Acoustic global FWI for shallow P-wave velocity field estimation

In this paper we present the results of an acoustic Full Waveform Inversion applied on the first arrivals and some far offset reflections of a seismic data set acquired for near-surface investigations. The estimated velocity model obtained with a GA-FWI optimization procedure which employs a two-grid strategy and makes use of the full seismic wavelet, is used to pre-stack depth migrate the data where direct and refraction events have been muted. We use two a priori information with different constraint strength, loose for the semblance velocity field and more tight for model estimated by refraction analysis to mitigate the complexity of the misfit function and reducing the number of unknowns. Due to the absence of well-log measurement in the area, we use pre-stack depth migration to asses the reliability of the final result. The computed CIGs show a fair horizontal alignment of the observed reflections, granting a good accuracy of the proposed solution.