Reproducibility and future projection of monthly maximum number of consecutive dry days in Japan: Improvement by a 5-km-mesh regional climate model

In order to evaluate the regional changes for future weather extremes in the vicinity of Japan in the warm season (June-October), time-slice experiments are in progress using a 5-km-mesh cloud-system resolving regional climate model developed in Japan Meteorological Agency (JMA) and Meteorological Research Institute (MRI) (NHM-5km) nested within a 20-km-mesh atmospheric global climate model developed in MRI (MRI-AM20km). Climate simulations will be performed for present-day (1979-2003), near-future (2015-2039), and future (2075-2099) climate. To evaluate the improvements by NHM-5km compared to MRI-AM20km in reproducing the monthly maximum number of consecutive dry days (MCDD), their reproducibility of MCDD over Japan during June–October in 2002–2006 is statistically examined. MRI-AM20km reproduces poorly seasonal march of MCDDs during June-September and local features, e.g., significant underestimation in some regions during June-August. On the other hand, NH-5km well reproduces observed seasonal march of geographical MCDD distribution: northward shift of low MCDDs (< 6 days) during June-July, and high MCDDs (> 10 days) in western Japan and low MCDDs in northern Japan during September-October. Local features of the distribution, such as high MCDDs in coastal area of the Seto Inland Sea, are also well reproduced. The simulated MCDDs averaged for 11 sub-regions of Japan agree well with those observations (correlation coefficient, 0.769; root mean square error, 1.16; mean error, 0.54). Moreover, interannual variations in MCDDs averaged over Japan are well reproduced. These results demonstrate the advantage of using a high resolution cloud-system resolving model for reproducing regional climatology of MCDDs in the warm season. In the presentation, future change of MCDD will be also shown and discussed.