Source depth classification by modal decomposition and correlation

Abstract

Various methods have been and are being used to classify and localize submarine targets using real-time information from acoustic sensors. We have investigated another technique which utilizes the environment and its variability to exclude (theoretically) all possible target depths except the correct one. The method involves synthesizing the normal modes using the best available sound speed profile and geoacoustic model as inputs to a complex normal mode model. Both single frequency and broadband fields can be modeled. These synthesized modes are sampled in depth (or range) to form a suite of replicas. The synthesized replicas are then correlated with the modally deconvolved incoming data having the same spatial sampling as the replicas. The ambiguity surface shows relatively high correlation for replicas having the target's true depth. At present, synthesized data containing noise has been used for comparisons. However, the method shows promise of being a useful tool for depth localization. Examples will be presented for both vertical and horizontal arrays.