Unexploded Ordnance (UXO) discrimination is achieved by extracting parameters from geophysical data that reflect characteristics of the target that generated the measured signal. Model-based parameters are estimated through data inversion, where the optimal parameters are those that produce acceptable agreement between observed and predicted data and satisfy any prior information we have of the target. These parameters are then used as inputs to statistical classification methods to determine the likelihood that the target is, or is not, a UXO. The task of accurately recovering model parameters is more difficult when sensor data are contaminated with geological noise originating from magnetic soils. In regions of highly magnetic soil, magnetometry and electromagnetic sensors often detect large anomalies that are of geologic, rather than of metallic origin. In this thesis I investigate different methods of recovering the dipole polarization tensor from time domain electromagnetic (TEM) data. The different data inversion methods are characterized by the amount of a priori information used. Different a priori information considered include target location and depth estimated from other data sets, and knowledge of the different types of UXO that can be expected at the site. In the first part of this thesis, I assume that the influence of background geology can be removed through a data pre-processing procedures such that the UXO can be assumed to sit in free space. In the second part of this thesis we take a closer look at the influence of viscous remnant magnetization on electromagnetic data. Several software and hardware based approaches are proposed for improving detection and discrimination of UXO in geologically magnetic areas.
展开▼
