Polarity Encoding Full Waveform Inversion with Prior Model Based on Blend Data

Blend data inversion is an effective way for full waveform inversion which is an effective method for velocity models, for reducing calculation amount. However, with the increasing of iterations, the convergence and stability reduce due to the influence of the direct wave. In order to solve these problems, the paper presents the theory of polarity encoded multi-shot full waveform inversion with prior model. Numerical tests on sample and complex models show that multi-shot FWI with prior model can accelerate the convergence, reduce the influence of low frequency noise, suppress the effect of direct wave, and then improve the quality of deep structures inversion.