In 1998, the Defense Science Board (DSB) estimated that it would require $15 billion to clean up only 5% of the sites contaminated with unexploded ordnance (UXO) and, with the current practices, 75% of funds would go toward excavation of benign objects (i.e., false alarms). A major recommendation presented by the DSB was to reduce the false alarm rate for UXO by a factor of 10 within the next decade. Discrimination is central to reducing the false alarm rate for UXO, and this project sought to enhance UXO discrimination.

Magnetometers and metal detectors are used widely for ordnance detection. Magnetometers detect only ferrous metals that are contained in most munitions. Metal detectors detect all metals, ferrous and nonferrous. A magnetometer is a passive sensor that uses the static earth's field as its source, while a metal detector generates its own varying electromagnetic (EM) field. The sensor improved in this project is a combination of the GEM-3 hardware and electromagnetic induction spectroscopy (EMIS). The GEM-3 measures the broadband EM spectrum over a buried target, and the EMIS software attempts to identify the target by matching the observed spectrum with those spectra stored in the library for known UXO targets.

The broadband EM sensor improvement and demonstration undertaken in this project took the prototype GEM-3 and evolved it into an operational sensor with increased bandwidth and dynamic range and with enough memory and processing power to allow efficient data acquisition while decreasing the weight for ease of operation. In addition to the original handheld configuration specified at the outset, a cart-mounted large-coil configuration also has been developed for surveying large open areas. Specifically, the frequency range was doubled from 24 kHz to 48 kHz; 24-bit ADC replaced the 16-bit ADC, providing 48 dB of increased dynamic range; digital signal processor allowed continuous operation at more than 10 frequencies simultaneously; and compact-flash memory accommodated internal storage of data from a day of surveying. Performance in moguls and woods was not consistent with prior experience with detection capability of the handheld GEM-3, and the project team believes a problem with the target geo-referencing may have occurred. Detection performance was more reasonable in other areas, but discrimination capability has yet to be realized.

Further improvements may help achieve a discrimination capability for this system, including an increased range capability for detection of deeper targets and improved detection channel algorithms and discrimination algorithms. The ultimate goal is to help lower the extensive UXO cleanup cost for military lands through enhanced UXO discrimination. (Project Completed - 2006)