Item

Abstract

The crustal stress accumulated due to inelastic deformation at depth in the interseismic time period is released as future inland earthquakes. Therefore, evaluating the inelastic deformation in the local crust is important. However, there is no standard method to separate it from the local deformation field so far. It is necessary to develop a new method to separate inelastic deformation from geodetically observed signals such as GNSS. We try to separate inelastic deformation in Central Japan, where major active faults are distributed and many hazardous inland earthquakes occurred. We select several GNSS sites to form an optimum closed boundary and describe the outer contributions as boundary conditions. By assuming perfectly elastic medium and plane stress conditions inside the boundary, the interior velocities caused by outer contributors can be predicted with a 2-dimensional interpolation approach using Green’s functions of an elastic body under plane stress conditions (Sandwell and Wessel, 2016). The residuals between predictions and observations indicate deformations induced by inelastic deformation sources inside.However, a single result is biased by crustal heterogeneity and inappropriate boundary selections. To cancel these biases, we generate a regular grid with 2 km intervals in Central Japan and batch-predict velocities inside each grid-centered boundary. This way, GNSS sites can be either boundary or interior sites. Averaged residuals for all sites are used to cancel the errors. The averaged velocities pattern shows similar surface movements as fault motions in central Japan; SE-NW compression around the northern Itoigawa-Shizuoka tectonic line; left-lateral strike-slip motion around the central ISTL.