Shallow landslides repeatedly occurred on the slopes of Aso Volcano, Central Kyushu, southwest Japan, during heavy rainfall in 1990, 2001, and 2012. This study aimed to identify the geomorphological causes of these intensive landslides covered with fall-out tephra layers via geographic information system analysis and geomorphological classification of their distribution in the Takadake and Saishigahana areas. Source areas of the abovementioned landslides were mapped on Red Relief Image Maps processed from light detection and ranging digital elevation model data (1m resolution). The fact that they were mostly distributed on slopes with gradients of more than25° and relative heights of more than 20 m in an area of a 50×50 m window indicates the minimum gradient of a rainfall induced landslide occurrence and the fall-down possibility of unstable material, respectively. More than half of the landslide locations in 2012 were different from those in 1990 and 2001 ; this was also found for landslides in 2001 and 1990 in both study areas. The slopes in contact with the rear and sides of the1990 and 2001 landslides collapsed in 2012 at rates of 32.7％ and 21.2％ in Takadake and Saishigahana, respectively, and those of the 1990 landslides collapsed in 2001 at rates of 19.1％ and 28.4％ in the respective study areas. In contrast, the occurrence in 2001 and 2012 at the inside of the former landslide slope was 3―4％ in Takadake and 1―7％ in Saishigahana. Approximately 50％ aerial occupancy of remnants of past landslides (RPL) revealed that shallow landslides are geomorphologically common phenomena on the slopes of Aso Volcano. The spatial relation of landslide occurrence area in 2012 to RPL was classified into seven types based on geomorphological mapping, namely 1) in contact with rear of RPL, 2) in contact with side of RPL, 3) inside of RPL without deposition of collapsed material, 4) below RPL, 5) in and around deposition of collapsed material in RPL, 6) on valley side slope below slope break, and 7) independent. Type 1 was dominant, and types 2 and 3 were subsequent. Deposition of tephra or transported materials from the head scrap on the collapsed material of a past landslide which increases the soil depth above the rupture surface might be the cause of the difference between fewer repeated landslide occurrences on former slopes of the 1990, 2001, 2012 landslides and type 3. Types 1 and 2 are concordant with frequent occurrences of landslides behind and at the side of former landslides that are caused by the instability of the slope toe or side where its counterbalance has decreased. Based on these findings, viewpoints for the detection of landslide-susceptible slopes caused by rainfall is also proposed.

  Nowadays, it has become easy for us to use the spatially distributed rainfall data observed by weather radars. Such high-resolution data allows us to predict groundwater rising in each individual slope due to torrential rain in a short period, using reliable numerical techniques such as finite element analysis of saturated-unsaturated seepage. However, usually a high-spec calculator and a long period of calculation time are required to carry out such rigorous simulations, and it means that such a method cannot satisfy the engineers who are concerned with real-time simulation used for early warning processes. In this study, a simple method for prediction of shallow groundwater rising in natural slopes in time history will be developed based on the parametric studies with the finite element analysis under the assumption of semi-infinite homogeneous slope. By referring the hydraulic head predicted by the proposed method and the observed shear strength of soil material in the slope, it becomes possible to facilitate a real-time evaluation for the degradation of the total factor of safety for slope failure from moment to moment.

  Shallow landslides densely occurred on tephra-covered slopes in the northeastern part of Aso caldera during the July 2012 northern Kyushu heavy rainfall. We estimated a thickness distribution of tephra deposits above the potential rupture surface of shallow landslide using spline interpolations of an isopach map, which was developed based on previous studies on the eruptive history of Aso Volcano, and compared it with the actual thickness measured at 11 sites. The results showed that ratios of the measured thickness to the estimated one (named residual ratio, Rr) ranged from 0.12 to 0.31, and had decreasing trends with increasing slope angle and curvature. Infinite slope stability analysis based on these characteristics captured the locations of approx. 84％ of the landslides induced by the July 2012 rainfall event (total 309 head scarps) at its maximum, demonstrating the accuracy of Factor of safety (Fs) calculated by this analytical method. On the other hand, the total area of landslide-prone sites, where the Fs was estimated to be lower than10.0, accounted for approx. 67％ of the analyzed area, which was higher than the area ratio of landslides in a single rainfall event (approx. 4-15％).

  Large-scale sediment disasters, induced by heavy rains in 1990, 2001, and 2012, occurred in the Aso caldera in Kumamoto prefecture, Japan. These natural disasters were caused by shallow landslides of unconsolidated slope deposits. Geomorphological and geological surveys of the Mt. Takadake slope in the Aso caldera showed an outcrop that exhibited characteristics of the initial stage of a shallow landslide (Sato et al. 2017). These deposits consist of tephra and “Kuroboku” (humus andosols) layers, deformed by a flow-type gravitational deformation. Sato et al. (2017) indicated that the deformed slope deposits exhibited significant risk of future shallow landslides. Therefore, it is necessary to investigate the ground strength of the deformed layers and evaluate the risk of future landslides. In this study, the ground strength of soil layers is measured using a vane shear cone test and a number of soil tests. The results of these tests reveal that the strength of the deformed layers is lower than that of the un-deformed layers.