PREWHITENING FOR IMPROVED DETECTION BY MATCHED-FIELD PROCESSING IN ICE-RIDGING CORRELATED NOISE

Correlated ambient noise at an array can severely impair the ability o f matched-field processing and conventional processing to detect and l ocalize an acoustic source. When components of the correlated noise fi eld are stationary, however, their effects can be reduced through the technique of prewhitening the data covariance matrix. Presented here a re the results of a series of simulations in which prewhitening was ap plied to synthetic data generated for a correlated noise field. This s tudy is relevant to prewhitening in noise fields that arise from dista nt localized biological sources, distant shipping, or ice ridging at l ow frequencies where the noise source remains spatially stable for hou rs. The present simulations modeled noise from ice riding in the Arcti c. In these simulations, a slanted linear array spanning the water col umn of a range-independent environment was used to enable full 3-D loc alization. The ability of prewhitening to enhance signal detection for low signal-to-noise cases was demonstrated for realistic conditions o f covariance estimation of both the noise and data matrices. Under con ditions where the intensity ratio of the ice-ridging noise to spatiall y uncorrelated noise was 20 dB, the sensor signal level required for d etection was found to be 16 dB lower if prewhitening was employed, eve n when the source was positioned directly below one of the ice ridges. These results were obtained for the Bartlett processor, realistic tim e-bandwidth products, and experimentally determined environmental cond itions. (C) 1995 Acoustical Society of America.