Lawrence Berkeley National Laboratory

High-resolution imaging of fault zone structures is essential for understanding earthquake physics and fault mechanics. As a major left-lateral strike-slip fault in northeastern Tibet, fine structures of the damage zone in the creeping (Laohushan) section of the Haiyuan fault remain unclear. To resolve geometry and velocity reduction of the damage zone, we deployed a dense temporary network of 110 seismic stations around the creeping section of the Haiyuan fault. Travel time delays from teleseismic P arrivals suggest an approximately 1-km-wide low-velocity zone, likely illuminating a broader damage zone around the Haiyuan fault. A catalog is constructed for local earthquakes based on phase picks identified from a machine learning technique. The amplification of waveforms from these local events and waveform modeling of fault zone trapped waves indicate a narrower inner damage zone with depth-dependent width (ranging from 150 to 50 m) that extends to a depth of approximately 4 km. These values are generally consistent with those found on other noncreeping faults in California, suggesting that these damage zone properties are not affected by fault slip behaviors at shallow depth. In addition, a clear bi-material velocity contrast across the fault is revealed by the analysis of teleseismic P arrivals. Assuming the contrast extends to a depth of 15 km, we find that P-wave velocity is approximately 5% slower in the crustal block north of the fault. Our study shows that a temporary dense seismic network is effective in illuminating cross-fault velocity contrast and fault geometry.