 mm (50mm).
6. Summary oI pre-storm sounding variables Ior the sixty-six storm
events. PW is the precipitable water computed at atmospheric
conditions, PW
sat
is the precipitable water computed assuming saturated
conditions, and LCL is the liIting condensation level.
7. Statistical summary oI 4 oI the 6 sounding variables presented in Table
6.

41

42

43

44

45-46

47-48

48

1
1. Introduction
Pueblo dam was constructed by the Bureau oI Reclamation as part oI the Fryingpan-Arkansas
project to divert water Irom the western slopes oI the Continental divide to the east. It is located on
the Arkansas River about six miles west oI Pueblo, CO and drains an upstream area oI 4686 square
miles (Fig 1). Besides providing Ior municipal water supply, irrigation, and recreation, it serves as a
Ilood control Iacility Ior the Pueblo community. Pueblo dam cannot, however, successIully pass
the current Probable Maximum Flood (PMF) without overtopping. To determine whether a
Reclamation Dam SaIety deIiciency exists, a comprehensive hydroclimatological analysis oI
storms that cause extreme Ilood is required.

Annual Ilood peak observations Irom the Arkansas River at Wellsville (1485 sq miles) and
Parkdale (2548 sq mi; approximately 50 river miles downstream oI Wellsville) illustrate key
Ieatures oI the Ilood hydrology oI the upper Arkansas River basin:

1. The seasonal distribution oI Ilood peaks is tightly concentrated around mid-June. Annual
Ilood peaks Irom the 40 year record at Parkdale range in occurrence Irom May 25 to August
4 and reIlect a range oI snowmelt, rainIall, and rain-on-snow Ilood events.

2. Historical snowmelt Ilood peaks are principally determined by the drainage area above
Wellsville. For many events, including the 1980 and 1983 Ilood peaks, the peak discharge
at the upstream station is oI comparable magnitude as the downstream station.

3. Annual Ilood peaks Ior the Arkansas River at Parkdale range in magnitude Irom less than
2000 cIs to slightly more than 6000 cIs. These historical Ilood magnitudes are modest in
comparison with Ilood peaks in smaller basins Irom the Plum Creek (17 June 1965), Big
Thompson (31 July 1976), Rapid City (9 June 1972), BuIIalo Creek (12 July 1996) and Fort
Collins (28 July 1997) events.

Storms that produce extreme rainIall in Colorado are concentrated in the summer months,
peaking between mid-July and mid-August (Fig 2). Low-level winds in the summer are oIten Irom
the southeast and cause warm moist air Irom the GulI oI Mexico to be transported up the Arkansas
River valley. Orographic liIting oI this moist air oIten leads to aIternoon thunderstorms which can
produce heavy rainIall and hail. |Bullard and Leverson, 1991|. A notable Ieature oI these systems is
the diurnal cycle associated with their genesis, growth, and decay.

The area drained by the Arkansas River above Pueblo consists oI a mix oI mountainous regions
with rugged topography and the western portion oI the Great Plains. Elevation ranges Irom about
1400 to 4400 meters above sea level (around 4600 to 14400 Ieet). South oI Colorado Springs, the
Ioothills Iollow a westward direction Ior about twenty miles. Near Canon City, it swings to the
south and southeast and Iollows this course Ior about Iorty miles (see Fig 1). Within this Ilexure is
a triangular tongue oI the plains country halI surrounded by mountain ranges |Gilbert, 1896|.

Analyses oI extreme rainIall and Ilooding in the upper Arkansas River basin are strongly
inIluenced by two extreme rainIall events that caused Iloods in Pueblo, 3-4 June 1921 and 16-17
June 1965. The 3-4 June 1921 Ilood at Pueblo was estimated to have a peak Ilow rate oI more than

2
100,000 cIs (2832 cms). The maximum stage height was recorded at 24.66 It. Damages to
structures and properties were estimated to be more than $19M (1921 values), not to mention the
loss oI lives which was pegged at the oIIicial count oI seventy-eight. The 17 June 1965 Ilood at
Fountain Creek, a tributary oI the Arkansas River, was estimated to have a maximum discharge oI
80,000 cIs |Snipes et al., 1974|.

This study was designed to quantiIy the occurrence oI extreme convective rainIall in the upper
Arkansas River basin Ior the summer months oI June, July and August and to provide the Bureau oI
Reclamation with datasets useIul Ior developing design hydrographs Ior dam saIety assessment. In
line with the second objective, it is oI interest to reconstruct the storm rainIall magnitudes oI the
June 1921 event that will serve as the basis Ior extreme storm probabilities and regional storm
transposition. The 1921 storm and radar storm catalog will then later be used to estimate extreme
storm and Ilood probabilities Ior Pueblo dam. The speciIic objectives oI this study are to answer the
Iollowing questions:

1. What are the characteristic temporal and spatial patterns oI extreme rainIall events in the
upper Arkansas River basin?
2. Can storm initiation, motion and evolution optimize rainIall distribution over the basin by
down-basin south to southeast storm propagation?
3. How are the storm properties controlled by terrain?
4. Can extreme storm rainIall occur above 7500 It?
5. Can extreme rainIall in the lower basin (below Wellsville) signiIicantly ampliIy Ilood peaks
produced by preceding storms above Wellsville?

To address these questions, rainIall data sets were developed Irom weather radar observations and
analyses oI storm initiation, storm structure and storm motion were carried out, with particular
emphasis on their relation to the terrain oI the Colorado Front Range.

2. 3-4 1une1921 Pueblo Storm

The tremendous Ilood oI June 1921 resulted Irom 'cloudbursts in very localized areas
along the Arkansas River. The synoptic scale environment oI these storms Ieatured a low pressure
area over the Southern Rocky Mountain plateau and a high pressure area over Manitoba and the
northern part oI North Dakota. The synoptic environment caused the storm to move primarily Irom
the east or northeast. Owing to the Iunnel shape oI the upper Arkansas valley, air convergence and
mechanical liIting produced strong vertical motions. Thunderstorms and hail reports pointed to the
convective nature oI the storm. The Iollowing description oI the timing oI rainIall and Ilood is
based on the report by Follansbee and Jones |1922| submitted to the U.S. Geological Survey.

As a result oI the storm`s evolution and dynamics, the heaviest rainIall was experienced in
the Arkansas Valley between Colorado Springs and Pueblo. According to bucket surveys oI
accumulation and resident observations, heavy rainIall was concentrated between 3 PM and 12
midnight local time. Throughout the morning oI June 3, storm clouds were visible along the top
and sides oI the Wet Mountain Range and also in the hills southwest oI Pikes Peak. Around 1 PM,
these clouds gradually dropped lower and started raining in the hills. By 3 PM, it had spread over

3
the upper and middle parts oI the valley, and between 5 and 7 PM it reached the lower end near
Pueblo. The hardest rain in Pueblo occurred at around 10 or 11 PM and continued intermittently
until aIter midnight. The isohyetal map Irom HMR 10 |USWB, 1939| Ior 12 hours ending at 3:00
AM oI June 4 shows accumulations as high as 11.84 inches around Pueblo (see Fig 3). Areas oI
heavy precipitation are shown in Fig 4.

Responding to the heavy rain the aIternoon oI the 3
rd
, the Arkansas River at Pueblo began to
rise rapidly at 5 PM. This rise was due to the contribution oI the tributaries within the upper oI the
two areas oI heavy rainIall shown in Fig 4. By 8:45 PM the levees were overtopped at a stage oI
18.1 Ieet Irom the water Ilowing Irom the tributaries near Pueblo. The river continued to rise until
it reached its peak stage oI 24.66 Ieet around midnight.

Pueblo experienced other large Iloods aIter the 1921 event but none as devastating and
widespread. The June 3, 1994 Ilash Ilood resulted in $5M in property damages. Thunderstorms
produced 5 inches oI rain in two hours which Ilooded much oI Pueblo. A peak Ilow rate oI 10,400
cIs was recorded Ior the Arkansas River at Pueblo. A major highway was closed, with over two
Ieet oI water. Another Ilood on August 13, 1994 with reported rainIall accumulations oI 4.87
inches in one and a halI hours produced by heavy thunderstorms. The June 3, 1995 Ilood resulted
in $17M in property damages. Torrential rains and small hail caused localized Ilash Ilooding in and
near the city oI Pueblo.
3. Data and Methodology
Volume scan reIlectivity observations Irom the WSR-88D (Weather Surveillance Radar -
1988 Doppler) radars at Pueblo and Denver were obtained Irom the National Climatic Data Center
(NCDC). The native Iormat oI the reIlectivity data is in polar coordinates with 1 azimuthal
resolution and 1-km range resolution. The data are processed to a Cartesian Iormat at 1 km
horizontal resolution using procedures developed at Princeton and which largely emulate
operational procedures used by the National Weather Service |Baeck and Smith, 1998; Fulton et
al., 1998|. The volume scan time scale ranges Irom 5 to 6 minutes. The Cartesian reIlectivity data
(in 3 dimensions) are stored in the MDV (Meteorological Data Volume) Iormat |Dixon, 1994|.
These MDV Iormat reIlectivity data are a core element oI the data system developed under this
project and delivered to the Bureau oI Reclamation. RainIall estimates are obtained using a Z-R
relationship, a power law equation relating the radar reIlectivity Iactor, Z, to rainIall rate, R. It has
the Iollowing Iorm

Z a R
b

where a and b are constants obtained Irom paired observations oI radar reIlectivity and raindrop
spectra. The constants a and b are thereIore rainIall-type dependent. To allow Ior signiIicant
evaporation which occurs between high cloud bases and the ground, as the case in the Front Range
oI the Rockies, a Kelsch Z-R relationship (Z500 R
1.3
) was used |Dixon, 1994|
.
Storm total accumulation Iields and time series oI rainIall rates were then obtained Ior all
the days in June, July and August oI 1995-2003 with radar data (a complete listing oI the days with
available MDV Iiles are presented in Tables 1 and 2). Because oI the strong diurnal cycle that

4
characterizes the convective activity oI the region, a 'day' is deIined Irom 16 UTC to 16 UTC oI the
Iollowing day (reIer to the Discussion and Analyses Ior more details). For simplicity, storms will
be designated by the beginning days only and not the two days oI coverage (e.g. June 26, 1997
storm instead oI June 26-27, 1997 storm). For rainIall analysis, a 200 km by 200 km area has been
deIined Irom 39.5 to the north to 37.7 to the south and -106.6 Irom the west to -104.3 to the
east. Terrain elevations range Irom 1400 to 4400 meters above sea level.

A storm catalog oI extreme events was developed based on area-averaged time series oI
rainIall rate, storm total accumulation and extent oI precipitation. Examples oI the area-averaged
rainIall rate time series and storm total accumulation Ior the July 15, 2003 storm event are
presented in Figs 5 and 6. From this catalog, sixty-six events were chosen to be Iurther analyzed as
representative oI the variability and heterogeneity oI warm season rainIall in the region oI interest.

Other auxiliary datasets that were used Ior the analyses were upper air radiosonde data Irom
the Denver Stapleton Airport archived by NOAA and cloud-to-ground lightning Irom the National
Lightning Detection Network. The upper air radiosonde data gives the vertical proIile oI
temperature, pressure and wind. Aspects oI the storm environment can be derived Irom the
sounding observations. The lightning data are used to examine the convective signature oI the
storms.

A storm identiIication and tracking algorithm (see Dixon and Wiener, 1993) was used to
perIorm Lagrangian analyses oI storm properties derived Irom volume scan reIlectivity data. Storm
identiIication is perIormed by speciIying a reIlectivity and a volume threshold. A contiguous
region oI reIlectivity above the thresholds is considered a storm cell. Tracking is solved as an
optimization problem where storms at time 1 are matched with their time 2 counterparts (see Figs 7
to 8). Some oI the storm properties that can be obtained Irom the tracking include the storm
initiation sites, the time history oI the storm location, precipitation area and Ilux, and storm cell
speed and direction (see Tables 3 and 4 Ior a listing oI the days with available tracking). This
algorithm has been used successIully by other investigative studies (see Ior example Smith el al.,
1996).

4. Storm Catalog

The development oI a storm catalog oI heavy rainIall events was undertaken to support both
the hydrologic modeling objectives oI the project and to support climatological analyses oI extreme
rainIall in the upper Arkansas River basin. An important issue in the development oI the storm
catalog was the presence oI data gaps in the radar scans. Only storms with complete radar coverage
were considered Ior the catalog. The Iinal sixty-six storm events chosen Ior Iurther analyses were
selected based on the storm total rainIall maps and the area-averaged rainIall rate time series.
Table 5 shows the list oI the sixty-six storm events considered with some oI the storm properties
included. These events represent the variability oI storms that are typically experienced in the
region. The majority oI the storms are convective in nature and exhibit a wide range oI mesoscale
Ieatures. Storm duration ranges Irom 5 to 20 hours. Areal precipitation coverage is usually
concentrated on the plains east oI the Ioothills and this is reIlected in the small values oI the
Iraction in the second column oI Table 5.

5

5. Discussion and Analyses

5.1 Diurnal Cvcle of Convective Activitv

An important Ieature oI the warm season climatology in the Front Range oI the Rockies is
the diurnal cycle associated with the convective activity oI the region. Several authors have studied
this phenomenon |Riley et al., 1987, Baeck and Smith, 1995| in and around Colorado.
Examination oI the area-averaged time series oI rainIall rate Ior the storms during the period oI
radar coverage shows that storms usually develop aIter 16 UTC (10 AM MST) and last eleven
hours on average.

Fig 9 shows the Irequency distribution oI the rainIall accumulation and lightning strikes oI
the sixty-six storm events as a Iunction oI time. Both plots peaked at approximately 2300 UTC
(1700 MST). Karr and Wooten |1976| observed a peak oI radar echoes at 1630-1730 MST around
Limon, Colorado. The convective nature oI the storms is also maniIested by the small delay
between the receding limbs oI both plots pointing to a weak stratiIorm rainIall signal.

5.2 Geographical Distribution of Storm Properties

Figs 10a through 10d show the geographical distribution oI storm properties. These
statistics were obtained by computing the aggregate properties oI the 66 signiIicant storm events.
The storm properties were obtained through the storm tracking and identiIication algorithm. A
reIlectivity threshold oI 35 dBZ and 50 km
3
volume threshold were used.

Fig 10a shows statistics on the spatial distribution oI storm initiation sites. A storm is
considered to create an event` at the grid point closest to its centroid even though the storm spans
several grids. The density oI the events is presented as storm count per km
2
. Hot spots Ior
initiation are along the slopes oI Pikes Peak and the mountains and the triangular junction Iormed
by Pikes Peak and the Wet Mountains. Storms usually originate in the higher elevation terrain
compared to the plains. With the storms originating Irom the mountains, they then propagate
eastward. Fig 10b shows this eastward propagation with the storms gaining speed when they clear
the mountains.

Storm activity is deIined as the number oI hours oI storms over a point divided by the total
number oI stormy hours in the region and is expressed as a percent. The algorithm considers all the
grid points covered by the storm and not just the centroid. This statistic highlights the areas with
the larger storms as they account Ior more time over each grid point. Fig 10c shows that the larger
storms occur more Irequently over the plains than over the higher elevation terrain.

The total rainIall accumulation plot is shown on Fig 10d. As can be seen, there is an area oI
heavy precipitation east oI Pikes Peak. There is an increasing gradient in the accumulation as the
elevation decreases.

To support these results, three oI the storm properties namely initiation, activity and
rainIall accumulation were plotted against elevation. Storm initiation sites usually occur in the

6
1500 to 3000 meters range. These are the elevations Ior the mountain slopes (see Fig 11 and Fig 1).
Storm activity and rainIall accumulation peaked in the lower elevation terrain corresponding to the
plains region (see Fig 12 and 13).

The preceding analyses support Jarrett and Costa`s |1988| Iindings that extreme storm
rainIall rarely occurs, iI at all, above 7500 It (around 2300 m). Their results were based on
precipitation, streamIlow and paleohydrologic studies oI channel Ieatures in the Front Range oI
Colorado. PaleoIlood investigations at elevations oI 7500 It point to insigniIicant Ilooding during
the last 10,000 years. This then points to the lack oI rainIall magnitude signiIicant enough to
produce Ilooding.

5.3 Storm Environment

Key elements oI the storm environment can be assessed through thermodynamic and
dynamic variables computed Irom radiosonde observations. Precipitable water, liIting condensation
level (LCL) and wind speed and direction were computed Ior each oI the 66 storm events.
Precipitable water (PW) is the vertically integrated water vapor density

=
:
w
d: PW
0
(1)

where
w
is the density oI the water vapor in kg m
-3
and : is the altitude in meters. The precipitable
water can be interpreted as the depth oI water at the earth`s surIace that would result iI all the water
vapor in the air column were condensed and converted to precipitation. Precipitable water is
usually expressed as a depth in millimeters (by dividing PW as kg oI water per square meter by the
density oI liquid water, 1000 kg per cubic meter and converting meters to millimeters.). It is a
measure oI how moist` the atmosphere is in a particular region. The liIting condensation level is
the lowest elevation at which a liIted parcel reaches saturation. Computationally, it is obtained by
determining at which level the temperature oI the air parcel equals that oI the environmental
temperature. The height oI the LCL will give inIormation on the storm initiation sties with relation
to the terrain.

Table 6 gives the summary oI the pre-storm atmospheric variables computed Irom the upper
sounding observations obtained Irom Denver Stapleton Airport Ior the sixty-six storm events. PW
is the precipitable water computed at the prevailing atmospheric conditions while PW
sat
is the
precipitable water computed assuming saturation conditions. The ratio oI this two variables,
PW/PW
sat
, indicates how close to saturation the ambient conditions were. Also given in the table is
the height oI the LCL and the surIace wind and steering-level wind speed and direction. The
surIace wind was taken to be the average oI the Iirst 500 meters above ground level and the
steering-level wind was taken at the 500 mb pressure height, which was the average value between
the 450-550 mb pressure height. The direction oI the surIace wind is an indication oI the source oI
moisture while the steering-level wind is the direction storms will move iI they are advected by the
upper level wind.

Table 7 gives the statistical summary Ior the precipitable water and the location oI the LCL.
The precipitable water and LCL values are not extreme Ior the 66 events. Comparing these results
with those oI the Colorado Extreme Storm Precipitation Data Study |McKee and Dosken, 1997|

7
showed that the PW and LCL usually were between the 50th and 75th percentile probability oI non-
exceedance. Only a Iew storms exceeded the 95th percentile. The PW/PW
sat
ratio indicates that on
average, these storms have moisture contents oI about 50 to 60 percent oI their saturation value.
The average height oI the LCL at 2700 meters also supported the Iindings earlier that the storms
usually initiate on mountain slopes. Fig 14 shows the direction and the magnitude oI the wind at
the surIace and at the steering-level. The surIace wind usually has a south/southeast origin with
slow to moderate speed. The steering-level wind on the other hand is predominantly easterly with
wind speeds exceeding 10 m/s.

5.4 3-4 June 1921 Storm Reconstruction

Ten storms Irom the storm catalog were chosen Ior rainIall maximization. The choice was
based on the storms` rainIall areal distribution and accumulation relative to the 1921 event (reIer to
Fig 3 Ior the 12-hour storm total map Ior the June 3-4 storm, ending at 3 A.M. MST). The map
shows the concentration oI heavy rainIall in and around Pueblo, with accumulations reaching as
high as 12 inches in some areas. Examination oI the bucket survey reports revealed that the heavy
rainIall in and around Pueblo was concentrated between 3 PM and 12 midnight local time oI June
3. It was around midnight that the devastating Ilood peak in Pueblo occurred.

It is desired to recreate the rainIall magnitudes oI the 1921 storm in the ten storm analogs as
a basis Ior later extreme storm probability studies and storm transposition as was mentioned earlier.
Two Iactors are believed to have contributed to the excessive rainIall accumulation, the slow
movement oI the storm and the intense rainIall rates Irom a series oI cloud-bursts that resulted Irom
heavy rain-bearing clouds striking against the mountains and being deIlected upward |Follansbee
and Jones, 1922|. The slow movement oI the storm was reproduced by making the storm analogs
have the same duration as the 1921 event. For eight oI the ten storms, this meant having their
duration extended. This was done by Iirst deIining a 75 km x 75 km (5625 km
2
) domain that
covers the area oI heavy rainIall Ior the 1921 storm (reIer to Fig 4). This domain is deIined by a
square whose southwest coordinates are -105.39 W and 38.00 N. The area-averaged rainIall rate
time series was obtained Ior each oI the analogs. The timing oI the radar scans were then altered so
that the storms span 9 hours, the same as the 1921 storm. These results are shown in Fig 15a, 16a,
to 24a. For some oI the storms like the July 15, 2003 their durations were extended by a Iactor oI
three or more.

The resulting storm total rainIall maps (not shown) are Iurther adjusted to match that oI the
1921 accumulation. An average rainIall depth oI 33 mm Ior 1921 was obtained Ior approximately
the same area as the 5625 km
2
domain. This value was then compared with the average rainIall
depth oI the 5625 km
2
domain Ior the ten storms. The ratio oI the two was then used as a
multiplicative adjustment Iactor Ior the ten storms. This Iactor ranged Irom about 1 to 5.10. These
resulting storm total maps are shown in Figs 15b, 16b to 24b. See also Fig 6 and Fig 24b to see the
result oI the maximization procedure on the July 15, 2003 event.

5.5 Depth-Area-Duration Analvses of the Ten Storms

The maximized rainIall accumulations Ior the ten storms were then used to come up with
depth-area-duration curves Ior 1, 3, 6, 9-hour accumulations Irom 25 to 1089 km
2
. A moving grid

8
oI 5 km by 5 km up to 33 km x 33 km domain in increments oI 1 km was used. The dataset was
originally in 1km x 1km spatial resolution. The maximum n-hour accumulation was then
determined. The results are shown in Fig 25a to Fig 25j. The spatial heterogeneity oI rainIall
between storms is evident in these curves. Although the average rainIall depths were maximized to
match that oI the 1921, the areal averages Ior a speciIic duration diIIered Irom storm to storm. The
1-hour accumulation at 25 km
2
ranged Irom 56 mm to 260 mm, almost a diIIerence oI a Iactor oI
Iive.

5.6 Hvdrometeorological Svnthesis

Storm based precipitation is the dominant Iorcing in extreme Iloods. RainIall-runoII models
rely heavily on the spatial and temporal distribution oI rainIall. Values oI extreme storm
production and associated probabilities along with the ability to characterize them in time and space
are needed in order to generate inputs to rainIall-runoII models. Proposed methods oI obtaining
these values involve storm rainIall maximization and storm transposition. A rainIall depth
maximization procedure was presented and had been done on 10 oI the storms in the catalog that
serve as analogs to the June 1921 storm. The spatial and temporal distribution oI extreme rainIall is
also presented together with how the storm properties are controlled by terrain. The heterogeneity
oI storms that occur in upper Arkansas River valley is evident in the storm catalog.

The Bureau oI Reclamation is currently looking into stochastic storm transposition (SST)
procedures. Storm transposition is a regionalization concept that involves moving storms within an
area to the watershed oI interest. The assumption here is that there exist meteorologically
homogeneous regions such that a major storm occurring somewhere in the region could occur
anywhere else in the region, with the provision that there may be diIIerences in the averaged depth
oI rainIall produced based upon diIIerences in moisture potential |NRC, 1988|. Stochastic storm
transposition extends this concept by incorporating the probability oI storm occurrence |Fontaine
and Potter, 1989|. The processed WSR-88D radar data will be the primary source oI inIormation
Ior the SST modeling. SST model needs: 1) storm rainIall distribution in space; 2) storm center
occurrence in space. Results Irom the radar analyses can be used to deIine a maximum area to
restrict averaged depth relationship. Results Irom the TITAN storm identiIication and tracking can
be used to put a limit on the storm center (x,y) location.

6. Summary and Conclusions
Extreme rainIall in the upper Arkansas River basin above the Pueblo Dam was analyzed
through analyses oI radar rainIall observations during the period 1995 to 2003. Analyses included
development oI high-resolution (1 km horizontal resolution, 5 minutes time resolution) rainIall data
sets and climatological analyses oI heavy rainIall over the upper Arkansas River basin.

Data sets that were delivered to the Bureau oI Reclamation either by submittal or electronic
download included 1) MDV and tracking Iiles Ior 1995 to 2003 Ior the months oI June, July, and
August, 2) 5-minute rainIall rate time series Ior each 1 km by 1 km bin in the 200 km by 200 km
domain, 3) storm total rainIall, lightning Irequency, area-averaged rainIall rate time series, and
sounding variables Ior the sixty-six storm events, 4) maximized storm total rainIall oI the ten 1921
storm analogs, and 5) depth-area-duration analyses oI the ten 1921 storm analogs.

9

A point-by-point examination oI the Iive questions posed in the Introduction Iollows.

1. What are the characteristic temporal and spatial patterns of extreme rainfall events in the
upper Arkansas River basin?
Extreme rainIall events in the Upper Arkansas River basin are driven by a strong diurnal cycle.
Storms usually initiate around 16 UTC and last eleven hours on average. The temporal maximum
oI rainIall accumulation occurs at about 23 UTC. The geographical distribution oI storm total
rainIall accumulation shows an increasing gradient oI heavy rainIall with decreasing terrain
elevation. An area oI maximum rain is observed on the plains east oI Pikes Peak.

2. How are the storm properties controlled bv terrain?
Storm initiation and propagation is tied to the terrain oI the Colorado Front Range. Advection
oI moisture by southerly/southeasterly winds results in orographic liIting and the start oI convective
activity. The triangular junction Iormed by the Wet Mountains and the Ioothills south oI Colorado
Springs is where storms usually initiate. These storms then propagate eastward towards the plains
where they grow and mature.

3. Can storm initiation, motion and evolution optimi:e rainfall distribution over the basin bv
down-basin south to southeast storm propagation?
As shown by the geographical distribution oI storm speed and direction, storm propagation is
Irom west to east. Because oI the storms` dynamics and evolution, heavy rainIall is distributed
mainly on the lower basin.

4. Can extreme storm rainfall occur above 7500 ft?
The upper portion oI the Arkansas River basin (above Wellsville) rarely experiences extreme
rainIall as evidenced by the geographical distribution oI storm total rainIall and storm activity.
Heavy rains and extreme storm occurrence are concentrated on the plains east oI the mountains.

5. Can extreme rainfall in the lower basin (below Wellsville) significantlv amplifv flood peaks
produced bv preceding storms above Wellsville?
As discussed above, the upper portion oI the basin (above Wellsville) rarely contributes to
extreme Ilooding because oI the dearth oI extreme rainIall occurring at those elevations.

The reconstruction oI the June 3-4, 1921 storm accumulation was accomplished two-Iold.
First, the duration oI eight oI the ten storm analogs were extended to match that oI the 1921 event.
Second, a magnitude adjustment Iactor was computed by looking at the ratio oI each oI the storms`
average rainIall depth over a speciIied domain to that oI the 1921`s. A depth-area-duration curve
was obtained Ior each storm Ior 1-, 3-, 6-, and 9-hour accumulation Ior areas Irom 25 to 1089 km
2
.
The curves highlight the temporal and spatial heterogeneity oI rainIall between storms.

10

References

Baeck, M.L., and J.A. Smith, Climatological analysis oI manually digitized radar data Ior the
United States, Water Resources Research, 31(12), 3033-3049, 1995.

Baeck, M.L. and J.A. Smith, Estimation oI heavy rainIall by the WSR-88D, Weather and
Forecasting, 13, 416-436, 1998.

Bullard, K.L. and V. Leverson, Pueblo Dam, Fryingpan-Arkansas Project Probable Maximum
Flood (PMF) Study. Flood Section, Bureau oI Reclamation, Denver, CO, 23 pp. and enclosures,
June 1991.

Dixon, M and G. Wiener, TITAN: Thunderstorm identiIication, tracking, analysis, and nowcasting
a radar-based methodology, J. oI Atmos. and Oceanic Tech., 10(6), 785-797, 1993.

Dixon, M.J. Automated Storm IdentiIication, Tracking and Forecasting A Radar-Based Method,
Doctoral Thesis, University oI Colorado and National Center Ior Atmospheric Research, 1994.

Follansbee, R. and E.E. Jones, Arkansas River Ilood oI June 3-5, 1921. U.S. Geological Survey
Water-Supply Paper 487, 44 pp., 1922.

Fontaine, T.A. and K.W. Potter, Estimating probabilities oI extreme rainIalls. J. Hydraulic Engr.,
ASCE, 115(11), 1562-1575, 1989.

Fulton, R.A., J.P. Breidenbach, D.J.Seo, and D.A. Miller, The WSR-88D rainIall algorithm,
Weather and Forecasting, 13(2), 377-395, 1998.

Gilbert, G.K., Underground water oI the Arkansas Valley in eastern Colorado: U.S. Geological
Survey Seventeenth Annual Rep., pt. 2, p.558, 1896.

Jarrett, R.D. and J.E. Costa, Evaluation oI the Ilood hydrology oI the Colorado Front Range using
precipitation, streamIlow and paleoIlood data, U.S. Geological Survey Water Resources. Inv. Rep,
87-4117, 37 pp., 1988.

Karr, T.W. and R.L. Wooten, Summer radar echo distribution around Limon, Colorado, Monthly
Weather Review, 104, 728-734, 1976.

McKee, T.B. and Doesken, N.J., Colorado Extreme Storm Precipitation Data Study, Climatology
Report #97-1, Department oI Atmospheric Science, Colorado State University, Fort Collins,
Colorado, 34 pp. and appendices, 1997.

National Research Council (NRC), Estimating probabilities oI extreme Iloods: Methods and
recommended research. National Academy Press, Washington, D.C., 141 pp., 1988.

11
Riley, G.T., M.G. Landin, and L.F. Bosart, The diurnal variability oI precipitation across the central
Rockies and adjacent Great Plains. Monthly Weather Review, 115, 1161-1172, 1987.

Smith, J.A., M.L.Baeck, M.Steiner, and A.J. Miller, Catastrophic rainIall Irom an upslope
thunderstorm in the Central Appalachians: the Rapidan Storm oI June 27, 1995, Water Resources
Research, 32(10), 3099-3113, 1996.

Snipes, R.J. et al., Floods oI June 1965 in the Arkansas River Basin, Colorado. U.S. Geological
Survey Water-Supply Paper 1850-D, 97 p., 1974.

United States Weather Bureau (USWB), Maximum Possible RainIall Over the Arkansas River
Basin Above Caddoa, Colorado. Hydrometeorological Report No. 10 (HMR 10),
Hydrometeorological Section, River and Flood Division, U.S. Weather Bureau, 52 p., 10 Iigs. and
appendices, with supplement, 1939.

.

12

Fig 1. Topography oI the Arkansas River drainage area above Pueblo, Colorado

13

0
1
2
3
4
5
6
7
8
9
10
123456789 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 123456789 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 123456789 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 123456789 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 123456789 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 123456789 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 123456789 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 123456789 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 123456789 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 123456789 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 123456789 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 123456789 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
N
u
m
b
e
r

o
f

O
c
c
u
r
r
e
n
c
e
s

Fig 2. Number oI extreme storms per day oI the year Irom 1860-2000. Data Irom the Colorado
Climate Center Database

14

Fig 3. Storm total rainIall map |in inches| Ior June 3-4, 1921 Ior 12 hours ending at 3:00 A.M.
MST oI June 4. Map is Irom HMR 10.

15

Fig 4. Shaded areas are locations oI heavy rainIall Ior the June 3-4, 1921 storm event at
Pueblo, Colorado.

16

Fig 5. Area-averaged rainIall rate time series Ior the July 15, 2003 storm event.

Fig 6. Storm total accumulation |mm| Ior the July 15, 2003 storm event

17

Storms at time 1

Storms at time 2

Fig 7. TITAN algorithm Ior a) storm identiIication; b) storm tracking. The black solid
lines are possible storm paths while the red dash lines are paths which are too long`.
Storm area projections are approximated as ellipses.
6 6 + +
7 7
3
6
3 3 3 5
2 2 2 2
1 1 1 1 1
Storm 1
Storm 2
a)
b)
6 6 + +
7 7
3
6
3 3 3 5
2 2 2 2
1 1 1 1 1
Storm 1
Storm 2
6 6 + +
7 7
3
6
3 3 3 5
2 2 2 2
1 1 1 1 1
Storm 1
Storm 2
6 6 + +
7 7
3
6
3 3 3 5
2 2 2 2
1 1 1 1 1
6 6 + +
7 7
3
6
3 3 3 5
2 2 2 2
1 1 1 1 1
Storm 1
Storm 2
a)
b)

18

Fig 8. An example oI how TITAN identiIies and tracks a storm. Shown is an image oI
the July 13, 2000 storm at 20:51 UTC

19

0
5
10
15
20
25
30
35
17 18 19 20 21 22 23 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
UTC hour
R
a
i
n
f
a
I
I

a
c
c
u
m
u
I
a
t
i
o
n
,

m
m
0
5
10
15
20
25
30
35
L
i
g
h
t
n
i
n
g

f
r
e
q
u
e
n
c
y
,

%
Rainfall
Lightning

Fig 9. Diurnal cycle oI rainIall accumulation and lightning Irequency based on the sixty-
six storm events.

20

a)

b)

Fig 10. Geographical distribution oI the aggregate storm properties oI the sixty-six storm
events: a) storm initiation sites; b) storm cell speed and direction.

21

c)

d)

Fig 10. Geographical distribution oI the aggregate storm properties oI the sixty-six
events c) storm activity ||; d) storm total rainIall |mm|.

22

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
1000-1500
1500-2000
2000-2500
2500-3000
3000-3500
3500-4000
4000-4500
E
I
e
v
a
t
i
o
n

[
m
]
Avg number of initiation sites per sq km

Fig 11. Average number oI storm initiation sites per km
2
as Iunction oI terrain elevation.
Based on the sixty-six storm events

23

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
1000-1500
1500-2000
2000-2500
2500-3000
3000-3500
3500-4000
4000-4500
E
I
e
v
a
t
i
o
n

[
m
]
Avg storm activity [%]

Fig 12. Average storm activity |expressed as a percent| as Iunction oI terrain elevation.
Based on the sixty-six storm events.

24

0 50 100 150 200 250 300 350 400 450
1000-1500
1500-2000
2000-2500
2500-3000
3000-3500
3500-4000
4000-4500
E
I
e
v
a
t
i
o
n

[
m
]
Avg rainfaII accumuIation [mm]

Fig 13. Average rainIall accumulation as a Iunction oI terrain elevation. Based on the

25

a)

b)

Fig 14. Pre-storm wind speed and direction at the a) surIace (taken at the Iirst 500 meters
Irom the ground level); b) steering-level (taken to be at the 500 mb pressure height)

26
0.0
1.0
2.0
3.0
4.0
5.0
0 100 200 300 400 500
Minute
R
a
i
n
f
a
I
I

r
a
t
e
6/26/1997
adj

a)

b)

Fig 15. RainIall maximization Ior June 26, 1997: a) duration adjustment; b) magnitude
adjustment.

27
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
0 60 120 180 240 300 360 420 480 540 600 660 720
Minute
R
a
i
n
f
a
I
I

r
a
t
e

[
m
m
/
h
r
]
7/13/2001
adj

a)

b)

Fig 16. RainIall maximization Ior July 13, 2001: a) duration adjustment. In this case it

28
0
1
2
3
4
5
6
0 100 200 300 400 500
Minute
R
a
i
n
f
a
I
I

r
a
t
e

[
m
m
/
h
r
] 7/14/2001
adj

a)

b)

Fig 17. RainIall maximization Ior July 14, 2001: a) duration adjustment. In this case, it

29

0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
0 100 200 300 400 500
Minute
R
a
i
n
f
a
I
I

r
a
t
e

[
m
m
/
h
r
]
7/26/2001
adj

a)

b)

Fig 18. RainIall maximization Ior July 26, 2001: a) duration adjustment; b) magnitude
adjustment.

30

0.0
0.5
1.0
1.5
2.0
2.5
0 100 200 300 400 500
Minute
R
a
i
n
f
a
I
I

r
a
t
e

[
[
m
m
/
h
r
]
6/25/2002
adj

a)

b)

adjustment

31

0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
0 100 200 300 400 500
Minute
R
a
i
n
f
a
I
I

r
a
t
e

[
m
m
/
h
r
]
8/28/2002
adj

a)

b)

Fig 20. RainIall maximization Ior August 28, 2002: a) duration adjustment; b) magnitude
adjustment

32

0.0
1.0
2.0
3.0
4.0
0 100 200 300 400 500
Minute
R
a
i
n
f
a
I
I

r
a
t
e

[
m
m
/
h
r
]
8/29/2002
adj

a)

b)

Fig 21. RainIall maximization Ior August 29, 2002: a) duration adjustment; b) magnitude
adjustment

33
0.0
1.0
2.0
3.0
4.0
5.0
6.0
0 100 200 300 400 500
Minute
R
a
i
n
f
a
I
I

r
a
t
e

[
m
m
/
h
r
]
6/10/2003
adj

a)

b)

adjustment

34

0
1
2
3
4
0 100 200 300 400 500
Minute
R
a
i
n
f
a
I
I

r
a
t
e

[
m
m
/
h
r
]
6/13/2003
adj

a)

b)

adjustment

35
0.0
2.0
4.0
6.0
8.0
10.0
12.0
0 100 200 300 400 500
Minute
R
a
i
n
f
a
I
I

r
a
t
e

[
m
m
/
h
r
]
7/15/2003
adj

a)

b)

Fig 24. RainIall maximization Ior July 15, 2003: a) duration adjustment; b) magnitude adjustment

36
0
50
100
150
200
250
300
0 200 400 600 800 1000 1200
Area [sq km]
R
a
i
n
f
a
I
I

a
c
c
u
m
u
I
a
t
i
o
n

[
m
m
]
1-hr
3-hr
6-hr
9-hr

a)

0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200
Area [sq km]
R
a
i
n
f
a
I
I

a
c
c
u
m
u
I
a
t
i
o
n

[
m
m
]
1-hr
3-hr
6-hr
9-hr

b)

Fig 25. Depth-Area-Duration curves Ior a) June 26, 1997; b) July 13, 2001.

37
0
20
40
60
80
100
120
140
0 200 400 600 800 1000 1200
Area [sq km]
R
a
i
n
f
a
I
I

a
c
c
u
m
u
I
a
t
i
o
n

[
m
m
]
1-hr
3-hr
6-hr
9-hr

c)
0
50
100
150
200
250
0 200 400 600 800 1000 1200
Area [sq km]
R
a
i
n
f
a
I
I

a
c
c
u
m
u
I
a
t
i
o
n

[
m
m
]
1-hr
3-hr
6-hr
9-hr

d)

Fig 25. Depth-Area-Duration curves Ior c) June 14, 2001; d) July 26, 2001.

38

0
50
100
150
200
250
300
350
400
450
0 200 400 600 800 1000 1200
Area [sq km]
R
a
i
n
f
a
I
I

a
c
c
u
m
u
I
a
t
i
o
n

[
m
m
]
1-hr
3-hr
6-hr
9-hr

e)

0
20
40
60
80
100
120
140
160
180
0 200 400 600 800 1000 1200
Area [sq km]
R
a
i
n
f
a
I
I

a
c
c
u
m
u
I
a
t
i
o
n

[
m
m
]
1-hr
3-hr
6-hr
9-hr

I)

Fig 25. Depth-Area-Duration curves Ior e) June 25, 2002; I) August 28, 2001.

39

0
50
100
150
200
250
300
0 200 400 600 800 1000 1200
Area [sq km]
R
a
i
n
f
a
I
I

a
c
c
u
m
u
I
a
t
i
o
n

[
m
m
]
1-hr
3-hr
6-hr
9-hr

g)

0
20
40
60
80
100
120
140
160
180
200
0 200 400 600 800 1000 1200
Area [sq km]
R
a
i
n
f
a
I
I

a
c
c
u
m
u
I
a
t
i
o
n

[
m
m
]
1-hr
3-hr
6-hr
9-hr

h)

Fig 25. Depth-Area-Duration curves Ior g) August 29, 2002; h) June 10, 2003.

40

0
10
20
30
40
50
60
70
80
90
0 200 400 600 800 1000 1200
Area [sq km]
R
a
i
n
f
a
I
I

a
c
c
u
m
u
I
a
t
i
o
n

[
m
m
]
1-hr
3-hr
6-hr
9-hr

i)

0
20
40
60
80
100
120
140
160
180
0 200 400 600 800 1000 1200
Area [sq km]
R
a
i
n
f
a
I
I

a
c
c
u
m
u
I
a
t
i
o
n

[
m
m
]
1-hr
3-hr
6-hr
9-hr

j)

Fig 25. Depth-Area-Duration curves Ior i) June 13, 2003; j) July 15, 2003.

41
Denver (KFTG) Total
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 davs
1995 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 30
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 29
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * 28
1996 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * 26
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * * * 28
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 31
1997 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 29
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 29
AUG * * * * * * * * * * * * * * * * * * * * * * 22
1998 JUN * * * * * * * * * * * * * * * 15
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * * 27
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * 28
1999 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 29
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * 26
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * 27
2000 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * * 27
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * * 27
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * 28
2001 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * * 27
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * * 27
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * 28
2002 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * 26
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * * 27
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 29
2003 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 29
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 30
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 30
Davs

* denotes MDV scan available Ior the day
Table 1.Summary table oI available MDV scans Ior Denver, Colorado (KFTG). A day is Irom 0:00 UTC to 23:59 UTC

42
Pueblo (KPUX) Total
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 davs
1995 JUN * * * * * * * * * * * * * * * 15
JUL * * * * * * * * * * * * * * * * * * * * * 21
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 31
1996 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 30
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 31
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 29
1997 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * * 27
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 30
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 31
1998 JUN * * * * * * * * * * * * * * * * * * * * * * 22
JUL * * * * * * * * * * * * * * * * * * * * * * * * 24
AUG * * * * * * * * * * * * * * * * * * * * * * * * 24
1999 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * * * 28
JUL * * * * * * * * * * * * * * * * * * * * * * * 23
AUG * * * * * * * * * * * * * * 14
2000 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * * * 28
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 30
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 30
2001 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * * * 28
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 29
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 29
2002 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 30
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * * * 28
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 29
2003 JUN * * * * * * * * * * * * * * * * * * * * * * * * 24
JUL * * * * * * * * * * * * * * * * * * * * * 21
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 31
Davs

* denotes MDV scan available Ior the day
Table 2.Summary table oI available MDV scans Ior Pueblo, Colorado (KPUX). A day is Irom 0:00 UTC to 23:59 UTC

43
Denver (KFTG) Total
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 davs
1995 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 29
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * * 27
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * 26
1996 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * * 27
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * 26
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 29
1997 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * * 27
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * * 27
AUG * * * * * * * * * * * * * * * * * * * * * 21
1998 JUN * * * * * * * * * * * * 12
JUL * * * * * * * * * * * * * * * * * * * * * * * * * 25
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * 26
1999 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * 26
JUL * * * * * * * * * * * * * * * * * * * * * * * * 24
AUG * * * * * * * * * * * * * * * * * * * * * * * * * 25
2000 JUN * * * * * * * * * * * * * * * * * * * * * * * * 24
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * 26
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * 26
2001 JUN * * * * * * * * * * * * * * * * * * * * * * * 23
JUL * * * * * * * * * * * * * * * * * * * * * * * * * 25
AUG * * * * * * * * * * * * * * * * * * * * * * * * * 25
2002 JUN * * * * * * * * * * * * * * * * * * * * * * 22
JUL * * * * * * * * * * * * * * * * * * * * * * * 23
AUG * * * * * * * * * * * * * * * * * * * * * * * * * 25
2003 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * * 27
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * * * 28
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * 28
Davs

* denotes tracking available Ior the day
Table 3. .Summary table oI available tracking Ior Denver, Colorado (KFTG). A day is Irom 16:00 UTC to 15:59 UTC oI the Iollowing day

44
Pueblo (KPUX) Total
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 davs
1995 JUN * * * * * * * * * * * * * * 14
JUL * * * * * * * * * * * * * * * * * * * 19
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 30
1996 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * * 27
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 30
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * 28
1997 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * 26
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 30
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 31
1998 JUN * * * * * * * * * * * * * * * * * * 18
JUL * * * * * * * * * * * * * * * * * * * * * * 22
AUG * * * * * * * * * * * * * * * * * * * * * 21
1999 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * * 27
JUL * * * * * * * * * * * * * * * * * * * * * 21
AUG * * * * * * * * * * * 11
2000 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * * 27
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * * * 28
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 30
2001 JUN * * * * * * * * * * * * * * * * * * * * * * * * * 25
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * 26
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * 28
2002 JUN * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 30
JUL * * * * * * * * * * * * * * * * * * * * * * * * * * * 27
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * 27
2003 JUN * * * * * * * * * * * * * * * * * * * * * * * 23
JUL * * * * * * * * * * * * * * * * * * * 19
AUG * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 31
Davs

* denotes tracking available Ior the day
Table 3. .Summary table oI available tracking Ior Pueblo, Colorado (KPUX). A day is Irom 16:00 UTC to 15:59 UTC oI the Iollowing day

45

Event Area 25, mm Area50, mm Mean rainIall
rate, mm/hr
Mean
accumulation,
mm
Duration, hrs

1995-08-04 0.0056 0.0017 1.96 0.0004 12
1996-06-12 0.0318 0.0014 3.37 0.0004 9
1996-06-14 0.0270 0.0048 3.46 0.0004 15
1996-07-24 0.0046 0.0001 1.05 0.0004 12
1996-08-08 0.0086 0.0013 1.77 0.0007 11
1997-06-13 0.0287 0.0140 2.78 0.0007 17
1997-06-26 0.0036 0.0001 1.19 0.0003 7
1997-07-27 0.0269 0.0016 4.23 0.0007 11
1997-07-29 0.0675 0.0158 5.71 0.0007 16
1997-08-10 0.0044 0.0004 2.53 0.0006 15
1998-06-14 0.0009 0.0002 2.24 0.0006 16
1998-06-29 0.0237 0.0021 2.03 0.0005 12
1998-07-03 0.0211 0.0026 2.21 0.0005 9
1998-08-09 0.0268 0.0036 3.55 0.0005 13
1998-08-11 0.0363 0.0023 3.90 0.0006 10
1998-08-19 0.0220 0.0038 2.59 0.0004 15
1998-08-25 0.0199 0.0032 3.71 0.0005 12
1999-08-28 0.0042 0.0004 1.70 0.0006 18
1999-08-29 0.0064 0.0016 1.29 0.0006 14
2000-06-05 0.0119 0.0010 1.52 0.0003 7
2000-06-27 0.0002 0.0000 1.39 0.0006 19
2000-07-02 0.0002 0.0000 1.06 0.0006 14
2000-07-20 0.0092 0.0002 1.61 0.0007 8
2000-08-06 0.0084 0.0004 1.83 0.0004 13
2000-08-20 0.0002 0.0000 1.57 0.0006 8
2000-08-21 0.0116 0.0001 4.77 0.0006 15
2000-08-26 0.0031 0.0000 2.53 0.0006 8
2001-06-04 0.0078 0.0000 1.27 0.0005 10
2001-06-08 0.0104 0.0001 1.83 0.0004 12
2001-06-22 0.0012 0.0000 1.61 0.0004 10
2001-07-09 0.0507 0.0065 5.78 0.0006 13
2001-07-10 0.0087 0.0000 2.96 0.0005 9
2001-07-13 0.0969 0.0139 8.62 0.0007 13
2001-07-14 0.0204 0.0003 3.74 0.0004 10
2001-07-23 0.0112 0.0003 3.33 0.0006 18
2001-07-25 0.0128 0.0002 2.72 0.0007 11
2001-07-26 0.0008 0.0000 2.16 0.0003 5
2001-08-11 0.0022 0.0000 1.58 0.0007 12
2001-08-14 0.0007 0.0001 2.56 0.0006 20
2001-08-16 0.0044 0.0001 2.44 0.0006 7
2002-06-14 0.0090 0.0001 1.98 0.0005 9
2002-06-19 0.0163 0.0024 1.87 0.0006 7
2002-06-25 0.0073 0.0000 1.33 0.0006 7
2002-06-27 0.0116 0.0000 2.38 0.0004 9
2002-06-28 0.0023 0.0000 1.65 0.0004 9
2002-07-09 0.0403 0.0035 5.33 0.0004 12
2002-07-10 0.0319 0.0035 4.21 0.0007 11
2002-07-26 0.0004 0.0000 0.91 0.0007 6
2002-07-29 0.0051 0.0001 0.85 0.0003 10
2002-08-02 0.0000 0.0000 0.41 0.0006 6
2002-08-05 0.0137 0.0013 2.74 0.0006 19
2002-08-21 0.0005 0.0000 1.81 0.0006 14
2002-08-28 0.0062 0.0000 2.92 0.0007 7
2002-08-29 0.0119 0.0001 3.26 0.0007 7
2003-06-10 0.0088 0.0000 3.26 0.0006 7
2003-06-13 0.0039 0.0000 2.50 0.0007 7

46

Event Area 25, mm Area50, mm Mean rainIall
rate, mm/hr
Mean
accumulation,
mm
Duration, hrs

2003-06-16 0.0146 0.0001 2.84 0.0007 10
2003-06-17 0.0361 0.0009 4.47 0.0006 10
2003-06-28 0.0146 0.0023 2.81 0.0007 8
2003-07-15 0.0327 0.0016 3.17 0.0005 7
2003-07-19 0.0685 0.0086 6.11 0.0007 9
2003-07-28 0.0383 0.0061 4.33 0.0007 11
2003-07-29 0.0042 0.0000 2.84 0.0007 12
2003-08-09 0.0132 0.0001 2.66 0.0007 8
2003-08-11 0.0293 0.0054 3.77 0.0007 13
2003-08-23 0.0099 0.0002 2.87 0.0007 19

Table 5. Summary oI the sixty-six events chosen Ior Iurther analyses. Area 25 (area 50) is the
Iraction oI the area (200 km x 200 km domain) with rainIall accumulations greater than 25 mm (50
mm).

47

Event PW, mm PW
sat
, mm PW/PW
sat
LCL, mb Wind spd, m/s Wind dir, (0
o

due N)
sIc 500mb sIc 500mb

1995-08-04 20 47 0.43 652 6.3 10.8 211 265
1996-06-12 17 35 0.49 694 4.1 10.1 338 271
1996-06-14 20 38 0.52 711 1.6 1.4 200 224
1996-07-24 19 37 0.51 733 3.3 13.0 345 310
1996-08-08 18 39 0.46 752 4.9 1.6 197 71
1997-06-13 18 28 0.64 747 5.7 1.1 179 227
1997-06-26 29 46 0.63 696 3.5 6.5 190 213
1997-07-27 32 43 0.75 784 5.5 5.1 207 217
1997-07-29 27 36 0.76 790 2.8 15.7 227 245
1997-08-10 15 29 0.52 696 9.8 13.7 283 320
1998-06-14 23 43 0.53 786 1.8 16.7 242 268
1998-06-29 23 47 0.49 734 6.4 3.8 10 235
1998-07-03 26 40 0.65 770 2.8 9.8 206 257
1998-08-09 25 45 0.56 767 4.3 6.0 312 277
1998-08-11 25 45 0.56 781 2.2 11.2 250 204
1998-08-19 29 44 0.67 826 4.6 8.5 232 240
1998-08-25 19 44 0.44 758 2.9 10.1 206 220
1999-08-28 13 34 0.38 770 2.2 9.0 195 253
1999-08-29 25 29 0.86 829 4.6 12.6 317 305
2000-06-05 18 48 0.37 694 2.8 14.0 345 270
2000-06-27 21 43 0.48 742 27.3 6.5 220 261
2000-07-02 19 42 0.45 757 4.6 12.7 351 266
2000-07-20 17 47 0.36 682 3.1 12.6 261 270
2000-08-06 16 40 0.40 756 2.9 9.6 217 250
2000-08-20 23 48 0.48 736 7.7 14.0 357 261
2000-08-21 21 27 0.77 802 2.6 5.0 180 304
2000-08-26 19 36 0.54 783 6.1 14.5 152 223
2001-06-04 17 42 0.41 769 4.2 14.8 180 287
2001-06-08 22 53 0.42 736 2.0 9.3 199 322
2001-06-22 24 45 0.54 758 3.9 3.6 298 128
2001-07-09 28 44 0.64 796 5.6 13.3 47 220
2001-07-10 26 43 0.60 821 2.9 5.9 203 303
2001-07-13 25 58 0.44 695 1.0 6.1 240 228
2001-07-14 20 46 0.44 738 1.8 6.4 269 208
2001-07-23 24 46 0.53 727 3.6 3.6 208 309
2001-07-25 19 40 0.48 828 4.5 7.7 193 220
2001-07-26 23 35 0.64 772 2.1 7.0 200 309
2001-08-11 12 29 0.42 741 2.6 9.9 157 311
2001-08-14 11 50 0.23 631 1.9 19.0 202 315
2001-08-16 18 52 0.34 663 2.6 9.7 230 208
2002-06-14 18 47 0.37 693 8.4 4.9 191 291
2002-06-19 17 48 0.36 664 6.9 3.6 188 261
2002-06-25 23 58 0.40 706 4.4 7.7 185 170
2002-06-27 27 47 0.57 775 1.8 6.2 189 45
2002-06-28 26 40 0.66 769 3.4 1.2 43 315
2002-07-09 14 45 0.30 665 4.9 16.3 273 244
2002-07-10 19 43 0.45 725 1.7 8.1 237 303
2002-07-26 31 46 0.68 791 1.5 9.2 150 205
2002-07-29 23 41 0.57 688 2.0 5.5 225 213
2002-08-02 24 35 0.68 824 3.8 6.6 205 226
2002-08-05 20 40 0.49 770 4.6 14.7 325 255
2002-08-21 19 32 0.59 781 3.0 2.9 209 207
2002-08-28 20 31 0.64 776 1.8 11.9 110 287

48

Event PW, mm PW
sat
, mm PW/PW
sat
LCL, mb Wind spd, m/s Wind dir, (0
o

due N)
sIc 500mb sIc 500mb

2002-08-29 20 47 0.43 652 3.6 0.8 339 193
2003-06-10 17 35 0.49 694 6.3 10.8 211 265
2003-06-13 20 38 0.52 711 4.1 10.1 338 271
2003-06-16 16 40 0.41 713 1.2 2.8 172 297
2003-06-17 22 30 0.73 817 2.3 12.1 291 315
2003-06-28 14 40 0.34 686 2.5 13.7 223 305
2003-07-15 17 59 0.29 689 5.6 5.9 193 313
2003-07-19 23 53 0.44 762 4.5 10.5 212 300
2003-07-28 29 46 0.62 790 3.9 1.6 332 342
2003-07-29 26 45 0.57 781 1.6 9.0 244 331
2003-08-09 21 52 0.41 724 1.6 5.1 250 341
2003-08-11 21 54 0.39 699 2.8 6.1 176 359
2003-08-23 23 50 0.45 693 4.7 6.5 218 177

Table 6. Summary oI pre-storm sounding variables Ior the sixty-six storm events. PW is the
precipitable water computed at atmospheric conditions, PWsat is the precipitable water computed
assuming saturated conditions, and LCL is the liIting condensation level.

Minimum Maximum average Std deviation

PW, mm 11 32 21 5
PWsat, mm 27 59 43 8
PW/PWsat 0.23 0.86 0.51 0.13
LCL, mb (MSL) 829 (1621) 631 (4022) 744 (2694) 48 (588)

Table 7. Statistical summary oI 4 oI the 6 sounding variables presented in Table 6.

APPENDIX C
FIELD DESCRIPTIONS OF SOILS PROPERTIES
Profile No.: AR1 Described by: Jeanne Klawon and Ralph Klinger Date: 10/05/2004
Map Unit: Slope: <1 Aspect: Northeast GPS Coordinates: 1045320.5; 381852.8
Location: Along the right bank of the Arkansas River upstream of Pueblo Dam and reservoir near the mouth of Little Red Creek.
USGS Quadrangle: Hobson 7.5 Township/Range: T20S/R67W Section: NE1/4, NE1/4, S16 Elevation: 4900 ft
Parent Material: 1) silty sand, 2) clayey sand, and 3) sandy gravel (river alluvium)
Horizon Depth
(cm)
Boundaries Structure Texture Clay
Films
Consistence
Stickiness Plasticity Dry
Gravel
%
CaCO3
Morphology
Color
(dry)
A 0-39 aw 1mgr - - - - so 0-<10 salt present 7.5YR4/2m
C 39-69 aw m mSL - so po lo 0 - 7.5YR5/3m
2C 69-87 aw m mSCL - s p - 0 - 7.5YR4/4
3C 87-100 - sg gS - so po lo 50 - -
Map Unit: Slope: <1 Aspect: Northeast GPS Coordinates: 1045357.9; 381915.0
Location: Along the right bank of the Arkansas River upstream of Pueblo Dam and reservoir near the mouth of Red Creek.
USGS Quadrangle: Hobson 7.5 Township/Range: T20S/R67W Section: Center SW1/4, S9 Elevation: 4910 ft
Parent Material: 1) silty sand (river alluvium)
Horizon Depth
(cm)
Films
Consistence
Gravel
%
CaCO3
Morphology
Color
(dry)
A 0-28 aw 3m-cgr vfSL - ss ps so 0 - 10YR5/3
B 28-63 cs 1csbk f-mSiL - s p sh-h 0 - 10YR4/3
1
Bk 63-78 cs 1msbk vfSiCL - vs p sh trace I- 7.5YR5/3
C 78-88+ - m mLS - so po lo 0 - 7.5YR4/3
1
Sample AR2-3JU collected from the interval 63-78 cm; yielded a radiocarbon age of 175040 (1740-1550 cal yrs B.P.).
Map Unit: Slope: 2-3 Aspect: East GPS Coordinates: 1052214.1; 382856.4
Location: Along the right bank of the Arkansas River downstream of the stream gaging station at Parkdale, CO.
USGS Quadrangle: Royal Gorge 7.5 Township/Range: T18S/R71W Section: SW1/4, NE1/4, S18 Elevation: 5740 ft
Parent Material: 1) sand and 2) sandy gravel (river alluvium)
Horizon Depth
(cm)
Films
Consistence
Gravel
%
CaCO3
Morphology
Color
(dry)
A 0-30 aw sg-1msbk mLS - so po lo-so 0-<10 - 10YR4/3
2,3
Bw 30-54 cw 1msbk mLS - so po so 0-<10 I- 7.5YR5/4
Bk 54-68 aw 2csbk mLS - so po so 0-<10 I, es 10YR5/3
2Bkb 68-134+ - 1vcsbk mSL - ss ps sh 75 II+, ev 10YR5/3
2
Sample AR3-1PI collected from 48 cm; yielded a radiocarbon age of 123040 (1260-1060 cal yrs B.P.).
3
Sample AR3-2PI collected from the interval 30-54 cm; yielded a radiocarbon age of 121040 (1250-1050 cal yrs B.P.).
Map Unit: Slope: 2-3 Aspect: East GPS Coordinates: 1052214.1; 382856.4
Location: Adjacent to and upstream of site AR3.
USGS Quadrangle: Royal Gorge 7.5 Township/Range: T18S/R71W Section: SW1/4, NE1/4, S18 Elevation: 5740 ft
Parent Material: 1) sand (colluvium), 2) sand, and 3) sandy gravel (river alluvium)
Horizon Depth
(cm)
Films
Consistence
Gravel
%
CaCO3
Morphology
Color
(dry)
4
Spoil 50-0 aw
A 0-20 as sg-1fsbk mLS - so po lo-so 10 - 10YR4/2
Bw 20-56 aw 2csbk mLS - so po sh 10-25 - 10YR5/4
2Bk 56-84 aw 2csbk mLS - so po so 0-<10 I, es 10YR5/3
3Bkb 84-150+ - 1vcsbk mSL - ss ps sh 75 II+, ev 10YR5/3
4
Profile is overlain by spoil from placer mining activity along the margin of the terrace.
Map Unit: Slope: <1 Aspect: Northwest GPS Coordinates: 1053939.8; 382303.4
Location: Right bank downstream of Cotopaxi, CO; in the Loma Linda fishing access.
USGS Quadrangle: Arkansas Mountain 7.5 Township/Range: T48N/R12E Section: SW1/4, NE1/4, S29 Elevation: 6310 ft
Parent Material: 1) sand (river alluvium), and 2) sandy gravel (glacial outwash)
Horizon Depth
(cm)
Films
Consistence
Gravel
%
CaCO3
Morphology
Color
(dry)
A 0-10 aw sg-1fgr mLS - so po lo 0 - 10YR4/3
Bw 10-25 aw 1f-msbk mLS - so po so <10 - 7.5YR4/3
5
2Btj 25-68+ - 1msbk mSL 2dco so po so 75 - 7.5YR4/3
5
Lowercase j was used to indicate very weak (or juvenile) argillic properties typically not developed well enough to meet minimum criteria for argillic
horizon.
Map Unit: Slope: 2-3 Aspect: Northwest GPS Coordinates: 1053936.8; 382305.5
Location: Right bank downstream of site AR6 in the Loma Linda fishing access area.
USGS Quadrangle: Arkansas Mountain 7.5 Township/Range: T48N/R12E Section: SW1/4, NE1/4, S29 Elevation: 6295 ft
Parent Material: 1) gravelly sand and 2) sand (river alluvium).
Horizon Depth
(cm)
Films
Consistence
Gravel
%
CaCO3
Morphology
Color
(dry)
A 0-20 aw sg-1fgr mLS - so po lo-so <10 e 10YR4/2
6
AB 20-52 cw 1msbk mLS - so po so <10 I-, e 10YR3/2
7
Bk1 52-80 cw 2csbk mSL - ss ps Sh <10 I, es 10YR5/3
Bk2 80-93+ - 1msbk mLS - so po sh 0 es 10YR5/3
6
Sample AR7-1PI collected from 20-52 cm; yielded a radiocarbon age of 83040 (790-680 cal yrs B.P.).
7
Sample AR7-2PI collected from 52-80 cm; yielded a radiocarbon age of 210040 (2150-1980 cal yrs B.P.).
Map Unit: Slope: <1 Aspect: Northeast GPS Coordinates: 1060356.6;383419.1
Location: Right bank about 0.75 miles downstream of U.S. 50 along County Rd. 163; in the Adobe Park fishing access area.
USGS Quadrangle: Salina West 7.5 Township/Range: T50N/R08E Section: SW1/4, NE1/4, S22 Elevation: 7160 ft
Parent Material: 1) sand and 2) gravelly sand (river alluvium)
Horizon Depth
(cm)
Films
Consistence
Gravel
%
CaCO3
Morphology
Color
(dry)
A 0-10 as sg-1fgr mLS - so po lo-so <10 e 10YR4/2
B1 10-27 aw 1msbk mLS - so po so <10 I-, e 10YR3/2
8
B2 27-42 as 2csbk mSL - ss ps sh 50 I, es 10YR5/3
B3 42-51+ - 1msbk mLS - so po sh 0 es 10YR5/3
8
Sample AR8-1PI collected from 42 cm; yielded a radiocarbon age of 40040 (520-420; 390-320 cal yrs B.P.).
Map Unit: Slope: 1 Aspect: North GPS Coordinates: 1045357.9; 381915.0
Location: Near the mouth of Red Creek; upslope of site LB2 and about 100 feet downslope of old Santa Fe railroad grade.
USGS Quadrangle: Hobson 7.5 Township/Range: T20S/R67W Section: NE1/4, SW1/4, S9 Elevation: 4910 ft
Parent Material: 1) sand and gravelly sand (alluvial fan), and 2) silty sand (river alluvium)
Horizon Depth
(cm)
Films
Consistence
Gravel
%
CaCO3
Morphology
Color
(dry)
A 0-6 aw 2cgr-1msbk fSL - ss po so 0 es 10YR5/3
AB 6-15 aw 2msbk fSL - ss po so 0 es 10YR5/3
B 15-22 aw 1msbk fSL - ss ps so 25 ev 10YR5/3
2B1 22-46 cs m vfSL - ss ps vh 0 ev 10YR5/3
2B2 46-51 as m vfSL - ss ps sh 0 ev 10YR5/3
9
2B3 51-70+ - 2msbk fSiL - s p so 0 es 10YR5/3
9
Sample AR9-1JU collected from 56 cm; yielded a radiocarbon age of 83040 (790-680 cal yrs B.P.).
APPENDIX D
MACROFLORAL ID
AND RADIOCARBON DATA
Table D-1: Summary of Radiocarbon Ages
Sample No.
(Lab No.)
Site
(NAD27 coordinates)
Soil
Horizon
(depth, cm)
Type of
Material
Sample
Weight
(g)
Radiocarbon Age
(14C age yrs B.P.)
Calibrated Age
(cal age yrs B.P.)
AR9-1-JU
(Beta-198216)
Pueblo State Park, CO
(N381915.0; W104
5357.9)
2B3
(56)
Juniperius
charcoal
0.125 83040 790-680
AR2-3-JU
(Beta-197337)
Pueblo State Park, CO
(N381915.0; W104
5357.9)
Bk
(63-78)
Juniperius
charcoal
0.004 175040 1740-1550
AR3-1-PI
(Beta-197338)
Parkdale, CO
(N382856.4; W105
2214.1)
Bw
(48)
Pinus
charcoal
0.006 123040 1260-1060
AR3-2-PI
(Beta-197339)
Parkdale, CO
(N382856.4; W105
2214.1)
Bw
(30-54)
Pinus
charcoal
0.045 121040 1250-1050
AR7-1-PI
(Beta-197340)
Loma Linda, CO
(N382305.5; W105
3936.8)
AB
(20-52)
Pinus
charcoal
0.038 83040 790-680
AR7-2-PI
(Beta-197341)
Loma Linda, CO
(N382305.5; W105
3936.8)
Bk
(52-80)
Pinus
charcoal
0.005 210040 2150-1980
AR8-1-PI
(Beta-197342)
Adobe Park, CO
(N383419.1; W106
0356.6)
B2
(42)
Pinus
charcoal
0.028 40040
520-420
390-320
EXAMINATION OF BULK SOIL AND DETRITAL CHARCOAL FOR RADIOCARBON DATABLE
MATERIAL FROM SITES ALONG THE ARKANSAS RIVER, COLORADO
By
Kathryn Puseman
Paleo Research Institute
Golden, Colorado
Paleo Research Institute Technical Report 04-113
Prepared For
Reclamation Service Center
Denver, Colorado
November 2004
INTRODUCTION
Nine sediment samples and twelve charcoal samples from sites along the Arkansas
River in southern Colorado were floated to recover organic fragments suitable for radiocarbon
analysis. These samples were recovered from river terraces adjacent to the Arkansas River
between Pueblo Reservoir near Florence, Colorado, and Adobe Park just west of Salida,
Colorado, for the Pueblo Dam Paleoflood Study. Botanic components and detrital charcoal
were identified, and potentially radiocarbon datable material was separated.
METHODS
The bulk samples were floated using a modification of the procedures outlined by
Matthews {, 1979 #327}. Each sample was added to approximately 3 gallons of water. The
sample was stirred until a strong vortex formed, which was allowed to slow before pouring the
light fraction through a 150 micron mesh sieve. Additional water was added and the process
repeated until all visible macrofloral material was removed from the sample (a minimum of five
times). The material that remained in the bottom (heavy fraction) was poured through a 0.5-
mm mesh screen. The floated portions were allowed to dry.
The light fractions were weighed, then passed through a series of graduated screens
(US Standard Sieves with 4-mm, 2-mm, 1-mm, 0.5-mm and 0.25-mm openings to separate
charcoal debris and to initially sort the remains. The contents of each screen were then exa-
mined. Charcoal pieces larger than 1-mm in diameter were broken to expose a fresh cross
section and examined under a binocular microscope at a magnification of 70x. The remaining
light fraction in the 4-mm, 2-mm, 1-mm, 0.5-mm, and 0.25-mm sieves was scanned under a
binocular stereo microscope at a magnification of 10x, with some identifications requiring
magnifications of up to 70x. The material that passed through the 0.25-mm screen was not
examined. The coarse or heavy fractions also were screened and examined for the presence
of botanic remains. Remains from both the light and heavy fractions were recorded as charred
and/or uncharred, whole and/or fragments. Charcoal samples were water-screened through a
250-micron mesh sieve and allowed to dry. The dried samples were scanned under a
binocular stereo microscope at a magnification of 10x. Charcoal fragments were separated
and examined under a binocular microscope at a magnification of 70x.
Macrofloral remains, including charcoal, were identified using manuals {Core, 1976
#81; Martin, 1961 #325; Panshin, 1980 #393; Petrides, 1992 #403} and by comparison with
modern and archaeological references. The term "seed" is used to represent seeds, achenes,
caryopses, and other disseminules. Because charcoal and possibly other botanic remains
were to be sent for radiocarbon dating, clean laboratory conditions were used during flotation
and identification to avoid contamination. All instruments were washed between samples, and
samples were protected from contact with modern charcoal.
3
DISCUSSION
The study sites are located at elevations between 4900 and 7160 feet along the
Arkansas River in southern Colorado. Local vegetation on the flood plain includes cottonwood
(Populus), willow (Salix), tamarisk (Tamarix), and other riparian plants. Pinyon pine (Pinus
edulis), ponderosa pine (Pinus ponderosa), and juniper (Juniperus) are found on the terraces
adjacent to the river, while the hillslopes and some of the higher and older glacial outwash
terraces support sagebrush (Artemisia) and chamise/rabbitbrush (Chrysothamnus). A total of
19 bulk sediment and charcoal samples were collected from sites along the Arkansas River for
the Pueblo Dam Paleoflood Study in southern Colorado.
Pueblo State Park Site
Sample AR1-1 represents charcoal-rich sediment from the A horizon at a depth of 0-39
cm in Pueblo State Park (Table 1). This sample contained one small fragment of possible
Asteraceae charcoal weighing 0.001 g and a single piece of conifer charcoal weighing less
than 0.001 g (Table 2, Table 3). The minimum requirement of charcoal for standard AMS
radiocarbon analysis reported by Beta Analytic, Inc., is about 3 mg or 0.003 g, so that at least
300 micrograms of final carbon is available for dating. The Micro-Sample AMS Counting
Service is now available for samples containing only 100-300 micrograms of final carbon;
therefore, it may be possible to date charcoal weighing only 1 mg or 0.001 g. The majority of
charcoal in sample AR1-1 was unidentifiable with a vitrified/mineralized appearance. The
charcoal was very hard, shiny, and no longer exhibited individual wood structure elements
such as vessels, tracheids, rays, etc., identifiable in the cross-section view. These elements
were somewhat present in the tangential view. A few pieces of possible coal might actually
represent charcoal fragments so vitrified/mineralized in appearance that no identifiable wood
structure remained. In addition, the sample contained an uncharred Chenopodium seed and a
moderate amount of rootlets from modern plants in the area.
Sample AR1-2 was collected from the C horizon at a depth of 39-45 cm. This sample
yielded four fragments of unidentifiable vitrified/mineralized charcoal weighing 0.023 g, a piece
of conifer charcoal weighing less than 0.01 g, and an unidentified piece of charcoal
representing a knot area and exhibiting smooth, rounded edges weighing 0.001 g. Nine
pieces of possible coal again might reflect charcoal too vitrified/mineralized for identification as
wood. The sample also contained a few uncharred Poaceae leaf/stem fragments, a few root
fragments, a moderate amount of uncharred rootlets, muscovite, and sand.
All of the charcoal fragments greater than 1 mm in size in bulk sample AR1-3 from the
2C horizon at a depth of 69-79 cm exhibited a vitrified/mineralized appearance. A moderate
amount of uncharred rootlets, three snail shell fragments, and sand were the only other
remains to be recovered.
Sample AR2-1 from the B horizon at a depth of 50 cm yielded one fragment of Pinus
charcoal weighing less than 0.001 g. Three uncharred Helianthus seed fragments and an
4
uncharred Solanum rostratum seed reflect modern sunflower and buffalo-bur plants. In
addition, the sample contained a few uncharred rootlets and sand.
The two charcoal fragments in sample AR2-2 from a depth of 40 cm in the B horizon
were too small for identification and weighed less than 0.001 g. The sample consisted of sand
with an uncharred Helianthus seed fragment, a few uncharred rootlets, and a small amount of
rock/gravel present.
Bulk sample AR2-3 was taken from the Bk horizon at a depth of 63-78 cm. This
sample contained two fragments of Juniperus charcoal weighing 0.004 g, five fragments of
Pinus charcoal weighing 0.005 g, and one fragment of Salicaceae charcoal weighing 0.003 g
that can be submitted for AMS radiocarbon analysis. One piece of Alnus and two fragments of
Quercus charcoal weighed less than 0.001 g. The sample also yielded an eggshell fragment,
two insect puparia fragments, several snail shells, a few worm casts, muscovite, a small
amount of rock/gravel, and a few uncharred rootlets from modern plants.
Bulk sample AR2-4 from the B horizon at a depth of 28-63 cm yielded a variety of
charcoal and other charred remains. Alnus charcoal weighing 0.009 g was the only identified
charcoal present in sufficient quantities for AMS radiocarbon analysis. Three pieces of
charcoal weighing 0.025 g were too vitrified for identification. Fragments of Amelanchier,
Artemisia, Atriplex, Chrysothamnus, and conifer charcoal each weighed less than 0.001 g.
Other charred remains present in the sample include a probable Atriplex seed, a Brassicaceae
seed, two Solanum rostratum seeds, an unidentified seed fragment, six monocot/herbaceous
dicot stem fragments, and six Salsola seed fragments. Salsola (Russian thistle) is reported to
have been introduced accidentally to North America from Russia in 1873 or 1874 in a shipment
of flax seed (Martin 1972:43). Charred Salsola seeds have been recovered from prehistoric
hearth features at archaeological sites in Utah, Wyoming, Colorado, and Nebraska (Cummings
1992; Puseman 1993; Roper 1996; Karin Guernsey, personal communication). Recovery of
these charred Salsola seeds raises the possibility that there might have been a native species
of Salsola in North America prior to the historic introduction. Uncharred Amaranthus,
Euphorbia, Helianthus, Iva, Solanum, and Solanum rostratum seeds, as well as numerous
uncharred rootlets, represent modern plants. Non-floral remains present in this sample include
two earthworms, four insect chitin fragments, a few insect puparia fragments, several snail
shells, and a moderate amount of rock/gravel.
Sample AR2-5 was recovered from the B horizon at a depth of 38 cm. This sample
contained seven fragments of Asteraceae charcoal weighing 0.004 g that can be submitted for
AMS radiocarbon analysis. In addition, the sample contained three charred Atriplex seeds and
seed fragments, four charred monocot/herbaceous dicot stem fragments, fourteen charred
Salsola seed fragments, five partially charred Salsola seed fragments, and two charred pieces
of fruity tissue. An uncharred Amaranthus seed, three uncharred Helianthus seed fragments,
an uncharred Solanum rostratum seed and seed fragment, and a moderate amount of
uncharred rootlets represent modern plants.
Sample AR9-1 from a depth of 56 cm yielded one larger piece of Juniperus charcoal
weighing 0.125 g that can be submitted for AMS radiocarbon analysis. A total of 40 smaller
fragments of Juniperus charcoal weighing 0.092 g also were present. The sample also
5
contained a few uncharred rootlets from modern plants, a snail shell, a small amount of
rock/gravel, and sand.
Charred remains in sample AR9-2 from a depth of 45 cm consisted of charred
monocot/herbaceous dicot stem fragments weighing 0.024 g. These stem fragments are of a
sufficient weight for AMS radiocarbon analysis. A few uncharred rootlets from modern plants,
a small amount of rock/gravel, and sand were the only other remains to be recovered.
Parkdale Site
Sample AR3-1 represents charcoal fragments from the Bw horizon at a depth of 48 cm.
Nine fragments of Pinus charcoal weighing 0.006 g can be submitted for AMS radiocarbon
analysis (Table 4, Table 3). The sample also contained sand and a few uncharred rootlets.
Bulk sample AR3-2 was collected from the Bw horizon at a depth of 30-54 cm. This
sample contained several pieces of Pinus charcoal weighing 0.045 g that can be sent for AMS
radiocarbon analysis. Numerous uncharred rootlets from modern plant, three insect chitin
fragments, a few worm casts, and a small amount of rock/gravel also were present.
Bulk sample AR3-3 from the Bk horizon at a depth of 60-68 cm contained one piece of
conifer charcoal weighing less than 0.001 g. The sample consisted mainly of sand and
numerous uncharred rootlets from modern plants. A few root fragments, an insect chitin
fragment, and a small amount of rock/gravel also were present.
Samples AR5-1, AR5-2, and AR5-3 represent charcoal fragments from slackwater
deposits. Sample AR5-1 at a depth of 38 cm contained one piece of Pinus charcoal weighing
0.003 g that likely can be submitted for standard AMS radiocarbon analysis. One piece of
charred fruity tissue weighing 0.006 g also was recovered, as well as a few uncharred rootlets
from modern plants and sand.
Sample AR5-2 also was collected from a depth of 38 cm and was comprised mainly of
sand and a few uncharred rootlets from modern plants. One piece of Pinus charcoal weighing
0.001 g might be datable using the Micro-Sample AMS Counting Service.
Sample AR5-3 from a depth of 49 cm contained 15 fragments of Pinus charcoal
weighing 0.038 g that can be sent for AMS radiocarbon analysis. The sample also contained a
few uncharred rootlets from modern plants and sand.
Loma Linda Site
Charcoal sample AR7-1 was recovered from the B horizon on a low terrace at a depth
of 25-35 cm. This sample consisted of Pinus charcoal weighing 0.038 g (Table 5, Table 3) that
can be submitted for AMS radiocarbon analysis, as well a few uncharred rootlets and sand.
6
Bulk sample AR7-2 from the Bk horizon at a depth of 52-80 cm contained nine
fragments of Pinus charcoal weighing 0.005 g that can be sent for AMS radiocarbon analysis.
Two pieces of Quercus charcoal weighing 0.001 g and unidentifiable vitrified charcoal
weighing less than 0.001 g also were present. In addition, the sample yielded one charred
bone fragment weighing 0.003 g, several insect chitin fragments, a small amount of
rock/gravel, sand, and a moderate amount of sclerotia.
Sclerotia are commonly called "carbon balls". They are small, black, solid or hollow
spheres that can be smooth or lightly sculpted. These forms range from 0.5 to 4 mm in size.
Sclerotia are the resting structures of mycorrhizae fungi, such as Cenococcum graniforme, that
have a mutualistic relationship with tree roots. Many trees are noted to depend heavily on
mycorrhizae and may not be successful without them. "The mycelial strands of these fungi
grow into the roots and take some of the sugary compounds produced by the tree during
photosynthesis. However, mycorrhizal fungi benefit the tree because they take in minerals
from the soil, which are then used by the tree" {Kricher, 1988 #282:285}. Sclerotia appear to
be ubiquitous and are found with coniferous and deciduous trees including Abies (fir),
Juniperus communis (common juniper), Larix (larch), Picea (spruce), Pinus (pine),
Pseudotsuga (Douglas fir), Acer pseudoplatanus (sycamore maple), Alnus (alder), Betula
(birch), Carpinus caroliniana (American hornbeam), Carya (hickory), Castanea dentata
(American chestnut), Corylus (hazelnut), Crataegus monogyna (hawthorn), Fagus (beech),
Populus (poplar, cottonwood, aspen), Quercus (oak), Rhamnus fragula (alder bush), Salix
(willow), Sorbus (chokecherry), and Tilia (linden). These forms originally were identified by Dr.
Kristiina Vogt, Professor of Ecology in the School of Forestry and Environmental Studies at
Yale University {McWeeney, 1989 #337:229`-`230}{Trappe, 1962 #509}.
Adobe Park Site
Charcoal samples AR8-1, AR8-2, and AR8-3 were collected from the B horizon on a
low terrace. Sample AR8-1 from a depth of 42 cm yielded two fragments of Pinus charcoal
weighing 0.028 g that can be sent for AMS radiocarbon analysis (Table 6, Table 3). The
sample also contained one uncharred Chenopodium seed, a few uncharred rootlets, and a
small amount of rock/gravel.
Sample AR8-2 from a depth of 47 cm and sample AR8-3 from a depth of 40 cm
consisted of sand with a few uncharred rootlets. These samples did not contain charcoal
fragments for radiocarbon analysis. Two uncharred Chenopodium seed fragments were the
only other remains present in sample AR8-2.
SUMMARY AND CONCLUSIONS
Examination of bulk sediment and charcoal samples from sites along the Arkansas
River in southern Colorado resulted in recovery of charcoal and other charred botanic remains.
The majority of charcoal in samples from AR1 in the Pueblo State Park was unidentifiable with
a vitrified/mineralized appearance. Samples AR2-3, AR2-04, AR2-5, and AR9-1 contained
7
charcoal from juniper, pine, willow/cottonwood, alder, and/or a woody member of the sunflower
family other than sagebrush or rabbitbrush that can be submitted for AMS radiocarbon
analysis. Sample AR9-2 yielded charred fragments of a monocot/herbaceous dicot stem that
are of a sufficient weight for AMS radiocarbon analysis; however, no charcoal was present in
this sample. Samples from the Parkdale, Linda Loma, and Adobe Park sites contained only
pine charcoal in sufficient quantities for AMS radiocarbon analysis.
8
TABLE 1
PROVENIENCE DATA FOR SAMPLES FROM SITES ALONG THE ARKANSAS RIVER, COLORADO
Site No. Sample
No. Horizon
Depth
(cm)
Provenience/
Description Analysis
Pueblo
State Park
AR1-1 A 0-39 Charcoal-rich sediment; low
terrace
Float/Charcoal ID
prior to C-14 analysis
AR1-2 C 39-45 Charcoal fragments; low
terrace
Float/Charcoal ID
AR1-3 2C 69-79 Bulk sediment; low terrace Float/Charcoal ID
AR2-1 B 50 Charcoal fragments; high
terrace
Charcoal ID prior to
C-14 analysis
terrace
Float/Charcoal ID
AR2-3 Bk 63-78 Bulk sediment; high terrace Float/Charcoal ID
AR2-4 B 28-63 Bulk sediment; high terrace Float/Charcoal ID
terrace
C-14 analysis
AR9-1 56 Charcoal fragments Charcoal ID prior to
C-14 analysis
AR9-2 45 Charcoal fragments Charcoal ID prior to
C-14 analysis
Parkdale AR3-1 Bw 48 Charcoal fragments; high
terrace
C-14 analysis
AR3-2 Bw 30-54 Bulk sediment; high terrace Float/Charcoal ID
AR3-3 Bk 60-68 Bulk sediment; high terrace Float/Charcoal ID
AR5-1 38 Charcoal fragments, from
slackwater deposits
C-14 analysis
slackwater deposits below
black bed
C-14 analysis
slackwater deposits
C-14 analysis
TABLE 1 (Continued)
Site No. Sample
No. Horizon
Depth
(cm)
Provenience/
Description Analysis
9
Linda Loma AR7-1 B 25-35 Charcoal fragments; low
terrace
C-14 analysis
AR7-2 Bk 52-80 Bulk sediment; low terrace Float/Charcoal ID
Adobe Park AR8-1 B 42 Charcoal fragments; low
terrace
C-14 analysis
AR8-2 B 47 Charcoal fragments; low
terrace
C-14 analysis
AR8-3 B 40 Charcoal fragments; low
terrace
C-14 analysis
10
TABLE 2
MACROFLORAL REMAINS FROM ALONG THE ARKANSAS RIVER, PUEBLO STATE PARK
Sample Charred Uncharred Weights/
No. Identification Part W F W F Comments
AR1-1 Liters Floated 0.10 L
0-39 Light Fraction Weight 1.13 g
cm FLORAL REMAINS:
Chenopodium Seed 1
Rootlets X Moderate
CHARCOAL/WOOD:
cf. Asteraceae Charcoal 1 0.001 g
Conifer Charcoal 1 <0.001 g
Unidentifiable - vitrified/
mineralized
Charcoal X 0.205 g
NON-FLORAL REMAINS:
cf. Coal 6
Rick/Gravel X Few
Sand X
cm FLORAL REMAINS:
Poaceae Leaf/Stem X Few
Roots X Few
Rootlets X Moderate
CHARCOAL/WOOD:
Unidentified knot area - with
smooth, rounded edges
Charcoal 1 0.001 g
Unidentified hardwood -
vitrified
Charcoal 4 0.023 g
NON-FLORAL REMAINS:
cf. Coal 9 0.028 g
Muscovite X
Sand X
TABLE 2 (Continued)
11
cm FLORAL REMAINS:
Rootlets X Moderate
CHARCOAL/WOOD:
Unidentifiable -
vitrified/mineralized
Charcoal X 1.011 g
NON-FLORAL REMAINS:
Sand X
Snail shell 3
AR2-1 Volume Water-screened <0.10 L
50 cm Water-screened Sample Weight 1.87 g
FLORAL REMAINS:
Helianthus Seed 3
Solanum rostratum Seed 1
Rootlets X Few
CHARCOAL/WOOD:
Pinus Charcoal 1 <0.001 g
NON-FLORAL REMAINS:
Sand X
AR2-2 Liters Floated
40 cm Light Fraction Weight 2.29 g
FLORAL REMAINS:
Helianthus Seed 1
Rootlets X Few
CHARCOAL/WOOD:
Unidentifiable - small Charcoal 2 <0.001 g
NON-FLORAL REMAINS:
Rock/Gravel X Few
Sand X
TABLE 2 (Continued)
12
AR2-3
Liters Floated 2.50 L
cm FLORAL REMAINS:
Rootlets X
CHARCOAL/WOOD:
Alnus Charcoal 1 <0.001 g
Juniperus Charcoal 2 0.004 g
Pinus Charcoal 5 0.005 g
Quercus Charcoal 2 <0.001 g
Salicaceae Charcoal 1 0.003 g
NON-FLORAL REMAINS:
Eggshell 1
Insect Chitin 2
Muscovite X
Rock/Gravel X Few
Snail shell > 1 mm 41 33 0.212 g
Snail shell < 2 mm 24*
Worm casts X Few
cm FLORAL REMAINS:
cf. Atriplex Seed 1 <0.001 g
Brassicaceae Seed 1 <0.001 g
Salsola Seed 6 0.001 g
Solanum rostratum Seed 2 0.008 g
Unidentified Seed 1 <0.001 g
Monocot/Herbaceous dicot Stem 6 0.005 g
Amaranthus Seed 16*
Euphorbia Seed 13 16*
Helianthus Seed 5 38*
Iva Seed 5 1
Solanum Seed 22 5
Solanum rostratum Seed 76 3
Rootlets X Numerous
TABLE 2 (Continued)
13
AR2-4 CHARCOAL/WOOD:
28-63 Alnus Charcoal 5 0.009 g
cm Amelanchier Charcoal 1 <0.001 g
Artemisia Charcoal 2 <0.001 g
Atriplex Charcoal 1 <0.001 g
Chrysothamnus Charcoal 1 <0.001 g
Unidentifiable - vitrified Charcoal 3 0.025 g
NON-FLORAL REMAINS:
Earthworm 2
Insect Chitin 4
Insect Puparia 2 8*
Rock/Gravel X Moderate
Orthoclase feldspar 1
Snail shell > 1 mm 10 31 0.47 g
Snail shell < 1 mm 16*
FLORAL REMAINS:
Atriplex Seed 3 3
Salsola Seed 14
Salsola Seed 5pc
Fruity tissue 2 0.003 g
Amaranthus Seed 1
Helianthus Seed 3
Solanum rostratum Seed 1 1
Rootlets X Moderate
CHARCOAL/WOOD:
Asteraceae Charcoal 7 0.004 g
NON-FLORAL REMAINS:
Sand X
TABLE 2 (Continued)
14
FLORAL REMAINS:
Rootlets X Few
CHARCOAL/WOOD:
NON-FLORAL REMAINS:
Rock/Gravel X Few
Sand X
Snail shell 1 0.002 g
FLORAL REMAINS:
Rootlets X Few
NON-FLORAL REMAINS:
Sand X
W = Whole
F = Fragment
X = Presence noted in sample
L = Liters
g = grams
* = Estimated frequency
pc = partially charred
15
TABLE 3
INDEX OF MACROFLORAL REMAINS RECOVERED
FROM SITES ALONG THE ARKANSAS RIVER, COLORADO
Scientific Name Common Name
FLORAL REMAINS:
Amaranthus Pigweed, Amaranth
Brassicaceae Mustard family
Atriplex Saltbush, Shadscale
Chenopodium Goosefoot
Euphorbia Spurge
Helianthus Sunflower
Iva Marsh-elder
Poaceae Grass family
Salsola Russian thistle
Solanum Nightshade
Solanum rostratum Buffalo-bur
Fruity tissue Fruity epitheloid tissues; resemble sugar-laden fruit
or berry tissue without the seeds, or succulent plant
tissue such as cactus pads
Sclerotia Resting structures of mycorrhizae fungi
CHARCOAL/WOOD:
Alnus Alder
Amelanchier Juneberry, Serviceberry
Asteraceae Sunflower family
Artemisia Sagebrush
Chrysothamnus Rabbitbrush
Atriplex Saltbush, Shadscale
Conifer Cone-bearing, gymnospermous trees and shrubs,
mostly evergreens, including the pine, spruce, fir,
juniper, cedar, yew, and cypress
Juniperus Juniper
Pinus Pine
Quercus Oak
Scientific Name Common Name
16
Salicaceae Willow family
TABLE 4
MACROFLORAL REMAINS FROM ALONG THE ARKANSAS RIVER, PARKDALE SITE
FLORAL REMAINS:
Rootlets X Few
CHARCOAL/WOOD:
NON-FLORAL REMAINS:
Sand X
cm FLORAL REMAINS:
Rootlets X Numerous
CHARCOAL/WOOD:
NON-FLORAL REMAINS:
Insect Chitin 3
Rock/Gravel X Few
Worm casts X Few
cm FLORAL REMAINS:
Roots X Few
Rootlets X Numerous
CHARCOAL/WOOD:
NON-FLORAL REMAINS:
Insect Chitin 1
Rock/Gravel X Few
Sand X
TABLE 4 (Continued)
17
FLORAL REMAINS:
Fruity tissue 1 0.006 g
Rootlets X Few
CHARCOAL/WOOD:
NON-FLORAL REMAINS:
Sand X
FLORAL REMAINS:
Rootlets X Few
CHARCOAL/WOOD:
NON-FLORAL REMAINS:
Sand X
FLORAL REMAINS:
Rootlets X
CHARCOAL/WOOD:
NON-FLORAL REMAINS:
Sand X
W = Whole
F = Fragment
L = Liters
g = grams
18
TABLE 5
MACROFLORAL REMAINS FROM ALONG THE ARKANSAS RIVER, LINDA LOMA SITE
AR7-1 Sample Weight 0.19 g
25-35 FLORAL REMAINS:
cm Rootlets X Few
CHARCOAL/WOOD:
NON-FLORAL REMAINS:
Sand X
cm FLORAL REMAINS:
Sclerotia X X Moderate
CHARCOAL/WOOD:
Quercus Charcoal 2 0.001 g
Unidentifiable - vitrified Charcoal 2 <0.001 g
NON-FLORAL REMAINS:
Bone 1 0.003 g
Insect Chitin 37
Rock/Gravel X Few
Sand X
W = Whole
F = Fragment
L = Liters
g = grams
19
TABLE 6
MACROFLORAL REMAINS FROM ALONG THE ARKANSAS RIVER, ADOBE PARK SITE
FLORAL REMAINS:
Chenopodium Seed 1
Rootlets X Few
CHARCOAL/WOOD:
Pinus Charcoal 0.028 g
NON-FLORAL REMAINS:
Rock/Gravel X Few
FLORAL REMAINS:
Chenopodium Seed 2
Rootlets X Few
NON-FLORAL REMAINS:
Sand X
FLORAL REMAINS:
Rootlets X Few
NON-FLORAL REMAINS:
Sand X
W = Whole
F = Fragment
L = Liters
g = grams
20
REFERENCES CITED
Core, H. A., W. A. Cote and A. C. Day
1976 Wood Structure and Identification. Syracuse University Press, Syracuse, New
York.
Cummings, Linda Scott
1992 Pollen, Phytolith, Parasite, and Macrofloral Analysis of Coprolites from Room 21
in Step House (5MV1285), Mesa Verde National Park, Colorado. Ms. on file with the
National Park Service, Mesa Verde National Park.
Martin, Alexander C.
1972 Weeds. Golden Press, Western Publishing Company, Inc., New York, New
York.
Martin, Alexander C. and William D. Barkley
1961 Seed Identification Manual. University of California, Berkeley, California.
Matthews, Meredith H.
1979 Soil Sample Analysis of 5MT2148: Dominguez Ruin, Dolores, Colorado.
Appendix B. In The Dominguez Ruin: A McElmo Phase Pueblo in Southwestern
Colorado, edited by A. D. Reed. Bureau of Land Management Cultural Resource
Series. vol. 7. Bureau of Land Management, Denver, Colorado.
Panshin, A. J. and Carl de Zeeuw
1980 Textbook of Wood Technology. McGraw-Hill Book, Co., New York, New York.
Petrides, George A. and Olivia Petrides
1992 A Field Guide to Western Trees. The Peterson Field Guide Series. Houghton
Mifflin Co., Boston.
Puseman, Kathryn
1993 Macrofloral Analysis of Samples from Site 5WL1794, Colorado. Ms. on file with
Centennial Archaeology, Inc., Fort Collins, Colorado.
Roper, Donna C.
1996 Toward a New Perspecitve on Upper Republican Life in the Medicine Creek
Valley: The Excavation of 25FT22, House 4, with Testing at Several Nearby Features.
Ms. on file with U.S.D.I. Bureau of Reclamation, Kansas-Nebraska Project Office.

FROM: Darden Hood, Director (mailto:mailto:dhood@radiocarbon.com)
(This is a copy of the letter being mailed. Invoices/receipts follow only by mail.)

November 30, 2004

Dr. Ralph E. Klinger
Denver Federal Center
D-8530
P.O. Box 25007
Denver, CO 80225
USA

RE: Radiocarbon Dating Results For Samples AR2-3JU, AR3-1PU, AR3-2PU, AR7-1PU, AR7-
2PU, AR8-1PU

Dear Ralph:

Enclosed are the radiocarbon dating results for six samples recently sent to us. They each
provided plenty of carbon for accurate measurements and all the analyses went normally. As usual, the
method of analysis is listed on the report with the results and calibration data is provided where
applicable.

As always, no students or intern researchers who would necessarily be distracted with other
obligations and priorities were used in the analyses. We analyzed them with the combined attention of
our entire professional staff.

If you have specific questions about the analyses, please contact us. We are always available to
answer your questions.

The cost of the analysis was charged to the MASTERCARD card provided. A receipt is
enclosed. Thank you. As always, if you have any questions or would like to discuss the results, dont
hesitate to contact me.
Sincerely,

Dr. Ralph E. Klinger Report Date: 11/30/2004
Bureau of Reclamation Material Received: 11/1/2004

Sample Data Measured 13C/12C Conventional
Radiocarbon Age Ratio Radiocarbon Age(*)

Beta - 197337 1700 +/- 40 BP -21.8 o/oo 1750 +/- 40 BP
SAMPLE : AR2-3JU
ANALYSIS : AMS-Standard delivery
MATERIAL/PRETREATMENT : (charred material): acid/alkali/acid
2 SIGMA CALIBRATION : Cal AD 220 to 400 (Cal BP 1740 to 1550)
____________________________________________________________________________________

Beta - 197338 1190 +/- 40 BP -22.3 o/oo 1230 +/- 40 BP
SAMPLE : AR3-1PU
____________________________________________________________________________________

Beta - 197339 1190 +/- 40 BP -23.5 o/oo 1210 +/- 40 BP
SAMPLE : AR3-2PU
____________________________________________________________________________________

Beta - 197340 790 +/- 40 BP -22.5 o/oo 830 +/- 40 BP
SAMPLE : AR7-1PU
____________________________________________________________________________________

Beta - 197341 2040 +/- 40 BP -21.3 o/oo 2100 +/- 40 BP
SAMPLE : AR7-2PU
2 SIGMA CALIBRATION : Cal BC 200 to 30 (Cal BP 2150 to 1980)
____________________________________________________________________________________



Beta - 197342 350 +/- 40 BP -22.1 o/oo 400 +/- 40 BP
SAMPLE : AR8-1PU
2 SIGMA CALIBRATION : Cal AD 1430 to 1530 (Cal BP 520 to 420) AND Cal AD 1560 to 1630 (Cal BP 390 to 320)
____________________________________________________________________________________
CALIBRATION OF RADIOCARBON AGE TO CALENDAR YEARS
( Variables: C13/C12=-21. 8:lab. mult=1)
La borato ry number: Bet a-1973 37
Convent io nal radiocarbon age: 175040 BP
2 Sigma calibrated result:
(95% probability)
Cal AD 220 to 400 (Cal BP 1740 to 1550 )
In tercep t data
Inter cept of r adiocar bon age
with calibr ation cur ve: Cal AD 260 (Cal BP 169 0)
(68% pr obability)
Cal AD 240 to 3 50 ( Cal BP 1710 to 16 00)
4 98 5 S. W. 7 4th Co ur t, Miami, Flor id a 33 15 5 Tel: (3 05 )66 7- 51 67 F ax: (3 05 )6 63 -09 64 E-Mail: b eta@r a dio car bo n. co m
Beta Analytic Radiocarbon Dating Labor atory
T alma, A. S., Vogel, J . C., 1993, Radiocarbon 35( 2), p317-322
A Sim pl ifie d Approac h to Calibratin g C14 Dates
Mathe matics
Stuiv er, M., e t. al., 1998, Radiocarbon 40( 3), p1041-1083
INTCAL98 Radiocarbon Age C al ibration
Stuiv er, M., v an de r Pl icht, H., 1998, Radi oc arbon 40( 3) , pxii -xi ii
Editorial Comm ent
Calibration Database
I NTC AL 98
Database u sed
References:
R
a
d
i
o
c
a
r
b
o
n

a
g
e

(
B
P
)
1600
1620
1640
1660
1680
1700
1720
1740
1760
1780
1800
1820
1840
1860
Char red materi al
1880
Cal AD
180 200 220 240 260 280 300 320 340 360 380 400
175040 BP
(95% probability)
Cal AD 690 to 890 (Cal BP 1260 to 1060 )
In tercep t data
1 Sigma calibrated results:
(68% pr obability)
Cal AD 720 to 7 40 ( Cal BP 1230 to 12 10) and
Cal AD 760 to 8 70 ( Cal BP 1190 to 10 80)
Mathe matics
Editorial Comm ent
I NTC AL 98
Database u sed
References:
R
a
d
i
o
c
a
r
b
o
n

a
g
e

(
B
P
)
1080
1100
1120
1140
1160
1180
1200
1220
1240
1260
1280
1300
1320
1340
Char red materi al
1360
Cal AD
660 680 700 720 740 760 780 800 820 840 860 880 900
123040 BP
(95% probability)
Cal AD 700 to 900 (Cal BP 1250 to 1050 )
In tercep t data
(68% pr obability)
Cal AD 770 to 8 80 ( Cal BP 1180 to 10 70)
Mathe matics
Editorial Comm ent
I NTC AL 98
Database u sed
References:
R
a
d
i
o
c
a
r
b
o
n

a
g
e

(
B
P
)
1060
1080
1100
1120
1140
1160
1180
1200
1220
1240
1260
1280
1300
1320
Char red materi al
1340
Cal AD
660 680 700 720 740 760 780 800 820 840 860 880 900 920
121040 BP
(95% probability)
Cal AD 1160 t o 127 0 ( Cal BP 790 t o 680 )
In tercep t data
(68% pr obability)
Cal AD 1180 to 1260 (Cal BP 760 to 6 90)
Mathe matics
Editorial Comm ent
I NTC AL 98
Database u sed
References:
R
a
d
i
o
c
a
r
b
o
n

a
g
e

(
B
P
)
680
700
720
740
760
780
800
820
840
860
880
900
920
940
Char red materi al
960
Cal AD
1140 1150 1160 1170 1180 1190 1200 1210 1220 1230 1240 1250 1260 1270 1280
83040 BP
(95% probability)
Cal BC 200 t o 30 ( Cal BP 21 50 t o 1 980)
In tercep t data
with calibr ation cur ve: Cal BC 110 (Cal BP 2060 )
(68% pr obability)
Cal BC 180 to 50 ( Cal BP 2 130 to 2000 )
Mathe matics
Editorial Comm ent
I NTC AL 98
Database u sed
References:
R
a
d
i
o
c
a
r
b
o
n

a
g
e

(
B
P
)
1960
1980
2000
2020
2040
2060
2080
2100
2120
2140
2160
2180
2200
2220
Char red materi al
2240
Cal BC/AD
220 200 180 160 140 120 100 80 60 40 20 0
210040 BP
2 Sigma calibrated results:
(95% probability)
Cal AD 1430 t o 153 0 ( Cal BP 520 t o 420 ) and
Cal AD 1560 t o 163 0 ( Cal BP 390 t o 320 )
In tercep t data
(68% pr obability)
Cal AD 1440 to 1500 (Cal BP 510 to 4 50)
Mathe matics
Editorial Comm ent
I NTC AL 98
Database u sed
References:
R
a
d
i
o
c
a
r
b
o
n

a
g
e

(
B
P
)
260
280
300
320
340
360
380
400
420
440
460
480
500
520
Char red materi al
540
Cal AD
1400 1420 1440 1460 1480 1500 1520 1540 1560 1580 1600 1620 1640
40040 BP

FROM: Darden Hood, Director (mailto:mailto:dhood@radiocarbon.com)
(This is a copy of the letter being mailed. Invoices/receipts follow only by mail.)

December 1, 2004

Dr. Ralph E. Klinger
Denver Federal Center
D-8530
P.O. Box 25007
Denver, CO 80225
USA

RE: Radiocarbon Dating Result For Sample AR9-1-JU

Dear Ralph:

Enclosed is the radiocarbon dating result for one sample recently sent to us. It provided
plenty of carbon for an accurate measurement and the analysis went normally. As usual, the method of
analysis is listed on the report sheet and calibration data is provided where applicable.

As always, no students or intern researchers who would necessarily be distracted with other
obligations and priorities were used in the analysis. It was analyzed with the combined attention of our
entire professional staff.

If you have specific questions about the analyses, please contact us. We are always available to
answer your questions.

The cost of the analysis was charged to the MASTERCARD card provided. A receipt is
enclosed. Thank you. As always, if you have any questions or would like to discuss the results, dont
hesitate to contact me.
Sincerely,

Bureau of Reclamation Material Received: 11/18/2004


Beta - 198216 770 +/- 40 BP -21.1 o/oo 830 +/- 40 BP
SAMPLE : AR9-1-JU
ANALYSIS : AMS-Advance delivery
____________________________________________________________________________________
(Variables: C13/C12=-21.1:lab. mult=1)
Laboratory number: Beta-198216
Conventional radiocarbon age: 83040 BP
(95% probability)
Cal AD 1160 to 1270 (Cal BP 790 to 680)
Intercept data
Intercept of radiocarbon age
with calibration curve: Cal AD 1220 (Cal BP 730)
(68% probability)
Cal AD 1180 to 1260 (Cal BP 760 to 690)
49 85 S .W. 7 4t h Cour t, Mi a mi , Flor ida 33155 Tel : (3 05 )6 67 -51 67 Fax: (3 05)663-0964 E-Ma il : bet a@radi oc ar bo n. com
Beta Analytic Radiocarbon Dating Laboratory
Talma, A . S., Voge l, J. C. , 1993, R adioca rb on 3 5(2) , p317- 322
A Simplifie d Approa ch to Calib ra ting C1 4 Dates
Mathematic s
Stuiv er, M. , e t. al., 19 98, Rad io carbon 40( 3), p1 041-1 083
INTCAL98 Radioc arbon Age Calibration
Stuiv er, M. , v an der Plicht, H. , 199 8, Radioc arbon 40(3 ), px ii- xiii
Editorial Comme nt
Calibratio n Databa se
In tc al98
Datab ase use d
References:
R
a
d
i
o
c
a
r
b
o
n

a
g
e

(
B
P
)
680
700
720
740
760
780
800
820
840
860
880
900
920
940
Charred material
960
Cal AD
1140 1150 1160 1170 1180 1190 1200 1210 1220 1230 1240 1250 1260 1270 1280
83040 BP

APPENDIX E
HEC-RAS MODEL CROSS SECTION DATA
E-1

Pueblo State Park: Seven Cross Sections

4870
4880
4890
4900
4910
4920
4930
4940
4950
4960
4970
0 200 400 600 800 1000 1200 1400
Distance from left endpoint, ft
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 1
150,000 cfs
Railroad
Embankment

4870
4880
4890
4900
4910
4920
4930
4940
4950
4960
4970
0 200 400 600 800 1000 1200 1400 1600
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 2
150,000 cfs
Railroad
Embankment

E-2
4880
4890
4900
4910
4920
4930
4940
4950
4960
4970
4980
4990
0 200 400 600 800 1000 1200 1400 1600 1800
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 3
150,000 cfs
Railroad
Embankment

4880
4890
4900
4910
4920
4930
4940
4950
4960
4970
4980
4990
0 500 1000 1500 2000 2500
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 4
150,000 cfs
Railroad
Embankment

E-3
4880
4900
4920
4940
4960
4980
5000
5020
0 500 1000 1500 2000 2500
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 5
Site AR9
150,000 cfs
Railroad
Embankment

4890
4900
4910
4920
4930
4940
4950
4960
4970
4980
0 200 400 600 800 1000 1200 1400 1600 1800 2000
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 6
150,000 cfs
Railroad
Embankment

E-4
4900
4910
4920
4930
4940
4950
4960
4970
0 200 400 600 800 1000 1200 1400 1600 1800
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 7
150,000 cfs

Parkdale: Eight Cross Sections

5692
5697
5702
5707
5712
5717
5722
5727
5732
0 50 100 150 200 250
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 1
20,000 cfs
30,000 cfs

Railroad Embankment
E-5
5690
5695
5700
5705
5710
5715
5720
5725
0 50 100 150 200 250
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 2
20,000 cfs
30,000 cfs

5690
5695
5700
5705
5710
5715
5720
5725
0 50 100 150 200 250
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 3
20,000 cfs
30,000 cfs

Railroad Embankment
Railroad Embankment
E-6
5690
5695
5700
5705
5710
5715
5720
5725
5730
5735
-50 0 50 100 150 200 250
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 4
20,000 cfs
30,000 cfs
Site AR3, AR4, AR5

5695
5700
5705
5710
5715
5720
5725
5730
5735
5740
-50 0 50 100 150 200 250 300
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 5
20,000 cfs
30,000 cfs

Railroad Embankment
Railroad Embankment
E-7
5695
5705
5715
5725
5735
5745
5755
0 50 100 150 200 250 300 350
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 6
20,000 cfs
30,000 cfs

5690
5700
5710
5720
5730
5740
5750
5760
5770
5780
0 50 100 150 200 250 300
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 7
20,000 cfs
30,000 cfs

Railroad Embankment
Railroad Embankment
E-8
5700
5705
5710
5715
5720
5725
5730
5735
5740
5745
0 100 200 300 400 500 600
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 8
20,000 cfs
30,000 cfs

Loma Linda: Nine Cross Sections

6305
6310
6315
6320
6325
6330
6335
6340
0 50 100 150 200 250 300 350
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 1
14,000 cfs
50,000 cfs
Railroad
Embankment

Railroad Embankment
E-9
6305
6310
6315
6320
6325
6330
6335
6340
0 50 100 150 200 250 300 350 400
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 2
14,000 cfs
50,000 cfs
Railroad
Embankment

6305
6310
6315
6320
6325
6330
6335
6340
6345
-200 -100 0 100 200 300 400 500
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 3
14,000 cfs
50,000 cfs
Railroad
Embankment

E-10
6300
6305
6310
6315
6320
6325
6330
6335
6340
-50 0 50 100 150 200 250 300 350
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 4
14,000 cfs
50,000 cfs
Site AR7
Railroad
Embankment

6300
6305
6310
6315
6320
6325
6330
6335
6340
6345
-100 0 100 200 300 400 500 600
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 5
14,000 cfs
50,000 cfs
Site AR6
Railroad
Embankment

E-11
6305
6310
6315
6320
6325
6330
6335
6340
6345
6350
6355
-100 0 100 200 300 400 500 600
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 6
14,000 cfs
50,000 cfs
Railroad
Embankment

6310
6315
6320
6325
6330
6335
6340
6345
6350
6355
-50 0 50 100 150 200 250 300 350 400 450
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 7
14,000 cfs
50,000 cfs
Railroad
Embankment

E-12
6305
6310
6315
6320
6325
6330
6335
6340
6345
6350
6355
-50 0 50 100 150 200 250 300 350 400 450
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 8
14,000 cfs
50,000 cfs
Railroad
Embankment

6305
6310
6315
6320
6325
6330
6335
6340
6345
6350
6355
-50 0 50 100 150 200 250 300 350 400 450
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 9
14,000 cfs
50,000 cfs
Railroad
Embankment

E-13
Adobe Park: Eight Cross Sections

7130
7135
7140
7145
7150
7155
7160
0 100 200 300 400 500 600 700
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 1
20,000 cfs
Road Embankment
Pond

7130
7135
7140
7145
7150
7155
0 100 200 300 400 500 600 700
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 2
20,000 cfs
Road Embankment
Pond

E-14
7134
7136
7138
7140
7142
7144
7146
7148
7150
7152
7154
7156
0 100 200 300 400 500 600 700 800
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 3
20,000 cfs
Road Embankment

7136
7138
7140
7142
7144
7146
7148
7150
7152
7154
0 50 100 150 200 250 300 350 400 450 500
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 4
20,000 cfs
Road Embankment

E-15
7137
7139
7141
7143
7145
7147
7149
7151
7153
7155
7157
-100 0 100 200 300 400 500
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 5
20,000 cfs
Road Embankment

7136
7140
7144
7148
7152
7156
7160
0 100 200 300 400 500 600
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 6
20,000 cfs
Road Embankment

E-16
7135
7140
7145
7150
7155
7160
7165
7170
-100 0 100 200 300 400 500 600 700
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 7
20,000 cfs
Site AR8
Road Embankment

7140
7145
7150
7155
7160
7165
7170
-100 0 100 200 300 400 500 600 700 800
E
l
e
v
a
t
i
o
n
,

f
t
Cross Section 8
20,000 cfs
Road Embankment

APPENDIX F
PEAK FLOW FREQUENCY INPUT FILES
'Arkansas River at Pueblo Run 6a flows in range, One paleoflood bound, Final
Estimates, BP+50'
'Historical Information to 1859, thresholds based on 1864 and 1893 floods 40,000
and 20,000'
nx
3
nh(i) neprim(i) tl(i) tu(i) nn(i) kk(i) kt(i)
694 0 0. 150000. 840 0 0 'paleohydrologic bound'
35 1 0. 40000. 146 2 0 'historical threshold 1864'
ns ne
82 4
alpha log
0.40 1
reg skew reg wt
0. 0.0
bias factor tolerance
1.0 1.d-6
run mode conflim type
1 1
number of discharges to estimate annual probabilities
2
list of discharges
100000 200000
year Qest tl tu
1864 45000 40000 60000
1893 22000 20000 25000
1894 39100 35000 40000
1895 6100 6100 6100
1896 16500 16500 16500
1897 4300 4300 4300
1898 7500 7500 7500
1899 8800 8800 8800
1900 7600 7600 7600
1901 11100 11100 11100
1902 30000 30000 30000
1903 10500 10500 10500
1904 8500 8500 8500
1905 8000 8000 8000
1906 11000 11000 11000
1907 6600 6600 6600
1908 7600 7600 7600
1909 5800 5800 5800
1910 8400 8400 8400
1911 3700 3700 3700
1912 10500 10500 10500
1913 7800 7800 7800
1914 7500 7500 7500
1915 17000 17000 17000
1916 8900 8900 8900
1917 6800 6800 6800
1918 9600 9600 9600
1919 6300 6300 6300
1920 8500 8500 8500
1921 100000 80000 103000
1922 8850 8850 8850
1923 25600 25600 25600
F-1
1924 6510 6510 6510
1925 4930 4930 4930
1926 4520 4520 4520
1927 12400 12400 12400
1928 7800 7800 7800
1929 10500 10500 10500
1930 6050 6050 6050
1931 3560 3560 3560
1932 4380 4380 4380
1933 8630 8630 8630
1934 2580 2580 2580
1935 9880 9880 9880
1936 11200 11200 11200
1937 9300 9300 9300
1938 11200 11200 11200
1939 2910 2910 2910
1940 3860 3860 3860
1941 7560 7560 7560
1942 10300 10300 10300
1943 3320 3320 3320
1944 5980 5980 5980
1945 9290 9290 9290
1946 7050 7050 7050
1947 7280 7280 7280
1948 10900 10900 10900
1949 12800 12800 12800
1950 8700 8700 8700
1951 9300 9300 9300
1952 4740 4740 4740
1953 6770 6770 6770
1954 10200 10200 10200
1955 11100 11100 11100
1956 8010 8010 8010
1957 9070 9070 9070
1958 4540 4540 4540
1959 2820 2820 2820
1960 5260 5260 5260
1961 5760 5760 5760
1962 3540 3540 3540
1963 8360 8360 8360
1964 2840 2840 2840
1965 23500 23500 23500
1966 10600 10600 10600
1967 5870 5870 5870
1968 5190 5190 5190
1969 6620 6620 6620
1970 6300 6300 6300
1971 3360 3360 3360
1972 3360 3360 3360
1973 6760 6760 6760
1974 5440 5440 5440
1975 10200 10200 10200
1976 12800 12800 12800
F-2
'Arkansas River at Parkdale Run 6a gage/historical with 1957 Q, Final Paleo
Estimates, BP+50'
'Historical Information to 1868, thresholds based on 1921? and 1957 floods
19,000 and 9,000'
nx
3
1113 0 0. 30000. 1250 0 0 'paleoflood bound 1150 -
1350 years'
ns ne
48 1
alpha log
0.40 1
reg skew reg wt
0. 0.0
1.0 1.d-6
1 1
2
list of discharges
19000 33000
year Qest tl tu
1921 20000 18000 22000
1946 4580 4580 4580
1947 5880 5880 5880
1948 4870 4870 4870
1949 5530 5530 5530
1950 3010 3010 3010
1951 3150 3150 3150
1952 5720 5720 5720
1953 4970 4970 4970
1954 3230 3230 3230
1955 2670 2670 2670
1957 10000 9220 10300
1965 4880 4880 4880
1966 4290 4290 4290
1967 2400 2400 2400
1968 3870 3870 3870
1969 3100 3100 3100
1970 5020 5020 5020
1971 3370 3370 3370
1972 3270 3270 3270
1973 4160 4160 4160
1974 2400 2400 2400
1975 5200 5200 5200
1976 2480 2480 2480
1977 1360 1360 1360
1978 4330 4330 4330
1979 5000 5000 5000
1980 5840 5840 5840
1981 2110 2110 2110
1982 3220 3220 3220
1983 6310 6310 6310
F-3
1984 5430 5430 5430
1985 5960 5960 5960
1986 4670 4670 4670
1987 5950 5950 5950
1988 2550 2550 2550
1989 2220 2220 2220
1990 4350 4350 4350
1991 2940 2940 2940
1992 1840 1840 1840
1993 4700 4700 4700
1994 3880 3880 3880
1995 6830 6830 6830
1996 4440 4440 4440
1997 4990 4990 4990
1998 3170 3170 3170
1999 3590 3590 3590
2000 3470 3470 3470
2001 2980 2980 2980
F-4
'Arkansas River at Loma Linda Run 6a gage, historical, two paleo bounds, Final
Paleo Estimates, BP+50'
'Historical Information to 1880, threshold based on June 1957 flood and
paleoflood bound 14,000'
nx
3
8500 0 0. 50000. 10000 0 0 'Pleistocene paleoflood
bound'
1375 0 0. 14000. 1500 0 0 'paleoflood bound'
84 1 0. 9000. 125 1 0 'historical to 1880 for
1957'
ns ne
41 0
alpha log
0.40 1
reg skew reg wt
0. 0.0
1.0 1.d-6
1 1
3
list of discharges
10000 14000 50000
year Qest tl tu
1957 9220 9000 10300
1961 2650 2650 2650
1962 3620 3620 3620
1963 1910 1910 1910
1964 2740 2740 2740
1965 4390 4390 4390
1966 2080 2080 2080
1967 2580 2580 2580
1968 3620 3620 3620
1969 4070 4070 4070
1970 4010 4010 4010
1971 3180 3180 3180
1972 3600 3600 3600
1973 3770 3770 3770
1974 2340 2340 2340
1975 3510 3510 3510
1976 2480 2480 2480
1977 1310 1310 1310
1978 4020 4020 4020
1979 4070 4070 4070
1980 6240 6240 6240
1981 1730 1730 1730
1982 2840 2840 2840
1983 5400 5400 5400
1984 5040 5040 5040
1985 6020 6020 6020
1986 4360 4360 4360
1987 4340 4340 4340
1988 2500 2500 2500
1989 2630 2630 2630
F-5
1990 3820 3820 3820
1991 2550 2550 2550
1992 1690 1690 1690
1993 3940 3940 3940
1994 3200 3200 3200
1995 5800 5800 5800
1996 4400 4400 4400
1997 4570 4570 4570
1998 2920 2920 2920
1999 3480 3480 3480
2000 3320 3320 3320
2001 2520 2520 2520
F-6
'Arkansas River at Adobe Park Run 6a gage, historical data, paleoflood bound,
Final Paleo Estimates, BP+50'
'Historical Information to 1880, threshold based on June 1957 flood 9,000 ft3/s'
nx
2
425 0 0. 20000. 550 0 0 'paleoflood bound'
48 0 0. 9000. 125 1 0 'historical threshold'
ns ne
77 1
alpha log
0.40 1
reg skew reg wt
0. 0.0
1.0 1.d-6
1 1
3
list of discharges
9000 20000 30000
year Qest tl tu
1895 2370 2370 2370
1897 3060 3060 3060
1899 4700 4700 4700
1900 3630 3630 3630
1901 4900 4900 4900
1902 1800 1800 1800
1903 2640 2640 2640
1910 2770 2770 2770
1911 3400 3400 3400
1912 3580 3580 3580
1913 1930 1930 1930
1914 4010 4010 4010
1915 2750 2750 2750
1916 3080 3080 3080
1917 4730 4730 4730
1918 4840 4840 4840
1919 2460 2460 2460
1920 3430 3430 3430
1921 4000 4000 4000
1922 2870 2870 2870
1923 4900 4900 4900
1924 5100 5100 5100
1925 1920 1920 1920
1926 3060 3060 3060
1927 3780 3780 3780
1928 3070 3070 3070
1929 3850 3850 3850
1930 3470 3470 3470
1931 3300 3300 3300
1932 2500 2500 2500
1933 2990 2990 2990
1934 1600 1600 1600
1935 4050 4050 4050
1936 3900 3900 3900
F-7
1937 2400 2400 2400
1938 3930 3930 3930
1939 2500 2500 2500
1940 1510 1510 1510
1941 3530 3530 3530
1942 3600 3600 3600
1943 2920 2920 2920
1944 2980 2980 2980
1945 2580 2580 2580
1946 3160 3160 3160
1947 4890 4890 4890
1948 4110 4110 4110
1949 4390 4390 4390
1950 2910 2910 2910
1951 2800 2800 2800
1952 4760 4760 4760
1953 4400 4400 4400
1954 1950 1950 1950
1955 1980 1980 1980
1956 3740 3740 3740
1957 9220 9220 9220
1958 5300 5300 5300
1959 2870 2870 2870
1960 3180 3180 3180
1961 3120 3120 3120
1962 3540 3540 3540
1963 1950 1950 1950
1964 4180 4180 4180
1965 5410 5410 5410
1966 2690 2690 2690
1967 2780 2780 2780
1968 2940 2940 2940
1969 4570 4570 4570
1970 3930 3930 3930
1971 3180 3180 3180
1972 3880 3880 3880
1973 4050 4050 4050
1974 2570 2570 2570
1975 3680 3680 3680
1976 2530 2530 2530
1977 1400 1400 1400
1978 3810 3810 3810
1979 3700 3700 3700
F-8
APPENDIX G
RAINFALL-RUNOFF MODEL INPUT
Prob 16 Flood Freq Simul Main SST
Storm NP2-23 is base, move location
KSIM 1 NROW 186 NCOL 209 DX 960 DY 960 TSTART 0.0
NDT 1
4.0 72.0
NPRINTOUT 1
0.0166666667 432
NPRINTGRID 1
0.166666667 432
ECHO ./fs16.eco
Data Group B: Hydrologic Parameters ***********************
MASK ../Input1/maskf.asc
ELEVATION ../Input1/e960elmod-9.asc
INFOPT 1
NSOILS 18
0.00000083608 0.2772 0.4550 0.500 BYV-SL
0.00000132223 0.2076 0.4500 0.500 CB-L
0.00000157500 0.1903 0.4070 0.500 CBV-SL
0.00000054446 0.2742 0.4260 0.500 CL
0.00000110831 0.2263 0.4180 0.500 CN-L
0.00000332500 0.1258 0.4110 0.500 FSL
0.00000083608 0.2375 0.3520 0.500 GR-COSL
0.00000140000 0.2024 0.4150 0.500 GR-SL
0.00000062223 0.3054 0.4630 0.500 GRV-L
0.00000058331 0.2981 0.4000 0.500 GRV-SL
0.00000070000 0.2621 0.4080 0.500 L
0.00001217223 0.0744 0.4220 0.500 LS
0.00000036946 0.3497 0.4130 0.500 SIL
0.00001090831 0.0775 0.4600 0.500 SL
0.00000435554 0.1074 0.3990 0.500 ST-SL
0.00000145831 0.2064 0.4700 0.500 STV-L
0.00000252777 0.1617 0.4180 0.500 STV-SL
0.00000050554 0.3141 0.4080 0.500 STX-L-STX-SL
SOILS ../Input1/soils_surftext3.asc
NLANDS 9
0.07 0.0 open water/perennial ice/snow (NLCD 11)
0.02 0.0 low/high intensity residential/commercial/industrial (NLCD 21)
0.03 0.0 bare rock/sand/clay quarries strip mines (NLCD 31)
0.52 0.0 forests - deciduous, evergreen, mixed (NLCD 41)
0.59 0.0 shrubland (NLCD 51)
0.2 0.0 grassland/herbaceous (NLCD 71)
0.46 0.0 pasture/hay (NLCD 81)
0.21 0.0 row crops/small grains/fallow (NLCD 82)
0.33 0.0 urban/recreational grasses (NLCD 85)
LANDUSE ../Input1/lulc960n3.asc
STORAGE_DEPTHS ../Input1/storagedepth960-1.asc
CHNOPT 1 FLDOPT 0
TPLGYOPT 0 OUTOPT 0
LINK ../Input1/links-e960.asc
NODE ../Input1/nodes-e960.asc
CHANNEL ../Input1/testarkchanwidthdepth_1005.out
INTITIAL_WATER_OVERLAND ../Input1/storagedepth960-1.asc
INTITIAL_INFILTRATION ../Input1/storagedepth960-1.asc
INTITIAL_WATER_CHANNELS ../Input1/initwaterchan0960.txt
RAINOPT 4
CONV1 0.0254 CONV2 0.0002777778 SCALE 1.0
G-1
SPACE_TIME_STORM ./storm16.in
DAD_GRID ./dadfs16.asc
NUMBER_OF_OVERLAND_FLOW_SOURCES 0
NUMBER_OF_CHANNEL_FLOW_SOURCES 1
CONV1 1.0 CONV2 1.0 SCALE 1.0
69 5 2 Outletbaseflow
100.0 0.0 100.0 50.0
NUMBER_OF_WATERSHED_OUTLETS 1
OUTLET1 134 185 0.012177 0
NQREPORTS 6
103 70 3135.2832 1 Salida
109 79 3818.1888 1 Wellsville
121 99 4818.1248 1 LomaLinda
116 129 6530.4576 1 Parkdale
114 139 8056.6272 1 CanonCity
134 185 11869.2864 1 Pueblo
Data Group E: Data Group E: Environmental Properties ***************
General_Environmental_Properties_NPROPG 0
Data Group F: Output Specification Controls ***************
EXPORT TIME SERIES OUTPUTS
WATEREXPORT ./fs16_watertimeseries.dat
POINT-IN-TIME GRID OUTPUTS
RAINFALL_RATE
RAINFALL_DEPTH
INFILTRATION_RATE
INFILTRATION_DEPTH
WATER_DISCHARGE
WATER_DEPTH
SIMULATION SUMMARY OUTPUTS
DUMP_FILE
MASS_BALANCE ./dummy_massbalance.txt
SUMMARY_STATISTICS ./fs16_summarystats.txt
Storm 16 NP 2-23 June 1964 Gibson Dam MT
NUM_DURATIONS 10 MAX_DURATION 10
NUM_AREAS 7 MAX_AREA 7
STORM_CENTER_LOCATION 462760.468750 4262771.500000
STORM_SHAPE_RATIO 2.0 STORM_ORIENT 141.0
1 3 6 9 12 15 18 24 30 36
50 1.14 3.08 5.86 8.24 10.38 11.77 13.36 14.65 15.44 16.04
100 1.14 2.98 5.71 8.09 10.18 11.62 13.16 14.45 15.24 15.84
200 1.09 2.83 5.51 7.80 9.88 11.27 12.76 14.10 14.85 15.39
500 1.04 2.63 5.06 7.15 9.04 10.43 11.72 13.11 13.70 14.25
1000 0.94 2.43 4.62 6.60 8.34 9.63 10.87 12.21 12.81 13.31
2000 0.84 2.18 4.17 5.96 7.60 8.79 9.88 11.22 11.77 12.21
5000 0.65 1.79 3.38 4.92 6.31 7.35 8.19 9.48 10.03 10.38
G-2

Pueblo Floodhazard Finalreport

Caricato da

Informazioni sul documento

Descrizione originale:

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Pueblo Floodhazard Finalreport

Caricato da

Copyright:

Formati disponibili

U.S.

Department of the Interior

A Radar Climatology oI Extreme RainIall in the Colorado Front

2. 3-4 1une1921 Pueblo Storm

3. Data and Methodology

6. Summary and Conclusions

Potrebbero piacerti anche