10:15 AM - 10:30 AM
[SSS04-06] Source parameter estimation of induced seismicity in the Weiyuan Shale Gas Field in China, and their implications on the hydraulic fracturing activities
Keywords:hydraulic fracturing, stress drop, corner frequency, seismic array, induced earthquakes
The hydraulic fracturing activities in the Weiyuan Shale Gas Field of Sichuan Province, southwestern China, have triggered an increasing amount of earthquake activities in the past 8 years compared to the silent historical record, which is accompanied by a deadly ML4.9 earthquake in 2019 and other ML5.0+ earthquakes afterward. How these earthquakes are related to hydraulic fracturing activities is still under debate. Earthquake source parameters such as stress drop have been found to vary spatially and temporally from anthropogenic activities, therefore in this study we aim at estimating the source parameters in the Weiyuan region to reveal the triggering mechanics of the large earthquakes and inspect the seismic hazard from shale gas exploration.
Our dataset covers the seismicity from March 2019 after the ML4.9 earthquake, to August 2020. Based on an enhanced local network, we detect nearly 28,000 earthquakes with magnitudes ranging from 0 to 5.2 (magnitude completeness: 1.5). These earthquakes are recorded on the local seismic stations, and a flexible dense array of short-period seismometers, with a sampling rate of 100Hz. They are first located based on Hypoinverse using a 1D velocity model, and then relocated with the double-difference method from P-wave arrival picks. It is found that small earthquakes concentrate at the depth of injection (around 3-4km deep) in northeastern Weiyuan; and after an M5.0 earthquake stroke at the depth of around 6km, most of the seismicity nearby are found closer to the M5.0 earthquake, potentially indicating fluid migration to deeper.
Next, we apply a spectral-decomposition-based approach to assess the stress drops of the earthquakes in this region. We first calculate the event spectra of the displacement seismographs in a one-second P-arrival window, and limit the SNR requirement to 3 for 0.5-30Hz information to minimize the influence of background noise on the stations; as a result, we obtain 41054 spectra recorded on 29 stations, among which 4033 earthquakes are selected that are recorded on at least 4 stations to ensure stability in the large-scale analysis. Then, the nearest 500 earthquakes of each ‘base’ earthquake are engaged in the spectral analysis to solve for their corner frequencies, and we combine the estimates from all ‘base’ earthquakes to obtain the final estimates for all the earthquakes. It is found that the stress drops range from 0.01MPa to 12.96Mpa with a median of 0.18Mpa, generally lower than the average stress drop level across China (0.1-10Mpa). To validate our stress drop estimates, as a comparison we implement a spectral-ratio method to evaluate the corner frequencies of M>3 earthquakes. We select EGFs within 1km near these earthquakes, the magnitudes are at least 1.5 lower than the main earthquake. Finally, 13 main earthquakes are selected and their corner frequencies are overall comparable to those from the spectral-decomposition method, indicating the stress drops obtained are reasonable. The stress drop level in northeastern Weiyuan is slightly higher than southwestern; it increases with depth above the injection zone, and nearly keeps constant at deeper. Besides, it visually scales with magnitude significantly, potentially indicating non-self-similarity of the induced seismicity. We will further analyze the spatiotemporal patterns of the stress drops related to the occurrence of major earthquakes, and will update our results during the meeting.
Our dataset covers the seismicity from March 2019 after the ML4.9 earthquake, to August 2020. Based on an enhanced local network, we detect nearly 28,000 earthquakes with magnitudes ranging from 0 to 5.2 (magnitude completeness: 1.5). These earthquakes are recorded on the local seismic stations, and a flexible dense array of short-period seismometers, with a sampling rate of 100Hz. They are first located based on Hypoinverse using a 1D velocity model, and then relocated with the double-difference method from P-wave arrival picks. It is found that small earthquakes concentrate at the depth of injection (around 3-4km deep) in northeastern Weiyuan; and after an M5.0 earthquake stroke at the depth of around 6km, most of the seismicity nearby are found closer to the M5.0 earthquake, potentially indicating fluid migration to deeper.
Next, we apply a spectral-decomposition-based approach to assess the stress drops of the earthquakes in this region. We first calculate the event spectra of the displacement seismographs in a one-second P-arrival window, and limit the SNR requirement to 3 for 0.5-30Hz information to minimize the influence of background noise on the stations; as a result, we obtain 41054 spectra recorded on 29 stations, among which 4033 earthquakes are selected that are recorded on at least 4 stations to ensure stability in the large-scale analysis. Then, the nearest 500 earthquakes of each ‘base’ earthquake are engaged in the spectral analysis to solve for their corner frequencies, and we combine the estimates from all ‘base’ earthquakes to obtain the final estimates for all the earthquakes. It is found that the stress drops range from 0.01MPa to 12.96Mpa with a median of 0.18Mpa, generally lower than the average stress drop level across China (0.1-10Mpa). To validate our stress drop estimates, as a comparison we implement a spectral-ratio method to evaluate the corner frequencies of M>3 earthquakes. We select EGFs within 1km near these earthquakes, the magnitudes are at least 1.5 lower than the main earthquake. Finally, 13 main earthquakes are selected and their corner frequencies are overall comparable to those from the spectral-decomposition method, indicating the stress drops obtained are reasonable. The stress drop level in northeastern Weiyuan is slightly higher than southwestern; it increases with depth above the injection zone, and nearly keeps constant at deeper. Besides, it visually scales with magnitude significantly, potentially indicating non-self-similarity of the induced seismicity. We will further analyze the spatiotemporal patterns of the stress drops related to the occurrence of major earthquakes, and will update our results during the meeting.