Time-lapse full waveform inversion: Synthetic and real data examples

Abstract

Time-lapse seismic data contains information about changes in the subsurface, and are nowadays used as an effective tool to characterize changes in reservoirs due to production of oil and gas or injection of CO2. The full waveform inversion (FWI) method can be used to detect these changes. Three different approaches for time-lapse FWI are presented: (1) performing independent inversion for each dataset, and then compare the inversion results; (2) first invert for one of the datasets and then use the end model from the first inversion as initial model for the inversion for the other datasets; (3) similar to (2), but the data consists of a combination of the optimal data of previous inversion and the difference between the observed datasets. The three approaches are applied on synthetic and real timelapse marine streamer seismic data. The approaches are able to detect and describe the changes in velocities in the subsurface.