Assessment of Seismic Hazard Potential for a Geothermal Field: A Case Study in West Texas

Abstract The El Paso (TX)–Ciudad Juárez (MX) metropolitan area, located within the tectonically active Rio Grande Rift, has historically recorded shaking from nearby high-magnitude earthquakes (e.g., the 1887 Mw 7.6 in Sonora, MX, and the 1931 Mw 5.8 in Valentine, Texas). Application of the machine-learning (ML)-based EarthQuake Compact Convolutional Transformer (EQCCT) algorithm to seismic data from 2008 to 2011 resulted in the detection of 645 seismic events in the area, lowering the magnitude detection threshold compared to public catalogs (e.g., U.S. Geological Survey ComCat with 35 events in the same period). Manual review and relocation using NonLinLoc and hierarchical clustering with GrowClust3D revealed seven seismic clusters: four clusters align with mapped Quaternary faults, whereas three clusters correspond to previously unrecognized seismogenic structures. These results demonstrate that advanced ML techniques can enhance earthquake detection and refine the understanding of regional seismicity. With new geothermal projects on the horizon in West Texas, the enhanced seismic catalog provides a more robust basis for assessing seismic hazard potential, which is critical for guiding safe geothermal development in the region.