Localization of impulsive sources in the ocean using the method of images

Impulsive transient signals produced by snapping shrimp dominate the high frequency ambient noise in warm shallow waters. A receiver not only observes the direct arrival of the snap from the shrimp, but also multipath reflections from scattering boundaries such as the sea surface, seabed, etc. We formulate a geometric model corresponding to one direct arrival and one surface reflection. The originating location of a snap can be extracted from the model if the exact position of each element in the receiver array is known. In practice, we only have coarse prior knowledge of the array depth and orientation, and the exact array element locations have to be estimated. The situation is further complicated by the fact that many snaps occur concurrently, and associating a snap with its reflection is a nontrivial problem. We use least logarithmic absolute (LLA) criteria to explore the unknown parameter in this joint estimation and association problem. The association can be further refined based on the complete geometric model. We outline our method and present a numerical simulation to test the performance of the estimator. Results from a dataset collected in Singapore waters using a broadband small-aperture phased array are illustrated. We compare the results with previous findings, demonstrating that accurate passive localization of known structures in the ocean using natural noise made by snapping shrimp inhabiting the structures can be achieved.