[S-SS11] Statistical seismology and underlying physical processes

[SSS11-05] Seasonal modulation in crustal seismicity driven by snow load

*Taku Ueda¹, Aitaro Kato¹, Christopher W Johnson², Toshiko Terakawa³ (1.Earthquake Research Institute, the University of Tokyo, 2.Los Alamos National Laboratory, 3.Earthquake and Volcano Research Center, Graduate School of Environmental Studies, Nagoya University)

It has been reported that seismicity rate at crustal depths correlates with phenomena that cause temporal changes of stress and fault strength at the surface and shallow subsurface, such as hydrological load, heavy precipitation infiltration, atmospheric pressure, and surface temperature (e.g., Heki, 2003; Amos et al., 2014). For example, in California, the seismicity rate varies seasonally in response to stress changes due to annual variations in hydrological loading (e.g., Amos et al., 2014; Johnson et al., 2017). Johnson et al. (2017) calculated the stress change which hydrological loading imposes on the fault plane based on GPS vertical displacement. They showed that seismicity in California is more likely to occur when shear stress increases. The seasonal variations in seismicity in Japan have also been pointed out, and its relationship to snow loading and precipitation has been discussed as the plausible cause (e.g., Heki, 2003; Ueda and Kato, 2019). However, few studies have directly compared the seasonal variations in seismicity with the stress changes brought by surface loading.
In this study, we estimated the spatial distribution of the surface load from the vertical displacement data in the inland Tohoku region and compared and examined the stress change in the subsurface caused by the surface load with the timing of earthquake occurrence. We used the final solution of the GEONET daily coordinates known as the F5 solutions during the 2004-2010 periods. First, we removed common-mode errors following a procedure based on the spatiotemporal filtering by principal component analysis (Dong et al. 2006) using the vertical displacements at the stations distributed in southern part of Japan. We separated common-mode removed vertical displacements into trend components, transient changes such as earthquakes and postseismic deformation, seasonal variation components, and residuals using the method of Bedford and Bevis (2018). We inverted the monthly spatial distribution of surface load from the extracted seasonal component of vertical displacements using the method of Johnson et al. (2017). The results show that the load distribution increases in winter and decreases in spring in the area along the Sea of Japan, matching snow depth change observed at AMeDAS sites.
We calculated the change of the Coulomb stress and stressing rate generated by the surface load onto the nodal planes where the shear stress is maximum in a tectonic stress field of the crustal earthquake occurrence area (Terakawa and Matsu'ura, 2008; 2010). We evaluated the percent excess seismicity that occur during a range of stressing intervals (e.g., Cochran et al. 2004; Johnson et al., 2017). The number of earthquakes is weighted by the probability that each earthquake occurred as a background event, estimated by the HIST-ETAS model (Ogata, 2004). As a result, we found the positive correlation between percent excess seismicity and both Coulomb stressing rate/stress, suggesting that background seismicity in inland Tohoku region is modestly modulated by snow load.

Acknowledgments: The GNSS vertical displacement data used was the F5 solution of GEONET from the Geospatial Information Authority of Japan. GRATSID (Bedford and Bevis, 2018) was used to extract the seasonal variation component of vertical displacement. The JMA catalog was used; the code of Ogata et al. (2021) was used to estimate the parameters of the HIST-ETAS model.