Two-Dimensional Inversion of Oklahoma EMAP Data with Smoothness Regularization

Abstract

Two-dimensional inversion has been applied to Electromagnetic Array Profiling (EMAP) data obtained in Oklahoma, USA. Since the data comprise scalar impedances measured along a survey line, a TM-mode inversion was performed on the original data, under the assumption that the survey line is perpendicular to the geologic strike. The inversion method applied is a linearized least-squares scheme with smoothness regularization. The “optimum smoothness” is selected based on a statistical criterion, ABIC (Akaike's Bayesian Information Criterion), which is derived from Bayesian statistics and the entropy-maximization theorem. Starting from a homogeneous earth as an initial guess, the inversion iteratively modifies the model structure until the observations are matched in a statistical sense and parameter modification becomes almost zero. The final two-dimensional models generally show a very conductive host formation (less than 10 Ω·m) with two large resistive bodies of approximately 100 Ω·m embedded near the middle of the survey line.