An iterative algorithm for conductivity image reconstruction from crosswell EM measurements

An efficient iterative modeling and inversion algorithm is developed to reconstruct conductivity images using data obtained from crosswell electromagnetic measurements. The forward modeling uses a distorted Born approximation, in which cylindrical conductivity distributions are considered to be deviations from a 1-D background conductivity. The inversion algorithm employs a constrained optimization method to achieve maximum stability when noise is present. Both synthetic and field-measured data are inverted in this paper. The method developed starts with an estimated homogeneous formation and reconstructs a 1-D conductivity image; the 2-D image reconstruction uses the reconstructed 1-D image as the initial input in the inversion process. Conductivity tomographs of two synthetic models and of the Devine Test Site are presented. The operating frequency used in the examples is 512 Hz and a range of 100 x 100 m(2) is imaged. The images reconstructed for the tested cases show a satisfactory agreement with the known geological properties.