Full Waveform Inversion and Multi-Diagonal Inversion for Monitoring of CO2 Storage at Sleipner

Abstract

Carbon capture and storage (CCS) can become necessary for mitigating climate change. However, storing CO2 in subsurface reservoirs requires thorough monitoring. The most accurate and widespread type of monitoring is seismic monitoring, which involves the acquisition, processing, and interpretation of seismic reflection data. A key step in seis- mic processing is the determination of subsurface parameters using inversion. The best inversion results are obtained using full waveform inversion (FWI). However, FWI meth- ods are computationally very expensive. Therefore, my thesis aims to obtain accurate inversion results at a relatively low cost and has two main objectives. The first objective is to present a new inversion method, which I call multi-diagonal inversion. This method provides results with an accuracy between more accurate FWI methods and conventional imaging, which uses the diagonal of the Hessian. The latter method is relatively cheap but not very accurate. I also present results for related inversion methods, which are variations of my proposed method with different levels of complexity and computational expense. All inversion methods presented in this thesis are based on forward modeling using the ray-Born approximation. Therefore, the second objective of my thesis is to validate the use of the ray-Born approximation. This is done by applying FWI using the L-BFGS algorithm and imaging to data generated by finite difference modeling as well as ray-Born modeling. I also compare the forward modeling results. I successfully apply multi-diagonal inversion, FWI using the L-BFGS algorithm, and imaging to a Sleipner- inspired model with CO2 layers to investigate the performance of these three methods for CO2 monitoring. A logical next step would be to apply the inversion methods to real data.