| Title | Delineating Faults Beneath Basalt at the Soda Lake Geothermal Field |
|---|---|
| Authors | Lianjie HUANG, Kai GAO, David LI, Yingcai ZHENG, and Trenton CLADOUHOS |
| Year | 2023 |
| Conference | Stanford Geothermal Workshop |
| Keywords | Denoise, fault, image noise, machine learning, migration image, migration artifacts, nested residual U-Net, Soda Lake geothermal field |
| Abstract | Extending geothermal wells beneath basalt at the Soda Lake geothermal field would increase geothermal production because of higher temperatures. Delineating faults beneath the basaltic unit is crucial for optimizing drilling into faults to maximize geothermal production. Because of the large velocity/impedance contrast between basalt and its surrounding formations and complex geologic structures, seismic signals reflected to the surface from geologic formations beneath basalt are very weak and the signal-to-noise ratios are extremely low, resulting in a poor and noisy seismic image. We apply a machine learning method based on nested residual U-Net to reverse-time migration images of a 3D surface seismic data acquired at the Soda Lake geothermal field to reduce image noises and migration artifacts and improve the image resolution, particularly beneath the basaltic unit. We then employ a nested-residual-U-Net fault-detection method to delineate faults on the enhanced migration images. Our procedure improves the reliability of fault detection on seismic migration images. The detected faults could provide valuable information for situating the best drilling locations beneath basalt at the Soda Lake geothermal field to increase geothermal energy production. |