ACDC2010: Ice Sheet - Ocean Interactions 8-19 June 2010, MIT FABLAB, Lyngen, NORWAY Advanced Lecture: Oceanographic regime of the West Antarctic Ice Shelves Adrian Jenkins British Antarctic Survey, Natural Environment Research Council, Cambridge, U.K. i. Forcing on continental shelf water temperatures ii. Ice-ocean interaction in the Ross and Weddell seas iii. Ice-ocean interaction in the Amundsen and Bellingshausen seas iv. Insights from observation in the cavity beneath Pine Island Glacier 1
(i) Forcing on continental shelf water temperatures Over the Southern Ocean the atmospheric pressure field and prevailing winds are undisturbed by land. Winds act directly on the surface water and drive a transport to the left. Mean sea level pressure (JJA) Mean sea level pressure (DJF) Westerlies to the north of the low pressure trough and easterlies to the south drive surface divergence an upwelling of deeper waters. (Source: http://en.wikipedia.org/wiki/atmospheric_pressure) Upwelling CDW is the source for all other Southern Ocean water masses. It also supplies nutrients to the surface waters where they can sustain an abundance of life. Shelf water properties are determined by the supply of CDW and the degree to which it is modified. The degree of modification depends on the efficiency of the sea ice export from the shelf. Speer et al., 2001 2
In the Ross and Weddell seas katabatic winds are funnelled in by the ice sheet geometry and steered north by major topographic barriers. The winds have high directional constancy and force continuous northward transport of sea ice off the continental shelf. Katabatic forcing and the resultant northward ice transport are much less effective in the Amudsen and Bellingshausen seas. [van Lipzig et al., 2004] Deep waters access the continental shelf via troughs carved by previous glacial advances. Total transports are not well known. In the Ross and Weddell seas the inflowing waters are cooled by mixing at the shelf edge. 3
In the Ross and Weddell seas virtually all the ocean heat brought onto the shelf is lost to the atmosphere. The water is cooled to the surface freezing point (approx. -1.9 C) and subsequent growth of sea ice raises the salinity. [Jacobs et al., JGR, 1985] Melting occurs beneath the ice shelves only because pressure lowers the freezing point. The product is Ice Shelf Water (ISW, T< -1.9 C). Dense waters flowing off the shelf form Antarctic Bottom Water. On the continental shelves of the Amundsen and Bellingshausen seas only the surface layer (top ~300 m) is close to the freezing point. [Hellmer et al., ARS, 1998] CDW is present beneath this layer, with temperatures > +1 C. Almost unmodified CDW survives because either more (as well as warmer) CDW comes on shelf, or less sea ice forms, or both. Ice shelf melt rates are much higher and no ISW or AABW forms. 4
West Antarctica East Antarctica 1992-2002 [Zwally et al., 2005] Ice shelves of the Amundsen/Bellingshausen seas are much smaller than those of the Ross/Weddell seas because of the high melt rates. They are also thinning rapidly, particularly in the Amundsen Sea. The Ross and Weddell sectors appear to be much less prone to change suggesting less variable as well as lower melting. (ii) Ice-ocean interaction in the Ross and Weddell seas First we consider the Filchner-Ronne Ice Shelf (FRIS). FRIS lies in the southern Weddell Sea, where the continental shelf is dominated by water at the surface freezing point, formed in winter. The patterns of melting and freezing are complex, but the basic pattern of the ice pump is still visible. The modifications arise because of the complex ice shelf and seabed geometry. [Joughin and Padman, GRL, 2003]. 5
Melting drives the mixed layer along the ice shelf base. Freezing is most intense where western coasts guide meltwater outflows. Melting is also strong where the main inflows enter the cavity. The strongest depth-mean flows follow routes of constant f/h. These routes guide the warm (-1.9 C) inflow to the deepest parts of the cavity. The seasonal production of dense shelf waters imposes an annual cycle on circulation and melting throughout most of the cavity. 6
The model reproduces the observed pattern of melting and freezing. However, the melting and freezing rates are too low unless tidal forcing is included. Melt rate (m/yr) without tides, mean = 11 cm/yr Tides generate only weak time-averaged currents, but if the buoyancy forcing is weak, tidal currents can dominate the instantaneous flow. Tides then determine the turbulent transport of heat to the ice shelf base. Melt rate (m/yr) with tides, mean = 22 cm/yr Depth (m) Latitude Meltwater fraction along 37.5ºW (Weddell Sea) Most meltwater exits the cavity at depth in the form of ISW. ISW spills off the continental shelf to contribute to AABW formation. Analogous processes take place in the Ross Sea. Depth (m) Latitude Meltwater fraction along 174ºW (Ross Sea) Melting and production of ISW seem to be slightly lower in the Ross Sea, probably because the ice shelf is thinner on average. 7
The ice shelves of the Ross and Weddell seas should be relatively insensitive to climate change. While enough sea ice is produced to maintain the shelf waters, the ocean temperature will be fixed at the surface freezing point. Also, where tidal currents are the dominant source of turbulent mixing, the response of the melt rate to moderate temperature changes will be linear. Ross/Weddell llsea (iii) Ice-ocean interaction in the Amundsen and Bellingshausen seas Next we consider the ice shelves of the eastern Amundsen Sea, where the continental shelf is flooded with CDW. Melt rates are much higher, averaging around 20 m/yr on Pine Island Glacier. Glaciological observations suggest that melting has been increasing. 8
The system of troughs guides the CDW towards the ice shelves. We would expect the CDW temperature to be unaffected by recent atmospheric temperature changes. But the inflow could vary in strength. There are few observations of variability. But a model suggests a seasonal input of CDW to Pine Island Bay (PIB) that varies in strength from year to year. 9
CDW is represented by layers 7 and 8. The thickness of these layers in PIB correlates with their thickness at the shelf edge. The model puts less CDW in PIB in the late 1980 s and early 1990 s. At that time the top of layer 7 was deeper in the water column at the continental shelf edge, allowing only a thin layer to flow onto the shelf. There were only minor changes in the CDW temperature, but major changes in the heat content of the water column. This suggests seasonal, interannual and decadal scale variability in the ocean forcing. There is little hint of a trend. 10
The source of the modelled variability is the wind forcing. On average there is a seasonal cycle driven by a combination of the motion of the Amundsen Sea Low and the Semi-Annual Oscillation. In winter and spring the pressure trough is further south on average. Westerly winds over the continental slope seem to drive the inflows of CDW. The seasonal cycle varies from year to year. In 1988, the westerlies were well to the north and there was a low input of CDW. Contrast this with 1994, when strong westerlies over the shelf edge gave a large input. There is decadal variability in the atmospheric circulation in this region, associated with ENSO and SAM. 11
[Turner et al., GRL, 2009]. Mean trend in 50 hpa geopotential height, 1979-2001, IPCC-AR4 models. Difference in 50 hpa geopotential height, pre-industrial minus year 2000 stratospheric ozone. Models suggest a deepening of the Amundsen Sea Low associated with a more positive SAM, caused by the loss of stratospheric ozone. However, it is not only the depth of the low, but also its location and variability that seem to be important in driving on-shelf flow of CDW. CDW properties should be relatively insensitive to recent climate change, since the water is drawn up from the deep ocean and insulated from the atmosphere by a cold surface layer. But delivery of CDW to the shelf seems to be influenced by regional wind forcing at the shelf edge and this is critical in the Amundsen Sea. In the Bellingshausen Sea cooling of the inflow, rather than its strength, may be the critical factor determining ice shelf melting. In both cases the link with the SAM is complex. Amundsen/Bellingshausen Sea 12
(iv) Insights from observation in the cavity beneath Pine Island Glacier All the glaciers draining into Pine Island Bay have accelerated, but behaviour differs in detail. The fastest part of Thwaites Glacier has shown little change while Pine Island has accelerated steadily. On Pine Island Glacier, acceleration has recently grown to about 7% per year. [Rignot, 2008] To understand how the ice-ocean interaction might cause this behaviour we need observations from beneath the ice shelves. Autosub3 is an Autonomous Underwater Vehicle (AUV) designed Autosub3 is an Autonomous Underwater Vehicle (AUV) designed and built in Southampton. It is 7 m long, 0.9 m in diameter and is powered by 5000 D cells. It navigates by dead-reckoning using heading and speed over the ground measured by ADCPs. It carries a CTD system and multi-beam echosounder. 13
It is launched and recovered using a purpose-built gantry. While on the surface a radio link allows mission programmes to be uploaded and data downloaded. Six sub-ice missions were completed, totalling about 500 km in about 100 hours. The longer runs went as far into the cavity as possible given a requirement for a minimum of 200 m of water. MODIS 28 Dec 2008 14
Missions were planned to track the (assumed flat) seabed on the way in, then track the ice shelf base on the way out. On longer runs the outer part of the cavity would be profiled. The seabed was far from flat, rising over 300 m to a ridge crest. The ridge presents a significant barrier to the inflowing CDW. As the ice shelf has thinned the gap has widened. 15
We can use the water properties to derive the fraction of meltwater. The main outflow is in the south. Some exits the cavity in a layer at 400 m depth, the rest upwells at the ice front. The main outflow carries a signature of high turbidity. This must be caused by sediment from the glacier bed. (1) (3) Multi-beam acoustic imagery of the ice shelf base shows crevassing. In between the smooth areas have a stepped structure. (5) 16
Multi-beam imagery of the seabed can tell us about the past configuration of the glacier. The outer cavity has rough topography, suggesting little sediment cover. On the ridge slope sediment flows from the crest have smoothed the topography. On the crest lineations parallel to ice flow indicate former grounded ice flow. When was the glacier last grounded on the ridge? Early satellite imagery shows a bump in the ice surface about 5 km upstream of the highest mapped point on the ridge. This could have been the last contact the glacier had with the ridge. The implication is that the current retreat is part of a longerterm change that was already underway in 1973. 17
The ridge is the most significant rise in the bed for 200 km. The glacier may have started a phase of unstable retreat when it lost contact with the ridge. Summary The temperature of the shelf waters is determined by the inflow of deep waters and the local loss of heat to the atmosphere. On the cold water shelves, local heat loss dominates and the shelf water temperature is fixed at the surface freezing point. On the warm water shelves the deep temperature is set by the relatively stable properties of the inflowing CDW. The ice shelves are more susceptible to changes in wind forcing affecting either the export of sea ice from, or the delivery of deep water to, the shelf than they are to changes in air temperature. Input of CDW to the Amundsen Sea shelf is subject to variability associated with changes in the Amundsen Sea Low. Despite variable forcing, changes on Pine Island Glacier appear to have been monotonic, probably because of the bed geometry. Downslope retreat from the ridge crest may have been selfsustained, while warming of the inner cavity as the gap over the ridge widened would have driven a rise in melting. 18