Sonar detection of riverine fish using the pulse pair convariance doppler frequency estimator

Detecting riverine fish by sonar is a difficult problem because of volume and boundary reverberation. For fish migrating upstream, counter to the flow of the current and debris, estimation of the Doppler shift can significantly increase the accuracy of the fish counts. We have developed a prototype salmon detection system that has both a range resolution of 1.5 m and a Doppler resolution of 60 Hz (equivalent to a velocity resolution of 10 cm/s). The Doppler frequency is calculated by the computationally efficient pulse pair covariance (PPC) technique, a time domain method of determining the mean frequency of a symmetrical spectrum. A three-tiered detection algorithm identifies the presence of a fish when thresholds for energy and frequency are exceeded simultaneously over several pings. This approach maintains good range resolution while providing a more statistically confident measure of the number of migrating fish than traditional echosounding methods. An advanced, cost-effective microcomputer system performs all signal processing tasks in real time. Detected fish are indicated on a monitor for operator verification, and a count of the fish is maintained automatically. The system gives good results except in rivers with extreme reverberation.