Research topics

Ongoing research projects include:

• UXOLab
  Advanced processing and inversion software library for UXO detection and discrimination. Magnetic and EM (time or frequency domain) data can be imported, inverted individually or cooperatively, and the resultant parameters used in classification. Written in Matlab, the  software is our platform for research and field data processing.

• Practical Discrimination Approaches for Application to Live Sites (ESTCP 0504) The goal is to validate how well our inversion and classification software performs in various field locations. Data from GPO sites are provided as training sets and for the development of workflows. Those workflows are then used to process data in blind test regions to generate dig lists. Success means validation; missing targets stimulates further research and development.

• Optimal Model Selection (SERDP 1637)
  The full electromagnetic forward problems are generally too computationally intensive to solve and hence approximate models are invoked. We investigate dipole and magnetic charge models, and their corresponding parameters, to determine which models are more robust and efficiently used in practical classification problems.

• Robust Statistics (SERDP 1629)
  Estimating UXO parameters from geophysical data requires assigning "correct" errors to the data that will be input into our minimization algorithms. Since the true data errors are never known in practise, our research focuses upon developing strategies so that statistics of the final misfit are consistent with the statistical assumptions used in formulating the inverse problem. This should increase the quality of inversion results and aid classification.
  

• Simultaneous Inversion of UXO Parameters and Background Response (SERDP 1573)
  Soils that exhibit VRM (Viscous Remanant Magnetization), soils that are highly conducting, and rocks or soils that have high magnetic susceptibility each distort the signals due to the UXO. Our goal is to develop practical procedures to extract meaningful information about the UXO in the presence of these disturbances.

• Multi-target Detection
  In field situations signals from adjacent targets can overlap. For inversion, it is essential to determine if an anomaly is due to a single or multiple targets, and, when necessary to invert for parameters for more than a single target.

• Survey Design
  New instruments, with more transmitters and receivers, or with different orientations, are constantly being proposed. Inversion methodologies and simulation tools are developed to determine the new configurations that have higher resolution and cost effectiveness.