This project evaluated the feasibility of combining the sensitivity to shape offered by electromagnetic induction (EMI) with the magnetometer sensors' proven location accuracy to dramatically reduce the false positive rate in UXO surveys. The modified Multi-Sensor Towed Array Detection System (MTADS) deploys arrays of total field magnetometers and enhanced EMI sensors. Magnetometer anomaly characterization yields accurate target location and depth and an approximate size estimate (± 25 percent). The measured EMI signal depends on the target shape and orientation, especially the length-to-width aspect ratio of the target. This provides a useful method for distinguishing UXO (typical aspect ratio of 4 to 5) from ordnance fragments (aspect ratios less than 1 or greater than 10). A composite fitting algorithm was developed to use data from both sensors to improve ordnance characterization and false target rejection. This algorithm accommodates the strengths and weaknesses of the individual sensors and weights the data from each sensor accordingly. The demonstration was performed in August 1999 on a live range, the "L" range, at the Army Research Laboratory's Blossom Point Research Facility in Charles County, MD. Known firings at this range include 81-mm mortars, 2.75-inch rockets, 60-mm mortars, 75-mm mortars, and various barrage rockets.

For a single ordnance item, 81-mm mortars, a roughly 60 percent reduction in false positives was achieved without impacting the probability of detection. In order to identify the small fuzes in this field as ordnance, a large number of clutter items had to be included, only approximately 25 percent of which were able to be rejected. In part, this is the inevitable result of trying to discriminate ordnance, ranging in size from fuzes to 5-inch rockets, from clutter. This difficulty may be mitigated by obtaining more data, hence better fit statistics, on the smaller ordnance items.

Reductions in the false-positive rate for UXO detection methods translate into significant cost and time savings as fewer non-UXO items are excavated. Compared to a traditional "Mag and Flag" survey, these methods cost 25 percent more on a per acre basis. Considering that "Mag and Flag" surveys only detect approximately 35 percent of deeper targets, these methods are far more cost effective on a detected target basis. In addition, the false alarm rejection is not applicable to a "Mag and Flag" survey.

MTADS can detect and locate ordnance with accuracies on the order of 15 centimeters. However, even with careful mission planning and preliminary training there are still significant numbers of non-ordnance targets selected. Thus, current surveys are subject to data interpretation by the user and more effective discrimination algorithms are required. (Project Completed - 2001)