Simultaneous flood risk analysis and its future change among all the 109 class-A river basins in Japan using a large ensemble climate simulation database d4PDF

Abstract

This study investigated simultaneous flood risk among all the 109 class-A river basins over Japan using the big data of (over 1000 years) annual maximum hourly flow simulated from a large ensemble climate simulation database for policy decision making for future climate change, and proposed a novel approach in its geospatial analysis by applying two informatics techniques: the association rule analysis and graph theory. Frequency analysis of the number of rivers with the annual maximum flow over the flow capacity in the same year (defined as simultaneous flooding here) indicated that simultaneous flood risk will increase in the future climate under 4-degree rise scenarios in Japan, whose increment is larger than the variation of sea surface temperature projections. As the result, the return period of simultaneous flooding in eight river basins (the same number as in a severe storm in western Japan, 2018, causing the second worst economic damage since 1962) is estimated at 400 years in the historical experiment, 25 years in the 4-degree rise experiment. The association rule and graph theory analyses for the big data of annual maximum flows in the future climate scenarios indicated that simultaneous flood occurrence is dominated by spatial distance at a national scale as well as by the spatial relation between mountainous ridges and typhoon courses at a regional scale. Large ensemble climate simulation data combined with the informatics technology is a powerful approach to simultaneous flood risk analysis.

Keywords

• simultaneous flood occurrence
• ensemble climate simulation
• d4PDF
• association rule analysis
• graph theory