Crustal movement estimation based on an unchanged-feature points method using TerraSAR-X intensity images: application to the 2011 Tohoku earthquake

This paper presents a method to estimate significant crustal movement caused by the 2011 Tohoku earthquake in Japan. The method is based on the assumption that independent of the dramatic changes of the ground surface, still many unchanged areas, which we call unchanged-feature points, can be found. First, we use the Scale Invariant Feature Transform (SIFT) algorithm to detect the unchanged-feature point candidates in pre- and post-event TerraSAR-X images at similar locations. Then, a sub-pixel alignment procedure is used to obtain accurately matched point pairs. Finally, the unchanged-feature points are detected by eliminating unstable point pairs. The method was tested on TerraSAR-X images covering the Sendai area in Tohoku, Japan. Compared with the global position system (GPS) records and two previously published methods, our method is found to be an effective tool to detect crustal movement at a sub-pixel level.