Evaluation of near-surface conditions simulated for Russia by the Community Climate System Model Version 3.0 (CCSM3.0)

Accurate prediction of soil temperatures, freezing and thawing of the active layer depth is important for ecosystem, biogeochemical process and high latitude hydrology studies. The soil temperature at different depths simulated by Community Climate System Model Version 3.0 (CCSM) is evaluated over Russia for three climate episodes (1951-1980, 1961-1990, and 1971-2000) by means of observations provided by the National Snow Ice Data Centre (NSIDC). The CCSM well captures the annual phase of the soil temperature cycle, but not the amplitude. It predicts slightly too high (low) soil temperature in winter (summer). Root mean square error (RMSE), on average, are less than 4 K. The predictions of near-surface temperatures well agree with those from the European Centre for Medium-Range Weather Forecasts reanalysis (ERA-40) data. Thus, the simulated atmospheric temperature forcing is not the cause of the discrepancy in soil temperature prediction. Evaluation of precipitation predicted by the CCSM by means of the observations obtained from Global Precipitation Climatology Centre suggests that an inaccurate prediction of the precipitation may be the reason of the underestimation of soil temperature in summer. Evaluation of simulated snow depth by means of observations indicates that overestimation of snow depth may contribute to the overestimation of soil temperature in winter. The lower boundary conditions of the model also have great impact on the overall performance of the model. An accurate prediction of the soil temperature requires an accurate lower boundary condition and stability between the precision and the computational time of the model. Future investigations will focus on the sensitivity study of the soil type alone with its effect on inaccurate predictions of the soil temperature and an evaluation of the soil moisture.