Technical Approach

This project will evaluate probability of detection (Pd) by inserting synthetic targets directly into geophysical survey data and processing them through subsequent stages. The synthetic targets will include a large number of challenging cases (deep burials, weak signals), allowing extrapolation of failure rates to high-confidence cases, which are otherwise virtually impossible to obtain since examples of failures are rare for shallow targets with strong signals. The targets will be inserted at a variety of depths and orientations. By superimposing synthetic targets on the measured EMI data, researchers will be able to determine the spatially dependent probability that one or more UXO were missed during the detection phase. Uncertainty in the estimated values of Pd will be determined through the observed binomial proportions of the synthetic target trials. In the case of Pd, a successful trial is one in which the target picker detects a new anomaly within some radius of the true synthetic TOI. Residual risk can be recalculated for any desired definition of this threshold, which also governs the number of excavations required. Sets of synthetic targets will be grouped based on the value of the independent variable, and uncertainty will be estimated through confidence intervals placed on the observed binomial proportions of their results. The variability of intrinsic polarizabilities extracted from the combined signatures also will be documented.