Contrary to prior expectations that warming would cause mass addition averaged over the Greenland and Antarctic ice sheets and over the next century, the ice sheets appear to be losing mass, at least partly in response to recent warming. With warming projected for the future, additional mass loss appears more likely than not.

Changes in the surface elevation of the Greenland and Antarctic ice sheets and ice shelves caused by variations in the rate of firn compaction are calculated with a time-dependent firn densification model driven by two decades (1982–2003) of satellite-observed monthly surface temperatures. The model includes the effects of melting and refreezing, both the direct changes in density and the subsequent effects on the densification rate. As previously shown, the temperature-dependent rate of densification is largest in summer, but changes in winter temperatures also have a significant effect. Over the last decade, climate warming has enhanced the rate of compaction and lowered the average surface elevation of Greenland by 1.8 cma-1 and most of West Antarctica by 1.9 cma–1. In East Antarctica, a small cooling raised the average surface elevation by 0.14 cma–1.

The accumulation rate on Potsdam Glacier, East Antarctica, and its spatial and temporal variations are examined using ground-penetrating radar, snow samples and firn-core studies. Physical properties in snow samples and along firn cores provide distributions of density with depth, showing only small spatial variation. Counting of peaks in δ18O along the firn cores yields an age–depth distribution that is transferred to the stratigraphy of isochronal internal layers observed with radar. From two radar horizons we determine the spatial accumulation pattern, averaged over the periods 1970–80 and 1980–2004. The shape of internal layers indicates an ablation area at the eastern margin of the investigation area. Accumulation rates show a very high spatial variability, with a mean value of 141 kgm–2 a–1 for the period 1970–2004 and a standard deviation of almost 50%. Mean temporal variation of only a few per cent throughout the investigated area for the observed time interval is much less than the spatial variations. The mean accumulation values are somewhat less than values reported before from this region. Accumulation pattern and surface topography are linked in a way indicating that wind-borne redistribution of snow significantly contributes to the observed spatial variations of accumulation rates. The accumulation data and their variability complement and validate present and future satellite studies of Antarctica’s mass balance.

Variations in the depth of radar-detectable englacial layers (isochrones) are commonly used to assess past variability in accumulation rates, but little is known about the effect of internal and basal flow variations on isochrone deflections (e.g. bumps, troughs). In this paper, we show how the isochrones are affected by such variation using a three-dimensional flow model to investigate changes in the flow mode and in increased basal melting. We also investigate how transverse flows with lateral velocity gradients affect the development of isochrones. We use the model to visualize how such variations will be seen in radar lines which cross the flow direction. We show that in the presence of lateral gradients in the flow field we can produce bumps and troughs when viewed along transects perpendicular to the flow. The model results show that the influences of flow convergence, melting and changes in flow mode, when coupled together, affect isochrones over the whole depth of the ice sheet. Finally, changes in the near-surface layers cannot be solely attributed to spatial variation in the accumulation rate; there can also be a strong signal from changes in the flow mode.

This paper aims at presenting a new dataset including the melt events derived from microwave remote sensing occurring in Antarctica from summer 1979/80 to 2005/06. The method for detecting melt events and sources of error is presented, and then trends in melt duration for every pixel are extracted from the dataset, mapped and analyzed. The analysis focuses on two particular cases, and the main results show that: (1) the trends over the period 1980–2006 in the Antarctic Peninsula match with lengthening of the melt season on the ice shelves and, surprisingly, shortening of the melt season in the mountainous area of the peninsula; and (2) the trends over the period 1996–2006 on the entire continent show a dipolar pattern, with the western regions experiencing decreasing melt duration, whereas East Antarctica and the Ross Ice Shelf experience increasing melt duration. This pattern closely mirrors the temperature pattern expected when the Southern Annular Mode is in a decreasing trend, as it is over the period 1996–2006. For further analysis and validation, the dataset has been made available at http://www-lgge.obs.ujf-grenoble.fr/~picard/melting/.

A new surface classification algorithm for monitoring snow and ice masses based on data from the moderate-resolution imaging spectroradiometer (MODIS) is presented. The algorithm is applied to the Greenland ice sheet for the period 2000–05 and exploits the spectral variability of ice and snow reflectance to determine the surface classes dry snow, wet snow and glacier ice. The result is a monthly glacier surface type (GST) product on a 1 km resolution grid. The GST product is based on a grouped criteria technique with spectral thresholds and normalized indices for the classification on a pixel-by-pixel basis. The GST shows the changing surface classes, revealing the impact of climate variations on the Greenland ice sheet over time. The area of wet snow and glacier ice is combined into the glacier melt area (GMA) product. The GMA is analyzed in relation to the different surface classes in the GST product. The results are validated with data from weather stations and similar types of satellite-derived products. The validation shows that the automated algorithm successfully distinguishes between the different surface types, implying that the product is a promising indicator of climate change impact on the Greenland ice sheet.

Johnsons and Hurd Glaciers are the two main glacier units of Hurd Peninsula ice cap, Livingston Island, South Shetland Islands, Antarctica. They presently cover an area of about 10 km2. Johnsons is a tidewater glacier, while Hurd Glacier ends on emerged land. In this paper, we estimate the changes in ice volume during the period 1956–2000, and compare them with the regional meteorological records. The volume-change estimates are based on the comparison of digital terrain models for the glacier surface, constructed from aerial photographs taken by the British Antarctic Survey in 1956 and from our geodetic measurements in 1999/2000. The total volume estimates are based on an ice-thickness map constructed from radio-echo sounding profiles (18–25 MHz) done in 1999–2001, showing maximum ice thickness of about 200 m. We estimate the changes in ice volume during the period 1956–2000 to be –0.108±0.048km3, which represents a 10.0±4.5% decrease from the 1956 total volume of 1.076±0.055km3 and is equivalent to an average annual mass balance of –0.23±0.10mw.e. during 1956–2000. Ice-thickness changes range from –40 to +20 m, averaging –5.5±4.4 m. Most areas show ice thinning; the thickening is limited to a small area within Johnsons Glacier. All glacier fronts, except Johnsons’ calving front, show retreat. These changes are consistent with the regional meteorological records for mean summer temperature, which show a trend of +0.023±0.005˚Ca–1 during the period 1956–2000.

In Antarctica, low-elevation (<1000 m) blue-ice areas (LEBIAs) may experience melt–freeze cycles due to absorbed solar radiation and the small heat conductivity in the ice. In some cases, LEBIAs can contain significant amounts of subsurface liquid water. Since the spatial extent of blue-ice areas depends on climatic conditions, they have been seen as good indicators of warming in Antarctica. A two-dimensional (x-z) model has been developed to simulate the formation and water circulation in the subsurface ponds. The model results show that for a reasonable parameter set, the formation of liquid water within the ice can be reproduced. Vertical convection and a weak overturning circulation is generated which acts to stratify the fluid and transport warmer water downward, thereby causing additional melting at the base of the pond. In a multi-year integration, a global warming scenario mimicked by a decadal-scale increase (3˚C per 100 years) in air temperature, leads to a general increase in subsurface water volume and changes in pond shape and depth. Even before melting at the surface is reached, heat that accumulates below a certain depth can no longer be removed during winter and leads to disintegration of the ice.

Drastic changes have been detected in glacial systems of the Antarctic Peninsula in the last few decades and are well documented in numerous scientific publications. However, the spatial and temporal distribution of glacier changes on the Antarctic Peninsula remains largely restricted to ice fronts. To expand the current monitoring of a few glaciers, unevenly distributed along the peninsula, to a representative set, we developed a method to simplify the detection of boundaries between glacier zones using satellite SAR data. The evolution of glacier zones is greatly influenced by local and regional climatic and meteorological settings. Their variations in response to changes in energy or mass balance are considered as good indicators of climatic changes. In this paper, we describe the results of knowledge-based image analysis algorithms on test areas located at Trinity Peninsula and near Marguerite Bay. In general, the two analyzed areas show different patterns of glacier zone development. The bare-ice zone occurs mainly on glaciers located on the eastern side of Trinity Peninsula. Its upper boundary shows a good correlation with the mean summer air temperature. Finally, the position of the dry-snow line shows different spatial patterns of change in both study areas.

The mass balance of polythermal ice masses is critically dependent on the proportion of surface-generated meltwater that subsequently refreezes in the snowpack and firn. In order to quantify this effect and to characterize its spatial variability, we measured near-surface (<10 m) snow and firn densities at an elevation of ~1945ma.s.l. in the percolation zone of the Greenland ice sheet in spring and autumn 2004. Results indicate that local snowpack depth above the previous end-of-summer 2003 melt surface increased by ±5% (7.6 cm) from spring to autumn while, over the same period, snowpack density increased by >26%, resulting in a 32% increase in net accumulation. This ‘seasonal densification’ increased at lower elevations, rising to 47% 10 km closer to the ice-sheet margin at 1860ma.s.l. Density/depth profiles from nine sites within 1 km2 at ~1945ma.s.l. reveal complex stratigraphies that change over short spatial scales and seasonally. We conclude that estimates of mass-balance change cannot be calculated solely from observed changes in surface elevation, but that near-surface densification must also be considered. However, predicting spatial and temporal variations in densification may not be straightforward. Further, the development of complex firn-density profiles both masks discernible annual layers in the near-surface firn and ice stratigraphy and is likely to introduce error into radar-derived estimates of surface elevation.

An ice-flow model is used to simulate the Antarctic ice-sheet volume and deep-sea temperature record during Cenozoic times. We used a vertically integrated axisymmetric ice-sheet model, including bedrock adjustment. In order to overcome strong numerical hysteresis effects during climate change, the model is solved on a stretching grid. The Cenozoic reconstruction of the Antarctic ice sheet is accomplished by splitting the global oxygen isotope record derived from benthic foraminifera into an ice-volume and a deep-sea temperature component. The model is tuned to reconstruct the initiation of a large ice sheet of continental size at 34 Ma. The resulting ice volume curve shows that small ice caps (<107 km3) could have existed during Paleocene and Eocene times. Fluctuations during the Miocene are large, indicating a retreat back from the coast and a vanishing ice flux across the grounding line, but with ice volumes still up to 60% of the present-day volume. The resulting deep-sea temperature curve shows similarities with the paleotemperature curve derived from Mg/Ca in benthic calcite from 25 Ma till the present, which supports the idea that the ice volume is well reproduced for this period. Before 34 Ma, the reproduced deep-sea temperature is slightly higher than is generally assumed. Global sea-level change turns out to be of minor importance when considering the Cenozoic evolution of the ice sheet until 5 Ma.

Ice-shelf melting (ISM) removes heat from and injects fresh water into the adjacent ocean and contributes significantly to the freshwater balance and water mass formation in the Antarctic marginal seas. The thermodynamic interaction between ocean and ice shelf is a complicated process and usually not adequately included in the ocean–ice climate models. In this paper, the ISM from all major ice-shelf areas around Antarctica is added to a global coupled ice–ocean model ORCA2-LIM following the parameterization proposed by Beckmann and Goosse (2003). Using interannual forcing data from 1958 through 2000, the impact of ISM on Southern Ocean hydrography and sea-ice distribution is investigated. The model also shows global signatures of the Antarctic ISM.

Ice-stream velocities can change rapidly. Understanding the spatial and temporal pattern of these changes and the forcings responsible is essential for predicting ice-sheet mass balance. Inland migration of the onset location will lead to more efficient drainage of inland ice. One way to monitor the stability of the onset location is to investigate changes in the velocity field. We report on the velocity near the onset of Bindschadler Ice Stream, West Antarctica, in 2002 and compare these data to the velocity measured in 1996. Mean annual velocities were determined by measuring the GPS position of markers during consecutive seasons. We compare our results with similar measurements from 1996 to investigate temporal changes in this ice-stream onset. Our results indicate that only minimal changes have occurred in the speed of the ice stream between 1996 and 2002.

In three-dimensional numerical ice-sheet models that use finite-difference schemes, the position of ice margins is poorly represented with a regular quadratic grid. As a result, in a centered difference scheme, the surface gradient term and the flux divergence term computed for the gridpoints next to the ice margin may be inaccurate. In this paper, an improved scheme is presented that computes the horizontal gradients at the ice-sheet margin using an asymmetric (upstream) second-order difference scheme in order to avoid using information from the zero-thickness gridpoints. The model is applied to an idealized synthetic geometry to obtain a steady-state ice-sheet topography. The improved model shows a realistically smooth thickness distribution near the margin. Thermomechanical coupling is found to enhance the error near the margin. The error in simulated thicknesses with the centered-difference method was significantly reduced with the new upstream scheme.

The model used by Lingle and Clark (1985) to approximate the deformation of the Earth under a single ice stream is adapted to the purposes of continent-scale ice-sheet simulation. The model combines a layered elastic spherical Earth (Farrell, 1972) with a viscous half-space overlain by an elastic plate lithosphere (Cathles, 1975). For the half-space model we identify a new mathematical formulation, essentially a time-dependent partial differential equation, which generalizes and improves upon the standard elastic plate lithosphere with relaxing asthenosphere model widely used in ice-sheet simulation. The new formulation allows a significantly faster numerical strategy, a spectral collocation method based directly on the fast Fourier transform. We verify this method by comparing to an integral formula for a disk load. We also demonstrate that the magnitudes of numerical errors made in approximating coupled ice-flow/Earth-deformation systems are significantly smaller than pairwise differences between several Earth models. Our implementation of the Lingle and Clark (1985) model offers important features of spherical, layered, self-gravitating, viscoelastic Earth models without the computational expense.

This study uses older topographic maps made from high-oblique aerial photographs for glacier elevation change studies. We compare the 1936/38 topographic map series of Svalbard (Norwegian Polar Institute) to a modern digital elevation model from 1990. Both systematic and random components of elevation error are examined by analyzing non-glacier elevation difference points. The 1936/38 photographic aerial survey is examined to identify areas with poor data coverage over glaciers. Elevation changes are analyzed for seven regions in Svalbard (~5000 km2), where significant thinning was found at glacier fronts, and elevation increases in the upper parts of the accumulation areas. All regions experience volume losses and negative geodetic balances, although regional variability exists relating to both climate and topography. Many surges are apparent within the elevation change maps. Estimated volume change for the regions is –1.59±0.07km3 a–1 (ice equivalent) for a geodetic annual balance of –0.30ma–1w.e., and the glaciated area has decreased by 16% in the 54 year time interval. The 1936–90 data are compared to modern elevation change estimates in the southern regions, to show that the rate of thinning has increased dramatically since 1990.

Determining whether increasing temperature or precipitation will dominate the cryospheric response to climate change is key to forecasting future sea-level rise. The volume of ice contained in the ice caps and glaciers of the Arctic archipelago of Svalbard is small compared with that of the Greenland or Antarctic ice sheets, but is likely to be affected much more rapidly in the short term by climate change. This study investigates the mass balance of Austfonna, Svalbard’s largest ice cap. Equilibrium-line fluxes for the whole ice cap, and for individual drainage basins, were estimated by combining surface velocities measured using satellite radar interferometry with ice thicknesses derived from radio-echo sounding. These fluxes were compared with balance fluxes to reveal that during the 1990s the total mass balance of the accumulation zone was (5.6±2.0)×108m3 a–1. Three basins in the quiescent phase of their surge cycles contributed 75% of this accumulation. The remaining volume may be attributable either to as yet unidentified surge-type glaciers, or to increased precipitation. This result emphasizes the importance of considering the surge dynamics of glaciers when attempting to draw any conclusions on climate change based on snapshot observations of the cryosphere.

A new calving criterion is introduced, which predicts calving where the depth of surface crevasses equals ice height above sea level. Crevasse depth is calculated from strain rates, and terminus position and calving rate are therefore functions of ice velocity, strain rate, ice thickness and water depth. We couple the calving criterion with three ‘sliding laws’, in which velocity is controlled by (1) basal drag, (2) lateral drag and (3) a combination of the two. In model 1, velocities and strain rates are dependent on effective pressure, and hence ice thickness relative to water depth. Imposed thinning can lead to acceleration and terminus retreat, and ice shelves cannot form. In model 2, ice velocity is independent of changes in ice thickness unless accompanied by changes in surface gradient. Velocities are strongly dependent on channel width, and calving margins tend to stabilize at flow-unit widenings. Model 3 exhibits the combined characteristics of the other two models, and suggests that calving glaciers are sensitive to imposed thickness changes if basal drag provides most resistance to flow, but stable if most resistance is from lateral drag. Ice shelves can form if reduction of basal drag occurs over a sufficiently long spatial scale. In combination, the new calving criterion and the basal–lateral drag sliding function (model 3) can be used to simulate much of the observed spectrum of behaviour of calving glaciers, and present new opportunities to model ice-sheet response to climate change.

Surface mass-balance and geometry data are key to quantifying the climate response of glaciers, and confidence in data synthesis and model interpretations and forecasts requires data from as wide a range of locations and glacier types as possible. This paper presents measurements of surface elevation change at the Svalbard surge-type glacier Finsterwalderbreen, by comparing a 1990 digital elevation model (DEM) with a surface GPS profile from 2003. The pattern of elevation change is consistent with that previously noted between 1970 and 1990, and reflects the continued quiescent-phase evolution of the glacier, with mass loss in the down-glacier/receiving area of up to –1.25mw.e. a–1, and mass gain in the up-glacier/reservoir area of up to 0.60 mw.e. a–1; the area-weighted, mean change for the whole glacier is 0.19mw.e. a–1. The spatial pattern of elevation increase and decrease is complex, and the boundary between thickening and thinning determined by combining GPS and DEM data does not appear to correspond with the equilibrium-line altitude determined from surface mass-balance measurements. There is no evidence yet of a decrease in the rate of reservoir area build-up driven by mass-balance change resulting from the warmer winter air temperatures, and decreased proportion of snowfall in total precipitation, noted at meteorological stations in Svalbard.