Improved adaptive regularization for simulated annealing inversion of Transient Electromagnetic

Abstract

Abstract Geophysical inversion usually involves ill-posed problems. Regularization is the most commonly used method to mitigate the ill-posedness faced in geophysical inversions. Adaptive regularization methods have been more widely used in linearized inversion than in stochastic inversion methods. In this paper, we consider the contri-butions of the data fitting term and the stabilizing term in the objective function and give an improved adaptive regularization method for very fast simulated annealing (VFSA) inversion of transient electromagnetic (TEM) data. The optimized regularization method adjusts two terms dynamically in the objective function to make them in balance. In TEM inversion, we compare the results with that of the original SA and the widely used regularization techniques. For synthetic models where layered Earth is considered, we find the improved regularization can balance the data misfit functional and the stabilizing functional. We use arbitrary initial regularization values in the process. We also compare the computational efficiency and find the improved regularization method runs much faster due to the adaptive adjustment of the regularization. The inversion results are similar to that of the conventional ones. Finally, we apply this method to field data, and the inversion results show very good agreements with nearby borehole data.