Demonstration of Advanced EMI Models for Live-Site UXO Discrimination at San Luis Obispo, California

Abstract : This demonstration is designed to illustrate the discrimination performance at a challenging live-site of advanced electromagnetic induction (EMI) modeling approaches, that are based on the normalized surface magnetic source NSMS model. The model is an extension of the simple dipole model, and provides better accuracy and discrimination ability. The approach combines 1. the NSMS for modeling targets EMI responses; 2. the differential evolution (DE) algorithm for nonlinear optimization to locate the target, and 3. Statistical classification approach for classifying targets as UXO and non-UXO. The study used cued data sets collected at San Luis Obispo, California using three next-generation EMI sensors, the Geometrics MetalMapper (MM), the Time-domain Electro-Magnetic Towed Array Detection System (TEMTADS) developed by the NRL and G&G Sciences, and Berkley UXO Discriminator (BUD) developed at the Lawrence Berkeley National Laboratory (LBNL). The site was contaminated with 60 mm, 81 mm, 2.36 inch and 4.2 inch munitions. During this study, first targets extrinsic (location and depth) and intrinsic (the total NSMS, which depends on its size, shape and material properties) parameters were estimated from the data. Then, the inverted intrinsic parameters were used to classify the targets, and finally, sensor-specific dig-lists were generated for each EMI instrument and submitted to the Institute of Defense Analyses (IDA) for independent scoring.