Time-lapse full-waveform inversion of limited-offset seismic data using a local migration regularization

Conventional methods for quantifying time-lapse seismic effects rely on a linear assumption that is easily violated. Therefore, more sophisticated methods are necessary. The full-waveform inversion (FWI) method is an inverse method that is able to reveal time-lapse changes in the image domain, in which the conventional methods break down. We investigated the behavior of FWI using different approaches for applying FWI on limited-offset time-lapse data. We compared acoustic and elastic inversion schemes. We introduced a method for constraining the model update for the monitor model to remove time-lapse artifacts. This method was based on migration of the residuals in the time-lapse data, which, in combination with a local contrast estimation algorithm, formed the update constraint. We found that for limited-offset data, elastic theory was necessary for the success of FWI and that FWI was able to quantify the time-lapse changes in the parameter models. The local migration regularization approach was able to remove time-lapse artifacts.