The task of discriminating unexploded ordnance (UXO) from non-UXO items is more difficult when sensor data are collected at sites where an electromagnetically active host contributes a large background signal. In general, approaches to UXO discrimination assume that the object of interest is in free-space. Any influence of the background medium is assumed to have been removed by preprocessing or filtering the data. However, the spatial variations of the background signal can be the same size as UXO anomalies. These signal variations can result from small scale topography (due to bumps or dips in the surface), changes in the orientation and height of the sensor relative to the ground, and spatial variations in the electromagnetic properties of the soil (i.e. conductivity and magnetic susceptibility). A high pass filter to remove the geologic signal is often unable to remove smaller scale spatial variations. In addition, this filtering step often introduces artifacts that can bias estimates of the target parameters.
This project addressed the problem of reduction of geological interference on UXO discrimination procedures through a new and innovative discrimination technology to substantially reduce the impact of geological effects on data collected by magnetic and electromagnetic (EMI) sensors.
This project developed methods for processing data collected at sites with magnetic geology. Instead of relying on filtering data directly to remove the background response, instrument position and orientation information were used to model the sensor response due to soils. Computational routines and techniques to simultaneously invert for the parameters of the buried object and the response of the background were developed. EMI data were simultaneously inverted for geologic properties of the background response in addition to the target’s dipole parameters. The response of the magnetic geology was calculated by using an approximate multi-dipole representation of the transmitter. Numerical modeling code for solving Maxwell's equations was developed to quantify the effect of the topography on the EMI signal and to determine the validity of the assumption that the response of a target buried in a magnetic and conductive half-space is additive.
This project’s methods were verified by processing both simulated and field data for the Geonics EM63 time domain electromagnetic sensor and the newly developed Man Portable Vector time domain electromagnetic sensor. Using data acquired over magnetic soil test pits at the Defense Research and Development Canada (DRDC) research facility in Suffield, Alberta, this project demonstrated an ability to accurately recover dipole parameters using the simultaneous inversion method. Numerical modeling code was used to study the two key assumptions of our approximate forward modeling: (1) the response of a target buried in a magnetic and conductive half-space is additive and (2) topography would not be essential. These studies confirmed that these assumptions would not significantly affect the accuracy of recovered target parameters.
This research will provide tools that will improve upon the existing capabilities for discriminating between hazardous UXO and non-hazardous metallic items sites with magnetic geology. Numerical modeling code has been tested that allow for modeling different aspects of EMI at sites with magnetic geology.