Detection, imaging, and especially discrimination of buried unexploded ordnance (UXO) are among the most urgent environmental needs for the Department of Defense (DoD).

UXO of concern include items of widely varying size from 20 mm shells to large bombs, varying depths from the surface to many feet deep, and buried in unknown orientations in earth of varying electromagnetic and magnetic properties. Common systems for UXO applications use electromagnetic induction (EMI) or magnetometer sensors. While EMI and magnetometer sensors have demonstrated relatively high detection rates under many conditions, the ability to discriminate between harmless buried metal objects and UXO is insufficient.

The objective of this project was to evaluate, modify, and test existing magnetic and EMI prototype systems originally designed for other geophysical applications to demonstrate that a combination of modified instrumentation and new interpretation algorithms, considered separately and together, can result in high probability of detection with reduced probability of false alarms.

The project team developed two prototype multi-axis systems, a time domain electromagnetic induction (TEM) system whose excitation waveform is on all the time (ALLTEM) and the Tensor Magnetic Gradiometer System (TMGS). ALLTEM is an active EMI system and has an ability to detect and, in many cases, to distinguish between ferrous and non-ferrous targets. The TMGS is a passive tensor magnetic system that uses magnetic gradients to greatly reduce noise effects from the Earth’s field including motion-induced noise. The researchers developed methods for data processing and inversion for target parameters for both systems. Multiple polarization data have advantages for target parameter inversion, which support target classification.

The project team concluded that it is possible to collect survey-mode (moving platform) data with a multi-axis system such as ALLTEM with position errors and sensor noise levels sufficiently low that dependable inversions for target parameters can be obtained. Therefore target classification with a high level of confidence can be accomplished in many cases without having to stop and collect data at a fixed location. The researchers concluded this based on data gathered at Yuma Proving Ground, Arizona over the Calibration Grid and Blind Test Grid, as well as from subsequent test stand data studies. A key to the success of accurate target classification from moving platform data is accurate position data either integrated in real time into the main data stream as was done with ALLTEM or time-tagged for post-processing as was done for the TMGS. The studies indicate that on flat ground the present generation of real-time kinematic (RTK)-GPS units provide sufficient accuracy when the “fixed” data quality is achieved. A laser total station such as the Leica would be even more accurate, and the cost of these units will decrease and their use will increase. A roll, pitch, and yaw orientation sensor is being added to ALLTEM to handle cases where the ground is not flat. These additional data will improve results from the classification algorithm.

Typically, in excess of 70% of removal action project costs are for non-UXO items using current technology. The achievement of high probability of detection with decreased probability of false alarms by means of enhanced EMI and magnetic sensors and new appropriate modeling and interpretation algorithms can save the DoD billions of dollars.