[9-5]

Real Data Assimilation for Optimal Estimation of Frictional Parameters and Prediction of Afterslip in the 2003 Tokachi-oki Earthquake by an Adjoint Method

M. Kano (University of Tokyo), S. Miyazaki (Kyoto University), Y. Ishikawa, Y. Hiyoshi (JAMSTEC), K. Ito (University of the Ryukyus), and K. Hirahara (Kyoto University)

Various type of fault slips, such as earthquakes, slow slip events, steady plate motions and afterslip, occur at plate subduction zones, depending on the frictional parameters on the plate interface. In this work, an adjoint data assimilation system is developed for estimating a spatially inhomogeneous frictional parameters in afterslip areas by using inverted slip velocity on the plate interface.

Before applying this method to the real data, a synthetic experiment is conducted to check the feasibility of estimating frictional parameters in the afterslip area. It is confirmed that the system is capable of estimating frictional parameters by adopting the physical constraint based on a numerical model if observations capture the acceleration and decaying phase at plate boundaries. On the other hand, it is unlikely to constrain frictional parameters in the region where little afterslip are estimated. Next, real data assimilation for the 2003 Tokachi-oki earthquake is conducted to incorporate slip velocity data inferred from time dependent inversion of GPS observation. Optimized frictional parameters yield the better fit to the observation and the better prediction skill of slip velocity afterwards. It will contribute to further understanding of frictional properties on plate boundaries and lead to the possibility for "terra forecasting".