| dc.description.abstract | According to the slip models from the finite-fault inversion technique, we understand that spatial distribution of slips on faults for a large earthquake may be heterogeneous. It causes that the largest ground shaking (and damage) on the surface may not be related to the epicenter but to an area with the largest slip. This area also indicates information of the seismic source (e.g., which fault sliding). It is important to efficiently determine the area for further seismic hazard assessments. However, since determing a stable finite-fault model requires a long computing time (couple days), we cannot have details of the source characteristic immediately when an earthquake occurring. To fill this window gap, we use the source-scanning algorithm (SSA) technique and consider data from a Taiwan earthquake early warning system, called P-alert. In this study, we analyze 7 earthquakes with the finite-fault models, including the 2016 M6.6 Meinong Earthquake and the 2018 M6.3 Hualien Earthquake, and 4 events with no fault models in Taiwan since 2013. We use the SSA approach, whose concept is similar to beamforming, to search for the area where the maximum slip occurred. The Taiwan 3D velocity model is applied. We then compare the determined maximum slip from SSA with it from the finite-fault model for each event. Our results show that the resolution of the earthquakes with a magnitude greater than 5.5 is better due to their stronger S waves and more available seismic records. Furthermore, number of seismic stations and their epicentral distance may be important. After a test, we find that applying stations with an epicenter distance ranging from 50 to 70 km is the best choice. Considering the calculation efficiency, we choose 50 km in the present study. In addition, the coverage of the seismic stations also affects the SSA results. The events occurred offshore show worse results. In addition, we discover that the results have the good spatial resolution in horizontal location, but the slightly poor resolution in vertical location. To solve this problem, increasing data weightening of the stations near the epicenter seems to a potential method. The results of this research show that the earthquake early warning seismic network can be used to obtain the maximum ground motion within a few minutes after the earthquake. We plan to combine SSA into the P-alert earthquake early warning system to provide reference for academic and disaster prevention units. | en_US |