11:00 AM - 1:00 PM
[SCG52-P02] Estimation of Time-dependent Stress Fields from CMT Data Inversion Using Gaussian Processes
We present a new method of the CMT data inversion (Terakawa and Matsu’ura, 2008) to produce time-dependent stress field models. Gaussian processes are applied to resolve a computational difficulty of the basis function expansion method in analyzing high-dimensional data. For this purpose, the relationship between the covariance functions of stress and centroid moment tensor is analytically derived from an observation equation. This formulation enables us to perform a time-dependent modeling of stress fields.
This method was applied to CMT solutions in and around Japan after the 2011 Tohoku earthquake. The estimated stress field exhibits a small-scale (~20 km) heterogeneity, and the spatial pattern is consistent with previous studies that analyzed CMT solutions before the 2011 event except for its focal region (Terakawa and Matsu’ura, 2010). The estimated time scale of stress variation is longer than the data period (8.8 yr), which suggests a long-term stability of stress state in time. Nevertheless, significant temporal variations are observed around the margin of the focal region of the 2011 event, the sense of which is opposite in the inner and outer portions.
This method was applied to CMT solutions in and around Japan after the 2011 Tohoku earthquake. The estimated stress field exhibits a small-scale (~20 km) heterogeneity, and the spatial pattern is consistent with previous studies that analyzed CMT solutions before the 2011 event except for its focal region (Terakawa and Matsu’ura, 2010). The estimated time scale of stress variation is longer than the data period (8.8 yr), which suggests a long-term stability of stress state in time. Nevertheless, significant temporal variations are observed around the margin of the focal region of the 2011 event, the sense of which is opposite in the inner and outer portions.