Synthetic Aperture Sonar (SAS), the underwater sound application of the aperture synthesis method borrowed from Synthetic Aperture Radar (SAR) is investigated by means of oceaa nitdium effects, platform motion effects, and signal processing effects. Synthetic aperture methods, both in radar and in sonar, allow large synthesized apertures, or equivalently synthesized arrays, to be formed from the combined use of signal processing and a small, real aperture transducer (e.g. a radar or a sonar transducer). Large synthetic apertures give rise to relatively narrow synthesized beams, which in turn allow very high resolution focused images to be formed from radar or sonar data. The target in the synthetic aperture sonar case could be the ocean bottom, and the objective could be, for example, ocean-bottom mapping or searching for a sunken vessel.While synthetic aperture radars have been extensively developed and used since the late 1950's, synthetic aperture sonars have been rare except for a number of academic papers and a few experimental systems. A computer-based model has been created to evaluate the imaging potential of a synthetic aperture sonar system within a horizontally stratified ocean. The model includes ocean refraction, spatial and temporal coherence, surface and bottom influences via multipath, deterministic and random platform motion, and a variety of processing options. This model, the most extensive ever used for SAS simulations, has been verified by comparison to a SAS ocean experiment performed by others. By the use of the model, the three dominant influences of ocean, platform, and signal processing may be studied on the performance of synthetic aperture sonar imaging.The ocean influence on SAS imagery is shown to depend on (1) the extent of spatial/temporal coherence which limits the useful aperture length that may be used for sharp imaging, and (2) the refraction profile which must be accounted for in image reconstruction to ensure a match between the real ocean and the estimated ocean in the computer model; this turns out to be a geometric matched fi!tcr.