Double multiple signal characterization active large band (D-MUSICAL) for extraction of observable in acoustic tomography

Abstract

In the ocean, the acoustic propagation is characterized by different arrivals. Each arrival is characterized by its observables: amplitude, arrival time, emission, and reception angles. To achieve ocean acoustic tomography, a method to separate arrivals and extract observables has been developed. Two vertical arrays are used: a source and a receive array. Until now, the double-beamforming algorithm has been used to extract observables. It allows to go from the initial (time–source depth–receiver depth) domain to the beamformed domain (time–source angle–receive angle). This algorithm is efficient thanks to the 3-D domain of estimation but shows resolution limitation in the angle domain. To improve wave separation, we develop and apply a high-resolution algorithm based on MUSIC principle, in the same 3-D domain. The proposed method, called Double-MUSICAL, is the extension on the emission angle of the 2-D MUSIC Acitve Large band method (MUSICAL). Its principle is to separate the recording space onto the two orthogonal signal and noise spaces. We present results on real data recorded in an ultrasonic tank and show that the developed method better detects and separates the first arrivals than double beamforming.