Elastic Full Waveform Inversion for CO2 Sequestration monitoring - 1D Synthetic Data Investigations

In greenhouse gas geological sequestration frame, time-lapse seismic has been demonstrated to be a powerful tool for monitoring CO2 injection. So far, it has provided essentially qualitative insights, such as gas cloud extent or reservoir compartment identification. Getting quantitative information requires accessing subsurface elastic properties. This is commonly done through Amplitude Versus Offset analysis. Such methodology is limited in terms of angular aperture and only takes into account reflected data. Use of elastic full wave form inversion (FWFI) can overcome such limitations. In this study we investigate strategies to assess FWFI ability to retrieve some 1D models from synthetic noise free data. Models properties are based on the Sleipner injection site, including shale and gas sand thin beds. Low to high frequencies inversions of pre-injection data retrieves accurately the baseline model. Inversion of post-injection data is more difficult as gas introduces very large velocity decreases. Two approaches are then proposed: first one is based on weak "time-lapse" assumptions and include a rough data pre-conditioning; second one assumes overburden as fixed and invert only high frequency data in the injection layer. This gives encouraging results showing that FWFI is able to provide quantitative insights in such context.