In this study, based on the ideas such as statistical downscaling method (Wilby et al., 2004) which estimates regional precipitation amounts from atmospheric conditions, we check the reproducibility of atmospheric conditions around Japan by CMIP5 models and estimate the future change in extreme precipitation in Japan with those data (not precipitation data itself).
Firstly, we made a composite map of daily 850hPa water vapor flux of the reanalysis dataset JRA-55 (Kobayashi et al., 2015, Ebita et al., 2011) on days which corresponds to the top 5 percent of the extreme precipitation in June in Kagoshima for the period from 1958 to 2014. Kagoshima is located in the most southern part of the main islands in Japan and often experiences much rain in the rainy season. Next, we made the composite map of daily 850hPa water vapor flux for each CMIP5 model on days corresponds to the top 5 percent days of daily precipitation in June for the period from 1979 to 2005 at the grid where is the nearest from Kagoshima observatory, similarly.
By comparing these composite maps, we checked the basic characteristic flow from the south of Kagoshima and the reproducibility of the CMIP5 models. As a result, we found that heavy precipitation at Kagoshima in the present climate occurs under slightly different atmospheric conditions in CMIP5 models.
In the further study, based on the reproducibility, we intend to assess the future change in extreme precipitation in Japan with statistical downscaling approach.