Abstract

Inversion of Controlled Source Electromagnetic (CSEM) data, for 1D model has been attempted as an optimum and overparameterized inverse problem. For these tests two scenarios has been considered. First scenario, when the number of layers in true model is available and second scenario is when no such information is available. For first scenario attempt is made to solve problem with optimum parameterization while for second scenario problem is solved using overparameterization. To discuss these two scenarios we have utilized two schemes of inversion, first (scheme 1) only invert for conductivity and second (scheme 2) which invert for both conductivity and thickness of layers. In case of first scenario, inversion scheme 1 needs a systematic trial and error approach to guess the thickness and depth of resistive layers. In inversion scheme 2 inverted model parameter can be estimated in one pass. For second scenario, due to overparameterization, problem is solved for the smooth (minimum information) solution. Due to over-parameterization parameter (conductivity) estimation developed mild oscillation around sharp conductivity contrast and in deeper parts of model space. These observations are valid for both the schemes of inversion. It has also been noted that inversion scheme 2 required more educated guess than scheme 1.