Sea Surface Simulator for Testing A Synthetic Aperture Sonar

With the move to use side-looking imaging sonars in very shallow waters as a component part of MCM operations, synthetic aperture sonars (SAS) appear to have some advantages over a conventional real aperture side-looking sonars. One significant advantage of SAS is that it is quite resiliant to image degradation caused by surface backscatter and surface multipath. The processing in all SAS imaging algorithm assumes the only thing moving between transmitted pings is the sonar platform. Since the algorithm uses coherent integration to assemble the final image, any movement of the sea surface between pings destroys the ping-to-ping coherence of the surface multipath as well as the ping-to-ping surface backscattered return. To move towards understanding just how effective a SAS is at supressing backscatter and surface multipath, we first need to model the moving sea surface in a believable way and establish just how the sound reflects off the undersurface of the sea. This paper first describes a commonly-used physically justifiable sea-surface autocorrelation function that accounts for wind direction, wave height, wave period and wave velocity. From this autocorrelation function, a statistically appropriate random wave surface is generated which evolves in both time and space. Finally in a first attempt to model the shallow-water sea surface multipath problem, a set of impulse responses are generated from this wave-surface as it evolves in time increments equal to the pulse repetition period. Here we model an isotropic one-way (reflected) acoustic path from the target at a depth of seven metres to the sonar platform at a depth of five metres separated by 25 m with the surface above the path covering an area of 160 m (cross-track) by 60 m (along-track) and we ignore any seafloor multi-path. Two sea-surface reflection/scattering mechanisms are used in this model. In the first, each surface facet acts as a diffraction-limited aperture and in the second, each facet acts as a Lambertian reflector. These descibe two limiting situations 1) when the acoustic wavelength is small compared with the roughness of any facet and 2) when the surface roughness is a significant proportion of the acoustic wavelength. Concentrating on the diffraction-limited model, we show the effect of surface multipath on the raw data collected by a SAS and its effect on the processed image. We also make some estimates of the signal to clutter ratio improvements as a function of the number of hits on target.