We began a three year study in 1987 to test the feasibility of using sonar in the Togiak River to estimate salmon escapements. Current methods rely on periodic aerial surveys and a counting tower at river kilometer 97. Escapement estimates are not available until 10 to 14 days after the salmon enter the river. Water depth and turbidity preclude relocating the tower to the lower river and affect the reliability of aerial surveys. To determine whether an alternative method could be developed to improve the timeliness and accuracy of current escapement monitoring, Bendix sonar units were operated during 1987, 1988, and 1990. Two sonar stations were set up opposite each other at river kilometer 30 and were operated 24 hours per day, seven days per week. Catches from gill nets with 12, 14, and 20 cm stretch mesh, a beach seine, and visual observations were used to estimate species composition. Length and sex data were collected from salmon caught in the nets to assess sampling bias.

In 1987, sonar was used to select optimal sites and enumerate coho salmon. In 1988 and 1990, the sites identified in 1987 were used to estimate the escapement of five salmon species. Sockeye salmon escapement was estimated at 512,581 and 589,321, chinook at 7,698 and 15,098, chum at 246,144 and 134,958, coho at 78,588 and 28,290, and pink at 96,167 and 131,484. Sonar estimates of sockeye salmon were two to three times the Alaska Department of Fish and Game's escapement estimate based on aerial surveys and tower counts. The source of error was probably a combination of over-estimating the total number of targets counted by the sonar and by incorrectly estimating species composition.

Total salmon escapement estimates using sonar may be feasible but several more years of development are needed. Because of the overlapped salmon run timing, estimating species composition appears the most difficult aspect of using sonar for management. Possible improvements include using a larger beach seine or selecting gill net mesh sizes evenly spaced between 10 and 20 cm stretch mesh.

Salmon counts at river kilometer 30 would reduce the lag time between salmon river entry and the escapement estimate to 2-5 days. Any further decrease in lag time, however, would require moving the sonar operations downriver into less desirable braided portions of the river.