Recent Advances in Data Selection and Data Conditioning for FWI

In the latest years FWI started to be used to improve the velocity models in a commercial way. But it is still an expensive process. In this paper we show the application of batch sampling as a tool to decrease the FWI run time. Moreover, we present the comparison between Gaussian filter and Structural filter used to smooth the gradient field in FWI process. The tests show that batch sampling can reduce the FWI run time without damaging the velocity update, while structural smoothing preserves the geological characteristics in the gradient field.