Ocean and Atmospheric Sci. Div., Code 541, NRaD, San Diego, CA 92152-5000
Characteristics of Arctic ambient noise include a wide dynamic range, non-Gaussian distributions, nonstationary intervals and a high variability in spectral shape. However, there are times during which these noise fields reflect stationary normally distributed processes. This variability demands that data sets be analyzed for independence, time stationarity, and distribution to verify the assumption of valid statistical ensembles prior to processing, as these properties are prerequisites for many algorithms. A statistical analysis of noise samples at 20 and 75 Hz over a 10-day period during which the omnidirectional levels varied significantly indicate that the change in statistics of the noise field correlates with local ice activity. The quiet intervals are dominated by long-range propagation where the noise exhibits stationary Gaussian characteristics and the noisy intervals are affected by local ice dynamics where the transient nature of the events tend to invoke non-Gaussian as well as nonstationary properties.