The IEEE Seventh Working Conference on Current Measurement Technology

Current and Wave Monitoring and Emerging Technologies

March 13-15 | Bahia Hotel | San Diego, CA, USA

  
     

Initial River Test of a Monostatic RiverSonde Streamflow Measurement System

Calvin C. Teague

Status: Accepted

2046 Kent Drive

Los Altos , California USA
94024-7027

Phone: 650-723-3596
Email: cal@alpha.stanford.edu

Co-Authors:
Donald E. Barrick, Peter Lilleboe, Ralph T. Cheng

A new RiverSonde streamflow UHF (350 MHz) radar system was tested for two days during May, 2002 along the San Joaquin River at Vernalis, California. In contrast to the first bistatic RiverSonde tests, a monostatic geometry was used for this experiment with the radar antenna placed on one bank of the river. The RiverSonde is derived from a standard SeaSonde system normally used at HF to observe currents on the ocean; it was retuned to a higher frequency consistent with the shorter water wavelengths expected in the river. A four-element yagi was used for the transmit antenna, with a broad azimuthal pattern illuminating the water both upriver and downriver from the radar location. A 3-bay array, consisting of the 4-element transmit yagi and two 5-element yagis displaced from the central transmitting yagi by about 0.5 m, was used for the receiver. The yagis were designed using genetic algorithms to simultaneously optimize the directional response and feed-point impedance, and their performance was verified by field calibration using a small transponder. MUSIC direction finding utilizing both the amplitude and phase responses of all 3 elements was used to locate the echoes. The monostatic geometry with a wide viewing angle proved almost ideal, with the downstream motion of the water spreading the echoes widely in Doppler frequency and almost all energy at a particular frequency bin coming from a single direction. Consequently, nearly all MUSIC direction solutions were single-angle, which generally are more robust than dual-angle solutions.

Data were recorded for about an hour at each of several signal bandwidths; results using 5 m range resolution are presented here. The hour-long run was divided into 20 segments of about 2.5 minutes each, with delay-Doppler and MUSIC direction finding applied to each segment. Each segment yielded about 2500 radial current estimates spread from one bank to the other over about 130 degrees in azimuth. Various data processing techniques were applied to reduce the data, ranging from fitting to a uniform flow to modeling the flow as normal modes with constraints at the banks and boundaries of the analysis region, with the mode coefficients calculated from the data. Cross-channel velocity profiles were calculated for several along-channel positions, and the temporal behavior of these profiles was investigated. The radar-estimated flow ranged from about 0.7 to 1.0 m/s, depending on the location in the river channel, with differences in the details depending on the particular model used. Comparisons between several radar data processing techniques and with in-situ measurements carried out simultaneously by the U. S. Geological Survey will be presented.

Submitted on October 15, 2002