17:15 〜 18:45
[SSS06-P05] Challenge for detection of earthquake precursor in Taiwan revealed by GNSS analysis
キーワード：GNSS, Earthquake Precursor, Baseline , Strain migration , Triggered Earthquake
Challenge for detection of earthquake precursor in Taiwan revealed by GNSS analysis
Central Weather Administration
Taiwan is located at the plate boundary of Philippine Sea Plate and Eurasian Plate with active tectonic, rapid crustal deformation and frequent earthquakes. There were many times disastrous earthquakes happened in Taiwan area. Therefore, related research on earthquake precursors is particularly important. The high precision Global Satellite Navigation System (GNSS) survey technique provides an efficient tool to study active tectonics and geodynamics even earthquake precursors. A dense continuous GNSS site array composed of more than 500 stations in Taiwan was established by the Central Weather Administration and Other Institutes. The data GNSS stations are processed with the GAMIT/GLOBK software. By removing the common errors, the precision of GPS data has been further improved to 2.3 mm, 1.7 mm, and 4.1 mm in the E, N, U components, respectively.
These high-precision GNSS data will be applied to observe earthquake precursors in this study. First, we use Routine-methods like the velocity filed variation analysis, strain field analysis, and strain variation field analysis. The basic concept is a regular precursor observation method with a stable time interval. We calculated the velocity field and strain field obtained from the GNSS data, and estimated the results for time span is one year for shifting month every time. The Huanlien-Shoufeng earthquake which is occurred at Apr. 18, 2021 is a doublet events, 2 earthquakes (ML 5.8 & 6.2; depth is 15 & 13.9 km) occurrence time interval is only 3 minutes. Form the Routine-method, we found the velocity field variation is represented clearly slow-down more than 2 months before the earthquake at near-filed epicenter area, it may suggest the strain is rapid accumulated to induce this earthquake. Another case is 2019 Xiulin-Shoufeng earthquake with the ML is 5.8 and the depth is 18.9 km. Before earthquake happened about 6 months-long, the strain variation is indicated the strain is slowly accumulated and also small earthquakes frequent occurred at the same time in the epicenter area, it can be regarded kind of precursor signal. Unfortunately, we can’t observed any precursor signal by Routine-method when time interval between major earthquakes is very short. For example, there are three disaster earthquake events with ML > 6 happened at Mar., Jun., and Sep., respectively. The short time interval between 3 events only represent the co- or post- seismic behavior but nor precursor signal.
Second, we use the Baseline – observation method to directly monitoring fault activity for probability of large earthquakes. We select GNSS sites with high precision and complete data, interconnect them to form a network with 80 baselines Taiwan area. From the case 2013 Rusui earthquake (ML=6.4, depth=15km) and 2016 Meinong earthquake (ML=6.4, depth=16.7km), we all observed the rate-slow-down anomalies which means the linear regression velocity for each baseline will slow-down before the earthquake happened. We have 9 and 26 baselines used in 2013 Rusui and 2016 Meinong area, among them, there are 3 and 8 rate-slow-down anomalies observed, respectively. We also estimate the the strain fields of these two earthquakes in different time periods. The strain migration phenomenon. The strain accumulated will slowly migrate to target area before the earthquake happened. We can take these two signals to be precursor signals, but they require a long observation time. Basically, these are phenomena that have only been observed over than data time span more than 4 years. Hence, the truth we have to face is a challenge when major earthquakes occurred very frequently, it is almost impossible to have inter-seismic period to observe the precursor signal. Especially there are more than 80% earthquake occurred in eastern Taiwan area. The effect of large earthquake post-seismic variation will add the difficulty for precursor monitoring. It makes all signals mixed together and accelerate the velocity field in who Taiwan area. Furthermore, some active faults will be triggered and form some crust deformation. It means there will be more difficult to predict where the next target area where larger earthquake happened is. In any case, this study comprehensively summarizes all precursor phenomena and challenges we have to face in order to achieve earthquake prediction in the future.
Central Weather Administration
Taiwan is located at the plate boundary of Philippine Sea Plate and Eurasian Plate with active tectonic, rapid crustal deformation and frequent earthquakes. There were many times disastrous earthquakes happened in Taiwan area. Therefore, related research on earthquake precursors is particularly important. The high precision Global Satellite Navigation System (GNSS) survey technique provides an efficient tool to study active tectonics and geodynamics even earthquake precursors. A dense continuous GNSS site array composed of more than 500 stations in Taiwan was established by the Central Weather Administration and Other Institutes. The data GNSS stations are processed with the GAMIT/GLOBK software. By removing the common errors, the precision of GPS data has been further improved to 2.3 mm, 1.7 mm, and 4.1 mm in the E, N, U components, respectively.
These high-precision GNSS data will be applied to observe earthquake precursors in this study. First, we use Routine-methods like the velocity filed variation analysis, strain field analysis, and strain variation field analysis. The basic concept is a regular precursor observation method with a stable time interval. We calculated the velocity field and strain field obtained from the GNSS data, and estimated the results for time span is one year for shifting month every time. The Huanlien-Shoufeng earthquake which is occurred at Apr. 18, 2021 is a doublet events, 2 earthquakes (ML 5.8 & 6.2; depth is 15 & 13.9 km) occurrence time interval is only 3 minutes. Form the Routine-method, we found the velocity field variation is represented clearly slow-down more than 2 months before the earthquake at near-filed epicenter area, it may suggest the strain is rapid accumulated to induce this earthquake. Another case is 2019 Xiulin-Shoufeng earthquake with the ML is 5.8 and the depth is 18.9 km. Before earthquake happened about 6 months-long, the strain variation is indicated the strain is slowly accumulated and also small earthquakes frequent occurred at the same time in the epicenter area, it can be regarded kind of precursor signal. Unfortunately, we can’t observed any precursor signal by Routine-method when time interval between major earthquakes is very short. For example, there are three disaster earthquake events with ML > 6 happened at Mar., Jun., and Sep., respectively. The short time interval between 3 events only represent the co- or post- seismic behavior but nor precursor signal.
Second, we use the Baseline – observation method to directly monitoring fault activity for probability of large earthquakes. We select GNSS sites with high precision and complete data, interconnect them to form a network with 80 baselines Taiwan area. From the case 2013 Rusui earthquake (ML=6.4, depth=15km) and 2016 Meinong earthquake (ML=6.4, depth=16.7km), we all observed the rate-slow-down anomalies which means the linear regression velocity for each baseline will slow-down before the earthquake happened. We have 9 and 26 baselines used in 2013 Rusui and 2016 Meinong area, among them, there are 3 and 8 rate-slow-down anomalies observed, respectively. We also estimate the the strain fields of these two earthquakes in different time periods. The strain migration phenomenon. The strain accumulated will slowly migrate to target area before the earthquake happened. We can take these two signals to be precursor signals, but they require a long observation time. Basically, these are phenomena that have only been observed over than data time span more than 4 years. Hence, the truth we have to face is a challenge when major earthquakes occurred very frequently, it is almost impossible to have inter-seismic period to observe the precursor signal. Especially there are more than 80% earthquake occurred in eastern Taiwan area. The effect of large earthquake post-seismic variation will add the difficulty for precursor monitoring. It makes all signals mixed together and accelerate the velocity field in who Taiwan area. Furthermore, some active faults will be triggered and form some crust deformation. It means there will be more difficult to predict where the next target area where larger earthquake happened is. In any case, this study comprehensively summarizes all precursor phenomena and challenges we have to face in order to achieve earthquake prediction in the future.