Automated processing of data recorded by mobile stations deployed after a Tonkhil's strong earthquake

Abstract

On March 20, 2020, at 11:03:12 a.m. local time (UTC+8), 2020, an earthquake with a magnitude of M_L6.0 occurred in Tonkhil soum, Govi-Altai province, located in the western region of Mongolia. The tremor was strongly felt across the western provinces. This earthquake was the strongest event in the region since the Takhiin Shar earthquake (July 4, 1974, M=7.0). According to the National Data Center, approximately 1,000 aftershocks had been recorded in the epicentral area by 2020. Following the mainshock, seven temporary seismic stations were deployed for three months to monitor and record the increased seismic activity. Data collected from these mobile stations were automatically processed to identify aftershocks. Phase picking was conducted using the PhaseNet software based on a convolutional neural network deep learning approach. Phase association and event determination were performed using the GaMMA software, which is based on a Gaussian mixture model. As a result of the automatic data processing, a total of 4,550 earthquake hypocenters were identified. Following the fault geometry, seven cross-sections were constructed, and the depth distribution of seismicity was analyzed for each section. Based on this analysis, interpretations were focused on two cross-sections where significant seismic activity was observed. The first active zone, Section E–E′, is characterized by a large number of low-magnitude repetitive earthquakes, accounting for approximately 60% of the total recorded events. In terms of depth distribution, 98% of the earthquakes identified in this section occurred at depths shallower than 18 km. The second notable active zone, Section C–C′, exhibits seismic activity aligned along the fault plane. With respect to depth distribution, 95% of the earthquakes identified in this section occurred at depths ranging from 10 to 18 km.