**3. Tectonic Conditions before and after the Tohoku Earthquake**

On 15 September 1998, an earthquake of magnitude (M) 5.0 occurred at a 12.4 km depth, on the deepest part of the Nagamachi-Rifu fault [14]. The mechanism of this earthquake and its aftershocks was reverse faulting, and the P-axes of the events described a line oriented east–west to northwest–southeast; the maximum principal stress direction in the deep part of the active fault was northwest–southeast.

On the basis of GPS data from an observation point about 6 km southeast of the dam, the Geospatial Information Authority of Japan (GSI) reported crustal movement of 2.55 m eastward and 16 cm ground subsidence during the Tohoku earthquake (e.g., [15]). The GSI also measured eastward crustal movements of 1.85 m in Tendo city, 30 km to the west, and 3.98 m in Higashi-Matsushima city, 70 km to the east (GSI, http://www.gsi.go.jp/index.html; Figure 1a). From these data plus the assumption of an elastic rock mass in the area, the strain change resulting from the Tohoku earthquake was estimated to be 2.13 × <sup>10</sup>−<sup>5</sup> in the study area. Eastward crustal movement around Sendai continued after the earthquake, reaching about 1 m during the following year (GSI; http://www.gsi.go.jp/cais/chikakuhendo40012.html).

It appears that the direction of maximum principal stress in the study area was roughly east–west both before and after the Tohoku earthquake [7]. However, at the Kamaishi mine, ≈170 km northeast of Sendai, horizontal displacement of 3.3 m was observed during the Tohoku earthquake [16]. In situ measurements of tectonic stress at the mine documented a large change in the stress field in the shallow crust in the two years after the Tohoku earthquake, with the principal stress decreasing in the east–west direction and increasing in the north–south direction [16].

The evidence suggests that although the Tohoku earthquake caused large changes of the stress field in some parts of the shallow crust, changes in the orientation of the stress field at seismogenic depths were small in the study area, which implies that the stress field in the study area still favors reverse faulting.
