[S-SS16] Crustal Deformation

[SSS16-09] GNSS time series analysis of postseismic deformation induced by 2016 Kumamoto earthquake (M7.3)

*Satoshi Fuwa¹, Mako Ohzono¹ (1.Hokkaido University)

Keywords:2016 Kumamoto earthquake, postseismic deformation, time series analysis

In order to investigate the afterslip and the viscoelastic relaxation mechanisms induced by the 2016 Kumamoto earthquake (M7.3), which occurred on 16 April 2016 in Kyushu Island, Japan, we conducted times series analysis of GNSS data. Although there are several models for this postseismic deformation, they treat early postseismic period with only velocity or displacement data. Several time series analysis for other large earthquakes is also proposed, but there has been little consideration about whether it reflects their mechanisms. We use 48 sites around the focal area of daily coordinates (F3 solutions) provided by the Geospatial Information of Japan (GSI) for two years from the earthquake. Because most of the afterslip is thought to be finished during the first 0.25 years in postseismic period, we fitted logarithmic decay function assuming the afterslip to 90-days postseismic time series data using weighted nonlinear least squares method and estimated their time constants. Then, we re-fitted both of logarithmic decay function and exponential decay function, which assuming viscoelastic relaxation, together to whole two years postseismic time series fixing the time constant of logarithmic decay function. As a result of this time series analysis, each afterslip and viscoelastic time evolution are estimated at each GNSS site. Using the cumulative deformation field of logarithmic decay function, afterslip distribution is estimated. The afterslip is located around the coseismic fault, and major slip area is assumed complementary to the coseismic slip area. Cumulative viscoelastic deformation signal is also used for rheological parameter estimation. We set stratified rheological structure, which composed of upper elastic layer and lower Maxwell viscoelastic layer, and searched appropriate thickness of elastic layer and viscosity of the viscoelastic layer. As a result, we found the elastic thickness with 20 km and viscosity of viscoelastic layer with 1×10¹⁹ Pa s. These estimated values roughly correspond to the results of previous studies. There are several remaining issues, such as the dependency of fitting period of the logarithmic time decay function, and application for the complicated fault geometry and rheological structure for modeling. However, this time series analysis and estimation of each mechanism using those results are effective to the interpretation for the postseismic deformation after this earthquake.