These new methods provide accurate measurements of surface elevation, ice thickness, glacier velocity, surface temperatures and snow wetness and albedo over vast regions, as well as high-resolution imagery over entire ice sheets. Time series of many of these measurements began in the early 1990s – they show both ice sheets to be shrinking as climate warms. Floating ice shelves and glacier tongues are thinning and even breaking up because of warming ocean waters and increasing summer air temperatures. Much of this floating ice fills coastal bays, and is pushed seawards by tributary glaciers, which speed up as much as eight-fold soon after ice-shelf break-up.
Ice shelves are acting like loosely fitting corks in a tilted jar of treacle, regulating leakage according to how well they fit. Loosening the corks – i.e. thinning of the ice shelves – increases leakage, and total removal increases it more. The amount is determined by how far the bottle is tilted – equivalent to how far inland the glaciers "feel" the effects of ice-shelf removal. At the same time, warmer summers are extending the zone and intensity of summer melting to higher elevations, particularly in Greenland. This increases melt-water runoff into the ocean and melt-water drainage to the glacier bed, where it lubricates glacier sliding – equivalent to tilting the bottle further.
Measuring the balance
The mass balance of an ice sheet – the rate of change of its mass – affects global sea level. Because of the enormous size of the ice sheets, measurement of mass balance is difficult. Three techniques have been developed to get an overall picture:
• the mass-budget approach, comparing total snowfall with losses by ice discharge and melt-water runoff
• repeated altimetry, to estimate volume changes
• monitoring temporal changes in gravity, to infer mass changes.
The first two methods provide estimates of mass balance for surveyed regions, whereas the third gives estimates for very large areas or even entire ice sheets.
Mass-budget calculations compare two very large numbers: small errors in either can result in large mass-balance errors. Total snowfall – obtained from shallow ice cores combined with weather-model analyses – is 1850 ± 130 Gt/year over Antarctica and 500 ± 25 Gt/year over Greenland. (One Gt equates to 1 billion tonnes while 360 Gt is equivalent to a 1 mm change in sea level.) Total-discharge estimates use satellite Interferometric Synthetic Aperture Radar (INSAR) to measure ice velocities and aircraft radar sounding to measure grounding-line ice thickness. However, incomplete data coverage results in total-loss uncertainty of probably more than 5% for Antarctica, increasing to at least 8% for Greenland because of uncertain model estimates of melt-water runoff. As a result, mass-budget uncertainty is ˜ ± 160 Gt/year for Antarctica and ± 35 Gt/year for Greenland.
And now for altimetry. Rates of surface-elevation change, after correcting for snow/ice density and bedrock uplift, reveal changes in ice-sheet mass. Data from satellite radar altimeters is available from the late 1970s, from laser altimeters in aircraft over Greenland since the early 1990s and from satellites globally since 2003. Uncertainties arise from the assumed snow/ice density, corrections applied for basal uplift and possible changes in snow-settling rates, interpolation between comparatively sparse laser surveys, temporal changes in radar penetration into surface snow, and problems with interpretation of wide-beam radar data over the sloping and undulating surfaces typical of coastal glaciers where elevation changes are most pronounced. Collectively, these uncertainties can result in errors of a similar magnitude to those for mass-budget estimates.
Finally, let's take a look at the gravity technique. Since 2002, NASA’s GRACE satellite has measured Earth's gravity field and its variation over time. After removing the effects of tides, atmospheric loading and so on, high-latitude data contain information on temporal changes in the mass distribution of ice sheets and underlying rock. Because GRACE is at a high altitude, resulting mass-balance estimates are at a coarse resolution – several hundred kilometres – and most reliable for entire ice sheets, which are extremely difficult to calculate figures for using other techniques. Error sources include measurement and data-interpretation uncertainty, leakage of gravity signals from other regions, and local causes of gravity change other than the ice sheets. Of these, the most serious are mass changes from bedrock motion: model estimates suggest these are ˜ 10 Gt/year for Greenland, and > 150 Gt/year for Antarctica, but these values contain large uncertainties and are continually under revision. Total mass-balance errors are probably similar to those for the other approaches.
The figures shown here, taken from Global outlook for ice and snow, UNEP, 2007, summarize recent mass-balance estimates for Greenland and Antarctica. It’s clear that published error estimates are generally smaller than presented here. This is partly because I have tended towards more pessimistic estimates of error components while many authors have tended toward the optimistic. The remarkably low errors assigned to some of the radar-based estimates result primarily from a failure to include some of the uncertainties, such as that in the density used to convert elevation change to mass balance and neglect of the possibility that warming-induced changes in the surface snow may affect radar-penetration depths.
Although the three techniques for measuring mass balance all have significant errors, they offer an increased level of confidence in their collective conclusions by providing independent estimates of ice-sheet behaviour since the early 1990s, namely that:
• Higher-elevation parts of the Greenland ice sheet thickened at accelerating rates, probably because of increasing snowfall, and there has been a slow thickening of parts of East Antarctica.
• Despite this, total losses increased substantially since the mid 1990s, partly because of increased melt-water runoff during warmer Greenland summers, and partly because of increased outlet-glacier discharge from Greenland and the West Antarctic ice sheet.
•There is a strong correlation between increased glacier discharge and thinning or breakup of floating extensions of the glaciers.
The slow, measured behaviour long associated with the Greenland ice sheet is being transformed to the more rapidly changing characteristics typical of big glaciers in Alaska and Patagonia. A zone of glacier acceleration is progressively moving northward, leaving Greenland’s southern ice dome under threat from both increased summer melting near the coasts and increased ice discharge down glaciers that extend their influence far inland. If this situation continues, it is quite possible that the ice dome in southern Greenland will reach a tipping point, with accelerating positive feedback causing its ever more rapid decline and an associated sea-level rise of about 85 cm. What’s more, continued northward-migration of the zone of glacier acceleration would make the far larger northern dome also vulnerable.
On the rocks
In Antarctica, disintegration of the West Antarctic ice sheet is of most concern. The bed beneath most of this ice is far below sea level, and removal of surrounding ice shelves in a warmer climate could initiate ice-sheet disintegration and a 5-metre rise in sea level. We don’t know how long this would take, but there are clues: ice-shelf break-up along the Antarctic Peninsula was rapidly followed by massive acceleration of tributary glaciers; ice-shelf thinning further south also caused glacier acceleration and widespread ice-sheet thinning (the pronounced red-coloured region in the figure). Here, acceleration is more modest but the glaciers are far bigger, so total losses are large. No one knows how far inland the zone of glacier acceleration will spread or why the ice shelves are breaking up. But their thinning is probably caused by increased basal melting, implicating the ocean. And final break-up is accelerated if there is sufficient surface melt water to fill and over-deepen crevasses in the ice shelves, effectively wedging the ice shelf apart into fragments.
Because of these changes, ice sheet contributions to sea level rise have increased from ˜0.2 mm/year in the early 1990s to perhaps 0.8 mm/year since 2003, compared to a total sea level rise of approximately 3 mm/year during the 1990s. Some of the thinning glaciers extend tens to hundreds of kilometres inland. Whether or not ice losses continue to accelerate will depend partly on whether ice shelves continue to thin out and partly on how far inland the zones of glacier acceleration can extend. These questions represent a major challenge to scientists and the answers could have a profound impact on all of us. Research planned for the International Polar Year in 2007–2008 will help provide answers, but it is clear that longer-term measurements will also be needed in order to separate short-term variability from long-term trends and to develop models that can reliably predict future behaviour in a warming climate. There is an urgent need to establish an international programme to make these measurements. It should include:
• Early warning of high-latitude changes in factors such as surface temperatures, snow-melt extent, glacier velocities and ice-shelf breakup from: passive and active-microwave data; INSAR measurements; and high-resolution imagery, all from satellites; and Automatic Weather Stations (AWS).
• Although absolute mass-balance estimates from gravity measurements depend on modelled crustal motion that is extremely difficult to validate, this approach provides more reliable estimates of temporal changes in mass balance, alerting us to the need for more focused, local investigations.
• Mass-budget estimates provide the ability to investigate individual catchment basins, and give details of local changes. In addition to routine INSAR data, improved ice-thickness surveys will be needed.
• Satellite laser altimetry provides an excellent assessment of volume change over the central 90% or more of an ice sheet, with more spotty coverage near the coast, where orbit spacing is generally too large to catch many outlet glaciers undergoing changes.
• Satellite radar-altimeter surveys should be continued in order to extend the already long time series, but there may be serious limitations to data interpretation because of radar penetration into surface snow.
• Information from these sources will reveal "hot spots" where important changes are occurring. These should be targeted for more detailed measurements, including aircraft surveys of elevation-change rates and ice thicknesses along specified routes within the basin at any desired cross-track spacing, as well as field measurements addressing specific problems.
With these tools available over an extended time period, it would be possible to develop a clear strategy for both the measurement and understanding of ice-sheet mass balance. This will be ever more crucial as the planet warms.