Numerical simulations are carried out for sea ice with four different advection schemes to study their effects on the simulation results. The sea ice model employed here is the Sea Ice Simulator （SIS） of the Geophysical Fluid Dynamics Laboratory （GFDL） Modular Ocean Model version 4b （MOM4b） and the four advection schemes are, the upwind scheme originally used in the SIS, the Multi-Dimensional Positive Advection （MDPA） scheme, the Incremental Remapping Scheme （IRS） and the Two Step Shape Preserving （TSSP） scheme. The latter three schemes are newly introduced. To consider the interactions between sea ice and ocean, a mixed layer ocean model is introduced and coupled to the SIS. The coupled model uses a tri-polar coordinate with 120x65 grids, covering the whole earth globe, in the horizontal plane. Simulation results in the northern high latitudes are analyzed. In all simulations, the model reproduces the seasonal variation of sea ice in the northern high latitudes well. Compared with the results from the observation, the sea ice model produces some extra sea ice coverage in the Greenland Sea and Barents Sea in winter due to the exclusion of ocean current effects and the smaller simulated sea ice thickness in the Arctic basin. There are similar features among the results obtained with the introduced three advection schemes. The simulated sea ice thickness with the three newly introduced schemes are all smaller than that of the upwind scheme and the simulated sea ice velocities of movement are all smaller than that of the upwind scheme. There are more similarities shared in the results obtained with the MPDA and TSSP schemes.
An annual cycle of atmospheric variations for 1989 in the Arctic has been simulated with the Weather Research and Forecasting （WRF） model. A severe cold bias was found around a cold center in surface air temperature over the Arctic Ocean, compared with results from ERA-Interim reanalysis. Four successive numerical experiments have been carried out to find out the reasons for this. The results show that the sea ice albedo scheme has the biggest influence in summer, and the effect of the cloud microphysics scheme is significant in both summer and winter. The effect of phase transition between ice and water has the biggest influence over the region near the sea ice edge in summer, and contributes little to improvement of the severe cold bias. The origi- nal crude albedo parameterization in the surface process scheme is the main reason for the large simulated cold bias of the cold center in summer. With a different land surface scheme than in the control run, cold biases of simulated surface air temperature over the Arctic Ocean are greatly reduced, by as much as 10 K, implying that the land surface scheme is critical for polar climate simulation.
LIU Xiying ZHAO Jiahua XIAHuasheng BAI Tonggui ZHANG Tao
