A Bayesian approach to monitoring and assessing unexploded ordnance remediation progress from munitions testing ranges

We present a statistical methodology which aims to monitor and assess the progress of unexploded ordnance remediation. We explicitly quantify the probability that each buried sensor-identified anomaly is not a target of interest conditional on the information gleaned from anomalies which have been dug and identified. We provide a measure of confidence that the anomalies which remain onsite after remediation are not unexploded ordnance—this measure of confidence is gleaned through Monte Carlo methods. The methodology is iterative in that, at any point in the remediation process, we can assess remediation progress and compute the probability that no targets of interest remain given the available dig information.