Sources of uncertainty in Doppler sonar measurements of fish speed

A 250-kHz, 30-kHz bandwidth coherent Doppler sonar was evaluated to determine sources of uncertainty in fish speed measurements. Three separate tests were undertaken: (1) towtank tests using styrofoam balls to simulate fish, (2) tank tests with live free-swimming fish, and (3) field tests with wild free-swimming fish. The standard deviation in a single speed estimate was 9 cm s super(-1) for styrofoam balls, 10-11 cm s super(-1) for swimming fish observed from a dorsal aspect, and 19 cm s super(-1) for swimming fish observed from a caudal aspect. The variation in precision was primarily due to the different signal-to-noise ratio (SNR) in each test: a larger SNR resulted in a smaller standard deviation. Doppler speed estimates were compared with independent estimates of target speed where possible. An accuracy of plus or minus 4 cm s super(-1) was typical of Doppler speed estimates in all the experiments.