We evaluated the reproducibility of two elastic models, by comparing forward calculation results with the coseismic data observed during the 2011 Tohoku earthquake. The two models are the halfspace model which consists of an isotropic and homogeneous half-space medium, and the spherical model which considers radial heterogeneity, surface curvature, and self-gravitation. The coseismic fault slip distribution was first inverted from the near-field GNSS data using each elastic model, and then the coseismic displacement and gravity change were calculated using two fault slip results and two elastic models. The calculated displacement of the half-space model exceeded 150% of the observed one at > 500 km epicentral distance, since the half-space model oversimplifies the real Earth. The near-field gravity change calculated using the half-space model also exceeded 200% of the observed one, although the near-field displacement data were used for the fault slip inversion. In contrast, the calculated displacement of the spherical model agreed, within approximately 30% in residual, with the one observed at all ranges of the epicentral distance in Japan, even though the far-field GNSS data was not used for the inversion. This is mainly because the far-field crustal deformation can be accurately reproduced by the existence of radial heterogeneity in the spherical model. Furthermore, the spherical model reproduced the coseismic gravity change observed in Japan at the smallest root-mean-square (15.0 μGal), because the spherical model considers the depth-dependent structure of density and elastic parameters in the Earth. From these points of view, the spherical model should be chosen in calculating ground gravity change as well as far-field crustal deformation. In future studies, the effects of lateral heterogeneity and topography need to be considered additionally, to reduce the residual in displacement of approximately 30% which still remains in the calculation results using the spherical model.
抄録全体を表示