Resultados del aula virtual

Two disparate mechanisms of frazil ice transformation into pancakes were identified: the classical accretion process by which the frazil first begins to agglomerate; and the

Chapter 5: Discussion

significantly faster subsequent frazil over-topping which occurs once platforms exist to support it. A ‘family’ of pancake types was identified which was consistent with the two processes, including a composite congelation-frazil structure (Type-D). It is suggested that these were examined during earlier studies of the area (Lange et al. 1989; Lange and Eicken 1991) but were not recognised at the time.

Many uncertainties have emerged during the course of this study, centred on the modification of ocean-atmosphere heat fluxes by the evolving pancake-frazil ice cover. The treatment of Chapter 2 makes no assumptions about any reduction in heat flux, since the frazil production modelled there simply represents a maximum rate for comparison with the observed pancake and frazil thicknesses. The redistribution model presented in Chapter 4 does assume that (a) only the interstitial frazil area contributes to ice production; and (b) the rate is not modified from the ‘free surface’ figure. These two assumptions have opposite effects on the ice production. Chapter 4’s model produces a frazil slick which is far thicker than the limited field observations would suggest, though the overall volume (and hence equivalent thickness) of the ice cover is in close agreement with the figure suggested by field measurements. It is left to the ‘scavenging efficiency’ tuning factor to reduce the frazil thickness to values which are perceived to be less extreme, shifting the volume to be dominantly pancake ice as observed.

Little literature exists to aid our understanding of these effects. Most measurements and theory development have been focussed on frazil formed at low area concentrations in relatively small leads or polynyas, which is subsequently herded downwind until it collects against the edge of the lead or polynya and freezes into a solid ice sheet. Important considerations in this type of modelling are wind speed and fetch (lead width), which indirectly parameterise the turbulence which both limits frazil production and mixes the frazil crystals down into the water column, determining their volume concentration.

These parameters have little relevance to the vast frazil/pancake fields of the Antarctic: turbulence levels there are largely determined by the high amplitude swell impinging on the ice cover from the Southern Ocean. Turbulence is thus determined by non-local

wind forcing (distant storms) and the dimensions of the ice cover (thickness, area concentration, distance) between the measurement site and the open ocean, plus the properties of the waves themselves (amplitude, period). Pedersen and Coon (2004) nonetheless presented a non-physical best-fit of Alam and Curry’s (1998) wind speed to ice thickness relation for the Odden. This is partly justified since the relatively small- scale of the Odden implies a closer proximity to storm wind forcing and a far less significant degree of damping of the resulting ocean waves by the ice cover. They found that an empirical ‘lead width’ of 1.5 km fitted the observed pancake thickness best, though the scatter was considerable.

Additional disconnections between the lead-derived and MIZ frazil treatments are the lack of a ‘dead zone’ in the MIZ, where the frazil damps the short period waves and consolidates, other than at the limit of the frazil-pancake zone itself. Langmuir

circulation appears to play little or no role in the frazil-pancake cover, which invariably forms a homogenous mixture and is not observed to organise into downwind rows, except in very low areal concentrations at the ice edge.

Conceptually, it seems unlikely that only the interstitial frazil area contributes to ice production. Indeed, it is difficult to see how 35% of the productive area can produce ice at 58% of the overall free-surface rate – as demonstrated by modelled top layer growth in Chapter 3 - with this assumption intact. The pancakes are small and highly mobile and it is therefore not unreasonable to assume that the entire area “sees” the cold

atmosphere at the integrated timescales over which heat loss from the ocean occurs. The situation is analogous to the heat exchange over leads, where the area-integrated heat flux is initially very sensitive to lead width and spacing, but becomes markedly less sensitive as ice concentrations drop below about 70% (Worby and Allison 1991). Worby and Allison point out that the air over the additional open water area has “already been modified” by the open water immediately upwind. Additionally, the pancakes are porous, unlike congelation ice, and the water within them is therefore less insulated from the cold air than would be the case under a congelation ice sheet.

Chapter 5: Discussion

An opposite effect is the modification of ocean-atmosphere heat fluxes by the presence of a frazil/pancake ice cover. Turbulent fluxes dominate heat exchange during frazil formation and the various methods of calculating these each use parameters that are modified by an ice cover. Bulk formulae use an exchange coefficient, while Monin- Obukov similarity theory requires a roughness length, which is certainly different from the open water value. Very thick frazil slicks, such as that observed in the Odden (Figure 4.2) will modify the area-integrated surface temperature from the freezing point of seawater, as will the presence of pancakes, the cooling of whose top surfaces will also account for some fraction of the heat flux which would otherwise contribute to ice formation (Leonard et al. 1998).

Given the supposed over-production of Chapter 4’s frazil model, the rate reduction effect may be expected to dominate. This expectation is enhanced by the tank

measurements of Smedsrud (2001), which determined a rate less than one third of the unmodified production with similar fluxes, though various reservations were expressed (Chapter 3) as to the equivalence of tank and field conditions, centring on the relative turbulence and mixing levels.