Abstract
Landslide Information Delivery System to deliver alert message to all users staying inside the area at risk where landslide occurrence possibility was assessed as high. The system was reviewed by comparing the operation results with actual landslide cases, in terms of “hit rate” or “missing rate”, to understand if the actual landslide was properly forecasted or not. The system was also compared with other existing systems, and was studied in terms of “whiff rate” as well as the time lag between the assessment time of landslide occurrence and actual landslide occurrence time. It was confirmed that, comparing with the grid information of Sediment Disaster Warning published by Japan Meteorological Agency during the heavy rain in Hiroshima (2014), Kanto-Tohoku (2015) and Kumamoto (2016), the proposed system could identify an area at risk on a smaller scale, corresponding better to the actual landslide cases.
It was revealed that time lag between the assessment time and actual landslide occurrence time leaves a certain amount of time that can be used for evacuation. The study also showed that more than 80 percent of actual landslide cases were assessed in the system, although there were some “missing” cases where landslide occurrence was not assessed beforehand.
