5:15 PM - 7:15 PM
[SGD03-P15] Development of a comprehensive catalog of long-term slow slip events along Nankai Trough using GNSS data
Keywords:slow slip event, GNSS, GEONET, Nankai Trough
The spatiotemporal distribution of slow slip events (SSEs) is crucial to understand stress release process along the subducting plate boundaries. SSEs can be categorized into short-term SSEs and long-term SSEs based on their duration. In the Nankai Trough subduction zone, comprehensive catalogs of short-term SSEs have been created based on systematic analyses of GNSS and tilt meter data. In contrast, the catalogs of long-term SSEs rely on a compile of several research using different methodology. This study aims to systematically detect long-term SSEs along the entire Nankai Trough using GNSS data and develop a comprehensive catalog of long-term SSEs.
We use daily GNSS coordinates from GEONET (F5 solution by Geospatial Information Authority of Japan). As the surface displacements of long-term SSEs are on the order of a few millimeters, data cleaning process is necessary. The cleaning process consists of six steps: (1) removing offsets caused by antenna maintenance, (2) reducing common-mode error by taking relative displacements, (3) removing coseismic offsets, (4) removing postseismic deformation, volcanic deformation, and local noise due to the reference site by principal component analysis (PCA) (5) removing annual variations, and (6) removing outliers. The details are as follows: (1) The offsets due to antenna maintenance are removed using the offset data provided by the Geospatial Information Authority of Japan. (2) We compute relative displacements by subtracting the displacements at the Mitsushima station from daily coordinates at each station. (3) The coseismic offsets are removed if the theoretical displacement expected by the Global CMT solutions exceeds 0.5 mm and the difference 10- day medians before and after the earthquakes exceeds 4 times the standard deviation for the 10 days prior to the earthquakes. (4) We apply the PCA for the data within four sub regions Kyushu, Shikoku, Kii, and Tokai, and subtract the first principle component from the timeseries at each station. (5) Since the characteristics of seasonal variations change at the 2003 and 2012 maintenances, we divided the timeseries into three segments and removed annual and semiannual components by fitting trigonometric functions to three segments independently. (6) The outliers are detected if the deviation from the 90-day moving averages exceeds 0.1 m. The cleaning process reduced the root-mean-square error from 4.87 mm to 1.62 mm.
The stacked timeseries along 20 survey lines parallel to the relative plate motion show transient deformations consistent with the compiled catalog by Ozawa et al. (2024). Moreover, we detected transient signals which may not be listed in the catalog, which may be attributed to small long-term SSEs. We will apply a systematic detection method using the elastic Green’s function and develop a comprehensive catalog to understand detailed long-term SSE activity.
We use daily GNSS coordinates from GEONET (F5 solution by Geospatial Information Authority of Japan). As the surface displacements of long-term SSEs are on the order of a few millimeters, data cleaning process is necessary. The cleaning process consists of six steps: (1) removing offsets caused by antenna maintenance, (2) reducing common-mode error by taking relative displacements, (3) removing coseismic offsets, (4) removing postseismic deformation, volcanic deformation, and local noise due to the reference site by principal component analysis (PCA) (5) removing annual variations, and (6) removing outliers. The details are as follows: (1) The offsets due to antenna maintenance are removed using the offset data provided by the Geospatial Information Authority of Japan. (2) We compute relative displacements by subtracting the displacements at the Mitsushima station from daily coordinates at each station. (3) The coseismic offsets are removed if the theoretical displacement expected by the Global CMT solutions exceeds 0.5 mm and the difference 10- day medians before and after the earthquakes exceeds 4 times the standard deviation for the 10 days prior to the earthquakes. (4) We apply the PCA for the data within four sub regions Kyushu, Shikoku, Kii, and Tokai, and subtract the first principle component from the timeseries at each station. (5) Since the characteristics of seasonal variations change at the 2003 and 2012 maintenances, we divided the timeseries into three segments and removed annual and semiannual components by fitting trigonometric functions to three segments independently. (6) The outliers are detected if the deviation from the 90-day moving averages exceeds 0.1 m. The cleaning process reduced the root-mean-square error from 4.87 mm to 1.62 mm.
The stacked timeseries along 20 survey lines parallel to the relative plate motion show transient deformations consistent with the compiled catalog by Ozawa et al. (2024). Moreover, we detected transient signals which may not be listed in the catalog, which may be attributed to small long-term SSEs. We will apply a systematic detection method using the elastic Green’s function and develop a comprehensive catalog to understand detailed long-term SSE activity.