Based on extensive assessments of other sensor technologies carried out at Naval Surface Warfare Center Panama City Division (NSWC PCD) for underwater Navy applications, sonar is expected to play an indispensable role in underwater unexploded ordnance (UXO) remediation. Acoustics can be used to probe for targets over a significant range and, being a wave phenomenon, can be used to image buried targets for discrimination from clutter. However, environmental factors can make detection and discrimination problematic, often making imagery insufficient to discriminate targets from clutter.

The objective of this research was to work towards resolving issues that affect sonar detection and classification/identification (C/ID) of underwater UXO using sonar.

Technical Approach

This work leveraged on-going Navy sponsored sonar tests to collect data to further the model development and validation needed to keep sonar models and simulations such as Personal Computer Shallow Water Acoustic Toolset (PC SWAT) and the more recent finite-element-based models up-to-date for UXO applications. This modeling capability and test data was then used both to build a database of sonar target signals useful for developing and evaluating C/ID algorithms that separate UXO from bottom clutter and to look for and understand target signatures that appear sufficiently unique for classification.


Work carried out during 2009-2011 covered four primary areas: controlled pond and tank measurements, finite element (FE) development and modeling, classification analysis, and a laboratory study of muddy sediments. Sonar target data in both monostatic and bistatic configurations were collected at NSWC PCD’s freshwater test pond in 2009-2010, which were processed to provide representations of target intensity in a variety of spaces (coordinate space for imagery, frequency vs. target aspect, frequency vs. time, etc.). Data were also collected within NSWC PCD’s small-scale test tank on a 1/16-scale UXO and other simple target shapes in 2011 to study more diverse scattering configurations not accessible in the full-scale measurements. FE algorithms were developed and used to generate plots of target strength as a function of aspect angle and frequency to compare with the experimental results. FE represents another cost-effective alternative to field measurements for creating databases of real target and clutter responses needed for classification analysis. Classification analyses performed on target acoustic data collected in the freshwater pond demonstrated the feasibility of class separating different targets using features derived from non-image representations of the target. Unlike image-based classification, this methodology was shown capable of discriminating between targets of the same size and shape but different material composition. Finally, research carried out by Boston University to estimate or measure the environmental parameters of muddy sediments needed in sonar simulations resulted in an electro-chemical model of mud that explains sound speed and attenuation trends seen in data.


A substantial database of target responses is often required to train and test C/ID algorithms. This project’s data augment existing databases and the models developed and validated enable further augmentation through simulation. Furthermore, this work supports a physics-based understanding of target responses to enable better selection of classification features, which would be more robust against environmental factors.