Double-wavelet Double-difference elastic full-waveform inversion
Xin Fu, Scott Keating, Kristopher A. H. Innanen, Qi Hu
Full-wave inversion (FWI) based on the wave equation has been employed extensively in geophysics. Time-lapse FWI that can detect time-lapse property changes of the subsurface with a high resolution has become an important tool. As a popular inversion time-lapse strategy, double-difference FWI (DDFWI) contains twice inversions, the first inversion is the baseline inversion, in which the input elements are the baseline data and a reasonable initial model, in the second monitoring inversion, DDFWI uses the starting model of the inverted baseline model and a composited data as an alternative of the monitoring data, that is, the difference data (the difference between the monitoring data and the baseline data) plus the synthetic data of the inverted baseline model. Since DDFWI is using the difference data, which helps it to focus on the target time-lapse area, thus DDFWI has fewer coherent errors in the inverted time-lapse model. But DDFWI also is of the shortcoming of requiring good repeatability of baseline and monitoring surveys, especially, when the wavelets of baseline data and monitoring data are different, the coherent errors are very heavy. To solve this problem, Fu et al. (2020) have developed a double-wavelet DDFWI method and implemented for the acoustic FWI. In this study, we will expand the double-wavelet method to the elastic FWI.