Spatial-Temporal Evolution Pattern of Earthquakes Illuminated by Machine Learning Method and Waveform Correlation in Weiyuan Shale Gas Field, Sichuan China
Session: Mechanisms of Induced Seismicity: Pressure Diffusion, Elastic Stressing and Aseismic Slip [Poster]
Type: Poster
Date: 4/20/2021
Presentation Time: 04:15 PM Pacific
Description:
The rapid development of hydraulic fracturing activities in the Sichuan Basin, China, has immediately followed by a surge of seismicity since 2015. In 2019, 3 magnitude 4+ earthquakes and 2 magnitude 5+ earthquakes struck the Weiyuan Shale Gas Field, causing casualties and damages. Albeit multiple research groups have conducted extensive studies on these major earthquakes, the centroid depth and inducing mechanisms are still in controversies. In contrast to most of the previous analysis based on a diffusive earthquake catalog constructed from phase arrivals on the local network, we aim to construct a high-resolution catalog using machine learning detector and waveform cross correlation, so as to derive spatial-temporal evolution of seismicity. In our study, we use the PhaseNet, a U-shaped neural network to automatically pick P and S arrivals from the 9 permanent stations across the Weiyuan area. The P and S arrival onset selection by the machine learning method has shown systematic improvements when compared to the catalog phase picks. The arrivals are associated with a grid search method and the events are located by VELEST. Then they are relocated by the double-difference algorithm with the arrival cross-correlation. The newly constructed catalogs detected one-third more earthquakes than the network catalog, despite using one-fourth of the available data from August 2018 to March 2019. The distribution of the earthquakes has delineated unmapped subsurface fault geometry possibly led to the 2019 Mw 5.0 earthquake in September, by far the largest earthquake in the region. The seismicity patterns display high spatial-temporal correlation of multiple earthquake clusters with the hydraulic fracturing wells, confirming a causal link between the emerged seismicity and the hydraulic fracturing in the region.
Presenting Author: Wing Ching Jeremy Wong
Student Presenter: Yes
Authors
Wing Ching Jeremy Wong Presenting Author Corresponding Author jwjeremy@link.cuhk.edu.hk The Chinese University of Hong Kong |
Jinping Zi zijinping@link.cuhk.edu.hk The Chinese University of Hong Kong |
Pengcheng Zhou pengcheng.zhou@link.cuhk.edu.hk The Chinese University of Hong Kong |
Hongfeng Yang hyang@cuhk.edu.hk The Chinese University of Hong Kong |
Jinrong Su sujr0816@163.com Sichuan Earthquake Agency |
|
|
|
|
Spatial-Temporal Evolution Pattern of Earthquakes Illuminated by Machine Learning Method and Waveform Correlation in Weiyuan Shale Gas Field, Sichuan China
Category
Mechanisms of Induced Seismicity: Pressure Diffusion, Elastic Stressing and Aseismic Slip