Characteristics of the Simulated East-Asian Summer Monsoon Circulation in the RegCM3

In this study, we investigate the characteristics of the simulated regional climate over East Asia during the summertime using a regional climate model (RegCM3). RegCM3 has been run for the period of June-July-August for 1982-2003, creating three-month long simulations from initial and boundary conditions provided by the NCEP/NCAR Reanalysis II datasets, which are available at 6-h intervals with a resolution of 2.5° × 2.5°. The ability of the model performance from diurnal to inter-annual variation is accessed. Observational data sets from weather station to global atmospheric datasets over timescales from diurnal to annual are utilized for validation of the simulated climate. Resulting climatology with the dynamic model is discussed in comparison with that using a statistical method combining the EOF (empirical orthogonal function) and SVD (singular value decomposition). The RegCM3 simulated realistically not only the summer mean climatology in precipitation but also its inter-annual variation. Less precipitation over the oceans and over the Korean peninsula is apparent, however. Precipitation over South China and Mongolia-Siberian regions is overestimated. Overall, the precipitation bias is about 0.5 mm/day. Large-scale features are also well reproduced with a bias in temperature at about 0.5K in the mid- and upper tropospheres. A cold bias of about -2 K is apparent in the lower troposphere, which is consistent with the positive bias in precipitation. Accordingly, low level jet is over-intensified. Inter-annual variation of seasonal mean precipitation anomalies generally follows what was observed. Intra-seasonal variation of simulated precipitation is well captured by the model for June and July, whereas in August the precipitation in sub-tropics is missed. Daily variations of temperature and precipitation over South Korea are also reasonably simulated by the RegCM3, with the time correlation of abut 0.7 and 0.5, respectively. Diurnal variation of simulated temperature over South Korea is too week, whereas the variation of precipitation is too strong for precipitation, which is due to the fact that the RegCM3 tends to overestimate the daytime precipitation. In the comparison of downscaled surface temperature and precipitation over South Korea, for temperature, the correlation coefficient from the dynamical downscaling method is larger than that from the statistical approach, whereas the coefficient in precipitation is larger by the statistical method than by the dynamical approach.