Numerical Study On The Arctic Intermediate Water

The Arctic Circumpolar Boundary Current is another large-scale circulation flowingaround the earth’s axis. It brings the Arctic Ocean warm North Atlantic water, whichoccupies the whole intermediate depth and is usually named the Arctic IntermediateWater (AIW). Since it has the relatively high temperature, it may have importantclimatic significant and becomes a hot issue of the Arctic physical oceanography.However, owing to lack of in-situ data due to the harsh environment, and missing ofsatellite data far below the surface, a systematic knowledge of the AIW has not beenachieved so far, though many studies on this topic have been accomplished in recentseveral years. Nevertheless, numerical models have the advantage of huge coverageboth in time and space. So General Circulation Model (GCM) has become aneffective tool for Arctic researches. But according to reports from AOMIP, most ofmodels in it failed to reproduce the AIW’s characters.In the paper, three published data, climate model CCSM3/CCSM4from IPCC,reanalysis data with assimilation SODA and GLORYS are analyzed using the AtlanticWater Core Temperature (AWCT) as a benchmark, aiming to evaluate the models’performances on AIW referring to a hydrographic climatology PHC. The results showthat AWCT and its depth are poorly represented in CCSM. The most significant erroris misunderstanding the component parts of the AIW (i.e. between FSB and BSB),and failing to reproduce the fact that the warm Atlantic water loses much of the heatcontent in shelf seas Barents Sea and Kara Sea. While SODA succeed in reproducingthe spatial pattern of the AWCT, which means the AWCT decrease and deepen gradually along its pathway from the Fram Strait to the Canada Basin. SODA is thebest of the three on the AIW. But the AWCT distribution differs to a surprising extentbetween the neighboring months. The Author concludes that it is caused by thenonuniform assimilation data used in this reanalysis. In GLORYS, no assimilationdata are in the Arctic Ocean. As a result, the free simulated results show an apparenttrend with errors during the less than20years’ simulation time. By comparing relativeerrors of heat content, both CCSMs are found that they have positive errors in allbasins; SODA is the best of the three data, and GLORYS in1993also reproduces areasonal AIW while in2009there are negative errors in every basin, which means ahuge amount of heat is lost druing the run.A coupled ice-ocean model under NEMO framework is used in this paper with aclimatological forcing. Outputs of the10thyear are analyzed and compared with theobservations, which show that the model is acceptable on the global SST, sea-ice innorthern hemisphere and circulation of upper Arctic Ocean. However, like othermodels, there are significant errors on the AIW, especially in the Eurasian Basin,where the AWCT is over0.5degree lower than PHC and the depth of the AWCT isseveral meters deeper.By comparing the terms in the thermal equation, the author points out that theisopycnal diffusion in the GM90parameterization may lead to an excessive verticalmixing at the slope of the Eurasian Basin. While this excessive mixing is due to thespecial thermohaline structure there, and not the uncertainties of the model, which isalso proved in the PHC. So the author concludes that the GM90parameterization isno longer suitable, though it is widely used in general circulation models in otherregions. A sensitive experiment is done to illustrate the mechanism of the AIW, inwhich the vertical mixing from GM90is constrained artificially. Although the AWCTis not improved, the magnitudes of the terms in the thermal equation are changed. In this sensitive experiment, the equilibrium of the advection term and lateral diffusionterm is successfully established. And lack of advection is considered as the mainreason for the AIW problem in numerical models.Moreover, some other factors which may affect the simulation of AIW are alsodiscussed briefly, such as spin-up time and climatological forcing.