Australia: The Land Where Time Began

A biography of the Australian continent 

Pine Island Glacier Ice Shelf Melt Distributed at Kilometre Scale

Ice streams in West Antarctica are contributing about 10 % of the observed global sea level rise by thinning and accelerating. The Pine Island Glacier has been thinning since at least 1992 and much of this ice loss has been driven by ocean heat transport changes beneath the Pine Island Glacier ice shelf and retreat of the grounding line. However, details of the processes driving this change have remained largely elusive, which has hampered the ability to predict what the behaviour of this and similar systems will be in the future. In this paper Dutrieux et al. have developed a Lagrangian methodology to measure oceanic melting of ice that is advecting so rapidly. Airborne and high resolution satellite observations of the velocity and elevation of the ice surface are used to quantify patterns of basal melt beneath the Pine Island Glacier ice shelf and the associated adjustments to ice flow. At the broad scale, near the grounding line melt rates of up to 100 m per year occur, reducing to 30 m per year 20 km downstream. Basal melting was largely compensated by ice advection between 2008 and 2011, which allowed Dutrieux et al. to estimate an average ice loss to the ocean of 87 km³ per year, which is in close agreement with estimates that were oceanographically constrained in 2009. At smaller scales, a network of basal channels that are typically 500 m to 3 km wide has been sculpted by concentrated melt; with anomalies at kilometre scale that reach 50 % of the broad-scale basal melt. The channels are enlarged close to the grounding line by basal melting; though melting tends to diminish them further downstream. A key component of the complex ice-ocean interaction beneath the ice shelf is kilometre-scale variations in melt, which implies that greater understanding of their effect, or very high resolution models, are required to predict the sea level contribution of the region.

Thinning of ice shelves (Pritchard et al., 2012; Shepherd et al., 2010) and the corresponding restraint decrease on the inland ice flow (Flament & Rémy, 2012; Joughin et al., 2010; Payne et al., 2004; Pritchard et al., 2009; Zwally & Giovinetto, 2011) are recognised to be major drivers of the current loss of ice in Antarctica. In West Antarctica the change of the ice shelf is particularly pronounced, where the grounded part of the Pine Island Glacier (PIG)  has thinned, accelerated and retreated over recent decades (Rignot,2008; Shepherd et al., 2001), in response to ocean heat transport that has increased beneath its floating ice shelf and the resulting feedbacks (Jacobs et al., 2011). The detailed patterns and rates of basal melt on specific ice shelves are known only on relatively coarse scales (Payne et al., 2007), though there have been some efforts to relate basal melt to ocean temperature and the ice shelf geometry on the broad scale (Holland et al., 2008). Without a thorough understanding of the processes controlling the dominant scales of ice shelf melt, projections of change in the future of the Pine Island Glacier and similar glaciers will be dependent on melting parameterisations that are poorly constrained by observations (Joughin et al., 2010; Katz & Worster, 2010),

The unsteady nature of the Pine Island Glacier is a difficulty encountered when studying it. The ice geometry of the Pine Island Glacier is rapidly evolving (years), and the ice is being advected at speeds of more than 4  km per year while the underlying ocean is expected to respond at sub-annual scales to both local and remote forcing (Steig et al., 2012; Thoma et al., 2008). There are also many reasons to expect the spatial pattern of melt to be complex. There is a series of both longitudinal and transverse channels under the floating tongue of the Pine Island Glacier (Bindschadler et al., 2011), and basal melt was found to be concentrated strongly along subglacial longitudinal channels (elongated in the ice flow direction) on a similar ice shelf (Rignot & Steffen, 2008). Basal crevasses beneath the Pine Island Glacier shelf are located above the apex of each channel. Dutrieux et al. suggest the formation of such crevasses may be in response to basal melting, which suggests that changes in channel-scale ice-ocean dynamics could affect the structural integrity of such ice shelves indirectly (Vaughan et al., 2012). Conversely, a recent study that used a complex ice-ocean model suggested that the presence of melt channels allow ice shelves to survive sub-ice ocean temperatures that are higher than they otherwise would (Gladish et al., 2012). The development of the ability to measure the spatial patterns of melt rate beneath the Pine Island Glacier is an important step towards improving understanding of these processes.

Conclusions

High melt rates near the grounding line of the Pine Island Glacier have been indicated by previous work (Payne et al., 2007; Rignot & Jacobs, 2002) and the presence in the ice of basal channels (Bindschadler et al., 2011: Mankoff et al., 21012; Vaughan et al., 2012). According to Dutrieux et sl. the pattern of melting on the Pine Island Glacier ice shelf was shown by their observations to be highly complex. The melt rate within 10 km of the grounding line is at least 100 m per year, reducing to 30 m per year 20 km downstream. Basal melting was largely compensated by ice advection between 2008 and 2011, which allowed the average ice loss to the ocean to be estimated, 87 km³ per year, which agrees closely with 2009 oceanographically constrained estimates. The melting is concentrated in the basal channels close to the grounding line and carves out those channels at 80 m per year. Further downstream, on the central part of the ice shelf that is dominated by longitudinal structures that are kilometres long, melting on the keels is 30 m per year faster than in the channels, which Dutrieux et al. say explains the gradual loss of channels in the downstream part of the ice shelf and the inversion of the surface elevation anomalies relative to free floatation. The method used by Dutrieux et al. does not give significant results for transverse or smaller, less than 1 km, structures, though over such smaller scales large spatial variations in the melting are likely to occur as well (Stanton et al., 2013).

A possible explanation for the gradual regime shift in channel melt could be the initial formation of buoyant meltwater plumes near the grounding line that rise up the ice base and entraining heat to the channel crests most efficiently, and a decreasing heat entrainment efficiency downstream as the slope weakens, the ice base shallows and the source of warm water gets further away. In this scenario, at some stage the plumes within the channels deliver less heat to the ice shelf than the deeper waters, that are warmer, that bathe the channel keels. But more complex scenarios are not excluded.

Dutrieux et al. suggest that with the advent of ice surface DEMs of even higher resolution  (few metres) taken at regular time intervals, it can be expected that the methodology that was developed for this study will reveal details that have not been expected about changes in the distribution of surface elevation, and by inference basal melt, where the underlying assumptions are valid, which would increase understanding of the interaction dynamics of the atmosphere-ice-ocean and the temporal and spatial variability.

Melt rates that are 80 % higher in channels than on neighbouring keels are indicted by the observations of Dutrieux et al. of the area close to the grounding line, and point to high spatial variability in the melt rates across the ice shelf, which indicates strong modulation of ice-ocean interactions at kilometre scales. It is implied by this that in situ observations need to be interpreted within their contextual position relative to the channels. Dutrieux et al. suggest that what is possibly the most important implication of this work concerns the modelling of sub-ice cavities in the shelf. It is challenging to represent accurately sub-kilometre scales using conventional ocean models, even in the case of dedicated regional studies, and will remain impossible for global coupled models for some time. An approach to solving this problem is the use of unstructured computational meshes to focus the resolution of the model on features of interest, such as these channels (Kimura et al., 2013; Timmermann et al., 2012). Parameterising their effect on larger scales that models are able to resolve would be a more conventional alternative. An essential pre-requisite for either of these approaches to be successful is a detailed observational understanding of the channels, and this study provides a significant advance.

Sources & Further reading

1. Dutrieux, P., D. G. Vaughan, H. F. J. Corr, A. Jenkins, P. R. Holland, I. Joughin and A. H. Fleming (2013). "Pine Island glacier ice shelf melt distributed at kilometre scales." The Cryosphere 7(5): 1543-1555.