Abstract
This progress report is the first of three overviews of important fields of study within contemporary glaciology, illustrated with reference to exemplar research contributions and highlighting key developments in each field since 2005, the date of the last review of ‘Glaciology’ in this journal. The topics covered in this paper are: (1) the mass balance of glaciers and ice sheets and their contributions to global sea level change; and (2) atmospheric melt-induced influences on the dynamics of the Greenland Ice Sheet. The paper aims to provide an overview of key developments over the last six years from each topic, with discussion of the continuing debates and unresolved issues which constitute the foci for ongoing and future research.
Keywords
I Introduction
The United Nations’ Intergovernmental Panel for Climate Change (IPCC) presented glaciology with a major challenge in 2007 by specifically highlighting, in their Fourth Assessment Report (IPCC, 2007), the major uncertainties in assessments of glacier and ice sheet surface mass balance and the influence of ice dynamics on mass changes. The uncertainty in measurements of ice sheet mass balance was effectively deemed too great and the time series too short-lived or spatially inconsistent to be able to discern significant trends. In addition, the uncertainties surrounding glaciological processes which govern the response of ice sheets to climate change precluded confidence in the ability of prognostic ice sheet models to deliver plausible estimates of their contribution to global sea level by the end of the 21st century. A huge global effort has continued over the last decade aimed at reducing these measurement and process uncertainties.
This progress report covers two topics of glaciological research: (1) measuring the mass balance of glaciers and ice sheets, and (2) atmospheric influences on the dynamics of the Greenland Ice Sheet (GrIS). The first is most directly related to efforts to reduce the uncertainty in our measurement of contemporary rates of change in the mass of glaciers and ice sheets throughout the world. The second topic is an example of how diverse glaciological research has focused around a topic which has received widespread public exposure and sparked considerable academic debate since the potential relevance of melt-induced dynamic thinning to the mass balance of the GrIS was first proposed a decade ago. These two topics of research will be discussed in separate sections, although partitioning the literature in this way is a somewhat arbitrary task. Future papers in this series will focus on topics which are obviously closely related, for example ice-ocean interactions. The paper does not attempt to provide an extensive review of research undertaken in each of these subjects as this would require a series of major review papers. Instead the rationale, methods and examples of key developments over the last six years from each topic are covered. Continuing debates over unresolved issues and uncertainties are identified which constitute the foci for ongoing and future inquiry.
II Mass balance
Glaciology is ultimately concerned with understanding the growth and decay of glaciers and ice sheets and the relationship between climate and glacier change. Thus the change in mass of a glacier or ice sheet is perhaps its most important characteristic and one of the most fundamental topics of glaciological study. There are three main methods for measuring the mass balance of a glacier or ice sheet: the mass budget method, the volume change method, and satellite gravimetry. An explanation of the approach and recent developments of these three methods follows. The final section of the mass balance overview assesses our current understanding of the contribution of glaciers and ice sheets from different regions of the world to global sea level change.
1 Mass budget method
The mass budget method requires all sources of mass input (accumulation) and output (ablation) to be estimated and summed to determine the overall mass change of the glacier. In the past efforts were focused on long-term field-based monitoring programs where (groups of) individuals would make repeated annual excursions to measure the accumulation and ablation along representative stake transects from which extrapolation and interpolation was used to determine the mass balance of the whole ice mass. By virtue of the vast human effort required and the often expensive logistics incurred, such efforts were focused at a relatively small number of case study valley glaciers and even more occasionally across ice caps. A classic example of this approach, which continues today, are the measurements of the Glaciology Section of the Geological Survey of Canada across the ice caps of the Canadian High Arctic (Koerner, 2005), most notably across Devon Ice Cap (Boon et al., 2010).
Technology has played a huge role in changing the type of research undertaken to determine mass balance across ice masses of all scales. Eisen et al. (2008) provided a comprehensive review of approaches to measuring accumulation across east Antarctica, many of which have been closely associated with the development and application of new technology over the last 20 years. Ground penetrating radar (GPR) measurements (see Woodward and Burke, 2007, for a review of GPR in glaciology) can be used to identify firn and ice layers seen in ice cores. These layers can be followed between core sites to help determine accumulation accurately (Arcone et al., 2005; Wadham et al., 2006). Long-term (multidecadal) measurements of accumulation can now be made in situ at multiple locations over a short space of time by employing new shallow ice coring/borehole technologies. Recent examples include: the use of Neutron probes to identify annual layers through measured density fluctuations (Hawley and Morris, 2006; Morris and Cooper, 2004); the use of borehole optical stratigraphy to identify annual stratigraphic layers (Hawley et al., 2008); and down borehole gamma spectrometry to identify known stratigraphic layers such as the 1963 radioactive ‘bomb’ layer (Mair et al., 2005).
Recent developments in remote sensing (RS) have improved the mass budget method by enabling the extrapolation of point field measurements across large spatial extents, based on some physical property that can be monitored from airborne or spaceborne platforms. An example of this is the analysis of airborne radar measurements of surface and near-surface reflection horizons in conjunction with ground-based snow pack depth measurements across transects of the GrIS to determine variability in snow accumulation (de la Peña et al., 2010; Helm et al., 2007). The potential exists to utilize satellite radar near surface reflections (for example from the European Space Agency’s (ESA) CryoSat2 satellite) to provide similar data. This could help reduce the uncertainties that exist across regions of the accumulation areas of the GrIS and Antarctic Ice Sheets that are not well constrained by ice core or snow-pit measurements (Bales et al., 2009). Remote sensing can provide insight to the extent and duration of surface melting which can be used to validate model estimates of surface ablation. Satellite-borne Ku- and C-band radar scatterometers can detect surface melt on the GrIS (e.g. Steffen et al., 2004). Data from QuikScat, including enhanced resolution data, have been used to: detect the dates of melt onset and freeze-up; determine annual melt extent and duration; and map snow and ice facies variability across ice caps of the Canadian High Arctic (Wang et al., 2005; Wolken et al., 2009), the GrIS (Nghiem et al., 2005; Wang et al., 2007) and the Eurasian Arctic ice caps (Rotschky et al., 2011; Sharp and Wang, 2009). Tedesco (2007) developed a technique for monitoring snowmelt over the GrIS based on multiple frequency, diurnal amplitude variations in brightness temperatures measured by the Special Sensor Microwave Imager (SSM/I).
Regional climate models can provide inputs to surface mass balance (SMB) calculations incorporating varying degrees of complexity. This has been particularly useful in providing estimates of SMB across the Earth’s two continental scale ice sheets in Greenland and Antarctica where the temporal and spatial coverage of field or RS measurements can be a major limiting factor. The calibration and validation of regional climate model output over polar regions is possible with close integration between model output and in situ automatic weather station measurements (Box et al., 2006; Burgess et al., 2010; Ettema et al., 2009; Van de Berg et al., 2006). Climate reanalysis data is also now widely used to provide input for ice sheet mass budget estimates (Gardner et al., 2011; Hanna et al., 2006, 2008; Monaghan et al., 2006); however, they must be used cautiously because of time-varying biases that can induce artificial trends (Screen and Simmonds, 2011).
For many larger ice masses the calving of tidewater glaciers directly into the ocean to form icebergs is a major source of mass loss. Until recently this term was a major uncertainty in mass budget calculations on account of a paucity of reliable estimates of the calving flux. Once again, remote sensing techniques have been vital in providing: (1) spatially extensive measurements of surface velocity approaching floating tidewater margins through interferometry and image correlation (Burgess et al., 2005; Joughin et al., 2010; Luckman et al., 2006; Rignot and Kanagaratnam, 2006; Rignot et al., 2008); (2) measurements of ice thickness across grounding lines (e.g. Dowdeswell et al., 2008), and (3) measurements of changing termini positions through time (e.g. Joughin et al., 2008b). All these factors have to be considered to be able to make an assessment of the mass flux lost at the tidewater margin (e.g. Mair et al., 2009; Williamson et al., 2008).
The advantage of the mass budget method for determining glacier and ice sheet mass balance is that some insight into the processes responsible for the mass change can be deduced by tracing the overall mass change to changes in accumulation, ablation or tidewater calving. A further advantage of the mass budget method has been the generation of long-term time series of annual mass balance in some regions of the world, for example in Scandinavia (Nesje et al., 2008). This has provided important data for the calibration and validation of mass balance models. The disadvantage in the method is that the overall mass balance can have large uncertainty associated with it since it is calculated from the residual of individual terms which may each have considerable error and uncertainty (Thomas et al., 2006).
2 Volume change method
Altimetry has been applied extensively to measure the mass balance of ice caps and ice sheets. Volume change can be determined from measurements of elevation change along spatially extensive transects, assuming negligible change in thickness due to mass loss at the base of the ice mass. Converting volume change to mass change requires knowledge of subsurface densification changes. In Antarctica, where surface melting is negligible, firn densification usually has to be estimated from dry firn compaction modelling (e.g. Arthern et al., 2010). Across the GrIS, where surface melt, percolation and subsurface refreezing are widespread, a refreezing term has to be incorporated (e.g. Li and Zwally, 2011). The volume change method is therefore largely limited by the uncertainty of densification models. Given that the percolation zone (Benson, 1962) covers as much as 40% of the GrIS each year, improved understanding of the physical processes controlling subsurface densification is a field ripe for further research. The overall impact of summer melting on densification rates has been studied through changes in snow pit and near surface firn core stratigraphies before and after summer melting on the GrIS (e.g. Parry et al., 2007) and on Devon Ice Cap (Bell et al., 2008). Continuous monitoring of the thermal profile of snow and firn through the lower regions of the percolation zone on the GrIS (Humphrey et al., 2012) has improved understanding of the relative importance of different refreezing processes in causing higher rates of firn densification.
The two main methods of determining surface elevation are by laser and radar altimetry from both airborne and space-borne platforms. Since 2005 there have been few attempts to determine ice sheet mass balance from satellite radar altimetry. Extracting reliable elevation data over undulating topography requires particularly refined radar backscatter processing (Roemer et al., 2007). Such topography is found near ice sheet margins where mass loss is often concentrated. In addition, temporally varying radar backscatter and penetration from the surface and near surface of snow and firn needs to be accounted for to provide reliable measurements of elevation change through time. The recent availability of ice sheet elevation data from ESA’s CryoSat2 satellite will be expected to address these limitations (Wingham et al., 2006). It benefits from a small along- and across-track resolution meaning it can cope with steeper gradients and undulating topography of ice caps and the margins of large ice sheets (Wingham et al., 2006). Importantly, radar backscatter signals from an airborne equivalent of the satellite altimeter have been calibrated and validated by an extensive international ‘ground-truthing’ effort over the last seven years (e.g. Brandt et al., 2008; Scott et al., 2006) which will significantly reduce the uncertainties associated with deriving the surface from complex time- and space-variant backscatter signals.
NASA’s ICESat satellite laser altimetry mission provided estimates of glacier and ice sheet elevation until October 2009. The spatial resolution of ICESat was severely compromised by a failure of one of its two lasers shortly after launch (Thomas et al., 2008). However, elevation change estimates have still been determined for regions of particular interest, such as the southeast coast catchments of Greenland where dramatic thinning rates have been observed in the last decade (Howat et al., 2008). Spatial interpolation between elevation data from proximal satellite tracks has allowed the successful examination of elevation change across huge swathes of Antarctica including most of the West Antarctic Ice Sheet (WAIS), and all the major drainage basins of the GrIS (Pritchard et al., 2009). The importance of the mission to understanding the response of the cryosphere to climate change ensured that NASA has committed to the launch of ICESat-2 in 2015 (Abdalati et al., 2010). ICESat-2 will carry a significantly improved laser system and be able to provide ice change measurements across more than a 15-year time span. It will provide observations with much greater spatial and temporal resolution, and accuracy, than has been achieved before.
Airborne laser altimetry has proved an effective method in helping to determine elevation changes across smaller ice caps (Moholdt et al., 2010; Raper et al., 2005) and large mountain glaciers and icefields (Rignot et al., 2005). Arendt et al. (2006) continued their measurements of long-term elevation change across Alaskan glaciers. They calculated the differences between laser measurements of elevation along glacier centre-lines, and elevations from US Geological Survey (USGS) maps from the 1950s to calculated elevation-dependent thickness changes. The method relies on extrapolating measurements of centre-line elevation change across glacier area at specific elevations to obtain glacier volume changes, and then upon extrapolation of measurements from a small sample of glaciers across a larger region to estimate mass change across a whole mountain range or icefield. Airborne laser altimetry can produce good estimates of mass change over large regions but has poor temporal coverage since mass changes are usually given for decadal time periods.
3 Satellite gravimetry
The Gravity Recovery and Climate Experiment (GRACE) satellite mission was launched in March 2002. GRACE measures the time-variable component of the Earth’s gravity field. The effects of redistributions of mass caused by glacial isostatic adjustment (GIA), Earth/ocean tides and variations in the hydrosphere and atmosphere have to be accounted for before the gravimetric measure of glacier mass change can be identified. Although each of these corrections involves modelling uncertainties (particularly for GIA across Antarctica, Riva et al., 2009), results from GRACE have provided compelling and completely independent evidence to support and strengthen the evidence showing widespread mass loss from the world’s ice sheets (GrIS – Luthcke et al., 2006; Van den Broeke et al., 2009; Antarctica – Chen et al., 2009; Velicogna and Wahr, 2006) and other heavily glaciated regions (Global estimates – Jacob et al., 2012; Patagonia – Chen et al., 2007; Alaska – Arendt et al., 2008; Luthcke et al., 2008; Canadian Arctic – Gardner et al., 2011). An advantage of GRACE is the high 10-day temporal resolution (Rowlands et al., 2005); however, with its low spatial resolution (2 arc degrees, which equates to approximately 49,000 km2 in Alaska), GRACE is not able to resolve mass changes of individual glaciers. The GRACE time series now covers the decade since 2002, but this is too short to make a strong link with recent interdecadal climate variability.
4 Glacier and ice sheet contributions to global sea level rise
It is beyond the scope of this progress report to provide a rigorous global and regional review of the global sea level rise (GSLR) contribution from loss of glacial ice cover, but some recent contributions towards determining the current state of terrestrial ice masses, globally and from selected regions are now summarized with examples of GSLR estimates quoted.
The combined contribution to GSLR of mass loss from the Earth’s mountain glaciers and ice caps (MG&IC), not including Greenland and Antarctic ice sheets, has been the subject of several studies over the last six years (e.g. Cogley, 2009; Kaser et al., 2006; Meier et al., 2007; Zemp et al., 2009). There are significant limitations attached to any assessment of the MG&IC contribution to GSLR. For example, glacier mass balance measurements are unevenly distributed and global mass change estimates are based on measurements from a tiny sample of all the world’s glaciers. Large areas remain without any measurements and the influence of iceberg calving and ice dynamics is still poorly understood. Ways of addressing this undersampling problem include modelling global glacier mass balance with the use of globally available climate gridded data (e.g. Hirabayashi et al., 2010; Hock et al., 2009) and using statistical upscaling methods (e.g. Bahr et al., 2009; Radić and Hock, 2011).
Despite the limitations in all these approaches, most of the published research has estimated that, over the last several decades, MG&IC have made a larger contribution to eustatic sea level rise than that from the GrIS and Antarctic Ice Sheets combined (IPCC, 2007; Meier et al., 2007). For example, Cogley (2009) estimated that MG&IC contributed 1.1–1.4 mm a−1 to GSRL from 2001 to 2005. However, this consensus has recently been challenged by a satellite gravimetry estimate of just 0.4 mm a−1 for the period 2003–2010 (Jacob et al., 2012). Since important glacierized regions of the world, such as Alaska and the high mountain ranges of Asia, display large interannual variability in mass change, and given the limitations in sampling described above, it is inevitable that comparisons of mass loss estimates will vary when determined over short time periods (less than a decade in many cases). Nevertheless, these discrepancies are large and the challenge to accurately assess the importance of MG&IC in balancing the global sea level budget remains a major scientific question to be resolved.
a Europe
Across Europe the rapid wastage of Alpine valley glaciers was for a time offset by positive mass balances across Scandinavia (Fealy and Sweeny, 2005). In the last decade Scandinavian glaciers have also switched to negative mass balance (Andreassen et al., 2005), making the overall European mass balance negative. However, because of the relatively low volumes of mass loss involved this has an almost imperceptible effect on GSLR. About half of the mass loss from the Swiss Alps over the last century may be attributed to natural climate variability (Huss et al., 2010). Paul et al. (2007) applied a simple model relating equilibrium line altitude (ELA) change to average temperature change to investigate the potential effect of warming on Swiss glaciers and proposed a 65% decrease in glacier area in the Swiss Alps by 2050 under a dramatic warming scenario of +2ºC. A rise in summer average temperatures of 3°C over the European Alps (which lies within IPCC scenario projections for the end of the 21st century) could reduce ice cover to as little as 20% of its current extent (Zemp et al., 2006).
b North America
Across the Canadian Arctic estimates of mass loss from ice caps have recently been revised upwards to 0.17±0.02 mm a−1 GSLR following evidence for increasingly negative SMB in response to increases in summer air temperatures (Gardner et al., 2011; Sharp et al., 2011). Iceberg calving has also been shown to be a major contributor to mass loss from the Canadian Arctic ice caps (Mair et al., 2009; Williamson et al., 2008), accounting for up to 40% of the mass loss across some of the largest ice masses (Burgess et al., 2005). The mass balance of the vast valley glaciers and icefields of Alaska and NW Canada have been assessed recently over several decades (1962–2006) using the volume change method from differences in sequential digital elevation models (Berthier et al., 2010). This study concludes that the overall rate of mass loss from this region may be 0.12±0.02 mm a−1 GSLR, approximately a third less than earlier long-term estimates (Arendt et al., 2002). Using a short (2003–2007) period of satellite gravimetry data, Luthcke et al. (2008) estimate mass loss of 0.24 ± 0.014 mm a−1 GSLR, but this compares with a slightly longer (2003–2010) and more recent GRACE-based estimate of mass loss from Alaska of just 0.13 ± 0.02 mm a−1 GSLR (Jacob et al., 2012). This is nearly half the value proposed by Arendt et al. (2002) a decade earlier for the period 1995–2001. Given the different methods and time periods adopted by these studies any assessment of the current contribution of all North American glaciers to GSLR is subject to considerable uncertainly. However, a value of 0.35 mm a−1 over the last decade would be a conservative estimate.
c Himalayas and Karakorum
Relatively few studies have focused on these regions of the world despite the vast scale of the glaciers and icefields. In the Himalayas, widespread marginal retreat of icefields and valley glaciers has been revealed using a variety of satellite imagery and aerial photography techniques (e.g. Ding et al., 2006). Evidence of long-term thinning at very high elevations has also been obtained from analyses of an ice core drilled from a glacier lying at over 6000 m elevation in the Central Himalayas (Kehrwald et al., 2008). However, glaciers across the Karakorum are healthier. Hewitt (2005) reported widespread glacier advance and thickening during the late 1990s throughout the region. The high incidence of surging glaciers in the region (Copland et al., 2009) makes the regional mass balance assessment from field observations difficult. However, Gardelle et al. (2012) calculated the regional mass balance of glaciers in the central Karakoram between 1999 and 2008, based on the difference between two digital elevation models, and concluded that the regional mass balance is marginally positive, contributing –0.01 mm a−1 to GSLR (although not significantly different from zero). A recent gravimetric study (Matsuo and Heki, 2010) also saw evidence for mass gain in the Karakorum from 2002 to 2006 but mass loss thereafter. However, this study estimated high mass loss rates across the Himalayas and concluded that the entire High Mountain Asia region contributed 0.13 ± 0.03 mm a−1 GSLR from 2003 to 2009. Such high rates of mass loss have recently been challenged by Jacob et al. (2012) who estimate a GSLR contribution of just 0.011 ± 0.056 mm a−1 from 2003 to 2010. Since nearly 800 million people depend on glacial meltwater runoff from this region for their water resources, the temporal and spatial effects of climate change across Himalayan and Karakorum glaciers is a question that merits rigorous investigation (Bolch et al., 2012).
d Patagonia and the Antarctic Peninsula Islands
In the Southern Hemisphere the mass balance of the icefields of Patagonia has been reassessed using gravimetry data from GRACE (Chen et al., 2007). Their estimate of 0.078 ± 0.031 mm a−1 GSLR confirms earlier work based on topographic and cartographic data (Rignot et al., 2003) and suggests that this region is contributing a disproportionately large freshwater flux to the ocean which is unlikely to be accounted for solely by changes in SMB. High rates of mass loss through calving into lacustrine environments are a likely explanation. Further south, attention has been focused on the mass balance of MG&IC around the Antarctic Peninsula. Hock et al. (2009) have developed and applied a grid-based, mass balance temperature and precipitation sensitivity model which can be extrapolated to glacierized regions lacking in direct measurements. They conclude that MG&IC around the Antarctic Peninsula may have provided 0.22 ± 0.16 mm a−1 to GSLR between 1961 and 2004, which is up to 28% of their calculated total for global MG&IC.
e Antarctica and Greenland
The Intergovernmental Panel for Climate Change (IPCC) in their Fourth Assessment Report of 2007 highlighted the major uncertainties associated with the potential contributions to GSLR over the next century from the Greenland or Antarctic Ice Sheets (Solomon et al., 2007). Unsurprisingly, there has since been a considerable international effort to improve estimates of mass change and assess the significance of temporal trends in the range of estimates.
A series of review articles in the last six years have built up an emerging consensus around the sign, magnitude and trends of mass change across the ice sheets (Alley et al., 2007; Shepherd and Wingham, 2007; Thomas et al., 2006; Van den Broeke et al., 2011). The annual surface mass balance variability across Antarctica is probably large enough to obscure any trend in the net mass balance, although this has recently been contested (Rignot et al., 2011). There is little evidence for a trend of increasing accumulation (Helsen et al., 2008) as has been hypothesized to result from atmospheric warming (Houghton et al., 2001). Dynamic mass loss from the WAIS, apparently driven mainly by ocean forcing, outstrips mass gain across the East Antarctic Ice Sheet (EAIS) resulting in an overall mass balance that is negative. Given the variability of estimates resulting from different spatial scales of measurement, different measurement approaches, different time periods for averaging, and different modelling approaches to SMB, GIA or subsurface densification, quoting a number for the overall mass loss is problematic, but the average value may lie between 0.4 and 0.5 mm a−1 GSLR. For the GrIS the issue is a little clearer. Mass gain in the interior is outstripped by mass loss closer to the margins. Annual SMB variability is very high since winters with low accumulation tend to be followed by summers with higher runoff on account of the earlier removal of snow and exposure of bare glacier ice. However, a downward trend in SMB has been evident since 1999 which has been combined with increased mass flux directly to the oceans resulting in a significant increase in mass loss from Greenland over the last decade. The GrIS is currently contributing about 0.7 mm a−1 to GSLR. Robust quantitative predictions of future ice sheet mass changes are still problematic since mass flux across calving margins of both Antarctica and Greenland represents major components of ice loss, yet oceanic, atmospheric, sea level and ice-dynamic processes controlling this flux remain poorly understood.
III Atmospheric influences on the dynamics of the Greenland Ice Sheet
Atmospheric warming is a major contributing factor to more negative or less positive SMB recorded across many glaciated regions on Earth over the last decade or longer. Notable exceptions are the ice sheets of Antarctica where summer temperatures are so low that surface melt is negligible across most of the continent with the important exception of the Antarctic Peninsula (Vaughan, 2006). However, across temperate valley glaciers and ice fields, across the Arctic ice caps, and across almost half of the GrIS, summer air temperatures are high enough to induce significant rates of surface melting. While the generation of increased melt and runoff has a direct influence on the SMB of glaciers and ice sheets, it may also have an impact on long-term mass balance because of the influence of glacier hydrology on glacier dynamics. Research continues into understanding glacial hydrological processes that control the well-known association between surface melting and enhanced summer velocities across temperate and high-latitude glaciers (e.g. Bartholomaus et al., 2008; Bingham et al., 2008; Fountain et al., 2005; Lappegard et al., 2006; Mair et al., 2008; Sugiyama et al., 2010), but the importance of the association to long-term glacier and ice sheet mass change is uncertain and has been the focus of a substantial body of research within the last few years. The debate is particularly pertinent to the future stability of the Greenland Ice Sheet with the probability of a continued trend of atmospheric warming (Solomon et al., 2007) raising the prospect of the ice sheet having an enhanced contribution to GSLR due to meltwater enhanced basal sliding during the 21st century.
Penetration of surface-derived meltwater to the bed of the GrIS may cause high subglacial water pressures, enhanced basal motion and increased glacier surface velocities (Zwally et al., 2002). This could promote enhanced ice sheet mass loss through more efficient transfer of ice down-gradient into the ablation zone, and/or across the marine grounding line. These processes could have the effect of ‘drawing down’ the interior elevations of the ice sheet (so-called ‘dynamic thinning’) thereby inducing a positive feedback since, all else being equal, reduced elevation will lead to greater surface melting. This melt-lubrication mechanism has been outlined, and its potential importance debated, in recent reviews and commentaries (Alley et al., 2005a; Kerr, 2008; Parisek, 2010; Witze, 2008). A summary of research activity and debate surrounding this issue with relation to the GrIS in the period since 2005 follows.
1 Mechanisms for surface meltwater penetration to the bed of the Greenland Ice Sheet
For melt-lubrication induced speed-up of the ice to have a widespread impact on the GrIS under a warming climate, surface meltwater must be capable of accessing the glacier bed through very thick ice, over 1 km thick. If this is not possible then the effect will always be limited to thinner, marginal regions. Zwally et al. (2002) interpreted the summer speed-ups at Swiss Camp to be the consequence of localized basal forcing caused by penetration of surface meltwater through 1 km thick cold glacier ice. Price et al. (2008) examined the local forcing hypothesis by applying a 2-D flow-line model, including longitudinal stress coupling, to an ice sheet transect through the Swiss Camp site to examine dynamic effects of local and distal basal forcing, relative to Swiss Camp. They suggested the speed-up that Zwally et al. (2002) measured could result from longitudinal coupling to a region of much thinner, crevassed ice approximately 15 km downstream. Since ice acceleration near the ELA could result from coupling to thinner, more crevassed marginal regions, they questioned whether the melt-lubrication theory really could have a widespread effect on large-scale ice sheet dynamics even with continued atmospheric warming.
The issue of how meltwater could access the ice sheet bed through cold, very thick glacier ice was tackled by Alley et al. (2005b). Applying a numerical model based on linear elastic fracture mechanics (LEFM, originally developed to examine magma-filled fracture through brittle crustal rock), they concluded that water-filled crevasses of several tens of metres depth could propagate to the bed as long as they were continually filled with water from the surface. Initiation of small crevasses represented the most difficult stage of the process but it was suggested that early overpressurization of small crevasses fed by drainage of supraglacial lakes was likely to be important in overcoming this physical barrier. Van der Veen’s (2007) model of water-filled crevasse propagation concluded that the rate at which a crevasse is filled with water is the main control on propagation rates. Rapid transfer of melt to the bed of a cold glacier or ice sheet could be achieved over hours to days depending on ice thickness and the availability of ponded surface water. Using LEFM theory, Krawczynski et al. (2009) focused on the crevasse geometry created during fracture propagation to calculate the dimensions of water-filled cracks beneath supraglacial lakes in order to place volumetric constraints on the amount of water required to drive crevasses through thick and cold glacial ice. They concluded that lakes larger than 0.25–0.80 km in diameter contain sufficient water to rapidly drive hydro-fractures through 1–1.5 km of subfreezing ice. Such lakes contain 98% of the meltwater volume held in supraglacial lakes in the central western margin of the GrIS.
2 The role of supraglacial lakes
The size, volume and spatiotemporal distribution of supraglacial lakes has consequently been the focus of several papers since their potential for delivering meltwater to the bed of the GrIS was established (e.g. Lampkin, 2011; Lüthje et al., 2006; Sneed and Hamilton, 2007). Box and Ski (2007) compared MODIS Terra imagery with ground-surveyed lake depth data to develop a depth-reflectance parameterization used to then estimate the volumes of meltwater that could access the bed via lake drainages. They concluded that an average lake drainage event could supply enough water to maintain a pressurized subglacial drainage environment for a period of hours to days. More recently, Tedesco and Steiner (2011) compared in situ depth and spectral measurements of supraglacial lakes with a variety of methods for deriving lake depth and volume from analyses of multispectral satellite reflectance data. This work allowed them to quantify the uncertainty of current procedures for multispectral bathymetry of supraglacial lakes. McMillan et al. (2007) analysed repeat satellite imagery of the growth and disappearance of supraglacial lakes on the western margin of the GrIS and found that the growth of lakes was controlled by the seasonal pattern of increasing surface melt, dependent on latitude, elevation and time. Lake drainages were most frequent in August. Sundal et al. (2009) extended this type of analysis to regions of the northwest and northeast GrIS over several melt seasons and found similar seasonal evolution in the size and extent of lakes and in the timing of their drainage. Years exhibiting high runoff and extensive lake coverage were generally characterized by low accumulation and high melt season temperatures, and vice versa. They concluded that the area and time period over which ice sheet surface to bed connections are established may increase in a warmer climate (Meehl et al., 2007).
Das et al. (2008) made a valuable contribution to the topic through field-based monitoring of the drainage of a supraglacial lake and recording the localized velocity response of the ice sheet to it. Immediately after a rapid lake drainage event there occurred concurrent horizontal speed-up and vertical uplift of the ice sheet immediately around the lake, which clearly demonstrated that surface water could penetrate to the base of the ice sheet through 980 m of ice; Das et al. (2008) proposed that crevasse hydro-fracture was most likely responsible. Catania et al. (2008) compared ice penetrating radar survey data from the region around Swiss Camp with output from a radar-simulation model that incorporated varying geometries of vertical conduits to simulate moulins. They found that measured radar profiles around lakes were typically not associated with features indicative of moulins, and thus proposed that many lakes drain over the ice sheet surface and that moulins are more common in the ablation zone where surface melt is higher and ice is much thinner. However, additional field evidence for direct coupling between the surface and bed of the GrIS through thick ice came from observations 37 to 72 km from the ice sheet margin on the land terminating Russell Glacier in west Greenland (Shepherd et al., 2009). Coincident fluctuations in meltwater volume, ice elevation and ice velocity were consistent with the presence of conduits linking the ice sheet surface and base. Diurnal fluctuations in velocity and uplift occur approximately two hours after the peak rate of surface melting in a region where ice thickness ranges from 900 to 1100 m, suggesting water was transferred rapidly to the bed as commonly occurs in Alpine glaciers (Nienow et al., 2005).
3 Magnitude, extent and timing of surface speed-ups
Joughin et al. (2008a) analysed temporal changes in spatial patterns of ice sheet surface velocity across a wide swath of the western Greenland ice sheet based on interferometric synthetic aperture radar (InSAR) and Global Positioning System (GPS) observations. They concluded that summer speed-up of the ice sheet margin was widespread with summer velocities ubiquitously 50–100% higher than during winter. Although they concurred that much of this increase was most likely due to atmospherically induced surface meltwater lubrication of the glacier bed, across tidewater catchments the relative speed-up was much less (<10–15%) and an order of magnitude smaller than other likely causes of speed-up such as reductions in back-stress at floating termini (Joughin et al., 2008a). One limitation of this study was that the velocity data coverage was overwhelmingly concentrated to within 30 km of the margin, below an altitude of 1000 m elevation, and so did not address the issue of what effect the melt-lubrication mechanism may have on ice flow in the ice sheet interior. An analysis of the displacement of the ‘K-transect’, a long-term mass balance stake and weather station transect, from 1991 to 2006 by Van de Wal et al. (2008) included data extending well above the ELA. The data supported evidence for short-term, melt-induced speed-up events during the summer melt season, but showed that the ice decelerated over the long term while ablation generally increased. The authors inferred that the long-term development of a more efficient subglacial drainage system could have prevented the development of a positive feedback between surface ablation and ice velocity.
The importance of subglacial drainage efficiency was highlighted by Bartholomew et al. (2010), who carried out a detailed ground based differential GPS ice velocity survey throughout a melt season on the Russell Glacier region of the western margin of the GrIS. Their high temporal resolution measurements showed higher than previously measured summer speed-up and modulation of flow controlled by seasonal evolution of subglacial drainage just like a temperate valley glacier (e.g. MacGregor et al., 2005; Mair et al., 2002; Nienow et al., 1998). However, this data only extended up to ∼1000 m elevation preventing an assessment of the impact of these processes on the catchment dynamics as a whole. A more spatially extensive analysis of surface velocity across this same region of the GrIS was undertaken by Palmer et al. (2011) using InSAR derived velocity data from a ∼10 000 km2 study area for seven day-long periods covering winter, early spring and late summer of 1995–1996. They showed that the late summer melt-induced speed-up occurred much further inland (∼100 km) and at higher elevations (∼1600 m) than previously recorded (e.g. Joughin et al., 2008a; Rignot and Kanagaratnam, 2006). There was a significant positive correlation between modelled runoff from individual catchments within the region and the late summer speed-ups. However, the early spring/summer data did not include elevations below ∼800 m where Bartholomew et al. (2010) had shown very high early summer speed-ups. There was a strong spatial association between areas of highest velocity and modelled surface drainage pathways and sinks which gave further support to the idea that surface hydrology and ice dynamics are closely linked even across high elevations where ice is >1000 m thick.
4 The role of subglacial drainage system evolution: theory and measurements
The potential for complex ice sheet flow variations to be driven by changes in glacier hydrology has provoked renewed interest in modelling subglacial hydrology beneath ice sheets (e.g. Hewitt, 2011; Schoof, 2005). Pimentel and Flowers (2010) explicitly attempted to qualitatively replicate patterns of ice flow speed-up observed by Shepherd et al. (2009) and Das et al. (2008) on the GrIS using a numerical model that coupled higher order ice dynamics to a subglacial drainage system that was able to switch between a slow distributed and a fast channelized water flow network. Although general seasonal patterns of behaviour were replicated, the model did not reproduce the large diurnal velocity variations observed by Shepherd et al. (2009) and overestimated the temporal impact of surface lake drainage on subglacial water pressures as compared to the field observations and interpretations of Das et al. (2008). However, better agreement might depend on better input parameters, which were poorly constrained by observational data, as well as developing the model to three-dimensional space.
An important development in the modelling of subglacial drainage beneath a conceptual ice mass was made by Schoof (2010). While still including the physics controlling the competing processes of conduit closure due to ice creep and enlargement due to frictional heating from turbulent water flow (e.g. following Röthlisberger, 1972), he unified the physics of cavities (slow distributed flow) and channels (fast channelized flow) in a single equation for the cross-sectional area of a subglacial conduit. The model showed how a spatially extended drainage system can switch from cavities to channels and back in response to changing rates of water input. The model qualitatively reproduced two features of subglacial drainage systems that have been inferred from ice sheet and glacier field measurements. First, channelization occurs above a critical rate of steady water input to the subglacial drainage system, leading to lower subglacial water pressures and hence glacier deceleration. Second, short-term increases in water input create transient peaks in water pressure that cause short-lived periods of ice acceleration. Thus variability in water input to the subglacial drainage system is a key control on ice dynamics. For the GrIS such variability could be driven by strong diurnal melt cycles, sudden rainfall events and surface lake drainage events.
The first of Schoof’s (2010) conceptual conclusions above was used by Sundal et al. (2011) to help understand their investigations into the influence of atmospherically induced melt on ice sheet dynamics across the ablation areas of a number of land-terminating glaciers in southwest Greenland. They compared satellite-derived surface velocities with modelled melt rates over contrasting melt seasons and concluded that although peak rates of ice speed-up were positively correlated with the degree of melting, mean summer flow rates were not. Their data showed that in warmer years the period of fast ice flow was shorter and overall summer velocity was slower. Earlier velocity slowdown was attributed to the earlier establishment of an efficient subglacial drainage system able to accommodate high meltwater fluxes without elevating subglacial water pressures and enhanced basal sliding. These findings were similar to those of earlier studies (Price et al., 2008; Truffer et al., 2005; Van de Wal et al., 2008) which proposed that the ice sheet subglacial drainage system may adjust to accommodate increased melting in a way that does not lead to proportionate increases in flow. Their analyses were limited by being concentrated on data pertaining to a narrow elevation band of 400–600 m elevation within 20 km of the ice sheet margin, and was based on motion averaged over periods of ∼35 days.
A detailed record of surface velocities throughout an annual cycle, including daily measurements throughout the summer melt season, was presented by Bartholomew et al. (2011a) from seven GPS sites along a 115 km transect, extending through the Russell Glacier catchment from the ice sheet margin to elevations of over 1700 m elevation, far into the accumulation area. In situ measurements were also made of air temperature and surface ablation, and satellite monitoring of ice surface albedo and supraglacial lake drainage were used to investigate hydrological controls on ice velocity changes. The data from the ice sheet were supported by simultaneous measurements of proglacial hydrological parameters (Bartholomew et al., 2011b) that were interpreted to investigate seasonal changes in the structure and behaviour of the hydrological system. Bartholomew et al. (2011a) supported the idea that the timing and rate of meltwater delivery to the ice–bed interface provides the key control on the magnitude of hydrologically forced velocity variations at each site. Although summer variation in ice motion was limited by subglacial drainage system evolution (supporting Schoof, 2010), the percentage enhancement of annual velocity caused by summer speed-ups was much greater at sites nearer the ice sheet margin, where surface melting is higher, than at sites above 1000 m elevation. At these higher elevations a shorter melt period and delayed establishment of a hydraulic connection between the ice sheet surface and its bed limited the timeframe for velocity variations to occur. Bartholomew et al. (2011a) concluded that if future atmospheric warming causes the behaviour they observed in the lower ablation zone to propagate further inland, then land-terminating sections of the GrIS will experience increased dynamic mass loss in a warmer climate.
5 Significance of melt-lubrication for long-term stability of the Greenland Ice Sheet
There is clearly considerable temporal and spatial variability in the sensitivity of ice dynamics to atmospherically derived meltwater inputs to the ice sheet bed. To date the collective results from all the above field-based and remote-sensing studies do not fully describe this phenomenon. More spatially extensive measurements of surface melt and dynamics, over several contrasting melt seasons, will be required to resolve the issue of whether the annual velocities across an entire catchment are enhanced or diminished by a warmer summer melt season. Moon et al. (2012) have recently presented a valuable record of decade-long, InSAR-derived maps of annual surface velocity across almost all of Greenland’s 200+ outlet glaciers, from 2000 to 2010, with annual sampling for the latter half of this period. They show very complex temporal and regional variability that point to highly complex relationships between catchment-wide ice dynamics and atmospheric or oceanic forcing that would be expected over what is still a climatically short time period. With a long-term, sustained combination of the large spatial coverage afforded by satellite derived data, and the high temporal resolution of continuous in situ field measurements of surface velocity, the relative importance of atmospheric and oceanic forcings on ice dynamics, and their climatic sensitivity, will become more evident. Enough is currently known to be confident that the melt-induced acceleration of the GrIS is unlikely to lead to a large-scale dynamic instability and mass loss capable of causing sea level rise of several decimetres by the end of this century (Kerr, 2008; Pfeffer et al., 2008).
The melt-lubrication effect on ice dynamics has a greater relative impact on annual velocities across land-terminating catchments (Joughin et al., 2008a). However, 50–60% of the ice mass lost annually from the GrIS is believed to be discharged through marine terminating outlet glaciers (Rignot et al., 2008; Van den Broeke et al., 2011). Although surface melt induced speed-ups do affect seasonal velocity variations on tidewater glaciers (How at et al., 2010; Sole et al., 2011), the cause of dramatic changes in the dynamics of tidewater glaciers is believed to be driven by processes at the tidewater termini. Understanding the interactions between the oceans and glaciers constitutes one of glaciology’s most important and pressing research issues today and will be the focus of the second progress report in this series of glaciological research updates. Once marine-terminating sectors of the GrIS retreat away from the ocean, then processes controlling the dynamics of land-terminating sectors will assume greater significance for the longer-term evolution of the ice sheet mass (Sole et al., 2008). Thus, over the longer term, quantifying the impact of warmer melt seasons on ice flux remains critical to reliable projections of rates of potential ice sheet decay in the future and for understanding processes responsible for ice sheet retreat in the past.
IV Conclusions
This progress report is the first of a series of glaciological research updates. Two research topics have been covered: (1) the mass balance of glaciers and ice sheets and their contributions to global sea level change; and (2) atmospheric melt-induced influences on the dynamics of the Greenland Ice Sheet. Three main approaches to determining the mass change of glaciers and ice sheets have been outlined: the mass budget method, the volume change method through altimetry, and satellite gravimetry. Each method has its advantages and disadvantages but, because they are independent of each other, there exists the potential to significantly reduce the uncertainty in estimates of contemporary mass change when temporally and spatially comparable measurements are made. The evidence indicates an increasingly negative global trend in mass balance over the last decade. This is particularly evident across High Arctic Canada and Greenland. Significant discrepancies still remain in estimating the GSLR contributions in many regions of the world, most notably across High Mountain Asia and Alaska where interannual variability is high and mainly driven by variations in precipitation. Determining the mass balance of the Earth’s glaciers and ice sheets and their contribution to the global sea level budget requires long-term, repeated and widespread measurement due to the spatially and temporally heterogeneous nature of the processes controlling mass balance.
Our understanding of one of the glaciological processes which may be an important secondary atmospheric control on the mass balance of the Greenland Ice Sheet, namely surface melt-induced acceleration of ice sheet flow, has advanced considerably through dedicated field-based and remote sensing measurements as well as through development of basal hydrological theory for application at the ice sheet scale. Field evidence suggests that surface meltwater can penetrate cold glacier ice in excess of 1 km thick and cause the overlying ice to accelerate. However, as summer melt continues this process becomes moderated by the evolution of an increasingly efficient subglacial drainage system. Quantifying the long-term effect of this process and determining its climatic sensitivity across the scale of entire glaciated catchments and ice sheets remain significant research objectives. However, looking to the future, this process is unlikely to be responsible for more than a few centimetres of GSLR by 2100.