Reproducibility of Snow by a Non-Hydrostatic Regional Climate Model (NHRCM) Nested in JRA-25

Abstract

 We used a 20-km-mesh non-hydrostatic regional climate model (NHRCM) nested in the JRA-25 reanalysis to reproduce snow phenomena of the present climate around Japan during 1985-2004 winter (December to March). To verify the NHRCM, we used monthly mean values for snow depth, surface air temperature, total precipitation, and snowfall derived from observations of the Automated Meteorological Data Acquisition System of the Japan Meteorological Agency. We obtained the following results: (1) Surface air temperatures were simulated over Japan, with cold biases of about 1°C on the Pacific Ocean side of Tohoku district and warm biases of 1-2°C on the Japan Sea side. (2) The model generally underestimated monthly total precipitation, snowfall, and snow depth, and the regression coefficients of the modeled values against observed values were only 0.3, 0.3, and 0.4, respectively. The NHRCM showed large negative biases along the Japan Sea coast from Hokkaido to the Chugoku district and small positive biases from eastern Hokkaido to the Pacific Ocean side of Tohoku district. (3) These biases decreased at elevations above 500 m. (4) Both modeled and observed surface air temperatures in winter were distributed bimodally. We derived statistical regression equations as a substitute for dynamical downscaling by the NHRCM to provide snowfall and snow depth according to region and elevation range.