Abstract
Acoustic Target strength (TS) of an object is a measure of the net reflection coefficient of that object and is generally a function of both incident and scattered directions as well as frequency. For each pair of directions at each frequency the TS is the ratio of the acoustic pressure incident on the object due to a harmonic plane wave to the acoustic pressure of the reflected far field observation direction (but referenced to 1 meter from the acoustic center of the object). Active sonar systems use scattered field measurements to detect the presence of an underwater object. For this reason target strength modeling is fundamental to the development of active sonar systems such as bathymetric profiling and mobile target detection. The model must accurately capture the echo response of the scattering body. This response can be simulated from a coherent summation of multiple individual feature scatterer responses known as highlights. Highlights can be represented in the form of simple shapes such as spheres and cylinders. Increasing the number of highlights for a given object can increase the fidelity of the model, at the cost of increased model complexity. Such complexities can additionally increase the uncertainty in the model validation process. This thesis advances the goal of an optimal level of model fidelity with an acceptable minimal level of model output uncertainty. This work compares observed underwater target strength measurements of a multi highlight scattering body against outputs from two different target strength scattering models while identifying key areas of uncertainty associated with random platform motion and possibly refraction effects that typically attend real acoustic measurement. Presented here are target strength measurements across the full range of incident and scattered look directions at small wavelengths relative to target size. We show quantitative and qualitative correspondence between modeled and observed target strength measures. Results reveal that the unmodeled effects of platform motion and multipath scattering as well as angle stability of the target frame degrade model fidelity. Additional sources of uncertainty are the necessary censoring of support cable reflections as well as the unknown scattering features of the target present in the observed data. These are qualitatively assessed. We report approximately 22% accuracy at a 90% confidence level, over an aspect range of 200 degrees. For aspect angles, 110 to 130 degrees, the models were most accurate. All frequencies, bandwidths, beam patterns, pulse lengths, and other sonar parameters used for this study are for experimental research only and do not reflect actual torpedo systems in use by the US Navy or its foreign partners and customers.