Non-Gaussian sonar clutter models

A new method for interpreting sonar signals in the presence of clutter is presented based on probability distributions following the formalism of the ideal gas. Likelihood detectors, used to evaluate the presence or absence of a desired target, are extremely sensitive to the tails of these distributions. The structure of the tails are highly sensitive to the underlying clutter statistics, implying that detection of a target is extremely sensitive to the underlying physics. We present simulations of non-Gaussian clutter and targets and explore the sensitivity of detection to these physical models and their parameters.