Antarctic sea ice plays a key role in the present climate system, providinga regulating balance between the atmosphere and ocean heat fluxes, as wellas influencing the salt fluxes and deep water formation over the continentalshelves. The severe winter environmental conditions of the Antarctic sea-icezone make it difficult to observe many of the physical characteristics in acomprehensive way. The inter-relations between the variables mean that muchcan be learnt from the observations of some features along with detailednumerical modelling of the whole system and the interactions between thevariables. This study therefore aims to use numerical modelling of theatmosphere, sea ice and surface mixed-layer ocean in the sea-ice zone,together with observations to simulate a comprehensive range of parametersand their variability through the annual cycle to provide a basis forfurther observations and model validation for the present climate.

The model includes a coupled atmospheric general circulation model with aninteractive dynamic and thermodynamic sea-ice model and surface mixed-layerocean. The deep ocean and ocean surface conditions outside the sea-ice zoneare constrained to the present mean climate conditions to ensure no climaticdrift. The sca-ice model is similar to previous published versions, bill hasrefined schemes for partitioning of the freezing of frazil ice within theleads and under the ice floes, and for rafting. These perform well in bothpolar regions with the same physics. The model simulates the annual cycle ofatmospheric and sea-ice features well in comparison with data from theglobal atmospheric analyses, the satellite sensing of sea ice, and thelimited in situ surface observations.

The output from the model also includes: all components of the heart fluxes,atmospheric profiles and surface temperatures for air, ice and ice-oceanmixtures, open-water fractions, surface snow and snow-ice depths, and thesea-ice convergence-divergence and drift. The comparison of these featureswith additional observations provides a means for further validating themodel and representing the present climate more closely.

The Sea-Ice Model Intercomparison Project (SIMIP) is part of the activitiesof the Sea Ice-Ocean Modeling Panel (SIOM) of the Arctic Climate SystemStudy (WMO) (ACSYS) that aims to determine the optimal sea-ice model forclimate simulations. This investigation is focused on the dynamics of seaice. A hierarchy of four sea-ice rheologies is applied, including aviscous-plastic rheology, a cavitating-fluid model, a compressible Newtonianfluid, and a simple scheme with a step-function stoppage for ice drift.

For comparison, the same grid, land boundaries and forcing fields areapplied to all models. Atmospheric forcing for a 7 year period is obtainedfrom the European Centre for Medium-Range Weather Forecasts (UK) (ECMWFanalyses), while occanic forcing consists of annual mean geostrophiccurrents and heal fluxes into a fixed mixed layer. Daily buoy-drift datamonitored by the International Arctic Buoy Program (IABP) and icethicknesses at the North Pole from submarine upward-looking sonar areavailable as verification data. The daily drift statistics for separateregions and seasons contribute to an error function showing significantdifferences between the models. Additionally, Fram Strait ice exportspredicted by the different models are investigated. The ice export of theviscous-plastic model amounts to 0.11 Sv. when it is optimized to the meandaily buoy velocities and the observed North Pole ice thicknesses. Thecavitating-fluid model yields a very similar Fram Strait outflow, butunderestimates the North Pole ice thickness. The two other dynamic schemespredict unrealistically large ice thicknesses in the central Arctic region,while Fram Strait ice exports are too low.

A one-dimensional (1-D), thermodynamic sea-ice model with parameterized icedynamics is coupled to a mixed-layer ocean model and driven with prescribedatmospheric forcings for the central Arctic. The model is used to calculatethe sensitivity of the ice pack to various parameterizations that havetraditionally been neglected or considered only implicitly in large-scalesea-ice models. The model includes melt ponds, leads (with summertimestratification), an ice-export term, a stability-dependent air–seaheat-exchange coefficient, a prognostic ocean–ice heat exchange, a crudeice-thickness distribution, and a sophisticated albedo parameterization.

The ice pack is sensitive to the partitioning of solar energy betweenlateral melting and mixed-layer warming, with the most realistic simulationsoccurring when the heat is nearly evenly divided between these twoprocesses. Conversely, ice thickness and coverage are fairly insensitive tothe amount of lateral mixing within the upper ocean, vertical mixing withinleads, and to the partitioning of mixed-layer heat content between warmingthe water and melting the ice bottom. The ice concentration during summer isstrongly dependent on the assumed ice-thickness distribution: the amount ofopen water during summer is less than half the size of the empirically baseddistribution used here, compared with one in which ice floes are distributeduniformly across a range of thicknesses.

The Arctic represents a relatively pristine frontier that is vulnerable topollution. Substances originating at mid latitudes are transported to theArctic by atmospheric processes, ocean currents and rivers. These pollutantscan accumulate in the Arctic environment. Testing of nuclear weapons,dumping of waste and operation of ships, and nuclear power plants also posethreats.

To investigate possible pollutant pathways we used a multi-levelprimitive-equation ocean model, coupled to a dynamic-thermodynamic sea-icemodel. Coupling included conservative transfer of momentum, heat and freshwater. Atmospheric forcing (wind stress, temperature, humidity, radiationand heat fresh-water fluxes) was supplied by datasets or bulk formulae. Openlateral-boundary conditions for the ocean model were supplied either bydatasets (temperature and salinity) or from a larger-scale ocean model(momentum). Two integrations were compared — one used a centred-differenceadvection scheme and viscous damping, while the other used a betterrepresentation of an advection scheme and a sub-gridscale eddyparameterization.

Tracer simulations showed (1) the importance of good representation ofnumerical advection, and (2) the role of eddy interacting with sea-floortopography (the neptune effect).

The effects on ice-thickness distribution of including ridge-relatedroughness in the drag parameterization is investigated using a sea-icedynamic and thermodynamic model. A long-term integration of the sea-icemodel forced by a 5 day running mean of daily geostrophic winds and steadybut spatially varying ocean currents was performed to constructclimatological thickness fields. Compared to a constant drag coefficientmodel, results that include roughness have a significantly different spatialdistribution with much thicker ice in the western Arctic. These resultsmatch observations better than those where ice roughness was omitted.

Sea ice on the large scale is characterized by leads and ridges thattypically have a given orientation. Because of various flaws, we wouldexpect that the ice will form oriented leads and ice-thicknesscharacteristics that control the heat and moisture fluxes into theatmosphere. Prediction of these oriented leads, ridges and slip lines isrelevant to understanding the role of ice mechanics in global climate changeas they can play a significant role in the ice-thickness distribution.

In this paper we develop a model for the dynamical treatment of leads andoriented flaws in large-scale sea-ice models. Two particular isotropicrealizations of this model relevant to climate studies are examined: (a) anisotropic composite with oriented leads in all directions imbedded in thickice, and (b) a simple "strain hardening" isotropic model where only orientedleads having the potential to open rapidly are allowed. Under applied stressboth models yield preferential deformation along a symmetric pair ofintersecting leads or ridges with the intersection angles dependent on theconfinement stress. The "uniform-orientation" model results in a yield curvethat approximates a sine lens, while the "strain hardening" model has ateardrop-like yield curve. How the resulting fracture-based yield curves andnon-normal flow rules may be cast in a form usable in numericalinvestigations of climate is discussed.

The surface radiation budget of the polar regions strongly influences icegrowth and melt. Thermodynamic sea-ice models therefore require accurate yetcomputationally efficient methods of computing radiative fluxes. In thispaper a new parameterization of the downwelling shortwave radiation flux atthe Arctic surface is developed and compared to a variety of existingschemes. Parameterized llnxes are compared to in situ measurements usingdata for one year at Barrow, Alaska. Our results show that the newparameterization can estimate the downwelling shortwave flux with mean androot mean square errors of 1 and 5%, respectively, for clear conditions and5 and 20% for cloudy conditions. The new parameterization offers a unifiedapproach to estimating downwelling shortwave fluxes under clear and cloudyconditions, and is more accurate than existing schemes.

Fluxes of heat and moisture over heterogeneous snow cover are studied usinga boundary-layer model. The performance of a “tile” model, suitable forcalculating gridbox-average surface fluxes in a GCM, is assessed incomparison with the boundary-layer model. The impact of using a tilerepresentation for heterogeneous snow cover in a single-column version ofthe Hadley Centre GCM is discussed.

A meteorological-analysis procedure allowing simulation of the snow cover inmountainous regions, where the general circulation model orography differsfrom the real orography, is described. This procedure uses model outputs toestimate the data needed to force a snow model. Temperature andprecipitation deduced from three five-year runs were compared to the snowclimatology of the French Alps. The snow cover simulations are verysensitive to temperature at middle elevations. At high elevations, thealtitude of the equilibrium line is simulated well.

We have conducted tests of the Canadian Land Surface Scheme (CLASS V2.5) forArctic tundra applications. Our tests emphasize sensitivities to initialconditions, external forcings and internal parameters, and focus on theAlaskan North Slope during the summer of 1992. Observational data from theNational Science foundation (NSF), Arctic Systems Science (ARCSS),Land/Atmosphere/Ice Interactions (LAII) Flux Study is available to serve asforcing and validation for our simulations.

Comparisons of the runs show strong sensitivities to the composition anddepth of the soil layers, and we find that a minimum total soil depth of 5.0m is needed to maintain permafrost. The response of the soil to diurnalvariations in forcing is strong, while sensitivities to other internalparameters, as well as to precipitation, were relatively small. Somesensitivity to air temperatures and radiative fluxes, particularly theincoming shortwave flux, was also present. Significant sensitivity to thespecification of the initial water and ice contents of the soil was found,while the sensitivity to initial soil temperature was somewhat less.

Although the snow albedo feedback mechanism has been shown to amplify globalwarming effects in nearly all models of global climate, it continues to berepresented as a simplistic parameterization. Here, we demonstrate howchanges in snow-pack energy-balance drive the seasonal fluctuations in snowalbedo for the Greenland ice sheet. For a detailed, point-basedinvestigation of the relationship between snowpack energy balance andalbedo, two models are coupled together; one that calculates snow grain-sizeand the other that uses those grain-size data as input to aradiative-transfer code to obtain spectral albedo. These data indicate thatin the near-infrared wavelengths, albedo values drop nearly 20%, during a 10day period during which grain-sizes increased dramatically. Satellite datawere used to map monthly changes in albedo over the entire Greenland icesheet during the spring and summer months. These monthly albedo imagesindicate albedo reductions of as much as 80% in coastal regions. Even inareas that experience little or no melt, albedo decreases of 10–20% werecommon. From these results, it is clear that snow albedo parameterizationsfor climate models must incorporate the dynamics of snowpack energybalance.

Significant improvements to the representation of climate forcing andmass-balance response in a coupled two-dimensional global energy balanceclimate model (EBM) and vertically integrated ice-sheet model (ISM) have ledto the prediction of an ice-volume chronology for the most recent ice-agecycle of the Northern Hemisphere that is close to that inferred from thegeological record. Most significant is that full glacial termination isdelivered by the model without the need for new physical ingredients. Inaddition, a relatively close match is achieved between the Last GlacialMaximum (LGM) model ice topography and that of the recently-described ICE-4Greconstruction. These results suggest that large-scale climate systemreorganization is not required to explain the main variations of the NorthAmerican (NA) ice sheets over the last glacial cycle. Lack of sea-ice andmarine-ice dynamics in the model leaves the situation over the Eurasian (EA)sector much more uncertain.

The incorporation of a gravitationally self-consistent description of theglacial isostatic adjustment process demonstrates that the NA and EA bedrockresponses can be adequately represented by simpler damped-relaxation modelswith characteristic time-scales of 3–5ka and 5 ka, respectively. Theserelaxation times agree with those independently inferred on the basis ofpostglacial relative sea-level histories.

A multi-layered snow model, including most physical processes governing theevolution of snowpacks, has been coupled to a global circulation model (GCM)to improve the representation of snow cover in climate simulations. The snowmodel (Crocus) includes original features to simulate the evolution ofsnowpack layering that allows a realistic calculation of snow albedo as afunction of the type and size of the crystals of the surface layer. Thecoupling scheme is based on a synchronous run of the GCM and of the snowmodel with an exchange of the surface fluxes at every time-step. It wastested in a five-year run at a T42 resolution. The impact on the atmospherewas important over most snow-covered regions and the snowpacks simulated inthe different regions present a layering that is realistic and very variablein connection with the climate. The simulated snow cover comparessatisfactorily with the present snow climatology.

The Laboratoire de Météorologie Dynamique (LMD) variable-grid atmosphericgeneral circulation model (AGCM) was used in this study for a five-yearhigh-resolution simulation of the Antarctic climate. The horizontalresolution is about 100 km over a large part of the ice sheet. This studyfocuses on the simulated surface mass balance (precipitation-evaporationsublimation-melt) and on the spatial and temporal variability of snowfall inAntarctica. The simulated annual mean surface mass balance for the wholecontinent is close to the observed value, and the model simulates well thespatial distribution of the surface mass balance. The annual cycle ofsnowfall exhibits a clear minimum in summer over the high interior plateauas well as for Antarctica as a whole, in agreement with the observations. Inthe interior of the continent, the model produces a permanent lightbackground snowfall that accounts for about 5% of the total annualprecipitation. The bulk of the snowfall is produced irregularly duringperiods that generally last only two or three days that are caused bycyclones off the coast.

Most spectral general circulation models (GCMs) use an envelope topographyto set on land surface elevations. The UK Universities Global AtmosphericModelling Programme General Circulation Model (UGAMP GCM) uses such aformulation for Antarctica, based on the US Navy 10 are minute charts of theregion. In the marginal regions of the continent, an envelope topographyconsistently overestimates the elevation leading to lower-than-observedsurface temperatures. Furthermore, errors in excess of 1000 m exist in theUS Navy data, and the UGAMP GCM treats the major ice shelves as sea ice,introducing a 12% reduction in the snow-covered area of the continent. Here,we use a new high-resolution, high-accuracy digital elevation model toimprove the representation of Antarctica in the UGAMP GCM. The effect ofchanging the land–sea mask and the topography on the surface temperature,precipitation and wind held is investigated for both summer and winter runs.Changing the land–sea mask had a dramatic effect on temperature, producing areduction of 13.3°C for the sector west of the Ross Ice Shelf. Using the newmean topography also introduces substantial differences in temperature, windspeed and precipitation for summer and winter.

Linearizations about two horizontal-dimensional ice sheets are proposed asmethods of generating normal mode initializations for ice-sheet models andfor computing the short-term response. Linearized models can be generateddirectly from balance-flux calculations without the need for tuning the ratefactor.

A linearized model is compared with the Eismint Benchmark, and the normalmodes for two coarse Antarctic digital elevation models are computed andcompared. Volumetric relaxation spectra are presented. The slowest mode hasa time constant comparable to that computed from scale theory.

Open-water leads in sea ice dominate the exchange of heat between the oceanand atmosphere in ice-covered regions, and so must be included in climatemodels. A parameterization of leads used in one such model is compared toobservations and the results of a detailed Arctic sea-ice model. Suchcomparisons, however, are hampered by the errors in observed lead fraction,but the parameterization appears to compare better in winter than in summer.Simulations with an atmospheric general circulation model (AGCM), usingprescribed sea-surface temperatures and ice extent, are used to illustratethe effect of parameterized lead fraction on atmospheric climate, and soprovide some insight into the importance of improved lead-fractionparameterizations and observations. The effect of leads in the AGCM islargest in Northern Hemisphere winter, with zonal mean surface-airtemperatures over ice increasing by up to 5 K when lead fraction isincreased from 1% to near 5%. The effect of leads on sensible heat loss inwinter is more important than the effect on radiative heat gain in summer.No significant effect on sea-level pressure, and hence on atmosphericcirculation, is found, however. Indirect effects, due to feedbacks betweenthe atmosphere and ice thickness and extent, were not included in thesesimulations, but could amplify the response.

The role of dynamics in modifying the response of the Arctic ice pack tointer-annually varying forcings and to climate perturbations is investigatedusing simulations from a two-dimensional ice model and a global climatemodel (GCM). Inter-annual variability in ice-covered area for 1985-93 isdominated by ice transport, and different transport regimes affectsubstantially the response of the ice pack to climate perturbations. Thethermodynamic-only simulations are more sensitive to initial ice conditions,and respond less than the dynamk-thermodynamic model to small perturbations,but with a greater response to larger perturbations. Comparisons of GCMsimulations that use different ice treatments highlights the importance ofconsidering the distribution of ice thickness and extent in assessingclimate-change responses.

The Climate System Model (CSM) developed at the National Center forAtmospheric Research (NCAR) consists of atmosphere, land and ocean models,as well as a dynamic-thermodynamic sea-ice model. The results of sea-icesimulation using the first coupled climate simulation with the CSM ispresented. It was found that the simulated total-ice areas in bothhemispheres compared well with observations for winter, but were too largefor summer. The numerical solution of the cavitating fluid dynamics wasfound to allow excessive ridging of ice, and an ad hoc correction wasimplemented. The ice velocities were realistic for the Antarctic, but forthe Arctic were turned toward Alaska and Siberia by modeled winds andcurrents. This ice-drift pattern was reflected by ice thickness, which lacksthe observed ridging near Greenland. The results illustrate the sensitivityof sea ice to the simulation of polar climate and the challenge of modelingthe entire climate system.

This paper investigates the long-term impact of sea ice on global climateusing a global sea-ice–ocean general circulation model (OGCM). The sea-icecomponent involves state-of-the-art dynamics; the ocean component consistsof a 3.5° × 3.5° × 11 layer primitive-equation model. Depending on thephysical description of sea ice, significant changes are detected in theconvective activity, in the hydrographic properties and in the thermohalinecirculation of the ocean model. Most of these changes originate in theSouthern Ocean, emphasizing the crucial role of sea ice in this marginallystably stratified region of the world's oceans. Specifically, if the effectof brine release is neglected, the deep layers of the Southern Ocean warm upconsiderably; this is associated with a weakening of the Southern Hemisphereoverturning cell. The removal of the commonly used “salinity enhancement”leads to a similar effect. The deep-ocean salinity is almost unaffected inboth experiments. Introducing explicit new-ice thickness growth in partiallyice-covered gridcells leads to a substantial increase in convectiveactivity, especially in the Southern Ocean, with a concomitant significantcooling and salinification of the deep ocean. Possible mechanisms for theresulting interactions between sea-ice processes and deep-oceancharacteristics are suggested.