Land Full Waveform Inversion for Dispersed Source Arrays

Abstract

Full Waveform Inversion (FWI) is one of enabling methods for estimating subsurface parameters such as velocity, impedance and density. Since Tarantola (1984) had introduced the back-propagation technique in the time domain, FWI has developed steadily in time, Fourier and Laplace-Fourier domains (Bunk et al., 1995; Pratt, 1998; Shin and Cha, 2009). FWI has been used generally with acoustic wave equation and has produced reliable results especially for marine datasets. However, in the application of real land datasets, several challenges remain, such as, near surface scattering noise, ground roll and source and receiver coupling and amplitude irregularities (Plessix et al., 2012). In this paper, we introduce an improved application of FWI on real land dataset acquired using inhomogeneous sources blending mode (i.e.,dispersed source arrays, Berkhout, 2012). The acoustic hybrid FWI methodology was used where the source signal spatial variations were incorporated during the inversion process. Additionally, we add a weighting function in the logarithmic objective function to suppress background noise and elastic waves having large amplitudes (i.e., ground roll).