Analysis of future precipitation change in Shikoku region usingstatistical downscaling

摘要

This study investigates the applicability of the statistical downscaling model (SDSM) in downscal-ing precipitation in Shikoku region, Japan, for both historical and future time periods. We calibrated the SDSM model using the National Center for Environmental Prediction (NCEP) reanalysis datasets and daily time series of precipitation in Shikoku acquired from nine surface daily observation points (SDP) and validated the NCEP model and general circulation model (GCM) outputs of HadCM3 under SRES A2 and B2 scenarios (1961-2099). To predict the future mean and extreme precipitation, we investigated the approach of downscaling the outputs of a GCM to a local scale using the SDSM to downscale precipitation in present and future climate scenarios. The results showed that in both climate scenarios: (1) the time series generated by SDSM indicates a clear increasing trend in the mean daily precipitation values in future winters and a decreasing trend in the summer; (2) the annual change in the number of days that exceed the 20 mm/day precipitation (R20) would significantly decrease in northern Shikoku; (3) the changes of R00 in the northern region were much more severe than in the southern region under both scenarios because the changes of R00 decreasesin the northern region occurred in spring, summer and autumn; and (4) in the future, the stable reservation of water for agricultural use becomes difficult in the northern Shikoku region. Annual maximum precipitation frequency was analyzed by fitting more than 50 continuous and discrete distributions. Distribution-estimated magnitudes of annual maximum precipitation were compared for studied return periods using H3A2 and H3B2 in the base period (1961-1990), 2020s (2011-2040), 2050s (2041-2070), and 2080s (2071-2099). We are confident about the impact of climate change on precipitation, and in the ability to simulate extreme precipitation events that may affect agriculture.