Three-month-long simulations are conducted using WRF/ARW for warm season over 11 years from 1993 through 2003 in which NCEP2 reanalysis data is given as initial and boundary forcing. The experiments revealed that the dynamical downscaling is very effective for Mongolian region to reproduce detailed spatial pattern of precipitation when SSiB is adopted as land surface scheme. The correlation coefficient for JJA precipitation for Mongolia is significantly higher than other experiments using different land surface scheme.
Most experiments show large bias in total precipitation for Mongolia. Sensitivity experiment using realistic initial soil moisture, in which GSWP dataset is adopted, leads to reduce precipitation bias significantly. This soil moisture correction is also effective to improve intra-seasonal variation of precipitation in East China area by improving southerly flow in lower troposphere by EASM. From these experiments, we conclude that land surface process is crucially important for simulating EASM and inland area in Eurasian continent.