Hydrological and nutrient run-off modelling of

Aiviekste river basin

Juris Senņikovs
Laboratory for mathematical modelling of environmental and
technological processes,
University of Latvia

Distributed hydrological model of the Aiviekste river basin has been

developed and calibrated to the typical, dry and wet hydrological regimes.
Conceptual model of nutrient load has been employed to the same
catchment.
Several scenarious of impact of climatic changes to hydrological and
nutrient load regime have been calculated.
Basin of the river Aiviekste

•Located in northeast part of Latvia

•Part of Daugava basin
•Area approx. 9000 km2
•Average discharge at the inflow to
Daugava – approx. 60 m3/s
•Aiviekste outflows from the lake
Lubanas situated in the centre of
the basin
Approach to hydrological model
Physically-based spatially and temporally distributed dynamic modelling
The catchment is divided into hierarchical subbasins downscalable up to the finite
element level. The hydrological cycle is resolved for the lowest hierarchical level, the
hydrological cycle modelling is coupled with the dynamic routing of the water flow
through the network of streams.

Principal components of model

•Surface water model – solves for surface water content
(intercepted+ponded).
•Groundwater model – solves for groundwater level
•Flow routing model
•Lake model – solves for waterlevel of lake
 Rivers
Surface
water
Mesh element
node

lout Lake
Apatch
Aelem

River
segment
Surface water model
Solves for surface water content (intercepted+ponded)

Principal components:
•Precipitation (in form of rain)
•Snowmelt
•Evapotranspiration
•Infiltration
•Groundwater saturation excess flow
•Overland flow - surface run-off

Finite volume model for triangular elements (variables at triangles)

Surface run-off path – according to surface level gradient
dw
 P  E  Vinf iltr  Vsurface  Vrunoff
( i )
 Vrunoff
(i )
 Vsnow
dt i i

P – precipitation (if rain)

w – surface water content (=wintercepted+wponded)
E – evapostranspiration
Vsurface – groundwater saturation excess flow
V- - surface runoff from element
V+- surface runoff entering element
Vsnow – snowmelt rate

 w  wint ercepted  S 0 
1 lout If w>wintercepted
2 / 3 1/ 2
Vrunoff n – Mannig coefficient for
n Aelem overland flow (land use
dependent)
S0 – surface slope
Evapotranspiration model

E  K e wesat T   e

Snow-melt model
Solves for water content in snow
Melt rate degree-day dependent

dS S – equivalent water content in snow

 Ps  Vsnow Ps – precipitation in form of snow
dt
Vsnow = CMELT (T-T2)
T2 – reference temperature (>0ºC)
Ground water model
Solves for piezometric head in upper layer (2D)

Principal components:
•Storativity
•Darcy flow
•Infiltration from surface
•Overflow to surface water and directly to rivers
•Flow to/from lakes

Finite element model, head at triangular grid nodes

 h
h  z g S s    K h  z g h  Vinf iltr  Vriver  Vsurface
t
h – head equal to water level
zg – level of aquitard
Ss – specific storativity
K - hydraulic conductivity
At lake node h=hlake
Outflow to rivers at river nodes
K h  hriver 
Vriver  If h>hriver
lriver ∆lriver – empiric
Outflow to surface water at surface elements

K h  hsurface  hsurface 
Vsurface  If h>(hsurface- ∆hsurface)
lsurface ∆lsurface – empiric
∆hsurface – surface level correction
Flow routing model
Solves for river level and discharge
St Venant equations
Staggered finite difference model (level at points, discharge at
segments)
Input data – river bed level, cross-sections are parametrized parabolic
profiles
Inflow sources – surface runoff, groundwater overflow
Lake model
Solves for lake level and discharge
Inflow – surface run-off, rivers, groundwater
Outflow – rivers, groundwater

dhlake
Alake   Qriverin
(i )
  Qriverout
(i )
  Vrunoff
(i ) (i )
lout ( w  wint ercept ) (i )
dt i i i

h
  K (h  z g )d { P  E}

n
Input data
•Digital terrain model + river network -> model grid, model river network,
model lake boundaries
•Land use data [National CORINE Land Cover 2000 in Latvia ] ->grouped to 6 main land
use types (agricultural, forests, bushes/grasslands, swamps, artificial and waterbodies),
model accounts for fraction of each type at each surface water element
•Observations of the river discharge at 7 stations [Ziverts]
•The daily meteorological data (precipitation and air temperature) at Rēzekne,
Zīlāni and Gulbene [LEGMA]
Model grid and river network
6380000

240

231.5

6360000 223

214.5

206

6340000
197.5

189

180.5
6320000
172

163.5

155
6300000
146.5

138

129.5
6280000
121

112.5

104
6260000
95.5

87

6240000 78.5

70

610000 620000 630000 640000 650000 660000 670000 680000 690000 700000 710000 720000
Land use according to CORINE
Distribution of area percentage of land use types

6380000 6380000

6360000 6360000
1 1

Gulbene Gulbene
6340000 6340000

6320000 0.67 6320000 0.67

6300000 6300000

0.33 0.33
6280000 6280000

Rēzekne Rēzekne
Zīlāni Zīlāni
6260000 6260000

0 0

6240000 6240000

6220000 6220000
600000 620000 640000 660000 680000 700000 720000 740000 600000 620000 640000 660000 680000 700000 720000 740000

Forests Agricultural
Location of hydrometric stations

300

270

240
Pededze
210

180
Kuja
Aiviekste-Lubāna
150

Iča
120

Aiviekste-HES
90
Malta
Rēzekne
60

30

0
Selection of calibration scenario

Three different Years are considered.

 Jul/76 to June/77 is dry Year, precipitation 518 mm average

run-off at Aiviekste HPP 35 m3/s
 Jul/77 to June/78 is average Year, precipitation 660 mm
average run-off at Aiviekste HPP 57 m3/s

 Jul/79 to June/80 is wet Year, precipitation 762 mm average

run-off at Aiviekste HPP 92 m3/s
Selection of calibration scenario

The most downstream gauge is Aiviekste HPP (basin area 8660

km2). The discharge at this gauge is, respectively 25%, 32%,
and 44% of the precipitation for three respective Years.
Dry Years – lower percentage due to the relatively higher
evaporation (high T, low e) and higher infiltraton (low soil
moisture, low groundwater level). Wet Years – higher
percentage doe to the lower evaporation (low T, high e) and
lower infitration (high sil moisture, high groundwater level).
Selection of calibration scenario

The hydrometric stations are located rather far upstream on the

tributaries. The particular tributaries with the available
discharge time series (Pededze, Rēzekne, Malta, Iča,
Balupe, Kuja) cover only 42% of the catchment area,
producing (on average) also 42% of the River Aiviekste run-
off. therefore the calibration of each particular tributary is
important for the adjustment of the land-use dependent
parameters, whilst the most of the river basin is accounted by
the total, i.e. Aiviekste HPP hydrometric station.
Selection of calibration scenario

Dry Years are characteristic with the single snow-melt flood.No

travel time of the flood signal may be distinguished. There is
no immediate response of the precipitation signal in the
discharge time-series.

Wet Year is characterised by multiple rain events during

summer and prolonged rainfalls during autumn. The later
results in the autumn high-water. Still neither travel time of
the flood signal may be distinguished not there is an
immediate response of the precipitation signal in the
discharge time-series.
 Aiviekste discharge. Dry Year.

250.00 25

200.00 20
AivHES
AivLub
W

150.00 15
Discharge

100.00 10

50.00 5

0.00 0
01.Jūl.76 31.Jūl.76 30.Aug.76 30.Sep.76 30.Okt.76 30.Nov.76 30.Dec.76 29.Jan.77 01.Mar.77 31.Mar.77 01.Mai.77 31.Mai.77 30.Jūn.77
Time
 Aiviekste discharge. Wet Year.

300.00 35

30
250.00

25
AivHES
200.00
AivLub
W
20
Discharge

150.00

15

100.00
10

50.00
5

0.00 0
01.Jūl.78 31.Jūl.78 30.Aug.78 30.Sep.78 30.Okt.78 30.Nov.78 30.Dec.78 29.Jan.79 01.Mar.79 31.Mar.79 01.Mai.79 31.Mai.79 30.Jūn.79
Time
 Kuja and Rēzekne discharge. Average Year.

30.00

25.00

20.00 Kuja-Aizkuja
Rēzekne
Discharge

15.00

10.00

5.00

0.00
01.Jūl.77 31.Jūl.77 30.Aug.77 30.Sep.77 30.Okt.77 30.Nov.77 30.Dec.77 29.Jan.78 01.Mar.78 31.Mar.78 01.Mai.78 31.Mai.78 30.Jūn.78
Time
Calibration strategy

• Calibration goal achieved by changing model parameters –

groundwater parameters and land-use dependent surface water
run-off, evaporation, melting and infiltration parameters

• Each three year model calibraton run is preceded by 90 year

run for stabilising the grounwater level in quasi-periodic state
(to avoid “initialization” effects)
 Surface level Typical groundwater level
240 240
6380000 6380000
235 235
230 230
225 225
220 220
6360000 215 6360000 215
210 210
205 205
200 200
195 195
6340000 6340000
190 190
185 185
180 180
175 175
6320000 170 6320000 170
165 165
160 160
155 155
150 150
6300000 6300000
145 145
140 140
135 135
130 130
6280000 125 6280000 125
120 120
115 115
110 110
105 105
6260000 6260000
100 100
95 95
90 90
85 85
6240000 80 6240000 80
75 75
70 70

600000 620000 640000 660000 680000 700000 720000 600000 620000 640000 660000 680000 700000 720000
Calibration results

300
Model - Aiviekste@HPP
AivHES
280
260
240
220
200
180 Aiviekste - HPP
m^3/s, m^3/s
140 160
120
100
80
60
40
20
0

01.09.1976 01.03.1977 01.09.1977 01.03.1978 01.09.1978 01.03.1979

 50
Model Malta
Malta

Malta – Viļāni
45
40
35
30
m^3/s, m^3/s
25 20
15
10
5
0

01.09.1976 01.03.1977 01.09.1977 01.03.1978 01.09.1978 01.03.1979

 80
Model - Iča
Iča

Iča - Kuderi
70
60
50
m^3/s, m^3/s
40 30
20
10
0

01.09.1976 01.03.1977 01.09.1977 01.03.1978 01.09.1978 01.03.1979

 90
Model - Pededze
Pededze

80
70
60 50
m^3/s, m^3/s
40 30
20
10
0
Pededze - Litene

01.09.1976 01.03.1977 01.09.1977 01.03.1978 01.09.1978 01.03.1979

 Impact of climate change to Aiviekste run-off
•Three climate change scenarious considered
•Climate change scenario gives seasonal change of precipitation and temperature
•The same three (normal, dry, wet) years calculated as in calibration

Input data for one of the climatic scenarious

Parameter/scenario Spring (III- Summer (VI- Autumn (X- Winter Annual

V) IX) XII) (I-II)
P / Central, % +2.5 +5.0 +5.0 +10.0 +5
T / Central, ºC +2.0 +1.5 +2.0 +3.0 +2.0
Impact of climate change to Aiviekste run-off

Calculated reference
Lowvs. climatic
vs. Central scenario discharges
scenario

300

Q-Central
250 Q-Low

200
Discharge

150

100

50

0
Jūl-76 Okt-76 Jan-77 Apr-77 Jūl-77 Okt-77 Jan-78 Apr-78 Jūl-78 Okt-78 Jan-79 Apr-79
Time
 Model for calculation of nutrient loads

•CATCHLOAD model, originally intended for a cacthment, is used at element

scale.
• Nutrient loading at element scale, nutrient transport following path of surface
water
• In rivers - nutrient transport with decay.
• In lakes – nutrient inflow/outflow balance and decay.

Input data
Monthly water quality observations and point-source load data [LEGMA]
P-PO4, N-NO3, N-NO2, N-NH4
The DTM, land-use, river network, surface mesh, other input data and the
modelling results from hydrological modelling was used.
 Model for calculation of nutrient loads
L = Qforest · cforest + Qfield · cfield
cforest = c0forest · f (lake percentage) · f (catchment area) · f (runoff) · f (soil frost) · f
(slope)
cfield = c0field · f (lake percentage) · f (catchment area) · f (runoff) · f (soil frost) · f
(slope) · f (plant cover type)

f (lake percentage) e(a pl) Pl a

f (catchment area) Ab A b A (km2)

f (runoff) (q / Mq)c q, Mq c (q / Mq)>l
f (soil frost) d SF + l SF d SF = 0 (no soil frost)
SF = l (soil frost)
f (slope percentage) (l + SL)f SL f

f (plant cover type) g PT + l PT g PT = 0 (plants, soil frost)

PT = l (no plants, no soil
frost)
Point sources

300

Alū

270

240

Gulbene
Balvi

210

Cesvaine 180

Lubā

Madona
150

120

Varakļā

Viļā 90

Rē

60

Malta

30

0
CALIBRATION
CATCHLOAD coefficients were found for the whole Aiviekste
catchment, distinguishing between agricultural and forest land-use
types.
The point-sources from the major towns were acounted for. We
assumed load proportional to population, pollution per capita was
calculated from observations up- and downstream Rēzekne.
The same time period (1976-1979) as for the hydrological model was
used.
Observed nutrient concentration vs. discharge

3.2 0.08

2.8 N P 0.07

2.4 0.06

2 0.05
N, mg/l

P,mg/l
1.6 0.04

1.2 0.03

0.8 0.02

0.4 0.01

0 0
0 20 40 60 80 100 120 140 160 180 200 220 240
Discharge, m^3/s
 Observed and calculated load of nitrogen

1800

1600

1400

1200

1000

800

600

400

200

0
07. 08. 10. 12. 02. 04. 06. 08. 10. 12. 02. 04. 06. 08. 10. 12. 02. 04. 06.
76 76 76 76 77 77 77 77 77 77 78 78 78 78 78 78 79 79 79