Statistical characterization of Arctic under-ice acoustic bursts and tonals

A 10-min time series of FRAM II Arctic under-ice noise data is analyzed, extending the results of an earlier study [J. G. Veitch, A. R. Wilks, and S.C. Schwartz, “Characterization of Arctic undersea noise,” pp. 45, Dep. Stat., Princeton University (June 1983), AD-A130 397/3 (No. 22, 1983)]. Specific analysis is conducted on the impulsive noise bursts and tonals that are characteristic of underwater noise when an ice cover is present. Point process modeling of the times of arrival of the impulse bursts is performed by testing the empirical distribution of the burst event interarrival times against an exponential parent distribution, characteristic of a Poisson process. Goodness-of-fit tests are invoked, as well as a test for serial correlation. Results suggest that the burst event point process may be modeled as a piecewise constant nonhomogeneous Poisson process. This model is accepted at > 5% confidence for all tests applied. Though a limited data set is utilized, the analysis methodology developed is applicable to larger sets of data. Spectral content of tonal components are also examined using Thompson's multiple window technique, and a harmonic relationship is found to exist.