fig03: Verification of the process for obtaining Δzk,triangulations.

Mentions: Though the DEMs before and after the earthquake were carefully prepared, comparing DEMs from the two essentially different methods can cause some inevitable bias to be included in the result. Therefore for the target area discussed hereafter, the obtained elevation changes were compared with those at points of triangulations in order to verify the result (Fig. 3). There were total six triangulation points whose movements were not affected by landslides in the target zone. In this figure, open squares show Eulerian displacements directly obtained from DEMs, while open circles show the vertical components of Lagrangian displacements estimated through the process mentioned above. Between vertical components of soil displacement Δzk,triangulation at the points of triangulation and Lagrangian expression of Δzk,DEMs (open circles) obtained from the two DEMs, a linear regression analysis was carried to best-fit b+Δzk,triangulation to Δzk,DEMs, yielding[4]For the range of Δzk,triangulation from 0.3 to 1.0 m, this −0.033 m bias is considered to be small enough to validate the process for converting Eulerian displacements Δzk,DEMs to those in Lagrangian expression. Thus the present method is considered to allow for describing the entire view of tectonic displacements. However, discussing displacements for particular locations will require careful attentions given the standard deviation of 0.34 m that may be a reflection of some local conditions.