DETERMINATION OF FISH NUMBER DENSITY BY A STATISTICAL-ANALYSIS OF BACKSCATTERED SOUND

A theory is presented for estimating the density of fish aggregations from the second normalized moment of intensity of overlapping backscattered acoustic pulses. The chief advantage over the normal echo-integration method is that no knowledge of fish target strength is required. The model assumes that the probability distribution function (PDF) of backscattered amplitudes from the individual fish is Rician, and that within regions of constant density the fish have a Poisson volume distribution. Problems arising from variations in density can largely be overcome by grouping together intensity values from regions of like density and estimating the density for each region separately. It is shown theoretically that the error in estimated density increases with the number of scatterers per resolution cell, which implies that the method is most useful at low fish densities. The theory has been tested at sea on 38-kHz backscattered pulses from two nonhomogeneous layers of small pelagic fish. The results are encouraging judged by their self-consistency and by comparison with estimates obtained by echo integration of the same signals. The method is considered to be a potential alternative or supplement to echo integration in low-density regions.