A Monte Carlo approach for capturing the uncertainty in sonar performance modelling

Abstract

Sonar performance modelling is an essential part of active sonar operations, as it is used both during planning of the operation as well as assessment of the expected coverage managed during the operation. Conventionally, the acoustic model has been input with a single realisation of the present environment in order to generate 2D coverage plots either as vertical cross sections in a selected direction from the sonar or as horizontal coverage plots encircling the sonar. The impact of environmental uncertainty on sonar performance is well known and documented. Errors in the input sound speed may give rise to large errors in the expected sonar performance. Monte Carlo methods are a well known method for capturing this uncertainty. This requires both a multitude of model runs and probability density functions representing the model input instead of single realisations. Here we propose using a fast raytrace model for estimating the sonar performance of a large number of different environmental and target realisations. The model results are then marginalized to collapse all but a few dimensions for efficient and concise presentation of the results for the sonar operator.