A global optimizing approach for waveform inversion of receiver functions

A global optimizing approach is developed and implemented to retrieve one-dimensional crustal structure by waveform inversion of teleseismic receiver functions. The global optimization for the inversion is performed using a Differential Evolution (DE) algorithm. This modeling approach allows the user to perturb, within a preset range of reasonable bounds, multiple parameters such as Vp, Vp/Vs, thickness and anisotropy of each layer to fit the receiver function waveforms. Compared with linear modeling methods, the global optimal solution can be achieved with fewer model parameters (e.g., a small number of layers) and hence eliminate potential artifacts in the final model. Receiver function bins with small ray parameter intervals are used in the inversion, which can reduce distortion caused by modeling a single receiver function stacked from many recordings spread over a wide range of epicenter distance. The efficacy of this global optimizing approach is demonstrated with synthetic datasets and real receiver functions from the permanent seismic station BJT.