In steep mountainous regions, deep catastrophic landslides that involve weathered bedrock as well as soils can cause serious damage. However, there is currently no widely used method for estimating spatial patterns of susceptibility to deep catastrophic landslides. We propose a new method to estimate landslide susceptibilities for many small catchments (~1 km2) over relatively large areas (hundreds of square kilometers). Our method identifies catchments prone to deep catastrophic landslides according to three criteria: (1) catchments with ancient deep catastrophic landslide scars, (2) catchments with faults and landforms caused by long-lasting mass movements, and (3) catchments with many steep slopes that have large upslope contributing areas. We demonstrated the applicability of this method using data from Mount Wanitsuka, Miyazaki Prefecture, Japan, where deep catastrophic landslides occurred during a typhoon in 2005.
Sediment monitoring systems consisting of a pressure type water level gauge, a turbidity meter, and a pipe hydrophone were installed in the main torrent of a 6.5-km2 drainage area and four tributaries in a mountain torrent in Japan. A pit bedload sampler was built at the main site to calibrate the hydrophone pulse data to the bedload transport rate. The sediment trapped in the pit provided information on the change in sediment grain size with time. Each field hydrophone was calibrated directly using a small chute. The results showed that sediment discharges differed widely from tributary to tributary due to differences in sediment production (erosion, debris flow, or landslide occurrence). The results also demonstrated that sediment transport equations cannot be applied to predict the sediment transport of mountain torrents because the discharged sediment is controlled not only by the sediment transport flow capacity, but also by the sediment production rate. The acoustic energy was recorded to cover the high sediment-transport range.
After the Tohoku (Japan) earthquake in March 2011, the National Institute for Land and Infrastructure Management interpreted and processed satellite images to determine the distribution of landslides caused by the earthquake. The study used multispectral images captured by the ALOS AVNIR-2 radiometer and panchromatic stereo images from the PRISM sensor provided by the Japan Aerospace Exploration Agency, along with satellite images with sub-meter ground resolution from Google Earth for coastal regions. The conclusions of this study are as follows. High-resolution (less than 1 m), large-area images from optical satellites were necessary for this sort of application. Automatic identification of landslides using satellite images led to some misidentifications. Visual interpretation of satellite images quickly revealed the spatial distribution of large-scale landslides. Acquiring recent nadir images and updating them frequently is essential.