Mapping lateral changes in conductance of a thin sheet using time-domain inductive electromagnetic data

Abstract

With the inductive electromagnetic geophysical method, the laterally varying conductance of thin sheet models can be estimated either through a direct transform of the measured data or through inversion. The direct transform (called the simplified solution) does not require grid or line data and is simple enough to be performed in the field because the conductance at a location is calculated directly from the ratio of two measured magnetic fields (the vertical spatial and temporal derivative of the vertical magnetic field) at that location. However, the simplified solution assumes that the secondary horizontal magnetic fields are zero and/or that the sheet has a uniform conductance. Our nonapproximate solution (called the full inversion) does not make these assumptions, but requires gridded data, measurements of the secondary horizontal magnetic fields, and more complicated inversion algorithms. Through forward modeling, we found that the full inversion provides better results than the simplified solution when the spatial gradient of the resistance is strong and/or when the horizontal magnetic fields are large. Because the simplified solution may be preferable due to its simplicity, we introduce two unreliability parameters, which assess the unreliability of the conductance calculated using the simplified solution. A comparison of the simplified solution and full inversion in a fixed in-loop survey collected overtop a dry tailings pond in Sudbury, Ontario, Canada, revealed that there were small differences around large conductance contrasts, which coincided with elevated unreliability parameters. The simplified solution is recommended if fast in-field interpretations are required, or additionally, as a first-pass survey that can be performed with sparse station spacing to identify areas of interest. Denser grid data can then be collected, for the more reliable full inversion, over areas of interest and/or zones where the simplified solution is expected to be unreliable as predicted by the unreliability parameters.