An improved differential evolution algorithm for nonlinear inversion of earthquake dislocation

Abstract: Geophysical inversion for earthquake dislocation source model from the observed crustal deformation field is a nonlinear multimodal problem. Although a lot of nonlinear algorithms have been developed, scientists are pursuing a rapid and accurate method to achieve more stable inversion solutions. Differential evolution, an improved Genetic Evolution algorithm, is implemented to solve the problem. The algorithm is fulfilled by Python 2.7 and tested for the 2004 Mw6.0 Parkfield earthquake, the 2009 M w6.3 L' aquila earthquake, and a virtual M w7.3 earthquake. The inversion results demonstrate that the differential evolution algorithm is not only simple and straightforward to implement, but also robust with impressive precision even if all the 9 model parameters are loosely constrained.