Distortions of magnetotelluric sounding curves due to a slope

This study considers a particular case of 2D modeling, namely the topographic effect on MT data. A simple way to remove topographic effects is to incorporate surface elevations, or depres­sions, into the forward model. In cases where a 2D approximation is valid, any 2D algorithm may be used for the calculations. Such a scheme allows model results and observations to be compared along a profile. This approach, however, does not necessarily provide useful answers since the topography may, and in fact nearly always does, couple with the geoelectric structure and/or varies on a more rapid scale than the observations. In order to understand the effect of topography on MT data the degree and characteristics of the distortion introduced by a simple slope are consid­ered. The slope model constitutes a general result with distortion characteristics that can be ex­tended to other features such as a depression or a hill. The introduction of additional features such as a superimposed geoelectric structure would greatly complicate the problem due to electro­magnetic coupling effects which would necessarily be a function of frequency. The single slope, or half-graben, is probably the simplest class of general near-surface distortions that affect MT data. The period range used is from 0.02 s (50 Hz) to 100 s (0.01 Hz), i.e. both audio and magneto­telluric observations are considered. The elevation of the slope is fixed at a 'throw' of 1 km and slope angles range from 0° to 90°. The spatial and frequency distortions due to the slope are considered in detail. At a particular location the distorting effect of the slope introduces a false gradient into an otherwise lD sounding curve. The topographic distortion introduced is investi­gated by applying a lD inverse procedure to the obtained results. The way in which false layers may be introduced into the geoelectric section is illustrated. An analytical solution to a purely DC topographic problem is also considered. The analytical results are compared with the correspond­ing low frequency results obtained with the numerical models of the present study.