High seismic potential areas along the collision front in southwestern Taiwan revealed from dense array analysis

Most of the ten destructive earthquakes that occurred in Taiwan in the 20th century were located in the deformation front area of the southwestern Taiwan. Therefore, it is necessary to conduct a detailed study on the seismic characteristics of this area. In order to effectively prevent earthquake-induced disaster risks in urban areas, this study aims to enhance high-resolution imaging of seismogenic structures and capture microseismic signals using a dense array at the front of the orogenic belt. Understanding the structure of the fault provides important data for modeling earthquake events and can help improve earthquake risk assessments in the region. Since deep learning neural network methods are widely used in earthquake-related research, phase picking is the most critical first step in seismic data processing. This study used a large number of microseismic events observed by a dense array deployed from 2020 to 2023 to explore possible potential seismic structures beneath the front edge of the foothill belt in order to understand the rupture mechanism and tectonic evolution process of the unknown seismic structure. This study used the initial earthquake catalog generated by AI automatic phase picking technology and used hypoDD to relocate microseismic events. By the way, the DBSCAN algorithm is used to find clusters where a large number of microearthquakes occur. The final relocation results showed that there was a west-dipping seismic belt and multiple east-dipping seismic belts at depths of 5 to 15 km, and an earthquake swarm was found at a depth of 15 km. According to the grouping algorithm and stress inversion results, the maximum stress axis in this area is mainly northwest-southeast, reflecting the direction of compressive stress since the orogenic process.