Study Of Melt Pond Fractions On Arctic Sea Ice Simulated Using CICE5.0

The albedo of melt pond is greater than open water but less than sea ice. In numerical modelling, melt pond fractions impact the calculation of sea ice surface albedo significantly, thus emphasizing the large effect accurate pond fraction simulation has on the heat budget in the atmosphere-ice-ocean system.Comparison is carried out of the three melt pond parameterization schemes in CICE5.0 based on satellite data and field observations. The results show that since the beginning of July, pond fractions decrease and deviate from MODIS results step by step. In particular, the extent that ponds cover and pond fractions simulated by the topo scheme (topographic scheme) are very small in sharp contrast to MODIS data. It imporves pond simulation greatly to switch conditions to that of the cesm scheme (simple and semi-empirical scheme) determing when for ponds to freeze.For all the three parameterizations, ponds last too long on first-year ice but coincide with observations on multi-year ice. When it comes to pond fraction evolution at a certain site, only the first and fourth stages are simulated correctly on the physical basis. After code bugs are fixed, pond water is allowed to percolate into the ocean through sea ice. But there are nearly no ponds on multi-year ice in cesm and Ivl schemes and the second and third stages are still missing. Simulation from newly released CICE5.1.2 shows similar results, indicating that the evolution of sea ice permeability or other physical processes remains to be improved in CICE model.Strengths and weaknesses of the three schemes are analyzed systematically. It is practical that the extent that ponds cover spreads from lower latitudes to higher latitudes for the cesm scheme and Ivl scheme (level ice scheme) but the range of inter-annual variability of pond fractions averaged over the Arctic Ocean do not agree well with observations. The inter-annual variability of topo averaged pond fractions falls into the normal range and the extent ponds cover is also practical but ponds on multi-year ice occurs much too early which may be settled by introducing snow infiltration from the other two. Relatively, the topo scheme can be based on for further study.Further experiments are implemented based on the topo scheme. It is highly recommended to reduce the ice classification of thin ice, which can better pond simulation in the western Arctic Ocean. The introduction of snow infiltration from the Ivl scheme makes it practical for topo ponds to spread from lower latitudes to higher latitudes. But three factors render the extent that ponds cover and pond fractions much smaller on multi-year ice, including too much water loss through the edge of ice floes, too strong Ivl snow infiltration and improper ice classification. Atmospheric temperature affects pond simulation to a large extent. In terms of the Arctic Ocean average, ponds freeze 15-20 days later if atmospheric temperature is raised 0.6 deg higher.