Effect of frequency combination on the resolution of 3D land-based CSEM and its optimal design

To maximize the parameter resolution of a 3D resistivity model, we present an experimental design algorithm to achieve a set of frequencies that is considered optimal for targeting the model by using a land-based controlledsource electromagnetic method (CSEM). An objective function (OF) is defined by combining the maximum energy and flatness of the singular values of a sensitivity matrix, which is used to evaluate the quality of one given survey layout. A genetic algorithm (GA) is employed to minimize the OF searching for the optimal frequency combination. The novelty of our algorithm is extending the optimization theory for CSEM survey system from 1D model to the case of 3D model and by using genetic algorithm to deal with the optimization of a huge-scale unknowns. The algorithm can be applied in two fundamental scenarios, i.e., to select (1) the optimal transmission frequencies before field data acquisition and (2) the optimal inverted frequencies after field data acquisition. The algorithm is first tested with one synthetic example, and the results show that more frequencies and a wider band do not ensure an improvement of the inversion result, but instead increase the computational cost. A further tested with real data acquired in a net array shows the advantages of the optimized frequencies that improve the spatial resolution of the resistivity anomalies in the inversion model. Our study provides theoretical and methodological support for the frequency optimization of a land-based CSEM.