Cubic Spline Regularization Applied to 1D Magnetotelluric Inverse Modeling in Geothermal AreasI

Abstract

Electrical resistivity interpretations in geothermal areas typically show a three domain geoelectric structure consisting in an upper resistive horizon, an intermediate and highly-conductive layer (the cap-rock) and a fairly resistive zone, potentially hosting the geothermal reservoir. The clay cap usually give rise to low resistivity anomalies that represent attractive targets for magnetotelluric exploration. Different algorithms are available to get 1D realistic pictures of the geoelectric structure from magnetotelluric soundings, such as blocky/sharp boundaries or smoothed, Occam-type approaches. Here we introduce a new method to model 1D resistivity profiles. In our approach, a cubic spline is assumed to be a realistic approximation of the true resistivity-depth distribution. The advantages of the method are twofold. First, cubic splines need a much smaller number of model parameters to be defined than the layered models used in smoothed inversion algorithms; this can be particularly useful in pseudo 2D/3D inversions or Monte Carlo approaches, where a large number of 1D models must be solved. Second, being the spline intrinsically a smooth 1D resistivity depth profile, there is no need of solution regularization. An example of application of this approach to MT single site data from a geothermal field is presented, and a comparison with blocky and smoothed models and data from a deep exploratory well is made.