Greenland ice-sheet mass balance and outlet glacier dynamics
Three satellite techniques, gravimetry, altimetry and SAR, together constrain how fast Greenland is losing ice and where. Each method has distinct error sources; reconciling them is the science.
Sensors
- GRACE-FO (Gravity Recovery and Climate Experiment Follow-On): Twin satellites measure inter-satellite range changes to the micrometre level using a laser ranging interferometer, resolving monthly gravity anomalies at roughly 300 km spatial resolution. Mass change is expressed in equivalent water height; the Greenland ice sheet signal is detectable at basin scale but not individual glacier scale.
- CryoSat-2 SIRAL (Synthetic Aperture Interferometric Radar Altimeter): Ku-band radar altimeter operating in three modes: low-resolution, SAR and SAR-interferometric. Over Greenland's interior the SAR mode achieves along-track resolution of roughly 380 m; the interferometric mode handles sloping margins. Repeat cycle is 369 days for exact repeats, but the dense sub-cycle coverage gives effective revisit of 30 days or less at high latitudes. Elevation precision over flat ice is a few centimetres.
- ICESat-2 ATLAS (Advanced Topographic Laser Altimeter System): Photon-counting lidar at 532 nm, six beams in three pairs, 91-day exact repeat, 70 cm along-track shot spacing. Over smooth ice the single-pass elevation precision is better than 3 cm. Cloud cover causes data gaps; roughly 30 to 40 per cent of Arctic passes are affected in winter. Provides a direct geometric measurement unaffected by radar penetration into the snowpack, which is a known ambiguity for CryoSat-2.
- Sentinel-1 A/B C-band SAR: Provides 6-day repeat (with both satellites) in Interferometric Wide Swath mode at 5 by 20 m resolution. Used for InSAR-derived surface displacement and intensity offset tracking to measure outlet glacier surface velocity. Phase coherence is lost over rapidly deforming or melting ice, so offset tracking is often preferred at fast-moving termini where velocities exceed several metres per day.
Three methods, three error budgets
No single instrument measures Greenland ice mass directly. GRACE-FO senses the gravitational pull of everything beneath its orbit: ice, meltwater, groundwater, the solid Earth. CryoSat-2 and ICESat-2 measure surface height, which must be converted to mass using an assumed density. Sentinel-1 measures surface motion at the glacier terminus, from which ice discharge can be estimated if ice thickness is known from airborne radar surveys such as NASA's Operation IceBridge.
The value of running all three in parallel is cross-validation. When gravimetric and altimetric mass-change estimates agree within their stated uncertainties, confidence in the total signal is high. When they diverge, it usually points to an unresolved process: variable firn compaction, anomalous snowfall, or an incomplete glacial isostatic adjustment (GIA) correction. Published reconciliation studies typically report Greenland mass loss in the range of 250 to 280 Gt per year averaged over 2003 to 2019, but annual variability is large enough that a single year can swing by 100 Gt or more.
The GIA problem: what the mantle remembers
Glacial isostatic adjustment is the slow rebound of Earth's crust and mantle in response to ice removed since the Last Glacial Maximum, roughly 20,000 years ago. The mantle is still moving. Over Greenland, GIA contributes a positive gravity signal that, if uncorrected, would cause GRACE-FO to underestimate ice loss. The correction is substantial: published GIA models for Greenland range from about 5 to 25 Gt per year in equivalent mass, and the spread between models is the single largest source of uncertainty in gravimetric estimates.
The problem is that GIA cannot be measured directly from orbit; it must be modelled from ice-load history reconstructions and mantle viscosity profiles derived from seismology and GPS uplift data. Different model families, ICE-6G, W12, and others, produce corrections that differ by tens of gigatonnes per year. Any gravimetric mass-balance figure published without a stated GIA model and its uncertainty range should be treated with caution.
Surface mass balance versus dynamic discharge: a distinction that matters
Greenland loses ice through two physically distinct pathways. Surface mass balance (SMB) losses occur when melt and runoff exceed snowfall accumulation; these are driven by atmospheric warming and are distributed across the ice sheet. Dynamic discharge losses occur when outlet glaciers accelerate and calve icebergs into the ocean faster than snowfall can replenish the interior; these are concentrated at a few dozen major outlet glaciers and are driven by ocean warming at glacier termini.
Satellite observations distinguish the two imperfectly. Regional climate models such as RACMO and MAR estimate SMB independently; the dynamic component is then inferred as the residual between total mass change and SMB. Sentinel-1 velocity mapping of outlet glaciers like Jakobshavn Isbræ, Helheim and Kangerdlugssuaq provides a direct check on discharge, but only where ice thickness from airborne surveys is available to convert velocity to flux. Jakobshavn, the fastest large outlet glacier, has at times moved more than 40 metres per day, a rate easily resolved by 6-day SAR pairs.
Radar penetration and the firn ambiguity in altimetry
CryoSat-2's Ku-band radar pulse penetrates the dry snowpack of Greenland's interior by a depth that varies with temperature, grain size and liquid water content. This means the instrument is not always ranging to the true surface; it ranges to a scattering horizon somewhere within the firn column. The penetration depth can reach a metre or more in cold dry conditions, and it changes seasonally. Elevation trends derived from radar altimetry therefore carry a firn-penetration correction that adds uncertainty, particularly over the high-elevation interior.
ICESat-2's 532 nm laser has no penetration issue; it scatters from the surface. The trade-off is cloud sensitivity. A comparison of CryoSat-2 and ICESat-2 elevation change over the same periods and locations is one of the most productive consistency checks available, and systematic offsets between the two instruments in the dry-snow zone are an active area of research. Neither instrument alone is sufficient for a definitive mass-balance estimate.
What an operational monitoring programme looks like
A credible Greenland monitoring workflow ingests GRACE-FO monthly mascon solutions (released with roughly two-month latency by JPL and CSR), CryoSat-2 elevation grids (ESA releases baseline-E products via the Cryoportal), and ICESat-2 ATL06 land-ice height products from NASA Earthdata. Sentinel-1 SAR scenes covering the major outlet glaciers are processed through offset-tracking or InSAR pipelines to produce quarterly velocity mosaics.
The outputs are not a single number. A well-constructed programme delivers basin-scale mass-change time series with explicit uncertainty bounds, velocity anomaly alerts when an outlet glacier accelerates beyond its historical envelope, and calving-front position change derived from SAR amplitude imagery. Satellize's analytics infrastructure runs on open constellations including Sentinel and can integrate GRACE-FO mascon data to produce these time series for government clients who need an independent, non-aligned view of ice-sheet state. The methodology is the same as that used in peer-reviewed literature; the difference is operational delivery rather than research publication.
Honest limits of the current observing system
GRACE-FO cannot resolve individual glaciers; its 300 km footprint blends signals from ice, bedrock and hydrology. A gap of roughly 11 months between the end of GRACE and the start of GRACE-FO in 2018 introduced a discontinuity in the long-term record that required careful bridging using altimetry. If GRACE-FO fails without a successor in orbit, gravimetric mass balance monitoring stops entirely.
ICESat-2's 91-day repeat means that seasonal surface height changes between passes are interpolated, not observed. Rapid dynamic events such as a surge or a sudden terminus retreat can occur and partially recover within a single repeat cycle. Sentinel-1's velocity precision is sufficient for most outlet glaciers, but phase-unwrapping errors in InSAR and speckle noise in offset tracking set practical floors on detectable velocity change of roughly 0.1 to 0.5 m per day depending on scene coherence. Ice thickness from airborne surveys remains the weakest link in discharge calculations; large sections of the bed are sampled at kilometre-scale spacing or coarser.
Typical figures
| GRACE-FO spatial resolution | ~300 km (mass anomaly footprint); individual glaciers not resolvable |
| GRACE-FO temporal resolution | Monthly gravity fields; data latency approximately 2 months |
| CryoSat-2 SIRAL along-track resolution (SAR mode) | ~380 m along-track; effective repeat coverage ~30 days at high latitudes |
| ICESat-2 ATLAS shot spacing and repeat | 0.7 m along-track shot spacing; 91-day exact repeat; 6 beams |
| ICESat-2 elevation precision (smooth ice) | Better than 3 cm single-pass over flat ice; degrades on steep slopes |
| Sentinel-1 IW SAR resolution | 5 × 20 m (range × azimuth); 6-day repeat with two satellites |
| Velocity detection floor (offset tracking) | ~0.1 to 0.5 m/day depending on scene coherence and baseline |
| GRACE-FO archive depth | GRACE 2002–2017; GRACE-FO 2018–present; ~11-month gap bridged by altimetry |
| CryoSat-2 archive depth | 2010–present; Baseline-E reprocessed products available via ESA Cryoportal |
| GIA correction uncertainty (gravimetric estimates) | Approximately 5 to 25 Gt/yr depending on model family; largest single uncertainty term |
Analytics Satellize can run
| Monthly Greenland mass-change time series | GRACE-FO mascon inversion with stated GIA model and uncertainty range | Time-series dataset with confidence intervals, delivered as CSV and GIS-compatible netCDF |
| Basin-scale elevation change maps | CryoSat-2 and ICESat-2 crossover analysis and repeat-track differencing | Quarterly gridded elevation anomaly maps at 1 km posting, GeoTIFF format |
| Outlet glacier velocity mosaics | Sentinel-1 SAR intensity offset tracking and InSAR phase analysis | Quarterly velocity maps for named outlet glaciers; anomaly alerts when velocity exceeds defined threshold |
| Calving-front position change | SAR amplitude image differencing and manual or automated front delineation | Seasonal front-position shapefile with retreat or advance distance relative to baseline year |
| SMB-discharge partition estimate | Residual method combining GRACE-FO total mass change with regional climate model SMB output | Annual report distinguishing surface-driven and dynamic-driven mass loss with stated uncertainty |
| Instrument cross-validation report | Systematic comparison of CryoSat-2 and ICESat-2 elevation trends at coincident locations | Bias and drift assessment report flagging anomalous divergence between sensors |
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.