Deep Learning Enhanced Joint Inversion for High-Resolution CO2 Plume Monitoring

Carbon dioxide (CO2) generated by fossil fuel combustion can have serious consequences for humans and the environment. Carbon capture and storage (CCS) is a key technological approach to slow down the CO2 accumulation in the atmosphere byseparating CO2 from industrial plant effluents and injecting it into an underground geological formation to permanently store the CO2. In order to monitor the potential leakage from the CO2 sequestration sites, real-time and long-term monitoring of geophysical and geochemical information of CO2 reservoirs are critical. Currently, the quality of deep subsurface monitoring system is low and the uncertainty is high. Existing data processing methods are incapable of providing high-resolution subsurface images and meeting both the long-term and real-time monitoring requirements to prevent potential CO2 leaks. In order to overcome this problem, we propose to develop an innovative multi-physics joint inversion algorithm by leveraging deep learning technology for identifying and monitoring CO2 plume in multi-resolution. This algorithm combines the measured data from EM, seismic and cross-well energized casing. It applies deep learning, geophysics, and mathematical principles into the joint inversion approach to achieve the high- resolution quantitative results for CO2 volume evaluation and plume monitoring. The goal of Phase I is to demonstrate the feasibility of our approach to build a reliable deep subsurface monitoring system for both long-term and real- time usages. In Phase I, Cyentech Consulting, in collaboration with the University of Houston, will 1) Design and verify the feasibility of using a deep learning framework to enhance the subsurface imaging of CO2 plume distribution by combining inversion results obtained via EM and seismic methods; 2) Conduct feasibility studies on cross-well energized casing measurements for CO2 monitoring. The success of this project will pave the way to produce a powerful, robust, and cost-effective CO2 sequestration monitoring system for deep subsurface sensing. This system can be applied to detect and predict the CO2 leakage for both long-term and real-time monitoring. In addition, the proposed technology and product can find many commercial applications in other areas such as oil and gas exploration, enhanced geothermal systems, and subsurface waste disposal.