BROADBAND SOURCE SIGNATURE EXTRACTION FROM UNDERWATER ACOUSTICS DATA WITH SPARSE ENVIRONMENTAL INFORMATION

Deterministic methods for source signature extraction require computation of the Green’s function (GF) for use in deconvolution of the measured time series. In shallow water environments, the GF can depend strongly on the sediment bottom geoacoustics parameters, which are usually not known accurately, and are sampled sparsely. Hence, it is desirable to consider alternatives to deterministic methods of source extraction in such environments. Such alternatives generally characterize the GF in a statistical manner, rather than require its explicit calculation. We illustrate with an example where the GF can be characterized by its kurtosis statistic, a measure of non‐Gaussianity.