Prediction of Catastrophic Large Landslide Sites and Estimation of Shallow Landslide Hazards Induced by Rainstorms — From the Experiences of Recent Landslide Hazards in Japan —

摘要

Prediction of potential landslide sites or estimation of landslide hazards induced by rainstorms has been one of the most important targets of researchers studying for the mitigation of landslide hazards for many years, but its methodology is not yet established. I propose to make different methodologies according to the scale of landslides; big ones should be specifically located and small ones are not located but their hazard potential are estimated area by area. Recent experiences of landslide hazards induced by rainstorms in Japan tell us that small, shallow landslides cannot be precisely located beforehand, but that large catastrophic landslides occurred on slopes with characteristic topographic landforms and geology, which could have been predicted. Most of small, shallow catastrophic landslides occurred on slopes with preferable geologic conditions made by weathering and related processes, so their hazard levels could be estimated from geology and previous history of landslides. The delineation of previous landslide distribution in a given area, which had been difficult to interpret from aerial photographs, became possible by airborne laser scanner, particularly in an area with coniferous trees or deciduous trees in moderate climate countries like Japan. The rainstorm hazards of 1972 Nishi-Mikawa, 1998 Fukushima, 1999 Hiroshima, 2004 Niigata, and 2004 Saijo indicate the consistency of hazard levels in given areas on the basis of airborne laser scanner. The logarithm of landslide density could be used as an index for Shallow Landslide Potential (SLP). Potential catastrophic large landslides, on the other hand, should be and could be located by closely examining aerial photographs. This is based on the careful examination of aerial photographs before and after landslide and geological investigation after landslide for the hazards of 2004 Miyagawa, 2004 Saijo, and 2005 Mimigawa; these were all induced by typhoons. Almost all of the catastrophic large landslides with volumes, say more than about 100,000 m3, were found to be preceded by slope deformation that could be topographically identified as scarplets. Some of the slope movements were due to high-angle tectonic shear surfaces at the upper part of slopes, some due to the nearly slope-parallel bedding planes or tectonic faults, and some due to impermeable shear zones. The relative heights of the scarplets are very small (10 m or less) in comparison with the size of landslides which exceeded a hundred meters in length, indicating that preceding deformation could have provided a condition just before the failure of whole slope, which would be followed by catastrophic stress drop.