Joint inversion of gravity and magnetic data using correspondence maps with application to geothermal fields

Abstract

SUMMARY Current geophysical joint inversion methods can be roughly divided into structurally or property driven. While structural coupling plays a key role in fully heterogeneous environments, it is blind to the actual physical properties, which limit its application. On the other hand, approaches based on direct property-correlations promise strong coupled models using petrophysical linkages derived from rock physics or through the search for cross-property relationships in the objective function directly, for example, fuzzy-clusters or correspondence maps. We apply a correspondence maps (CM) methodology to the joint estimation of density and magnetization in 3-D environments using gravity and magnetic data. CM joint inversion has yet to be applied to the specific combination of gravity and magnetic data, nor compared to other approaches like cross-gradient coupling or fuzzy-clustering. The CM searches for cross-property functional relationships; it has the potential to allow a gradual property variation that is not restricted to a predetermined number of homogeneous zones. We tested the algorithm using two illustrative synthetic examples to analyse the compositional and geometrical improvements and thoroughly investigate its capacity to recover a study model based on potential features in a geothermal scenario using polynomials. We investigated the case of using lower and higher polynomials than required for recovering the second synthetic model. The nature of the inversion suggested instability when high-grade polynomials were used, which was successfully overcome by gradually increasing the number of coefficients. The strategy was satisfactorily implemented to the field data. We applied the algorithm over the Los Humeros geothermal field in Mexico and fit a quadratic relationship between density and magnetization that exhibits two main zones consistent with previous works. The first one contains relatively low density and high magnetization. The second zone contains low density and low magnetization values. In this last zone, we emphasize that an observed link between low density, low magnetization, and major fault zones with high relevance to geothermal exploration is resulting from joint inversion, only. We achieved an enhanced petrophysical and structural integrated subsurface imaging in synthetic and field cases at the expense of slightly increased RMS misfit value.