Progress Report
Last updated:
Artificial generation of upstream maritime heavy rains to govern intense-rain-induced disasters over land (AMAGOI)[8] Damage Estimates
Progress until FY2025
1. Outline of the project
In this research item, we develop a methodology for estimating the effect of weather control on reducing economic damage (Fig.1). Specifically, we are developing; 1) a flood and inundation simulation model for the entire Japan for calculating the inundation area and its depth in the cases with and without weather control; 2) an exposure asset database based on various statistical data and a flood economic damage estimation model based on inundation depth and exposure information, and 3) a risk estimation model of landslide disasters triggered by heavy rainfall events. In FY2025, we newly started 4) developing the scenarios for implementing weather control in the actual society.
2. Outcome so far
1) Disaster Data Development for Model Regions (8-1)
In FY2025, Kyushu was selected as the model region. MLIT flood damage statistics were integrated to organize inundation frequency and losses by river system and inundation mechanism. Inundation-area data from public sources, including GSI and MLIT, were linked to heavy-rainfall events, and representative events were screened for reproduction analysis. With 4-1 and 8-2, the 2021 northern Kyushu heavy rainfall event was analyzed using weather-control model experiments. Changes in flood inundation and economic damage with and without intervention were estimated, confirming consistency among rainfall change, inundation scale, and damage (Fig.2).
2) Economic Damage Estimation Workflow (8-2)
An integrated framework and workflow were developed for economic damage estimation. The model provides one-stop estimates of human damage, private fixed-asset damage, and indirect losses, including business interruption and lost profit. A benchmark reproduction test for the 2015 Kinugawa River flood showed that mean estimated damage was within 70-90% of observed values. Further parameter tuning is needed, but benchmark performance was favorable. The team also contributed to analysis of the 2021 northern Kyushu weather-control experiment (Fig.2).
3) Hybrid Model for Slope Failure (8-3)
For rainfall-triggered slope failures, probabilistic-statistical and physical models were evaluated using the Hiroshima heavy rainfall case. The improved statistical model achieved ROC-AUC = 0.73, where 1 indicates high accuracy and 0.5 random prediction. Sensitivity analysis of the physical model identified soil and geological conditions as key factors. Relationships between the two model outputs were then developed to expand applicability through hybridization.
4) Natural Disaster Timeline Study (8-4)
Timeline cases for past natural disasters in Japan, including floods, sediment disasters, and storm surges, were collected and organized by hazard type, region, and meteorological conditions. Based on these cases and existing warning systems, preliminary timelines were drafted for residents, public agencies, and weather-control operators. Their timing was also compared with stages of assumed weather control, such as cloud seeding.
3. Future plans
For the flood inundation and economic damage estimation model, reproduction calculations of inundation areas and damage amounts will be conducted for major past flood events across Japan. For the slope failure model, the range of target cases will be expanded for further validation. For social implementation scenarios, timelines for residents, public agencies, and weather-control operators will be developed in model regions, assuming the use of weather control.