A quantitative approach for detection of subsurface voids using multi-sensor data fusion

Use of multiple analytic tools for a single geophysical problem is a conventional approach to reduce the degree of ambiguity in the interpretation. The multi-sensor approach finds one of the best applications in the detection subsurface voids since a single geophysical tool always produces some ambiguity in the estimated signal, due mainly to measurement errors and near surface geologic noise (clutter). The data integration is often done in a qualitative way whereby each dataset is interpreted independently, and the resulting individual models are compared and contrasted for the final geologic model. Alternatively, more robust data integration can be made by the fusion of geophysical data at various levels of processing (using fully quantitative methods) prior to the final geologic model. However, setting up novel methodologies for the latter requires a clear understanding of the similarities and differences of the geophysical methods. Starting from fundamental mathematical formulations, we analyzed the conventional geophysical methods in a comparative, quantitative approach for near surface geophysics with an emphasis on subsurface void detection problem. Analogies of the methods can be combined since all of the sensors are "seeing" the same geologic feature; while differences may be useful in a complementary manner in resolving different properties of the geologic target. We suggest that the quantitative combinations of gravity, magnetic, and ground penetrating radar (GPR) methods for void detection, can greatly improve the reliability of the estimation, and reduce the possibility of false alarms or misses.