Satellite Data Shows Major Ice Losses for Greenland and Antarctica: What the Gravity and Altimetry Records Reveal

Satellite data shows major ice losses for Greenland and Antarctica totaling roughly 6.9 trillion tonnes of ice between 1992 and 2020, according to the landmark IMBIE-2 assessment published in Nature in 2020. That combined loss raised global sea level by about 21 millimeters over the 28-year window. The figures come from three independent satellite techniques — radar altimetry, gravimetry, and input-output budgeting — that mostly agree within their stated error bars. What matters for a reader trying to interpret the headline is which instrument produced which number, where the losses concentrate, and why the two ice sheets behave very differently in the observational record.

How Three Independent Satellite Techniques Measure Ice Sheet Mass

Radar altimetry measures the height of the ice surface from orbit and converts elevation change into volume change. Satellites in this family include ERS-1, ERS-2, Envisat, CryoSat-2, and the ICESat and ICESat-2 laser altimeters flown by NASA. Elevation change is not the same as mass change, because snow compaction and firn density shifts can alter the surface height without adding or removing water mass.

Gravimetry uses the GRACE and GRACE-FO missions, launched in 2002 by NASA and the German Aerospace Center (DLR). These satellites detect tiny changes in Earth's gravity field as mass redistributes between the ice sheets and the ocean. Converting gravity anomalies into ice mass loss requires removing signals from post-glacial rebound, hydrology, and ocean circulation, and that modeling step introduces the largest single source of uncertainty in the method.

Input-output budgeting works differently. Researchers estimate how much snow accumulates on the ice sheet and subtract the ice discharged through outlet glaciers, then compare the two. Snow accumulation comes from regional climate models calibrated to weather stations; ice discharge comes from satellite radar interferometry measurements of glacier velocity and ice thickness from ice-penetrating radar. The method is conceptually simpler but depends on accurate knowledge of the ice sheet's bed topography.

The three techniques have different blind spots, which is why the IMBIE collaboration — the Ice Sheet Mass Balance Inter-comparison Exercise, coordinated by the European Space Agency and NASA — was built to reconcile them. IMBIE-2 combined 26 separate surveys from 1992 to 2020 and reported that Greenland lost 3.9 trillion tonnes and Antarctica lost 2.6 trillion tonnes over that period. The uncertainty on the combined figure is about 5% of the total, which is the number most researchers treat as settled.

Greenland's Mass Loss Divided Between Surface Melt and Glacier Calving

Greenland's loss is not evenly split between its two dominant processes, and the balance has shifted over the record. In the 1990s, the ice sheet was close to mass balance, with surface melt roughly offset by snow accumulation and glacier discharge only marginally exceeding replacement. The IMBIE data show Greenland entered sustained negative balance around 1997 and has been losing mass every year since, with the exception of a brief pause in 2017–2018 after an unusually wet summer.

The most complete attribution comes from the Ice Sheet Mass Balance Inter-comparison Exercise and from regional climate models like MAR and RACMO. Roughly half of Greenland's post-2000 loss came from surface meltwater runoff, and roughly half from increased discharge through marine-terminating glaciers, especially along the central west and southeast coasts. Jakobshavn Isbræ alone, near Ilulissat on the west coast, has thinned and accelerated since the late 1990s and accounts for a measurable share of discharge-driven loss.

The Greenland Ice Sheet's contribution to sea level has varied substantially by year. IMBIE-2 reports the 2010s decade averaged around 0.8 millimeters per year, up from roughly 0.2 millimeters per year in the 1990s. That is not a smooth trend — 2012 and 2019 were extreme melt years driven by persistent high-pressure ridges over the ice sheet, while 2018 was near-neutral. Any single-year figure should be read as noise around a rising baseline, not as a trend break.

Where the loss lands on the ice sheet matters for what happens next. Most of Greenland's mass sits in the interior, which is high enough to resist melt for decades under most warming scenarios. The losses in the observational record are concentrated at the margins, where the ice is already near the melting point. That pattern explains why the interior has so far served as a buffer against rapid loss — and why the coastal outlet glaciers are the place to watch.

Antarctica's Losses Concentrate in West Antarctica and the Antarctic Peninsula

Antarctica's record is far more uneven than Greenland's, and satellite data has made the geography of loss unmistakable. The East Antarctic Ice Sheet is, by most satellite estimates, close to mass balance or slightly gaining mass in the interior, though the uncertainty on that assessment is large enough that researchers argue about the sign. The West Antarctic Ice Sheet and the Antarctic Peninsula account for essentially all the continent's net loss.

The Amundsen Sea sector of West Antarctica is the epicenter. Pine Island Glacier and Thwaites Glacier have both thinned and accelerated since the 1990s, and satellite altimetry shows their grounding lines — where the ice lifts off the bed and floats — retreating inland. Thwaites in particular has been the focus of a joint US–UK field program because its bed sits below sea level and slopes downward inland, a geometry that in principle allows runaway retreat once the grounding line passes a threshold.

The Antarctic Peninsula has lost ice through the collapse of several ice shelves: Larsen A in 1995, Larsen B in 2002, and a partial collapse of Larsen C in 2017 that calved the A-68 iceberg. Ice shelves do not themselves raise sea level when they break up, because they already float. Their removal does remove the buttressing force that slows the glaciers feeding them, and the glaciers behind the former Larsen B have accelerated in the years since.

East Antarctica's interior is where the uncertainty lies. The IMBIE-2 assessment assigns a mass balance near zero for East Antarctica but with error bars larger than the central estimate. Regional snowfall variability, especially the huge precipitation swings captured by satellite microwave and gravity records, can flip the sign of East Antarctica's annual balance from year to year. That variability explains why single-year headlines about East Antarctica gaining ice should not be read as a refutation of overall Antarctic loss.

The 6.9 Trillion Tonne Figure and What It Means for Sea Level

Converting mass loss to sea level rise requires a simple ratio: roughly 360 gigatonnes of ice, or 360 trillion kilograms, raises global mean sea level by about 1 millimeter. That figure comes from dividing the mass by the ocean's surface area, and it is the standard conversion used in sea level assessments by the Intergovernmental Panel on Climate Change and in the IMBIE papers. Using that ratio, the 6.9 trillion tonnes of combined loss translates to about 21 millimeters of sea level rise.

The 21-millimeter and 19-millimeter figures should not be added directly to observed tide gauge records without accounting for ocean thermal expansion, glacier loss outside the two ice sheets, and groundwater extraction. In the IPCC's Sixth Assessment Report, ice sheets contributed roughly 0.6 to 0.8 millimeters per year to sea level from 1992 to 2020 on average, but that rate has accelerated. IMBIE-2 shows the combined ice sheet contribution rising from about 0.3 millimeters per year in the early 1990s to about 1.4 millimeters per year in the late 2010s.

Compare that to tide gauge and satellite altimetry records. Total sea level rise measured by satellite altimetry missions like TOPEX/Poseidon and its successors averaged around 3.3 millimeters per year from 1993 to 2020, according to the NASA Sea Level Change Portal. Ice sheets therefore accounted for roughly half of total sea level rise in the most recent decade, with the rest split between thermal expansion and smaller glaciers.

The 6.9 trillion tonne figure is also a reminder of scale. For comparison, the total mass of the Greenland Ice Sheet is approximately 2.85 million trillion tonnes, so the 28-year loss is a tiny fraction of the total reservoir. The concern is not that the ice sheet will vanish. It is that the rate of loss has been rising, and the processes that govern that rate — surface melt and outlet glacier discharge — are sensitive to ocean and atmosphere temperatures that respond to greenhouse gas emissions.

Why the Grounding Line and Not the Coastline Is the Key Metric

For a reader trying to track whether Antarctic losses are accelerating, the coastline is nearly meaningless as a metric. Ice shelf fronts calve icebergs regularly, and the position of the front moves without any change in ice discharge. The number that matters is the grounding line, the boundary where the ice sheet leaves the bed and begins to float, and that boundary is nearly invisible from the surface.

Satellite radar altimetry can detect grounding line migration because the ice surface bends slightly as it crosses from grounded to floating, and that flexure changes the elevation signal. NASA's Ice, Cloud, and land Elevation Satellite-2 (ICESat-2), launched in 2018, carries a photon-counting laser altimeter with enough precision to detect centimeter-scale elevation changes near the grounding line. Earlier ICESat (2003–2009) and CryoSat-2 have also contributed to grounding line records.

The satellite record shows grounding line retreat concentrated at Pine Island Glacier, Thwaites Glacier, and several glaciers in the Getz and Dotson sectors of West Antarctica. Pine Island Glacier's grounding line has retreated several kilometers per decade since the 1990s, according to ice-penetrating radar surveys and satellite altimetry combined. That retreat is a sign of ocean-driven melt: warm circumpolar deep water circulates beneath the floating ice shelf and melts its underside, thinning the shelf and reducing its ability to hold back the glacier.

What makes grounding line retreat consequential is the bed geometry. Where the bed slopes up toward the interior, retreat tends to stabilize because the ice sits on higher ground. Where the bed slopes down, as beneath Thwaites and much of the Amundsen Sea sector, retreat exposes more ice to warm water at the same time as the ice column thickens, which can accelerate flow. Satellite data cannot tell us the timing of these transitions, only that they are occurring.

Regional Patterns That Mass Loss Maps Make Visible

The maps that come out of GRACE and altimetry processing show very different spatial patterns for Greenland and Antarctica, and the differences hold lessons. Greenland's losses form a ring around the coastal margins, with the largest annual losses concentrated on the west central and southeast coasts. The interior shows negligible mass change, and some sectors show slight gains from increased snowfall in the far north.

Antarctica's map is dominated by strong negative signals in the Amundsen Sea sector and along the Antarctic Peninsula's eastern and western coasts. East Antarctica shows a mix of small positive and negative signals, with the largest positive anomaly in the sector draining into the Ross Ice Shelf and the largest negative at the Totten Glacier catchment in Wilkes Land. Totten is a concern because its bed also slopes downward inland and its ice shelf is being melted from below by warm water.

The regional maps also reveal a pattern that global totals obscure: losses move. During the 2015–2016 El Niño, Antarctic mass loss accelerated sharply, driven by increased snowfall deficits and ocean warming, then partially recovered the following year. The GRACE-FO record, which began in 2018 after a gap from 2017, documents this interannual variability clearly. That variability means that a single year's global mass loss figure can be misread as a trend.

Temporal patterns also matter for how the loss is attributed. GRACE data show Greenland's mass loss peaking in summer and slowing in winter, following the melt season. The seasonal cycle is small relative to the trend but large enough that a measurement campaign spanning only a few years needs careful seasonal correction. This is one reason long records like the 28-year IMBIE-2 window are more reliable than short studies.

Comparing Satellite Estimates: Where the Methods Disagree

The table summarizes what the IMBIE-2 reconciliation had to resolve. The disagreements are not random; they cluster around specific periods and regions. Before 2002, when GRACE data begin, the only long records come from altimetry and input-output methods, and their agreement improved substantially after 2010 as processing was refined. After 2018, GRACE-FO restored the gravimetry record, and early comparisons show it is consistent with the pre-gap GRACE data after calibration.

The most important disagreement for readers to understand is the Greenland altimetry-versus-gravimetry gap in the early 2000s. Altimetry-derived estimates reported smaller losses than gravimetry in those years, likely because firn compaction models misestimated how much surface lowering was due to density change rather than mass loss. Later work using improved firn models narrowed the gap, and IMBIE-2 combined both with weighted averaging rather than choosing one.

Cross-method agreement improves when the methods are combined rather than compared in isolation. The IMBIE approach of averaging 26 surveys gives uncertainty bounds that narrow to about 5% of the cumulative loss for the full period, down from 10–20% in earlier generations of assessment. That narrowing is the strongest evidence that the 6.9 trillion tonne figure is robust, and it is why climate researchers treat the combined estimate as the authoritative number.

What the Satellite Record Does Not Show and Where Uncertainty Persists

Satellite data cannot see what is happening beneath the ice shelves. Ocean temperature, melt rate at the ice shelf base, and circulation in the cavity beneath floating ice are inferred from models calibrated to sparse borehole and seismic measurements. The uncertainty on basal melt rates for Antarctic ice shelves is large — estimates for Thwaites' basal melt range from tens to over a hundred meters per year depending on the method — and that uncertainty propagates into projections.

The gravity record has a spatial resolution limit that matters. GRACE's resolution of roughly 300 kilometers means it cannot resolve individual glacier catchments; it reports mass change over regions the size of several counties. That is fine for tracking the West Antarctic total but insufficient for pinpointing whether Thwaites or Pine Island is losing more in a given year. Altimetry and radar interferometry fill that gap, but each has its own resolution-to-coverage tradeoff.

The record also contains a gap. GRACE ended in 2017 and GRACE-FO launched in May 2018, leaving roughly 11 months without gravimetry data. Input-output and altimetry methods bridge the gap, but the combined record across 2017–2018 has larger uncertainty than the surrounding years. This is a technical detail that rarely makes headlines, but it matters when someone compares a 2017 mass loss figure to a 2019 figure.

Looking ahead, the issue that satellite data has not yet resolved is how fast the ice sheets can lose mass under different emissions pathways. The observational record shows what has happened; projecting what happens next requires models that couple ice dynamics to ocean and atmosphere forcing, and those models still have wide uncertainty bands. The 2021 IPCC Sixth Assessment Report projected that ice sheet contributions to sea level rise by 2100 range from about 0.2 to 0.6 meters under intermediate emissions scenarios, and higher under lower-likelihood high-warming scenarios. Satellite data constrains the starting point; it does not determine the endpoint.

FAQ

How much ice did the two sheets lose in the most recent single year of satellite data?

IMBIE-2 reports the 2010s decade averaged combined losses of roughly 500 gigatonnes per year, or 0.5 trillion tonnes annually. That is the decade-scale average; individual years range widely, with strong melt years and near-neutral years mixed in. Use the decadal figure, not the single-year headline, when comparing to other periods.

Does Antarctica gaining ice anywhere mean the ice loss estimates are wrong?

No. East Antarctica's interior has shown small positive mass changes in several years, but the combined East Antarctic balance is near zero within uncertainty, and it does not offset the losses in West Antarctica and the Peninsula. The IMBIE-2 total already incorporates any East Antarctic gains, so the net figure accounts for them.

How is ice loss converted into sea level rise?

The standard conversion is 360 gigatonnes of ice equals about 1 millimeter of global mean sea level rise, computed from the ocean surface area. That ratio applies on average; regional sea level change varies because of ocean circulation, gravitational effects, and land subsidence. The global mean is not what any particular coast experiences.

Which satellite would I need to track Greenland's ice loss in near-real-time?

GRACE-FO provides monthly gravity solutions that capture mass change, and the European Space Agency's CryoSat-2 gives finer spatial detail on elevation change. Neither is a public dashboard in the sense of a weather app; the processed data appear in the journals and agency release cycles. The Danish Meteorological Institute publishes daily surface melt extent maps for Greenland that are updated faster than the mass balance records.

Is 6.9 trillion tonnes a lot compared to the total ice sheets?

The Greenland Ice Sheet holds about 2.85 million trillion tonnes, so the 1992–2020 loss is roughly 0.14% of that reservoir. The scale comparison cuts both ways: it reassures that the ice sheet is not disappearing quickly, and it warns that even a small percentage change in a very large reservoir is a large absolute mass.

Could the losses have been caused by something other than warming?

The patterns — coastal thinning, grounding line retreat, surface melt increasing in step with atmospheric temperature records, and the correlation with known El Niño variability — are consistent with warming-driven loss. Natural variability alone would not produce the sustained trend. Attribution studies separate these causes, but the satellite record is not by itself a causal proof.

What is the single best number to remember from the satellite record?

About 21 millimeters of sea level rise from the two ice sheets over 1992–2020, from roughly 6.9 trillion tonnes of combined loss, is the most-cited summary. It is the cumulative total, not the current annual rate, and the current rate contributes roughly 1.4 millimeters per year in the late 2010s.

This article is general educational information about climate science and satellite observation. It is not a prediction of future sea level at any specific location, and site-specific flood risk should be assessed with a qualified local coastal engineer or planning authority.