一般国道33号柳谷地区斜面崩壊の予測  [in Japanese] Forecasting of Slope Failures at Yanadani Area in National Route 33, Shikoku  [in Japanese]

Two slope failures occurred recently in a rock mass along National Route 33 which links the cities of Matsuyama and Kocbi. The first failure occurred on July 20, l979, with a collapse of 3,000 cubic meters of ground, while the second occurred on August 27, with 1,000 cubic meters of ground collapsing. A series of site investigations, which included slope movement forecasting, were conducted from April 20, and were finally able to predict successfuly approximate time of occurrence of the failtues, with the result that it was possible to hold damage to a minimum. Figure 2 shows the remote control extensometer that was employed in locating the point of failure and in estimating displacement. The fomula announced by SAITO (1968) was used in the slope failure forecasting operation. Figures 8, 9 and 14 illustrate the semi-logarithmic method for forecasting time of slope failure. The slope failure that occurred on July 20 was preceded by a major change in displacement due to rainfall (Figure l2). Because of difficulties in identifying secondary creep, the successful forecasting of the time of slope failure was accomplished 2 or 3 days before the failure, when the tertiary creep failure stage was reached. Figure 8 shows how successive semi-logarithmic forecasts were corrected, so that time of failure was predicted 3 or 4 hours before the event, with an error of about 30 minutes. The August 27 failure, which involved the movement from halfway down the slope of massive chunks of mountainside, began very abruptly due to rainfall that had begun the previous day. As shown in Figure 9, forecasting conducted at 10:00 on the day of the failure gave time of failure to be 12:30, traffic was halted at l2:00 and the actual time of failure was 12:50. While the entire operation involved hair-breadth timing, the forecast was accurate to within 20 minutes. As shown in Figure 6 the slope consists of Carboniferous to Jurassic slate and sandstone. Displacement occurred along these joints. Even though the Matsuyama side of the slope (extensometer lccations S-1, S-2 and S-3) showed greater displacement and the same acceleration tendencies, this part of the slope did not fail. The reason for the difference in behavior between the Kochi and Matsuyama sides of the slope may be attributed to the differing types of landslide tendencies shown on the respective sides (Figure 6). This case shows that the method by Saito is effective means for forecasting the time of slope failure. However, it is carefully make use of this method then not cleared as basic causative factor behind the slope movement and failure.