Objective
This project was undertaken by White River Technologies, Inc. to assess the feasibility of acquiring classification quality data during dynamic Electromagnetic Induction (EMI) surveys of Munitions Response Sites. A dynamic classification survey was performed using the One Pass Time domain EM Array (OPTEMA) at the former Southwestern Proving Ground near Hope, AR. Over the course of six days in September 2015 the team surveyed four acres of Recovery Field 15. Survey activities included mobilization of equipment to the site, initial calibration and instrument verification activities, dynamic data collection over the four-acre area, daily instrument verification and data quality checks, and demobilization.
Technology Description
The OPTEMA sensor comprises an array of five multi-directional transmitters and 14 receivers that are optimally configured to provide EMI characterization across the entire 1.8-meter sensor swath. This capability is the basis for effective dynamic classification since sensor position during dynamic surveys is based on survey transects rather than on a priori target location. In contrast to the sensor positioning requirements for cued surveys, it is likely that a large number of targets will be located at some lateral offset relative to the array center during a dynamic survey. Therefore, high quality EMI characterization across the array swath is critical for successful classification.
Demonstration Results
During the four-acre survey, a production rate of approximately 0.5 acre/hour was achieved. Instrument verification was performed three times daily in the Instrument Verification Strip (IVS) to confirm sensor functionality. Quality checks included evaluation of system positioning accuracy and data sample density, as well as evaluation of classification feature quality (obtained from IVS data analysis). These quality checks ensured that the team departed the site with classification-level data that met the following objectives:
1. 100% coverage of the four-acre site with 1.2 m transect spacing (33% sensor footprint overlap) and average along track sample spacing <8 cm;
2. Dynamic sensor noise levels low enough to enable classification of munitions as small as 20mm projectiles;
3. Positioning accuracy sufficient to determine target source locations to within 15 cm of actual ground truth locations.
Post-survey data analysis activities included dipole model-based inversion of the dynamic data to acquire classification features associated with anomalies. Analyzing these features, classified 2022 anomalies that were intrusively investigated by dig teams for ground truth verification. Of the 2022 anomalies, 29 were Targets of Interest (TOI) and the remaining items were non hazardous debris. Independent scoring of the classification analysis revealed that it correctly classified 100% of the TOI with a clutter rejection rate of 94% and it achieved a clutter rejection rate of 82% at the stop dig point.
Implementation Issues
These results indicate that dynamic classification methods provide an efficient data collection alternative to cued surveys and can produce classification results comparable in quality to those obtained with cued methods.