Projecting Multiscale River Flood Changes Across Japan at +2°C and +4°C Climates

Abstract

This study addresses computational challenges in high-resolution, large-domain, process-based flood quantile estimation, focusing on Japan's future flood risks at 150 m resolution. Using the Aggregating Grid Event (AGE) method, the Rainfall-Runoff-Inundation (RRI) model, and the Peaks-Over-Threshold (POT) approach, it incorporates 2,160-year precipitation data from a 5-km dynamically downscaled ensemble (d4PDF DDSJP) across three climate stages (historical, +2°C, +4°C). The AGE method identified critical precipitation events for the flood quantile estimations and the POT method was employed to estimate 100-year discharge (Q100) for over 2.2 million river grid cells. Key findings include: (a) Nationwide, 100-year discharge (Q100) is projected to increase 1.16 times (+2°C) and 1.37 times (+4°C), with equivalent return periods reduced to 45 years (+2°C) and 23 years (+4°C). Northern regions (Hokkaido and Tohoku) are particularly climate-sensitive, exceeding national averages in Q100 increases. (b) Small river basins and transition zones from plains to mountains exhibit higher increase ratios, necessitating targeted flood prevention measures. (c) Flash flood risks are expected to rise, with most national basins seeing flashiness increases of over 10% (+2°C) and 20% (+4°C). Southern Japan faces further flash flood intensification, while Northern Japan under +4°C stage anticipates emerging challenges related to flash floods. The study underscores the urgency of adaptive flood management strategies to mitigate increasing risks, offering a foundation for informed policymaking and public-engaged mitigation. Simulation data opens pathways for further research on cascading disaster scenarios in +2°C and +4°C climates.