Technologies are needed that can detect and discriminate unexploded ordnance (UXO), which range in size from 20- mm shells to 2000-lb bombs, from other items in the subsurface. The development of cost-effective detection and discrimination technologies is needed in the following three categories: (1) integrated systems that can cost-effectively survey large tracks of land, detect potential UXO, and discriminate UXO from clutter; (2) systems that are cued by other survey technologies that can cost-effectively and non-invasively interrogate the suspected item; and (3) processing technologies that can exploit the current state-ofthe- art sensors to improve discrimination capabilities.

The objective of this SERDP Exploratory Development (SEED) project was to demonstrate a matched-filter data processing technique applied to measure synthetic aperture ground penetrating radar (GPR) data.

This project involved the development of a matched filter processing technique to improve the discrimination of UXO targets detectable by an existing GPR. This processing technique exploits target-specific radar scattering features directly in the image formation process. The GPR used for this project was a mobile standoff system (i.e., the GPR antenna is remote from the ground surface) that uses synthetic aperture radar (SAR) techniques to form true threedimensional imagery of the subsurface environment.

A limited but reasonably diverse UXO target set was defined. For every UXO target in the set, an appropriate matched filter was generated using the NEC-4 computer code to model radar scattering. The anticipated performance was simulated by the matched filters. SAR/GPR data was collected under controlled conditions on inert UXO test targets. The measured GPR data was processed using three-dimensional SAR processing with and without the matched-filter processing. The performance was assessed by using the measured data to analyze the impact of the data collection and processing technique on reducing false positives.

The standoff GPR sensor was able to detect a variety of surface and buried UXO ranging in size from small (MK II grenade) to moderately sized (105-mm shell) targets. Furthermore, the matched filter processing technique was able to achieve significant processing gain for certain surface targets, indicating the potential for discriminating UXO targets. However, the matched filter processing, as currently implemented, was unsuccessful in achieving any significant processing gains when used against buried targets, resulting in negligible discrimination capability for this case. The discrepancy between the matched filter simulation modeling, which indicated a significant potential for discrimination, and the actual data for buried targets is attributable to limitations in the models underlying the matched-filter SAR image formation algorithm.

Development of an effective matched-filter processing technique for use with a SAR GPR offers the potential for improvement in UXO discrimination. When coupled with the high surveying productivity of a standoff GPR, the technique could produce site characterization reports featuring good detection and low false positive rates cost-effectively. (SEED Project Completed - 2001)