SEISMIC RESPONSES AND IMAGING FOR SUBSURFACE FLUID-INDUCED PROPERTIES

Abstract

Due to approaching climate change, natural gas and CO2 have become the target fluid contents for seismic exploration. Besides conventional seismic outcomes, the intrinsic attenuation is strongly beneficial to augment the seismic interpretation for gas/CO2 saturation. Current high-resolution Q model building (e.g., Q-FWI) or direct gas/CO2 illustration require tremendous computational and human resources. In this thesis, I propose new methods and applications for gas/CO2 illustration with higher efficiency, resolution and robustness than state-of-the-art technologies. For simulation, I propose new natural-attenuation absorbing boundary condition (naABC) to improve the efficiency of 3D viscoacoustic wave equation modeling; for processing, I develop a Q-interface imaging method for high-resolution reconstruction of low-Q; for interpretation, I train and apply a neural network (NN) to achieve an end-to-end mapping from 4D seismic images to 3D CO2 distribution probability varying along time. Synthetic examples and field data applications demonstrate the effectiveness and efficiency of our proposed methods.