Downscaling of CMIP5 ocean future projections around Japan using high-resolution regional ocean models

This study aims to create the downscaling dataset of CMIP5 ocean future projections around Japan, which will be useful for climate change adaptation research in the Japanese local governments. Since CMIP5 models basically have global and low-resolution framework, they are not good to reproduce oceanic fine structures such as coastal currents, Kuroshio/Oyashio, and their related fronts/eddies, which will be important factors in considering future climate of Japan coastal seas. Thus, we have developed North Pacific ocean model with 10km horizontal resolution (10km model) and regional ocean model around Japan with 2km horizontal resolution (2km model) for downscaling CMIP5. We have adopted several CMIP5 models and scenarios (historical, RCP8.5, and RCP2.6) to compose the surface atmospheric forcings for ocean future projection simulation. Using the CMIP5 atmospheric forcings with the 10km and 2km models, we have performed ocean future projection experiments from 1981 to 2100 with several cases and scenarios and produced the ensemble dataset (here called SI-CAT data). The basic spec of the 10km model is common to the ocean reanalysis dataset FORA-WNP30, and we have also prepared some reference dataset including the realistic historical simulation using the atmospheric reanalysis data JRA55. The practicability of cross validation with these realistic products is one important feature of the above ocean future projection products. From the validations, we have confirmed basically good reproducibility of historical fields and reasonable climate change response (e.g., SST, SSH, currents and transports around Japan) and also found some biases against the realistic reference dataset. Details of the experiments and validations will be shown in the presentation.