It is estimated that unexploded ordnance (UXO) may contaminate 15 million acres or more within the United States alone. A need for improved technologies for mapping and detecting UXO has led to development of a sequence of airborne reconnaissance systems using electromagnetic and magnetic sensors. The benefits of vertical gradient (VG) configurations in magnetometer systems are common knowledge, and these configurations are routinely used in ground-based UXO investigations. Overall, airborne systems provide a tool for wide area assessment (WAA) to support evaluation and reconnaissance over large Department of Defense (DoD) sites where only a portion of the site is contaminated with ordnance.

There were two objectives for this demonstration: 1) assessing the effectiveness of two vertical magnetic gradient systems for mapping and detecting small ordnance items and 2) assessing the effectiveness of the VG configurations for WAA applications. The demonstration site for this project was used for previous WAA demonstrations and therefore provided a basis for achieving this second objective.

In 2002, Battelle staff (then at Oak Ridge National Laboratory [ORNL]) evaluated a prototype airborne vertical magnetic gradient system for mapping and detecting UXO. At least two categories of magnetic noise influence the effectiveness of airborne systems for UXO mapping and detection—rotor noise and maneuver noise. Both have been shown to be effectively reduced by pairing magnetometers in a vertical magnetic gradient configuration. Based on the success of the 2002 tests, Battelle committed corporate funds to design and construct two new systems, both employing the VG concept. Both systems were intended as an improvement over earlier systems, which showed only moderate performance in the detection of mid-size ordnance, 81mm and smaller.

Airborne total field systems demonstrated detection rates of less than 50% for these ordnance types in earlier performance assessments. The vertical magnetic gradient systems of VG-16 and VG-22, differ in the number of magnetometers as well as the separation between magnetometer pods (where a pod houses two magnetometers) and in their swath width. The VG-16 system employs 16 cesium-vapor airborne-quality magnetometers, and has 1.7-m horizontal separation between magnetometer pods rendering a 12-m swath width. In contrast, VG-22 employs 22 cesium-vapor airborne-quality magnetometers in a similar vertical magnetic gradient configuration with 1.0-m horizontal separation between seven magnetometer pods in the foreboom structure rendering a 6-m swath width.

Both systems were developed with WAA in mind, with the expectation that VG-16 would provide an improved tool for airborne WAA surveys contaminated with large munitions. The VG-22 system was intended for WAA use at sites where smaller ordnance types could be detected more reliably. In practice, the suitability of a system for WAA or for detection of individual items has been determined on the basis of blind-seeded tests and post-survey validation of anomalies. Lower probability of detection (P_d) may be acceptable for systems used for WAA applications, but the detection systems must still demonstrate the ability to detect some portion of the ordnance types that are of interest at a site. No universal detection thresholds have been established for either type of application, but site-specific detection requirements have been implemented at some sites.

The primary focus of this project was a survey at the Kirtland Precision Bombing Range (KPBR) in New Mexico. The site was established by ESTCP for development and evaluation of WAA technologies. VG-16 and VG-22 were deployed at two areas of the KPBR. A 500-acre site located between the runways at Double Eagle Airport (within the KPBR) was selected as a blind test grid. Approximately 100 seed items were emplaced by ESTCP contractors. Detection of those items was assessed by Institute for Defense Analyses (IDA) staff based on dig lists provided by Battelle for both systems. Data were also acquired by both systems in an area north of the Double Eagle airport (referred to as the “North Area”). For VG-16, surveys within the North Area were conducted within two 500 acre plots, and for VG-22, data were acquired within two 250 acre plots located within the 500 acre VG-16 plots.

The VG-22 system achieved an overall detection (P_d) of 86%, and VG-16 achieved an overall detection of 55% for all ordnance types emplaced in the blind-seed grid, emplaced and assessed by ESTCP with support from IDA. When broken down by ordnance type, these results exceed the anticipated performance of both systems for all ordnance types except 60 mm and represent a significant improvement over the performance of airborne systems in previous blind-seeded assessments. The 60 mm emplaced at KPBR lacked tail fins and nose cones, resulting in lower mass than other ordnance types. Review of the “missed” anomalies indicated that nearly all anomalies were detected, but were not selected because the detection threshold was too high. This project would have achieved an overall P_d of 98% with VG-22 if a detection threshold of 2.0 nano Tesla per meter (nT/m) was chosen instead of the 2.5 nT/m that was used. If only the medium-to-large ordnance types are considered (81 mm and larger), the detection rates increase to 100% for VG-22 and 90% for VG-16, using the original 2.5 nT/m threshold.

The success of the VG-22 system in this WAA and the high quality of maps derived from VG-22 data have led the project team to consider whether the system might be suitable for applications that would normally be restricted to ground-based systems, where detection of individual items is required. The KPBR demonstration included only a few small ordnance items (e.g. 40 mm-60 mm) as it was assumed that the performance of airborne systems would be poor for such items. A more thorough assessment of VG-22 performance with small ordnance items is recommended in order to determine its suitability for broader applications. Until such an assessment is conducted, the applicability of the VG-22 system for broader use will remain an open question.

Validation results from the North Area at KPBR, where M-38 practice bombs are the predominant ordnance type, were largely inconclusive because there were few if any UXO-like objects recovered from 260 excavations. Pseudo-receiver operating characteristic (ROC) curves for the North Area have P_d represented by the fraction of point-like targets and probability of false positive (P_fp) represented by the fraction of non-point-like targets.

Issues related to this demonstration project center on the appropriate use of the technology. Clearly, the improved airborne system is unable to detect all UXO items of potential interest. This may not be critical for WAA surveys, where detection of a portion of the target ordnance items or detection of concentrations of small ordnance items is acceptable. Airborne geophysical systems continue to be constrained by the presence of tall vegetation and rough terrain that increases the distance between the system and the UXO items of interest, thereby limiting detection ability. This has been shown to be less problematic for VG systems than for total field systems. It remains apparent that application of the technology to small survey areas will not be cost-effective due to the large cost associated with mobilization/demobilization and considerable helicopter costs. Users should consider both the intended UXO targets and survey area (size, terrain, and vegetation) before considering the use of airborne systems for UXO detection, mapping, and WAA.