Microseismic Source Localization at the FORGE Geothermal Site using Fourier Neural Operator

The study addresses the significant threat to public safety posed by seismicity induced by human activities, emphasizing the critical need for precise and timely earthquake hypocenter localization. This research utilizes the Fourier Neural Operator (FNO) framework to localize microseismic events in the Enhanced Geothermal Systems (EGS) setting at the Utah FORGE site. By applying FNO to actual microseismic data from the EGS operational well, the study demonstrates the model’s capability to accurately predict hypocenter locations. First arrival traveltimes were used for the four selected microseismic events recorded during EGS activities at the Utah FORGE on April 24-28, 2019, including a perforation shot. The velocity model in this study was approximated using sonic log data from well 58-32. This work not only establishes the efficacy of FNO in source localization under realistic conditions, addressing challenges like partial data coverage, but also highlights its potential as a valuable tool for microseismic monitoring. The proposed method opens a pathway to real-time microseismic monitoring, as the trained FNO model can be evaluated instantly to determine the source location, facilitating real-time decision-making for the safe and effective development of subsurface operations.