09:00 〜 09:15
[MGI29-01] 西南日本のプレート境界すべり・ブロック運動の時空間発展の大規模同時推定に向けた未知パラメータ推定・分離精度評価
キーワード：プレート境界歪み収支、広域測地データの一括使用、状態空間モデル
Interseismic deformation in a plate subduction zone is attributed to not only slip and coupling on the single plate boundary fault but also strain partitioning on both overriding and subducting plates. The latter is often modeled by the rigid motion of small tectonic blocks and creep and coupling on their boundary faults. Although many deformation analyses for southwest Japan have revealed the spatial and temporal distribution of these phenomena, the interaction of these phenomena and their effect on the total slip budget on the plate boundary fault is a remaining important topic. This is mainly because previous studies individually analyzed different areas and periods with different estimation methods, arising individually different uncertainties of the results. It makes us difficult to combine these previous estimations.
Therefore, we attempt a comprehensive estimation of the spatio-temporal evolution of interplate slip and block motion around southwest Japan based on state space modeling, using GNSS time series for about 25 years since the mid 1990s. Toward this final goal, we first evaluated separation precision between fault slip and block motion by numerical experiment. We generated synthetic cumulative displacement time series for 770 GNSS sites around southwest Japan, assuming various contribution ratios for interplate back slip and block motion of the forearc region. We just assumed monotonically increase of back slip without any transient slip events. For block rigid motion, we assumed constant values of rotation vector through the analysis period. Then, we applied the formulation of block modeling suggested by Meade and Loveless (2009) to Kalman filtering. As a result, cumulative back slip on the plate interface always showed underestimation of about 10 to 40 % for known strongly coupled regions beneath the Tonankai and Shikoku regions. Correspondingly, amount of block rigid motion showed slight overestimation, although its direction was well recovered. Ratios between the estimated and true slip amounts are nearly constant except for the first few years in which signal to noise ratio is relatively low. It suggests that the occurrence of trade-off is mostly controlled by the setting of problem such as configuration and number of subfaults and blocks. In the presentation, we show results in case of including transient slow slip in time series. Then, we discuss the detailed structure and mechanism of the trade-off between fault slip and block motion through covariance matrix estimated by Kalman filtering. Additionally, we examine the impact of the joint use of seafloor geodetic data and/or GNSS sites operated by private company on the resolution between fault slip and block motion.
Therefore, we attempt a comprehensive estimation of the spatio-temporal evolution of interplate slip and block motion around southwest Japan based on state space modeling, using GNSS time series for about 25 years since the mid 1990s. Toward this final goal, we first evaluated separation precision between fault slip and block motion by numerical experiment. We generated synthetic cumulative displacement time series for 770 GNSS sites around southwest Japan, assuming various contribution ratios for interplate back slip and block motion of the forearc region. We just assumed monotonically increase of back slip without any transient slip events. For block rigid motion, we assumed constant values of rotation vector through the analysis period. Then, we applied the formulation of block modeling suggested by Meade and Loveless (2009) to Kalman filtering. As a result, cumulative back slip on the plate interface always showed underestimation of about 10 to 40 % for known strongly coupled regions beneath the Tonankai and Shikoku regions. Correspondingly, amount of block rigid motion showed slight overestimation, although its direction was well recovered. Ratios between the estimated and true slip amounts are nearly constant except for the first few years in which signal to noise ratio is relatively low. It suggests that the occurrence of trade-off is mostly controlled by the setting of problem such as configuration and number of subfaults and blocks. In the presentation, we show results in case of including transient slow slip in time series. Then, we discuss the detailed structure and mechanism of the trade-off between fault slip and block motion through covariance matrix estimated by Kalman filtering. Additionally, we examine the impact of the joint use of seafloor geodetic data and/or GNSS sites operated by private company on the resolution between fault slip and block motion.