Data Collection with Vehicular-Based Systems
Dr. Stephen Billings | Black Tusk Geophysics
Objectives of the Demonstration
The objective of this demonstration was to collect high quality MetalMapper sensor data in a cued interrogation mode over 2,370 anomalies previously detected during an EM61 survey. The data collection was intended to meet the following objectives:
- Demonstrate the transition of the new, but now commercially available, MetalMapper sensor to personnel engaged in production data collection (and not involved in research and development).
- Provide advanced electromagnetic induction (EMI) sensor data from the MetalMapper to a wide range of data analysts to test different processing and interpretation methodologies.
Sky Research’s SKY3D vehicular platform integrates a MetalMapper sensor, real-time-kinematic global positioning system and inertial measurement unit (in the form of the Novatel SPAN), and a custom user interface on a Kubota all-terrain vehicle. Cued interrogation data were collected by dwelling for 30 seconds over 2,370 anomalies previously identified via a full-coverage geophysical survey of part of the Bisbee Area of the Pole Mountain Target and Maneuver Area, near Laramie, Wyoming. The data collection was conducted in the field over a 14-day period in July and August 2011.
The production rate was as high as 39 points per survey hour with a maximum of 326 anomalies visited in a day. During the first eight field days, an electrical issue with the instrument orientation sensor resulted in a lower average production rate of 21 points per hour. During the last six field days, and after the orientation sensor problem was rectified, the production rate increased 64% to an average of 32 points per hour. Weather and summer thunderstorms impacted two days.
Eight performance metrics were evaluated, including four pertaining to the data collection (reliability/robustness, survey rate, percentage of site covered, MetalMapper sensor position accuracy) and four to the subsequent processing and classification of the collected data (percentage of munitions correctly identified, reduction in false-alarm rate, appropriate specification of stop-dig point, and minimization of “can’t analyze” anomalies). Six of the eight performance objectives were easily achieved, while one objective, 90% of MetalMapper positions within 30 cm of each Target of Interest (TOI), was missed by 1 cm (90th percentile at 31 cm). The last and only qualitative objective on reliability/robustness was only partially met due to the electrical issues with the orientation sensor on the Novatel SPAN.
At the final stop-dig point selected after analysis of the MetalMapper data, 369 anomalies were excavated and all 160 TOI were identified. A total of 209 of the 2,210 clutter items were excavated: this meant that 91% of the clutter could have been left in the ground, resulting in significant potential cost savings. The per anomaly costs (excluding mobilization and reporting) were $41.24 for data collection and $4.08 for data processing. The total cost per anomaly was $45.52. Using an often quoted rule of thumb that each excavation costs $100, the excavation costs would have been approximately $237,000 without the MetalMapper system. With the MetalMapper, only 369 anomalies needed to be excavated at a cost of $36,900. When combined with the MetalMapper mobilization, data collection, processing, and reporting costs of $168,270, the cost of clearance using the MetalMapper was $205,170: a savings of $31,830 (or 13%).
Comparing EM61 pick locations versus the actual ground-truth locations of TOI revealed large errors in position and demonstrated the importance of fine-tuning the cued-interrogation location using the (currently limited but functional) real-time position estimation capabilities of the MetalMapper. Nightly quality control of the collected data was another important determinant of cued-interrogation data quality, with 125 anomalies recommended for recollection.