As a substitute for the original displaced pole grids, a simple rotated spherical coordinate system was introduced into the Community Sea Ice Model version 4(CSIM4), which is a component of the Community Climate System Model(CCSM) of the American National Center of Atmospheric Research(NCAR), to deal with the pole problems.In the new coordinates, both the geographical North Pole and South Pole lie in the model equator and grid sizes near the polar region are more uniform.With reanalysis dataset of American National Centers for Environment Prediction(NCEP) and Levitus dataset without considering sub-mixed layer heat flux, the model was integrated for 100 years with thermodynamics process involved only in the former 49 years and both dynamic and thermodynamic processes involved in the left time.Inner consistency of model results was checked with no contradiction found.The results of last 10 years’ model output were analyzed and it is shown that the simulated sea ice seasonal variation is rational whereas sea ice extent in the Barents Sea in winter is larger than that of observation.Numerical experiment on influence of sub-mixed layer heat flux was also carried out and it is shown that the sub-mixed layer heat flux can modulate seasonal variation of sea ice greatly.As a model component, the sea ice model with rotated spherical coordinates was coupled with other models (the oceanic general circulation model is the LASG/IAP Climate System Ocean Model(LICOM) with reduced grid, other models are components of NCAR CCSM2) forming a climate system model and its preliminary results were also given briefly.