Pn‐Wave Tomography Using Deep‐Learning‐Based Phase Picking Reveals Fine‐Scale Structure of the Uppermost Mantle Beneath the Southwestern United States

The southwestern United States, with its intense tectonic activity and dense seismic networks, serves as a natural laboratory for studying crust‐mantle material exchange and energy transfer. In recent years, the rapid growth of seismic observational data and the development of deep learning‐based automatic phase‐picking techniques have provided new opportunities for high‐resolution seismic imaging research. This study utilized the Seis‐PickNet phase‐picking model to expand the Pn‐wave travel‐time database for the southwestern United States. Based on the newly picked travel times, the study constructed a high‐resolution model of the uppermost mantle velocity structure and anisotropy structure in the region. The inversion results show that the spatial distribution of active tectonic units in the area corresponds well to significant low‐velocity anomalies and strong anisotropic features in the uppermost mantle. Furthermore, the study reveals independent thermal upwelling conduits in the uppermost mantle beneath volcanic fields along the margins of the Colorado Plateau, including the Black Rock Desert Volcanic Field, the Uinkaret Volcanic Field, and the San Francisco Volcanic Field. Additionally, the results support the existence of ancient Farallon slab subduction remnants beneath the central Great Basin. This research not only deepens the understanding of mantle dynamics and complex subsurface structures in the region but also highlights the critical role of deep learning‐driven phase‐picking technology in enhancing the quality of seismic tomography.