Ice-sheet mass balance from satellite gravimetry
Twin satellites ranging each other to millimetre precision detect month-to-month shifts in Earth's gravity field caused by ice-mass loss over Greenland and Antarctica. The technique is the only operational method for whole-ice-sheet mass budgets, but its 300 km resolution floor means it cannot pinpoint which glacier is responsible.
Sensors
- GRACE-FO K-Band Ranging (KBR): Measures inter-satellite distance changes to better than 1 micrometre per second in range-rate; translates to gravity anomalies resolved at roughly 300 km spatial scale with monthly revisit. Launched May 2018, operational to present.
- GRACE (historical archive): Predecessor twin-satellite mission, April 2002 to October 2017. Same KBR measurement principle; provides the 15-year baseline against which current trends are assessed. Monthly gravity field solutions publicly archived.
- ICESat-2 ATL11 (complementary): Photon-counting lidar measuring surface-elevation change at roughly 17 m along-track footprint with ~91-day repeat. Used alongside gravimetry to partition mass change into dynamic ice flow versus surface mass balance, and to constrain firn-compaction corrections.
- Copernicus DEM (GIA correction input): 30 m global digital elevation model derived from TanDEM-X; used as a topographic input for glacial isostatic adjustment models that must be subtracted from the raw gravity signal before any ice-mass interpretation is valid.
How ranging two satellites to a micrometre reveals a melting continent
GRACE and GRACE-FO carry no camera, no radar, no optical sensor. Their instrument is the distance between them. The two spacecraft fly roughly 220 km apart in the same polar orbit. When the leading satellite passes over a region of slightly higher mass, its orbit is perturbed forward; the trailing satellite has not yet reached that mass concentration, so the gap between them changes. A microwave K-band ranging link measures that gap continuously to sub-micrometre precision. Ground processors invert thousands of these ranging observations each month into a global map of Earth's gravity field.
Ice has mass. When a glacier loses mass, the local gravitational pull weakens. A monthly gravity field solution from GRACE-FO will show that weakening as a negative gravity anomaly over Greenland's outlet glacier zones or over West Antarctica's Pine Island and Thwaites drainage basins. Convert the anomaly to equivalent water height, account for the density of ice, and you have a mass-change estimate in gigatonnes per month. The physics is elegant; the data-processing chain is not.
The correction that can swamp the signal
Glacial isostatic adjustment, GIA, is the slow rebound of bedrock that was depressed under the weight of the last glacial maximum ice sheets. That rebound adds mass to the solid Earth beneath Greenland and Antarctica at a rate that the satellite sees as a positive gravity trend. It is not ice. If you do not subtract it, your ice-loss estimate is wrong, possibly by tens of gigatonnes per year in Antarctica where GIA rates are largest and most uncertain.
The problem is that GIA models disagree with each other by amounts that are not trivial relative to the ice-mass signal. Published estimates of Antarctic GIA correction range from roughly 53 to 180 gigatonnes per year depending on the assumed mantle viscosity profile, a spread that directly propagates into uncertainty in the ice-mass trend. This is not a solvable problem with better satellites alone; it requires better seismic constraints on mantle structure beneath the ice sheets. Buyers of gravimetry-derived mass-balance products should always ask which GIA model was applied and what the stated uncertainty is.
The 300 km resolution floor and what it costs you
The spatial resolution of GRACE-FO gravity solutions is fundamentally limited by orbital mechanics and signal-to-noise, not by instrument quality. Meaningful mass-change signals are recoverable only at spatial scales above roughly 300 km. That is larger than most individual outlet glaciers. Jakobshavn Isbræ, one of Greenland's fastest-flowing glaciers and a major contributor to sea-level rise, is perhaps 50 km wide at its terminus. GRACE-FO cannot isolate its contribution; it sees a blurred aggregate of the entire southwest Greenland drainage.
This is where ICESat-2 surface-elevation data earns its keep. By combining the gravimetric mass budget with elevation-change profiles from ICESat-2's ATL11 repeat-track product, analysts can apportion the basin-scale GRACE-FO signal across individual glacier catchments. The combination is more powerful than either dataset alone, but it introduces its own assumptions about firn density and basal melt that must be stated explicitly. No single sensor gives you the full picture.
What the archive already tells us, and what it does not
The combined GRACE and GRACE-FO record now spans more than two decades. Published analyses using this archive, including work by the IMBIE consortium, have documented accelerating mass loss from both ice sheets. Greenland has been losing mass at rates that increased from roughly 34 gigatonnes per year in the 1990s to over 250 gigatonnes per year in the 2010s, though annual variability driven by surface mass balance is large. Antarctica's signal is dominated by West Antarctic losses, with East Antarctica remaining closer to balance, though with significant uncertainty.
What the archive cannot tell you is whether a specific policy intervention or a specific year's weather was responsible for an anomaly in a specific basin. The resolution and the GIA uncertainty together prevent that attribution. The record is authoritative for trend detection at ice-sheet and major-drainage-basin scale. It is not a monitoring tool for individual glacier management or for short-term operational decisions.
Translating gravity data into something a government can use
Most sovereign clients encounter GRACE-FO data indirectly, through sea-level rise projections that feed into coastal infrastructure planning, or through water-resource assessments that treat ice sheets as long-term freshwater reservoirs. The monthly gravity solutions are publicly archived through NASA's Physical Oceanography DAAC and through the GFZ Potsdam processing centre, but raw spherical harmonic coefficients are not a usable product for a planning ministry.
The analytic work lies in selecting the appropriate mascon or spherical-harmonic solution, applying a consistent GIA correction, propagating uncertainty honestly, and presenting trend and variability in terms that map onto a client's decision horizon. Satellize can run that processing chain against the public GRACE-FO archive, combining it with ICESat-2 elevation data where basin-scale attribution is needed. The Overhead column has covered the methodological debates around GIA correction for readers who want the technical background before commissioning a bespoke assessment.
For any government integrating ice-sheet mass balance into a national climate-risk framework, the honest starting point is acknowledging that the gravimetry record gives you the most reliable whole-ice-sheet mass budget available, at a resolution that is useful for strategic planning and genuinely insufficient for glacier-by-glacier attribution.
Typical figures
| Spatial resolution | ~300 km (fundamental limit of gravity field inversion from GRACE-FO orbit) |
| Temporal revisit | Monthly gravity field solutions; some mascon products at sub-monthly cadence with reduced accuracy |
| Ranging precision (KBR) | Sub-micrometre per second in range-rate; translates to ~1 cm equivalent water height at 300 km scale |
| Coverage | Global; polar orbit gives full coverage of Greenland and Antarctica each month |
| Archive depth | April 2002 to present (GRACE April 2002 to October 2017; GRACE-FO May 2018 to present; ~11-month gap between missions) |
| GIA correction uncertainty (Antarctica) | Approximately 53 to 180 Gt/yr depending on mantle viscosity model; dominant source of systematic error |
| Minimum detectable mass change | Roughly 10 Gt/month at basin scale under favourable signal conditions; smaller signals are within noise |
| Complementary lidar (ICESat-2 ATL11) | ~17 m along-track footprint, ~91-day repeat; used for basin-scale attribution and firn correction |
| Data delivery format (public archive) | Spherical harmonic coefficients and mascon grids (NetCDF, HDF5) via NASA PO.DAAC and GFZ ISDC |
Analytics Satellize can run
| Monthly ice-mass anomaly time series | Mascon or spherical-harmonic inversion of GRACE-FO Level-2 gravity fields with selected GIA correction applied | Time-series report (PDF and CSV) showing Gt/month anomaly relative to chosen baseline, with stated uncertainty bounds |
| Decadal mass-loss trend estimate | Linear and seasonal decomposition of combined GRACE and GRACE-FO archive; GIA model sensitivity analysis across published models | Trend report with confidence intervals and GIA sensitivity table; suitable for inclusion in national climate-risk assessments |
| Basin-scale mass-change attribution | Joint inversion combining GRACE-FO mascon solutions with ICESat-2 ATL11 elevation-change profiles and firn-compaction model | GIS layer (GeoTIFF and shapefile) of attributed mass change by major drainage basin; updated quarterly |
| Sea-level contribution estimate | Conversion of ice-mass loss to sea-level equivalent using standard density and area corrections; partitioned between Greenland and Antarctica | Structured data feed (JSON) compatible with coastal infrastructure planning models |
| GIA correction sensitivity assessment | Comparison of published GIA models (e.g. ICE-6G, W12, Purcell) applied to the same GRACE-FO mascon solution | Technical memo quantifying how model choice affects client's headline mass-loss figure; recommended for any sovereign policy use |
Who does the work
We can get this done for you. Satellize runs its own analyst desk and a strong science team. You do not buy a data feed and work out what it means; our people source the imagery, run the analysis described on this page, and hand you the answer with its confidence limits stated. Discuss this requirement.