MEC Classification Parameter Selection Based on Site‐Specific Munitions Characteristics

Parsons has participated in the Environmental Security Technology Certification Program (ESTCP) unexploded ordnance (UXO) Discrimination Study at Camp Sibert, Alabama, and most recently at Camp San Luis Obispo (SLO), California. In 2008, ground truth results from Camp Sibert were used to identify commonly used target parameters from EM61-MK2 data that would improve the classification of munitions and clutter. UXA-Size (size), a parameter based on instrument response across the three major axes of an object, and a decay constant were identified as having the greatest potential for classifying the 4.2-inch mortar seed items found at Camp Sibert. A simple rule-based classification was developed based on size and decay, and the resulting receiver operating characteristic (ROC) curve showed improvement over a previous rule-based classification of the Camp Sibert ground truth results. In 2009, Parsons participated in a similar target classification study for the SLO site. Unlike Camp Sibert, which included only the 4.2-inch mortar, the SLO site included a range of munitions types: 2.36-inch rockets, 4.2-inch mortars, 60-millimeter (mm) projectiles, and 81mm projectiles. This range of munition types made application of the size and decay based classification of anomalies detected using the EM61-MK2 at Camp Sibert more challenging for the SLO site. Because of the variety of munitions types at SLO, a single size parameter could not be identified for effective classification of all munitions types from clutter. Attempts to define unique rules involving size per munitions type were not as diagnostic for the SLO data set due the wider range of size values observed for munitions and clutter at this site. The most effective rule-based classification at the SLO site was to use two sets of decay values from the EM61-MK2 data to classify the munitions and clutter.