An iso-deviant strategy for efficient ambient noise predictions with EAGLE

Abstract

Transmission loss (TL) computations in littoral areas require a dense spatial and azimuthal grid to achieve acceptable accuracy and detail, which is very slow. This problem of accuracy versus speed led to a new concept, OGRES (Objective Grid/Radials using Environmentally sensitive Selection), which produces sparse, irregular acoustic grids, with controlled accuracy. Recent work to further increase accuracy and efficiency with better metrics and interpolation led to EAGLE (Efficient Acoustic Gridder for Littoral Environments). On each iteration, EAGLE produces an acoustic field with approximately constant spatial uncertainty (hence, iso-deviance), yielding TL predictions with ever-increasing resolution and accuracy. This work adapts EAGLE to ambient noise computations. The Dynamic Ambient Noise Model (DANM) allows accurate, detailed estimation of the mean and variance of ambient noise in both space and time, but its TL computations are too slow for many applications. In the present work, a series of EAGLE acoustic field predictions was used by DANM (and compared to the dense full-grid solution) to determine the relationship between transmission loss uncertainty and noise-field uncertainty for a complex littoral area. An example is presented where approximately an order of magnitude efficiency improvement (over regular grids) is demonstrated. [Work sponsored by ONR under the LADC project.]