1995 年 104 巻 7 号 p. 1032-1046
We show a couple of methods of inversion of geodetic data to interpret crustal deformations from the viewpoint of kinematics. Geodetic data do not have sufficient resolving power because of their distribution and uncertainty. Therefore inversion of geodetic data is to find a way to overcome nonuniqueness. Currently the Monte Calro approach and inversion with a priori data are widely used among several ways. Especially introduction of a priori data solved this problem and stimulated development of methods to estimate fault slip or tectonic block motions and slip deficits on interface of plates.
We show two illustrative examples of estimate of fault parameters by different schemes; the Kurile earthquake of October 4, 1994 and the Kobe earthquake of January 17, 1995. For the Kurile earthquake we apply a Monte-Carlo approach because of insufficient infor mation on fault geometry. We generate thousands of sets of parameters for fault geometry randomly in their possible ranges and estimate slip for each fault model and residual. So far obtained results suggest that a high-angled fault trending along arc is more preferable than that perpendicular to the trench on the basis of the consistency with CMT solutions and distribution of residuals. Fault parameters of the Kobe earthquake can be estimated by linear least squares method with a priori information, because of sufficient geodetic data and th eir geographically favorable distribution. We propose a six segment fault model trending along the known fault system to explain the observed coseismic displacements of the Kobe e arthquake. The inversion infers a couple of large slip zones (>1m) beneath the Nojima fault in Awaji island, north of the Akashi strait and northeastern Kobe.
We also introduce models with block motion and slip deficit for the interseismic deformation of tectonic zones. In these models interseismic crustal defor mation can be represented as the sum of relative motion of blocks and displacement due to the slip deficit on the locked part of faults. The results for the Japanese islands indicate that significant strain partitioning is occurring there, which means relative plate motions ar e partly accommodated by intraplate faults. Furthermore the analysis of Japanese data infers significant aseismic slips along the Japan trench.