There have been few modelling studies with spatial mesh sufficiently fine to represent baroclinicity adequately within the narrow channels of the Canadian Archipelago. The coupled ice-ocean model of the US Navy Postgraduate School, which has 1/12th degree resolution (about 9 km), has been used for pan-Arctic simulations of the period 1979-2002 (Williams et al. 2004). The results indicate that the Canadian Arctic through-flow is the greater contributor (relative to Denmark Strait) of oceanic fresh water to the North Atlantic. According to the simulation, the fresh-water flux through the Archipelago has increased over the period studied, a trend that has perhaps contributed to decreasing salinity in the Labrador Sea.
The goal of future work is a prognostic model with time evolving fields of temperature and salinity. This is not a trivial undertaking, particularly on a terrain-following mesh; methods with acceptable truncation error have been sought for many years. Such capability is essential for realistic simulation of baroclinic effects, including fresh-water and heat fluxes. An increase in resolution is also desirable, best accomplished for this area using the finite element method. The present best resolution is 1.1 km, barely adequate to represent important channels such a Hell Gate (4 km), Cardigan Strait (8 km) and Fury and Hecla Strait (1.8 km). Ocean circulation models need to be forced using wind fields that adequately reflect the important influence of topography and boundary-layer stratification on the mesoscale. Lastly, there is need for a realistic and fully interactive ice dynamics model; not only is pack ice an important element of ocean dynamics, but moving ice is itself a component of the fresh-water flux. The ice element may become a more important fraction fresh-water flux in a warmer climate, when ice of the Canadian Archipelago may be mobile longer each year (Melling 2002).
Moving pack ice transports a fresh-water flux disproportionate to its thickness, by virtue of its low salinity (less than one tenth that of seawater) and of its position at the ice-atmosphere interface where it moves readily in response to wind. Both ice thickness and drift velocity are needed to calculate the sea-ice fresh-water flux. At present, ongoing observations of ice thickness are not available for any of the gateways discussed in this chapter. Here we concentrate on using satellite-based sensors to measure the movement of pack ice through the Canadian Archipelago. With supplementary guesses of pack-ice thickness, approximate values for the accompanying fresh-water flux can be provided.
The geography of the Canadian Archipelago is too complex for effective use of satellite-tracked drifters to measure the through-flow of pack ice. Methods based on the tracking of features in sequential images from satellite-borne sensors are better suited to the task. Microwave sensors provide the least interrupted time series of ice flux at key locations because they are relatively unaffected by cloud and wintertime darkness. However, the tracking of ice movement may be error-prone at times when ice features have poor contrast or when the pack is deforming appreciably as it moves; the latter is a common circumstance during rapid drift through narrow channels.
The displacement of sea ice over the interval between two images is derived by the method of maximum cross correlation (Agnew et al. 1997; Kwok et al. 1998). The technique works with sub-regions or patches on the two images that are 5-50 pixels on a side, depending on resolution. The underlying premise is that difference between consecutive images is the result of displacement only, the same for all features. Any additional rotation and straining of the ice field or creation of new ice features (e.g. leads) degrade the correlation.
Two long-term studies of ice movement through the Canadian Arctic have been completed. One used scenes acquired by synthetic-aperture radar at 0.2-km resolution (Radarsat: Kwok 2005; Kwok 2006) and the other utilized images from a passive microwave scanner, which resolves ice features at approximately 6-km resolution (89 GHz AMSR-E: Agnew et al. 2006). Both approaches yield estimates of ice displacement and ice concentration at intervals of 1-3 days, constrained by the interval between repeated orbital sub-tracks.
The utility of AMSR-E is marginal in some parts of the Archipelago where channels are only a few pixels wide. Moreover, the 89-GHz channel is of little value during the thaw season (July-August) when the wet surface of the ice and high atmospheric moisture degrade image contrast; data acquired during the shoulder-months of June and September may also be poor at times. Microwave radar produces images of better contrast than microwave scanners during the thaw season, but the identification of floes and ice features from Radarsat can still be challenging during summer.
The flux estimates derived from microwave-emission images only incorporate ice motion that occurred during the cold months (October-May or SeptemberJune). Since this period overlaps significantly with fast-ice conditions within the Canadian Archipelago, the months of most active ice movement may have been missed. The flux estimates derived from Radarsat nominally span the entire year. However, it is noted that feature-tracking algorithms return a null result (low correlation) when the quality of images is poor or ice-field deformation is large; this fact may contribute a low bias to average displacement during the summer, when image contrast is poor and low ice concentration permits rapid movement and deformation of the pack.
Radarsat transmits microwaves and detects the energy back-scattered from the rough surface or upper few centimetres of the ice; it is not sensitive to ice thickness. AMSR-E detects natural microwave emission at several frequencies and polarizations, which can be manipulated to yield information on ice type and concentration. In general, satellite-based data on ice movement must be augmented by ice-thickness values from other sources if the flux of ice volume and fresh-water are to be estimated.
Kwok et al. (1999) calculated an area budget for Arctic multi-year ice during 1996-1997 using observations made from space by microwave scatterometer (NSCAT). They estimated an annual outflow from Nares Strait of 34 x 103 km2 by mapping multi-year ice in northern Baffin Bay, presumed to have arrived here via Smith Sound. Subsequently, Kwok (2005) has used Radarsat images over a 6-year period (1996-2002) to measure directly the drift of ice through a 30-km wide gate at the northern end of Robeson Channel (Fig. 9.20). During these years, the average annual flux of ice from the Lincoln Sea into Nares Strait was 33 x 103 km2, with an inter-annual span of ±50%. There was a strong annual cycle in ice drift, with the bulk of the transport during August through January; ice is typically fast in Nares Strait between mid winter and late July. The average volume flux of an assumed 4-m thickness of ice would have been 130 km3/year (4 mSv).
For the years 1997-1998 to 2001-2002, Kwok (2006) has estimated ice-area transport across the main entrances to Canadian Archipelago from the west (Fig. 9.20): Amundsen Gulf, M'Clure Strait, Ballantyne Strait plus Wilkins Strait plus Prince Gustaf Sea (cf. Queen Elizabeth Islands south) and Peary Channel plus Sverdrup Channel (Queen Elizabeth Islands north). His results are summarized in Table 9.4. On average during the 5-year study, Amundsen Gulf was a source of ice for the
Was this article helpful?