The Department of Defense (DoD) is involved in a number of unexploded ordnance (UXO) site remediation efforts where the application of advanced technologies can potentially improve UXO detection efficiency, lead to substantial cost savings by reducing false alarms, and significantly expedite the transfer of lands for reuse. One of the most prominent of these efforts is the UXO cleanup of the Kaho’olawe, Hawaii bombing ranges where significant magnetic anomalies from geologic sources and near-surface metal fragments make traditional magnetometer-based surveys impractical. Even using active electromagnetic induction (EMI) sensors for the Kaho’olawe remediation has resulted in only 2.7% of detected anomalies being due to UXO with the remainder attributable to geology (27%) and scrap metal (70.3%). The application of advanced technologies that could reduce this false alarm rate would offer tremendous savings.


The primary technical objective of this demonstration project was to evaluate the detection and discrimination capabilities (including production rates and costs) of advanced UXO systems in difficult magnetic clutter environments such as those encountered at Kaho’olawe, Hawaii. A preliminary demonstration was performed at Jefferson Proving Ground (JPG), Indiana. The evaluation objectives for the demonstrations were to:

  • Compare the performance of the advanced systems with the baseline mag-and-flag technologies at JPG and other technologies currently employed at Kaho’olawe.
  • Provide high-quality, well ground-truthed, georeferenced data for post-demonstration analysis and development of receiver operating characteristic (ROC) curves.

Demonstration Results

The detection performance of all demonstrated systems on Kaho’olawe was considerably lower than expected and significantly lower than achieved during the preliminary demonstration at JPG. None of the systems demonstrated any ability to discriminate ordnance from metallic clutter, much less to identify ordnance by size or type in this environment (extreme clutter density, geologic noise, and operationally difficult). It was not possible to evaluate the systems’ ability to discriminate ordnance from geologic anomalies because the demonstration area was so cluttered with unknown metallic objects that the effects of geology could not be reliably separated. None of the demonstrators reached the probability of detection required for Kaho’olawe Tier II clearance (85%). Most significantly, none of the advanced EMI systems demonstrated significant improved capability over the baseline EM-61 system operated in the EM-and-Flag mode.

Implementation Issues

Currently available sensors are not capable of acceptable detection or discrimination performance in difficult environments such as Kaho’olawe. Advanced sensors being developed by SERDP and demonstrated by ESTCP show promise for these tasks but will require validation at challenging sites before routine use.