← Back to Sessions

Deep Learning Analysis of Transient Signals Preceding the 2023 Mw 7.8 Kahramanmaraş Earthquake in Türkiye

Description: 

Laboratory studies indicate that faults undergo damage and stress evolution before rupture, potentially leaving discernible signatures in seismic data. Before the 2023 M_W 7.8 Kahramanmaraş earthquake in Türkiye, an extended period of ~ 8 months displaying seismicity around the epicenter and the entire rupture suggested an extended preparation phase (Kwiatek et al., 2023; Picozzi & Iaccarino, 2023). This study analyzes low-frequency (< 5 Hz) recordings from regional broadband seismic stations near the future epicenter of this earthquake. Using a deep neural network, we extract key features from the continuous seismic waveforms and their spectral characteristics and employ an unsupervised clustering analysis to identify distinct classes in the continuous waveform recordings. Our analysis reveals two main changes in the spectral characteristics about 8 and 6 months before the mainshock, respectively, coinciding with the onset of elevated seismic rates and changes in b-values. Of particular interest is the emergence of numerous (N=3724) temporal episodes lasting about 12~30 minutes containing enhanced radiated energy in the frequency band 1-5 Hz, starting 6 months before the mainshock. These episodes are observed solely at five seismic stations within 46 km epicentral distance from the future mainshock epicenter, all located on the east side of the East Anatolian Fault Zone. The episodes are likely of tectonic origin, as their occurrence is not correlated with time of day, weather changes, construction processes, telemetry-related events, or instrument-related factors. Each episode comprises numerous small transient pulses resembling extremely small earthquakes. The pulses have a similar moveout to that from a local earthquake near the main rupture epicenter. We present the statistical analysis of observed episodes as well as their source properties and investigate their physical origin.

Session: Advances in Operational and Research Analysis of Earthquake Swarms -II

Type: Oral

Date: 5/3/2024

Presentation Time: 05:15 PM (local time)

Presenting Author: Marco

Student Presenter: No

Invited Presentation: 

Authors