Caracterización de los hogares que perciben la CFB

1.5.1 A review of coupled ice-ocean models

W e now sum m arise som e o f the achievem ents o f past and curren t sea ice-o cean m odels, and their contribution to furthering our understanding o f the interactions in this com plex system. It is not intended to be exhaustive - for m ore com prehensive review s o f ice-ocean m odelling in the Arctic, the reader is referred to B arry e t al.

[1993] diWdMellor and H akkinen [1994].

W e have already described the im portance o f the A rctic sea ice-ocean system in influencing clim ate, through its control of the global therm ohaline circulation and su rface h eat b alance. The ultim ate aim o f d ev elo p in g m odels d escrib in g the exchanges o f heat and salt in the Arctic is therefore to enhance our understanding o f the global clim ate and clim ate change, through the coupling o f the A rctic system to m ore com plex global clim ate simulations.

The A rctic O cean is a difficult basin to model, due to its com plex bathym etry, and the high resolution required to sufficiently represent energetic baroclinie eddies, w hich are on the scale o f the local radius of deform ation o f 5-10 km [M aslowski e t a l ,

1998]. T he current m odels are approaching these resolutions, and once reached,

should significantly im prove the realism of their p redictions [JSC Study G roup on

A C SYS, 1992]. Zhang e t a l [1998] also recom mend an eddy-resolving m odel in order to adequately represent open w ater regions, w hich have a significant influence on the ocean to atm osphere heat fluxes.

(b)

\5-

Figure 1.11 Contours of mean ice draft (in metres) for (a) summer, and (b) winter

(from Bourke and McLaren [1992]).

S em tn er [1976] perform ed the first ocean m odelling study o f the A rctic Ocean. The m odel was based on the Bryan-Cox ocean m odel w ith prescribed ice properties, and

successfully sim ulated the general circulation pattern. M a yku t a n d U ntersteiner

[1971] developed a one-dim ensional therm odynam ic ice m odel and applied it to the central Arctic. Such models were found to be highly sensitive to the param eterisation o f surface albedo and the choice o f ice rheology, w hich describes the relationship betw een ice stress and strain rate.

H ib ler [1979] extended the m odelling horizontally into tw o-dim ensions on a 125 km grid, and included dynam ic as well as therm odynam ic processes. The patterns o f ice thickness produced by this m odel were realistic, w ith the thickest ice predicted to be lo c a te d no rth o f G reenland and the C anad ian A rch ip elag o . P a r k in so n a n d W a sh in g to n [1979] produced a sim ilar m odel to sim ulate the annual cycle o f ice

ex tent and thickness in both the A rctic and A ntarctic. M ello r and K antha [1989]

co u p led a one-d im ension al, turbulence closure m odel to an ice m o del, w hich explicitly param eterised leads, recognising their im portant role in heat exchange.

S e m tn e r [1987] developed a three-dim ensional coupled ice-o cean m odel w hich dem onstrated the im portance o f oceanic heat fluxes in determ ining properties o f the

ice cover. H akkinen and M ellor [1992] produced a three-dim ensional coupled ice-

ocean m odel w hich included the m ixed layer physics develo ped by M ello r and

K antha [1989]. This model was used to test the hypothesis o f A agaard and Carmack

[1989] that the G reat Salinity A nomaly discussed in section 1.2.1 was the result o f an increase in ice export out of the Arctic Ocean. The m odel results o f H akkinen [1993] do indeed suggest that the Arctic was the source o f the anomaly.

M ore recent m odelling efforts have included the effects o f ice deform atio n and ridging on ice thickness redistribution [Flato and H ibler, 1995; Steiner e t a l , 1998],

and studies o f the inter-annual variability of ice export through F ram Strait [Harder et

a l , 1998]. Webb et a l [1998] have produced a high resolution global ocean m odel w hich includes the Arctic, but w hich as yet has no coupling to sea ice. Fully coupled high resolution ice-ocean m odels are now being run, w ith resolutions approaching the

local radius o f deform ation. For exam ple, the m odel described by M aslow ski et a l

[1998] and Z hang et a l [1998] has a resolution o f 18 km , and has been used to study the interaction o f m esoscale current systems on sea ice.

1.5.2 Global climate models and their representation of the Arctic

C urrent clim ate m odels portray the A rctic as a region o f w idespread influence and high sensitivity to perturbations in clim ate [Battisti et a l , 1997]. These m odels are particularly sensitive to the ice albedo feedback m echanism discussed in section 1.2.3

[W ashington and M eehl, 1996] and are know n to underestim ate natural variability in the A rctic [Battisti e t a l , 1997].

T he m ajority o f global clim ate models represent sea ice in a very sim ple w ay, often as slabs o f u niform thickness on very coarse grids. The use o f clim ato lo g ical ice param eters, such as thickness and extent, and the requirem ent to include substantial heat and salinity flux corrections in order to m ake the m odels represent present day

conditions, result in a generally poor representation o f the Arctic. F or exam ple. Smith

et a l [1997] revealed the shortcom ings in the A rctic sea ice extent predictions o f the U .K . H adley C entre G C M by perform ing com parisons w ith data from the SSM /I instrum ent. It is w idely agreed that m ore realistic sea ice and ocean m odels are required, w ith an im proved param eterisation o f sea ice, in order to con strain the response o f sea ice to perturbations in clim ate [Rind et a l , 1995; Curry et a l , 1995]).

JS C Study G roup on A C SYS [1992] suggest that incorporating regional sea ice m odels into global clim ate models will have a large im pact on m odel skill and our ability to predict future clim ate change. It is also suggested that uncertainties in sea ice m odels can be reduced by initialising them w ith realistic ice thickness data, and by using accurate forcing fields, including w inds and ocean currents. C om parisons o f m odel predictions o f ice thickness, extent and velocity with observations are also highlighted as an im portant step in reducing uncertainties.

T he im po rtance o f developing ocean circulation m odels capable o f realistically reproducing the circulation o f the A rctic O cean, and w hich can be coupled to global

clim ate m odels, has also been recognised [JSC Study G roup on A C SY S, 1992]. An

ex p an sio n o f the ex istin g A rctic O cean data base fo r pro viding v aiid atio n and assessm ent o f these regional models is a necessary first step before using the results to

understand various aspects o f circulation [Stammer e t a l , 1996].