Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
net/publication/224234122
CITATIONS READS
14 365
8 authors, including:
Some of the authors of this publication are also working on these related projects:
SOWFIA project: Streamlining of Ocean Wave Farms Impact Assessment View project
All content following this page was uploaded by George Voulgaris on 01 June 2014.
K.-W. Gurgel
University of Hamburg, Institute of Oceanography
Hamburg, Germany
Abstract - Since April 2006, long range (8.3MHz) WERA HF Wind direction estimates from radar using manufacturer-
radars have been operated on the Southeastern United States supplied software were evaluated for a range of wind speeds and
coastline, as part of the U.S. Integrated Ocean Observing System direction (fetch limited or not) and by differentiating between
(IOOS) and in particular the national HF Radar network. These conditions where the predominant water waves satisfied the long-
radars measure currents operationally, and waves and winds wave assumption or not. For non-fetch limited wind speeds in
experimentally across the wide continental shelf of Georgia (GA) excess of the Bragg wave propagation speed, and wave fields for
and South Carolina (SC). Half-hourly data at 3km horizontal which the long-wave assumption is relevant, correlations between
resolution are acquired to a range of approximately 200 km, radar and in situ anemometer wind directions were good, at
providing measurements across the wide continental shelf and approximately 0.7.
into the adjacent Gulf Stream at the shelf edge. Radar HF-radar directional wave spectra estimates are also derived
performance in range and quality is discussed. Ease in siting of with aftermarket software from SeaView Sensing Ltd, and are
these space and cable intensive systems along populated compared to in situ estimates from a five beam ADCP. These
coastlines, and the feasibility of their operation by non-radar novel in situ measurements use the ADCP slant beam horizontal
specialists is also briefly discussed. velocities combined with vertical velocities from the vertically
Long term in situ measurements of currents, waves and oriented 5th beam to measure wave orbital velocities and derive
winds concurrent with the long-term radar measurements were directional wave spectra. Wave buoy accelerometer estimates of
available. These measurements were also acquired as part of the directional wave spectra are also available for comparison.
U.S. national coastal ocean observatory network, under the Radar significant wave heights and wind direction from SeaView
evolving auspices of SABSOON, SEACOOS, and SECOORA. software are also being examined.
Wind, wave and ADCP measurements from several instrumented Keywords - ocean currents; HF-Radar; waves; winds; ocean
Navy towers at the 27, 33, and 44 m isobaths are available, and observatories.
winds and wave information also exist at the 18m isobath from an
NDBC buoy. Comparisons between radar-derived estimates and
the in situ measurements are examined for a variety of I. INTRODUCTION
parameters, including (near) surface currents, significant wave With the growth of observational oceanography in the
heights, directional wave spectra, and wind direction and speed. United States, increasing numbers of shore-based high
Radar estimates of surface velocity compare quite well with frequency (HF) radars are being installed for the purpose of
in situ ADCP near surface current data, with complex vector measuring surface currents across wide areas of the coastal
correlation magnitude of 0.95, and phase angle of -0.2 degrees.
The negative sign is consistent with an expected counterclockwise
ocean. These systems are increasingly being deployed by
rotation with depth between the radar surface current estimates groups with no prior history of radar work, with assistance up
and the subsurface upper water column ADCP measurements. the rather steep learning curve being provided by more
Tidal amplitudes, which are large and predominantly semi- experienced users through direct interactions, working group
diurnal on the GA/SC coast are extremely well reproduced by the meetings and email list-servers.
radar estimates.
Many of the more recent installations have been CODAR
Radar significant wave height estimates from manufacturer-
supplied software are much noisier than measurements from in systems, with which the oceanographic community is
situ pressure sensors or wave buoys, but capture several-hour gradually becoming familiar. These direction finding systems
low-passed variability fairly well. The spectra from which the enjoy a reputation of good customer support from the
radar significant wave heights are estimated have been examined, manufacturer and a scientifically gregarious experienced user
also exhibiting higher variability than indicated by the in situ group. However other systems are available: another type of
estimates. radar operated in the U.S. is the WEllen RAdar (WERA) [1] to
978-1-4577-0022-4/10/$26.00©2010 IEEE
Proceedings of the IEEE/OES/CWTM Tenth Working Conference on Current Measurement Technology
[5]. WERA is a system capable operating in both beam Georgia Research Alliance funding. The systems are installed
forming and/or direction finding mode (depending on Rx on St. Catherine's Island in Georgia (31o 41.262' N, 81o 8.016'
antenna configuration) although most WERA users prefer the W), a privately held island in approximately the middle of the
beam forming mode. All systems are available at a range of state's N-S coastline extent, and on Pritchard's Island, SC (32o
operating frequencies and bandwidths, quantities that control 18.172'N, 80o 30.775'W), owned and maintained by the
range coverage. The purpose of the present paper is to University of South Carolina. They are 90 km apart, each
summarize performance of a pair of long-range WERA radars with nominal range of 200 km (actual range examined below).
that have been in operation since April 2006 on the coastline Both systems operate at a frequency of 8.35 MHz, with a
of Georgia and South Carolina in the Southeastern U.S. This frequency modulated continuous wave mode (FMCW) with a
pair constitutes the first deployment of long-range WERA in bandwidth of 40 KHz. The transmit power is quite low,
the U.S., and is an addition to a small number of higher approximately 40W total throughout the sampling interval.
frequency WERA installations nationwide. A further From April 2006 through Feb 2008, the two systems operated
distinction is that until this installation, WERA operation in in ping-pong mode, issuing 1024 consecutive chirps (with Rx
the U.S. had been the exclusive domain of experienced radar switched on continuously throughout) over approximately 8
oceanographers: the University of Miami group led by L. K. minutes. Pritchard’s (PRI) operated on the hour and half hour,
Shay, and the University of Hawaii group led by P. Flament, and St. Catherine’s (STC) sampling began on the 15 and 45
both operating higher frequency systems. As the first novice minute marks of every hour. Range and azimuthal resolution
users of this capable system, a summary perspective on the dependent on chirp width and Rx array length and frequency.
feasibility of its adoption by other oceanographers would be of The configuration described renders 3 km range and 2o
value. azimuthal resolution. At mid-shelf, this is about 3 km,
expanding to 6 km near the shelf edge. In late May - early
WERA systems are manufactured by Helzel Messtechnik
June 2007, additional manufacturer software was installed at
of Germany, after the design of [6]. Each system consists of a
both sites to reduce the influence of radio frequency
transmit array (Tx) and a receive array (Rx), both feeding to a
interference (RFI) [7]. After Feb 2008, the sampling time was
central Tx/Rx Radio unit controlled by a Linux-based
doubled, with 2048 chirps from PRI beginning at the top of the
workstation. The Tx array consists of four frequency-specific
hour lasting approximately 16 minutes, and identical sampling
tuned monopole antennae oriented in a rectangle with long
from STC commencing at the half hours. After Nov 2008,
axis parallel to the shoreline. Required separation between the
simultaneous operation of both STC and PRI began, using
antennae is frequency dependent; with the long axis corners
opposing sweep directions (STC chirped with descending
separated by 1/2 the wavelength of the transmitted radio wave,
frequency, PRI with ascending frequency), providing 16
and front to back spacing of 0.15 wavelengths. The Rx array
minute samples from each station and half hourly combined
in beam-forming configuration consists of several monopole
vector data. In Feb 2010, a third station was installed on
antennas oriented in a line parallel to the shore. These
Jekyll Island GA. The systems all operate simultaneously,
systems typically have 16 Rx antennas, but due to the large
separation required between them for lower frequencies, the
12 antenna option was chosen for GA/SC installations. Since
the system is modular, addition of 4 more antennae at a future
time is possible. Through the installation of tuning coils, the
length of the antennas is minimized for less obtrusive
installation in public areas. WERA system operation in
direction finding mode requires 4 Rx antennas in a square
configuration, minimizing the antenna footprint significantly.
However, the phased array beam-forming arrangement of the
Rx antennae allows unambiguous determination of range and
azimuth for current estimation, and permits use of the second
order returns for determining directional wave spectra on the
same areal grid as current estimates are made, as long as the
signal to noise ratio is sufficient. The latter degrades at ranges
greater than 50% of the range applicable for current estimates.
From the air-conditioned space housing the control computers,
a power cable and a single coax cable run to the Tx array. If
the array is sufficiently distant from the power source, a power
amplifier is installed at the Tx end. Conversely, each Rx
antenna has a separate cable connecting to the corresponding
channel on the Rx Radio unit. In order to maintain constant
phase delays on all cables, identical cables of equal length are
recommended as this ensures identical length variation due to
heating and cooling. Figure 1. Snapshot of surface velocities from WERA installations along the
U.S. Southeastern coast.
The two shore-based WERA HF-radar systems in GA/SC
(Fig. 1) were installed with SEACOOS (through ONR) and
Proceedings of the IEEE/OES/CWTM Tenth Working Conference on Current Measurement Technology
Figure 3. Radar range for each installation for returns within +- 30 deg of
boresite, for two periods before and after the installation of RFI software.
Figure 5. Estimated M2 (12.42 hour) tidal ellipses (using T-Tide) for radar
velocities and ADCP velocities from several depths in the water column.
Proceedings of the IEEE/OES/CWTM Tenth Working Conference on Current Measurement Technology
Proceedings of the IEEE/OES/CWTM Tenth Working Conference on Current Measurement Technology
heights are then estimated by integrating the sea surface Comparison of the R2 VADCP estimates to a estimates at a
spectra over the recorded frequency range. WERA wave WERA grid point located at tower R7 (same water depth as
estimates were constructed for May 2007-Feb 2008 (after the R2, but somewhat closer to the PRI radar, not shown here) are
installation of the WERA RFI reduction software). For somewhat better, especially for the frequency band between
comparison to the R2 tower subsurface in situ pressure based roughly 0.05 and 0.15 Hz.
wave measurements, WERA spectral estimates are
transformed to pressure signal at the depth of the pressure 3) Wave Comparison Summary
transducer using linear theory that applies a frequency Overall, the occasional close correspondence between the
dependent attenuation with depth. directional spectra and the low-frequency correspondence of
the radar and in situ significant wave height estimates suggests
Scatter plots of in situ vs. radar–derived wave estimates there are conditions under which reasonable radar
from the nearest WERA grid point show broad scatter for both performance can be expected. However, at this stage we
buoy 41008 and tower R2. This is somewhat improved by believe that the wave comparison is compromised both by
averaging WERA spectral estimates over a 3x3 grid point suboptimum radar configuration and by the the fact that the in
array nearest the in situ measurements before estimating situ measurements are outside of the footprint where high
significant wave heights (Fig. 8, upper panels). A time series quality radar wave estimates would be expected. The quality
subset of compared data from tower R2 (Fig. 8, lower panel) of radar second order return deteriorates with increasing angle
illustrates that while the half-hourly WERA wave height from the radar central boresight and increased range. In
estimates are noisier, relative to the in situ measurements, the addition the beamwidth of a 12 antennae array is wider than
lower frequency variability in the in situ records is tracked that of a 16 antenna array, which diminishes the angular
relatively well by the radar estimates. Implementation of resolution of the system. Efforts to improve radar signal to
spectral averaging over several grid points or several time noise or reduce sidelobe problems could improve results. We
steps prior to implementation of the wave algorithms in the will be assessing both these possible contributors in the future.
radar processing software may prove fruitful to reducing noise It will be possible to add additional antennas to our existing
in these estimates. installations, due to the flexibility of the WERA system. It is
also possible to vary the sampling scheme as well, including
2) Spectrum Inversion Method
sample timing and interval length, chirp length, and transmit
HF-radar directional wave spectra estimates are also frequency. Work on improving wave estimates from radars
derived with aftermarket software from SeaView Sensing Ltd, and identifying conditions when they are more reliable is
and are compared to in situ estimates from the VADCP. While ongoing [4], [19].
neither system can be verified as accurate ground-truth of the
actual directional wave spectra, the correspondence between
the two estimates at the R2 tower is not routinely good. B. Wind Direction
Both the WERA system manufacturer-supplied software
and the SeaView software use a sech2(ȕș) angular distribution
model [20]. The SeaView solutions, which have not yet been
evaluated for these radar installations, solve for both available
parameters ȕ and wind direction ș. In the WERA distribution
software, ȕ is currently set to a fixed value; future versions
will fit ȕ as well. Wind direction measurements collected at 4
in situ anemometers were compared with the WERA HF radar
over a 20 month period to assess the accuracy of the radar
system. The in situ measurements are from NOAA Buoy
41008, located in the Grays Reef National Marine Sanctuary,
and from the three instrumented U.S Navy Platforms; R2, R6
and R8. Buoy 41008 and tower R2 are in 18 and 27 m water
depths, as previously mentioned, towers R6 and R8 are in ~30,
and ~45 m water depth, respectively. Locations are shown in
Fig. 1. The Grays Reef Buoy is a 3m discus buoy with an
ARES 4.4 payload, and an R.M Young wind monitor 05103
installed at 5m above sea-level, measured at the top of every
hour with a unit-vector averaging time of 10 minutes.
Anemometers at the three tower locations were R.M Young
marine wind monitor 05106, elevation 50m (R2 and R6) or
34m (R8, all recording continuously with a unit-vector
averaging time of 6 minutes, of which hourly subsamples are
Figure 8. Comparison of radar and in situ measurements of significant wave used here. The radar data was interpolated to hourly values
height. Upper left: scatter plot from Grays Reef data. Upper right Scatter plot
from tower R2 data. Lower panel: Sample timeseries from tower R2 showing An unexpectedly large RMS error of 58.9° was calculated
scatter of radar significant wave height estimates (blue) about the in situ over all stations with a record-long complex correlation
pressure based measurements (red).
coefficient of 0.65. A fairly large (~30o) negative bias
between radar and anemometer measurements exists at all
Proceedings of the IEEE/OES/CWTM Tenth Working Conference on Current Measurement Technology
Proceedings of the IEEE/OES/CWTM Tenth Working Conference on Current Measurement Technology
Overall, the high quality of the surface current data has [14] Wyatt L.R. (1990), A relaxation method for integral inversion applied to
exceeded expectation. With it we are providing a rich and HF radar measurement of the ocean wave directional spectrum.
International J. of Remote Sensing, 11,1481-1494.
reliable realtime data resource for the operational aspirations
[15] Wyatt L.R., J.J. Green and A. Middleditch, “HF radar data quality
of the oceanographic community. Perhaps better still, we are requirements for wave measurement,” Coastal Engineering, in press.
accumulating research-quality surface current time series of [16] Gargett, A. E. (1994), Observing turbulence with a modified acoustic
high temporal and spatial resolution over a broad shelf and Doppler current profiler, J. Atmos. Ocean. Tech., 11(6), 1592-1610.
shelf-edge area, including, as with the higher frequency [17] Terray, E.A., Brumley, B.H. and Strong, B. (1999) Measuring waves
WERA systems deployed by the University of Miami, the and currents with an upward-looking ADCP. Proc. IEEE 6th Working
measurement of Gulf Stream variability at unprecedented Conference on Current Measurement. IEEE Press, 66-71.
resolution and temporal extent. We are sufficiently [18] Gurgel K.-W., H.H. Essen and T. Schlick (2006), An empirical method
encouraged by the wave and wind comparisons between radar to derive ocean waves from second-order Bragg scattering: prospects
and in situ sensors to begin investigating avenues for and limitations. IEEE Journal of Oceanic Engineering, 31(4): 804-811.
improvements. The flexibility in the system for both hardware [19] Wyatt, LR, JJ Green, KW Gurgel, JC Nieto Borge, K Reichert, K
Hessner, H Gunther, W Rosenthal, M Reistad (2003) Validation and
and software changes will accommodate this pursuit. intercomparisons of wave measurements and models during EuroROSE
experiments. Coastal Engineering 48: 1-28.
ACKNOWLEDGMENTS [20] Donelan, M.A., Hamilton, J. & Hui, W.H., 1985. Directional spectra of
wind-generated waves. In: Philos. Trans. Roy. Soc. Lon. Ser. A. Vol.
The GA/SC radar sites are operated with funding from 315, 509-562.
NOAA/IOOS as part of the South East Coastal Ocean [21] Tennekes, H. (1973), The Logarithmic Wind Profile, J. Atmos. Sci. V.
Observing Regional Association (SECOORA), and through 30 234-238.
NSF research grant NSF OCE-0536326 which also supported [22] Archer, M. (2008), WERA HF radar Measurements of wind direction in
VADCP measurements. the South Atlantic Bight, B.Sc Ocean Science Dissertation, U. Plymouth,
40 pages.
REFERENCES
[1] Voulgaris, G, N. Kumar, K.-W. Gurgel, J.C. Warner and J.H. List
(2011), 2-D Inner-Shelf Current Observations from a Single VHF
WEllen RAdar (WERA). Proceedings of the IEEE/OES Tenth Current,
Waves and Turbulence Measurement Workshop, This Volume.
[2] Savidge, D. K., J. Norman, C. Smith, J. A. Amft, T. Moore, C. Edwards
and G. Voulgaris (2010), Shelf Edge Tide Correlated Eddies along the
Southeastern United States, Geophys. Res. Lett., 37, L22604,
doi:10.1029/2010GL045236.
[3] Chavanne, C., P. Flament, and K.-W. Gurgel, (2010), Observations of
strong submesoscale anticyclone and associated frontogenesis near an
island, J. Phys. Ocean., V. 40, pp. 1802-1818.
[4] Haus, B.K., L.KBrian K., Lynn K. Shay, P.Paul A. Work, G. George
Voulgaris, R.Rafael J. Ramos and J., Jorge Martinez-Pedraja (2010),
Wind speed dependence of single site Wave-Height Retrievals from
Phased-array HF radars, J. Atmos. Oceanic Technol., 27(8), pp., 1381-–
1394. doi: 10.1175/2010JTECHO730.1.
[5] Shay, L.K., H.E. Seim, D. Savidge, R. Styles, and R.H. Weisberg
(2008), High frequency radar observing systems in SEACOOS, MTS
Journal, 42(3), 55-67.
[6] Gurgel KW, G Antonischki, HH Essen, and T Schlick (1999), Wellen
Radar (WERA): a new ground-wave HF radar for ocean remote sensing.
Coastal Engineering 37: 219-234.
[7] Gurgel K.-W. and Y. Barbin (2008), Suppressing radio frequency
interference in HF radars. Sea Technology, 39-42.
[8] Essen,H.-H., K.-W. Gurgel, T. Schlick, (2000), On the Accuracy of
current measurements by means of HF radar. IEEE JOE 25(4), pp 1-9.
[9] Seim, H., S. Haines and M. Muglia (2008) Examining the validity of the
Outer Banks HF radar system, AGU/ASLO Ocean Sciences Meeting,
Orlando FL, Abstract Volume, p. 364.
[10] Lee, T.N., J. A. Yoder and L. P. Atkinson (1999), Gulf Stream frontal
eddy influence on productivity of the southeast U.S. continental shelf, J.
Geophys. Res., Vol 96, pp 22191-22205.
[11] Wyatt L. R., J. J. Green, A. Middleditch, M D Moorhead, J. Howarth,
M. Holt, S. Keogh (2006), Operational wave, current and wind
measurements with the Pisces HF radar. IEEE Journal of Oceanic
Engineering, 31,819-834.
[12] Green, J.J. & L.R. Wyatt (2006), Row-action inversion of the Barrick-
Weber equations. Journal of Atmos. and Oceanic Technol., 23,501-510.
[13] Wyatt L. R. (2000), Limits to the inversion of HF radar backscatter for
ocean wave measurement. Journal of Atmos. and Oceanic Technol., V
17,1651-1666.