In this study, our aim is to use image recognition techniques to identify slopes at high-risk of slope failure. The feature of this study is that a three-dimensional terrain model is generated using the surrounding elevation values of each grid obtained from aerial laser surveying and the high-risk areas of landslide disasters are assessed by a Convolutional Neural Network (CNN) model trained on three-dimensional color images in which the three terrain factors are converted into color index. First, a standardized three-dimensional terrain model is constructed by extracting the elevation values of the surrounding grid and subtracting its own elevation values using the coordinate values of the grid data. We added per-grid slope and Laplacian terrain quantities to this three-dimensional terrain model. We have developed a method to quantitatively represent these three terrain quantities as color images by reflecting them in a 16-bit RGB color index. With the developed method, the three-dimensional terrain model becomes a single-color image. More than 100,000 images could easily be generated for each of the three watersheds included in the study area. A high-risk area was defined as the case where the center of the three-dimensional terrain model was included in the slope failure areas had occurred in the past. These images were used as explanatory variables to train the CNN. As a result, the occurrence and non-occurrence rates were balanced and could be evaluated at approximately 80%. Despite the high resolution of the evaluation unit being a 1 m grid, the proposed method resulted in high-risk grids forming clusters that matched the surrounding terrain, providing a terrain-consistent evaluation. Additionally, the results were consistent with the terrain characteristics of slope failure phenomena during heavy rainfall as previously indicated. This suggests that the proposed method can accurately capture the fine terrain irregularities compared to traditional methods.

View full abstract

This study is a statistical analysis of morphological characteristics. The characteristics are mainly derived by a standard GIS application on vector polygons representing the outlines of landslides (hereafter “landslide polygons”) delineated on a best available Digital Elevation Model (DEM) with 0.5 m resolution. The focus of the study is to gain new insights into the description of morphological features calculated by a simple GIS based on LiDAR data, which have improved dramatically in recent years. The study area is the area where more than 7,000 landslides were triggered by the 2018 Hokkaido Iburi Eastern Earthquake. The methods are as follows: (1) building materials in vector and raster format suitable for GIS calculation, (2) deriving statistics of each geomorphological feature, and (3) analysing statistical relationships between morphological features. As a result, some distinctive features were obtained. For example, (1) the average ratio of length to specific height was larger than that of previous landslides, and (2) the ordinal direction of the landslides was most frequent in the south-east direction, where crustal deformation was observed, and distributed in the east-west direction, where the amplitude of the velocity waveform of earthquake motion was larger. Other practical observations confirmed in this study are: (1) the applicability of valley openness estimated from contours for the evaluation of slope concavity instead of plan curvature calculated by GIS, and (2) the applicability of the ratio of soil surface thickness to the width of landslide blocks for preliminary risk assessment in the light of morphological characteristics. On the other hand, some important aspects were not clear enough to obtain conclusive results and remain for further study, such as (1) the subdivision of landslide polygons into collapse areas, run-out areas and accumulation areas, and (2) the study of hydrological processes in slopes and soil structures. The latter two have been removed from the current scope.

View full abstract

Since the introduction of the effective rainfall method by Suzuki and Kobashi (1981) in the field of sediment hydrology, there have been many studies on the occurrence of debris flow and slope failure using the effective rainfall method. These studies can be divided into two types: those related to the prediction of sediment-related disasters and analytical studies on the mechanism of sediment disasters. The advantage of using the effective rainfall method for analytical studies of slope failures is that the method is closely linked to practical tank modeling and is now an important model for groundwater analysis and for estimating water storage in hillslopes. However, the history, significance, and role of the effective rainfall method are still unclear. In this paper, the origin of effective rainfall, its relationship with the tank model, and the role of effective rainfall on slope failure are discussed. First, it is shown that the distinction between the effective rainfall and the antecedent-precipitation index (API) is based on whether or not the tank model is taken into account. Next, it is clarified that the effective rainfall can be derived by coupling the tank model to the unit hydrograph method. The tank model is also shown to be a model that represents the water storage at a representative point in a saturated subsurface flow in the soil layer of hillslopes. It is shown that the current effective rainfall corresponds to the water storage capacity by saturated subsurface flow. Finally, the significance of the effective rainfall and the one-stage tank model for slope failure is discussed by applying the model to sediment-related disaster cases. As an application, it is shown that slope failure can be classified into four types based on the occurrence of water storage and the drainage conditions of the soil-slide surface.

View full abstract

We calculated parameters for the landslide volume estimation equation and for the relationship between landslide area and landslide depth on slopes underlain by four types of geology where many landslides were occurred by heavy rains on August 2022 in Kaetsu Region, Niigata Prefecture. We discussed the appropriate regression method and the relationship between each parameter and geology. The results showed that a method that performs linearization transformation followed by regression has the potential to appropriately evaluate each parameter. The granite slopes in this disaster and the heavy rain disaster in Hiroshima on August 2014 had constant landslide depth regardless of the landslide area. The rhyolite slopes in Niigata and Hiroshima had different changes in the landslide depth relative to the landslide area, so it is necessary to investigate the lithology and weathering characteristics.

View full abstract