Inversion of magnetotelluric data with fuzzy cluster petrophysical and boundary constraints

Abstract

Inverse magnetotelluric (MT) problems are naturally ill-posed and smoothing criteria are typically added to stabilize the process. Smoothing and geo-electrical equivalency tend to produce unrealistic geological models. In reality the subsurface geology is differentiated by distinct rock units that are often better defined by boundaries rather diffuse or smooth boundaries. We present the application of fuzzy clustering as an added constraint within the inversion process to guide model updates toward earth models that resemble geological units. Fuzzy clustering divides the simulated model into clusters based on the similarity of model features. Moreover, fuzzy clustering enables the inclusion of additional prior information in the inversion process such as structural and/or petrophysical information. The inclusion of this information produces geo-electrical distributions that more closely reflect the true rock units and unit boundaries. This is demonstrated through several synthetic examples. The simulations show that by including prior petrophysical and/or boundary location information within the inversion the original conductivity distribution is well resolved.