Technical Paper: Inversion of CSEM Data with Model-based Inversion Algorithm

Society: EAGE
Paper Number: G004
Presentation Date: 2008

In this abstract, we present a model-based inversion scheme for processing the CSEM data, which reconstructs the conductivities and shapes of the regions of interest based on some a priori information. The required a priori information can come from independent measurements (e.g., seismic) or from the inversion results of the same data using a pixel-based inversion approach (Abubakar et al. (2006)). The model-based inversion presented in this paper, generalizes the so-called sharp boundary inversion approach (see Smith et al. (1999) and Hoversten et al. (2006)). This model-based inversion scheme is based on a Gauss-Newton minimization method with multiplicative regularization and a line search approach to stabilize the inversion process (see Habashy and Abubakar (2004)). The forward solver is based on a two-and-half dimensional frequency domain finite-difference simulator described in Abubakar et al. (2006). In this model-based inversion algorithm, the shape of the various regions, expressed as two dimensional (2D) polygons defined by their vertices, can be reconstructed along with their locations and conductivities. Some initial numerical tests of CSEM models show good reconstructions.

Request More Information