Scale correction, verification, and warming attribution of summer rainfall extremes over Japan for the d4PDF climate ensemble

Abstract

Historical climatological upper percentile rainfall over Japan from the Database for Policy Decision-making for Future climate changes (d4PDF) was compared with that from 121 rain-gauges for the 1952–2010 period. Percentiles calculated excluding or including values below 1 mm, “wet-extremes” and “all-extremes” were compared. The 5 km d4PDF hourly wet (all) extremes matched rain-gauges at 72% (45%) of the locations, but for daily extremes this was 39% (33%). Further comparison with radar-AMeDAS (Automated Meteorological Data Acquisition System) was made for hourly extremes in the 2006–2021 period. Wet (all) extremes matched for 84% (79%) of rain-gauges, and 80% (51%) also matched radar-AMeDAS at the same locations. Exceedance probabilities were calculated for historical and 4 K warmer climates using thresholds based on a non-warming climate. Historically, hourly (daily) extremes were more frequent over most of (eastern) Japan. In a 4 K warmer climate, daily extremes would become more frequent over most of Japan. Rainfall from 5 km d4PDF at hourly and daily resolutions are provided in prefecture and river basin subsets, with compressed sizes of each below 1 Gigabyte.