Canal and river tests of a RiverSonde streamflow measurement system

Abstract

Results of field tests of a RiverSonde streamflow radar are
compared with in-situ current measurements at a canal and a river in
central California during June, 2000. Typical water velocity in the
middle of the canal was about 0.45 m s^-1 and 0.30 ms^-1
at the edges. Velocity in the river was about 20% lower with
similar cross-channel variation. Differences between the RiverSonde and
in-situ velocities were 6-18% of the mean flow, with similar differences
among the various in-situ velocities. In addition to the surface
velocities, the total volume flow was estimated based on the in-situ
depth measurements. Volume flow for the canal was about 37 m³
s^-1 and for the river was about 64 m³ s^-1
, with differences between the various radar and in-situ
techniques of less than 10%

Full text

Page 1

Receiver
Transmitter
Distance Along River, meters
-10
Distance Along River, meters
102030 40 50 6070-20-30
-40
-50 -60-70
0
10
20
30
0
10
20
30
Canal and river tests of a RiverSonde streamflow measurement system
Calvin C. Teague, Donald E. Barrick, Peter Lilleboe
CODAR Ocean Sensors, Ltd., 1000 Fremont Avenue, Suite 145, Los Altos, CA 94024
650-941-5897, fax 408-773-0514; cal@alpha.stanford.edu, don@codaros.com, pete@codaros.com
Ralph T. Cheng
U. S. Geological Survey, 345 Middlefield Road, Menlo Park, CA 94025
rtcheng@usgs.gov
Abstract– Results of field tests of a RiverSonde streamflow
radar arecompared with in-situcurrent measurements at a canal
and a river in central California during June, 2000. Typical wa-
ter velocity in the middle of the canal was about 0.45 ms–1and
0.30 ms–1at the edges. Velocity in the river was about 20%
lower with similar cross-channel variation.
tween the RiverSonde and in-situ velocities were 6–18% of the
mean flow, with similar differences among the various in-situ
velocities. In addition to the surface velocities, the total vol-
ume flow was estimated based on the in-situ depth measure-
ments. Volume flow for the canal was about 37 m3s–1and for
the river was about 64 m3s–1, with differences between the
various radar and in-situ techniques of less than 10%.
Differences be-
INTRODUCTION
Recently a RiverSonde streamflow radar was developed
based on a standard SeaSonde radar [1]. The SeaSonde nor-
mally is operated over salt water at HF (3–30 MHz) to mea-
sure ocean surface currents [2]. It was modified to operate at
UHF (350 MHz) and wide FM sweep width (10–30 MHz) to
match the expected water wavelengths and channel dimensions.
Transmit power was about 1 W, and maximum range over fresh
water was a few hundred meters. A bistatic geometry was used
for the antennas, with transmit and receive antennas on oppo-
site sides of the water channel, as illustrated in Fig. 1. As with
ocean current observations, current is estimated from the shift
of the position of the first-order Bragg line from its still-water
position, taking into account the increase in Bragg wavelength
due to the bistatic scattering angle [3, 4]. The backscatter Bragg
wavelength for the water waves is 0.43 m, and water waves up
to about 0.55 m contribute to the bistatic scattered energy. For
bistatic scattering, the radar measures the component of current
perpendicular to the constant-delay ellipse. Total currents were
computed assuming that the flow was parallel to the channel
and that cross-channel flow was negligible.
A2-elementbroadsidetransmitantennawithagroundscreen
reflector provided a broad floodlight illumination of the wa-
ter surface, with the main lobe directed downriver and a null
directed toward the receiver to limit the dynamic range re-
quired of the receiver. A 3-element array was used at the re-
ceiver. Conventional time delay and Doppler processing pro-
vided range resolution of 10–30 m and velocity resolution of
Figure 1: Bistatic scattering geometry. Transmitter and receiver
are on opposite sides of a river or canal. Contours of constant
time delay are ellipses, and velocity is measured in a direction
perpendicular to the ellipses. A scattering cell at 40 m range is
highlighted.
about 2.5 cms–1. MUSIC direction finding [5] was used to de-
termine the bearing of the signals at the receive antenna. The
bistatic geometry complicates the processing somewhat but re-
sults in a received signal strength which is nearly independent
of scattering position across the water channel. In these early
tests the transmit antenna was connected to the radar system
by a coaxial cable stretched across the water, but in the final
version the transmitter is expected to be a stand-alone unit in-
dependently powered and synchronized to the radar controller.
EXPERIMENT
Two locations were used for the initial field tests in June
2000. The first site was on the Delta-Mendota Canal in cen-
tral California. The canal is about 30 m wide with a concrete-
linedchannelandflatbottom. Thewaterflowiswellcontrolled,
and a bridge provides convenient support for the transmit ca-
ble. Radar observations were made 20–60 m downstream of the
bridge. The bridge is supported by several pillars in the water,
resulting in some turbulence immediately downstream from the
pillars. This site provided an ideal test bed for the early tests.
The second site was on the American River in downtown
Sacramento, California. This is a natural river about 110 m
wide and represents a realistic environment for a typical de-
ployment. Radar observations were made about 150 m past a
bend in the channel. There were no obstructions in the water
at this site. The river bottom was irregular and included both
muddy and rocky areas.
Extensive in-situ measurements were made at both sites by
the U. S. Geological Survey. These measurements included
0-7803-7033-3/01/$10.00 (C) 2001 IEEE

Page 2

submerged flow meters suspended from a small boat at sev-
eral locations and depths across the water channel, floating ten-
nis balls optically tracked by a TV camera, an optical surface
flowmeter, bottom depth probing with a weighted line, and an
anemometer for wind measurements.
?7.5 ?5 ?2.50 2.55 7.5
Freq, Hz
?120
?100
?80
?60
Power, dB
Delta?Mendotarb 21
?7.5 ?5 ?2.502.557.5
Freq, Hz
?120
?100
?80
?60
Power, dB
American River rb 13
Figure 2: Spectrum of received energy at the Delta-Mendota
Canal on 5 June 2000 (top) and at the American River on 7 June
2000 (bottom). The waves are moving predominately toward
the radar at the canal and away from the radar at the river. The
short red lines indicate the estimated noise level and the region
of the spectrum selected for MUSIC direction finding.
Radar echoes from the water were seen out to several hun-
dred meters at both sites. Fig. 2 shows received power spectra
obtained at the Delta-Mendota Canal and the American River.
In both cases the scattering is predominately first-order even at
the short radar wavelength used for this experiment, with the
first-order energy clearly identifiable. The spectrum from the
canal is complicated by a nearly cross-channel wind. The spec-
trum from the river is simpler, resulting from the wind blow-
ing nearly downriver. In both cases the antennas were look-
ing downriver, and energy from the negative Bragg line (from
waves receding away from the radar) was processed. Spectral
broadeningisduebothtothebistaticgeometryandcurrentvari-
ation across the river.
VELOCITY
Fig. 3 shows the current velocities as a function of distance
across the channel for the Delta-Mendota Canal test on 5 June.
Results for 6 June were similar. The RiverSonde and in-situ re-
sultsareingoodagreementexceptforaregionwithinabout8m
of the receive antenna. The reasons for that disagreement are
not clear, but may be due to distorted antenna patterns. Fig. 4
shows the velocity profiles for the American River on 7 June.
The velocities for the river were similar to those for the canal,
with similar cross-channel variation.
5 10 152025 30
Distance, m
0.1
0.2
0.3
0.4
0.5
0.6
Velocity, m s?1
Delta?Mendota, 5 June: Surface Velocity
RiverSonde
Tennis balls
Pygmy?bridge
Pygmy?boat
Figure 3: Delta-Mendota velocity profiles for 5 June 2000. Dis-
tance is measured across the channel from the bank near the
receive antenna. The in-situ measurements were from a flow
meter suspended from a small boat about 60 m from the bridge
(Pygmy-boat) or with the boat tethered to the bridge (Pygmy-
bridge), and from optical tracking of floating tennis balls (Ten-
nis balls). The RiverSonde measurements are shown as a thick
dashed blue line.
Table 1 summarizes the RMS velocity differences between
the RiverSonde and in-situ measurements and between the var-
ious in-situ measurements for the 3 days of the experiment.
RMS differences between the RiverSonde and in-situ measure-
ments were 8–18% of the mean velocity, and differences be-
tween the in-situ measurements were 8–17% of the mean.
VOLUME FLOW
The total volume flow for each site was computed by in-
tegrating the product of velocity and depth as measured by a
weighted line dropped to the channel bottom at several points
across the channel. The volume flows are summarized in Ta-
ble 2. Because the various in-situ velocity measurements cov-
ered slightly different swath widths across the channel, the
RiverSonde volumes were computed for the same swath as each
of the in-situ measurements, as well as for the entire channel
width (bank-to-bank). Differences in the magnitude of the cur-
rent flow between in-situ and RiverSonde ranged from 0.3% to
10% of the in-situ flow for all 3 data sets.
SUMMARY
The initial tests of the RiverSonde radar were encouraging.
Signals with 20–50 dB SNR were achieved with less than 1 W
0-7803-7033-3/01/$10.00 (C) 2001 IEEE

Page 3

10 2030
Distance, m
40 506070 80
0.1
0.2
0.3
0.4
0.5
0.6
Velocity, m s?1
American River, 7 June: Surface Velocity
RiverSonde
Tennis balls
Pygmy?boat
Figure 4: American River velocity profiles for 7 June 2000.
Distance is measured across the channel from the bank near the
receive antenna. The in-situ measurements were from a flow
meter suspended from a small boat about 150 m from the radar
(Pygmy-boat), and from optical tracking of floating tennis balls
(Tennis balls).
of transmitter power at 350 MHz using a bistatic geometry.
Spectral analysis of the echoes indicates that the dominant scat-
tering process is first order. The frequency proved ideal, pro-
viding strong echoes even in morning periods when calm con-
ditions were expected. RMS velocity differences between in-
situ and RiverSonde measurements were 6–18% of the mean,
and volume flow differences were 0.3–10% when using the bot-
tom profile provided by direct sounding with a weighted string.
Some improvements in the system are required, including the
elimination of the cable between transmit antenna and radar
hardware, and improvements in both transmit and receive an-
tennas to reduce signals from unwanted directions. MUSIC di-
rection finding with a 3-element array appears to work well.
REFERENCES
[1] D. E. Barrick, M. W. Evans, and B. L. Weber, “Ocean sur-
face currents mapped by radar,” Science, vol. 198, pp. 138–
144, 1977.
[2] J. D. Paduan and M. S. Cook, “Mapping surface currents
in Monterey Bay with CODAR-type HF radar,” Oceanog-
raphy, vol. 10, no. 2, pp. 49–52, 1997.
[3] D. E. Barrick, “Theory of HF/VHF propagation across the
rough sea. Part II: Application to HF/VHF propagation
above the sea,” Radio Science, vol. 6, pp. 527–533, 1971.
[4] C. C. Teague, “Bistatic radar techniques for observing long
wavelength directional ocean-wave spectra,” IEEE Trans.,
Geoscience Electronics, vol. GE-9, pp. 211–215, Oct 1971.
[5] R. O. Schmidt, “Multiple emitter location and signal pa-
rameter estimation,” IEEE Trans. on Antennas and Propa-
gation, vol. AP-34, pp. 276–280, 1986.
Table 1: RMS Velocity Differences
Data Sets
Delta-Mendota, 5 June
RiverSonde–Pygmy boat
RiverSonde–Pygmy bridge
RiverSonde–Tennis balls
Pygmy boat–Pygmy bridge
Pygmy boat–Tennis balls
Pygmy bridge–Tennis balls
Delta-Mendota, 6 June
RiverSonde–Pygmy boat
RiverSonde–Pygmy bridge
RiverSonde–Tennis balls
Pygmy boat–Pygmy bridge
Pygmy boat–Tennis balls
Pygmy bridge–Tennis balls
American River, 7 June
RiverSonde–Pygmy boat
RiverSonde–Tennis balls
Pygmy boat–Tennis balls
ms–1
% of Mean
0.0668
0.0790
0.0542
0.0608
0.0482
0.0336
15
18
12
14
11
8
0.0370
0.0746
0.0805
0.0564
0.0714
0.0645
8
16
18
12
16
14
0.0581
0.0657
0.0654
15
17
17
Table 2: Volume Flow Measurements
Data Set
In-situ
m3s–1
Same-Swath RS
m3s–1
Delta-Mendota, 5 June
Pygmy meter boat
Pygmy meter bridge
Tennis balls
Bank-to-bank RiverSonde
Delta-Mendota, 6 June
Pygmy meter boat
Pygmy meter bridge
Tennis balls
Bank-to-bank RiverSonde
American River, 7 June
Pygmy meter boat
Tennis balls
Bank-to-bank RiverSonde
33.84
40.83
39.13
34.63
38.06
37.75
38.77
33.50
41.09
36.70
33.22
36.93
33.22
37.86
64.08
65.13
64.25
64.08
70.88
0-7803-7033-3/01/$10.00 (C) 2001 IEEE