Computation and accuracy trade-offs in applied reverberation modeling

Abstract

Reverberation modeling and prediction are a research area of intense interest, often involving large, high resolution databases and advanced numerical techniques, and stretching modern computers to their limits. Nevertheless, applications of underwater acoustic systems often need estimates of reverberation sooner than they can be provided by the research community. What results is that applied models for reverberation, in order to meet the pressing needs with available computer memory and allowable computation times, often employ approximations or shortcuts that fall short of current research modeling standards. This paper traces several of the approximations used in reverberation models in the past in order to meet system design and deployment needs. The results show an inherent conflict between basic research and applied interests. The traditional DOD paradigm is that basic research results transition to the applied world. History suggests the opposite, however: Applied research ignores the torpedoes and moves full speed ahead where needed, then forges the chain used to pull reluctant basic research in the needed direction.