Glacier surface velocity and ice-flow dynamics
SAR interferometry and optical feature tracking let analysts measure glacier surface motion from centimetres to kilometres per year. Sentinel-1, ALOS-2, Landsat and Sentinel-2 each cover different speed regimes, with honest limits around wet snow, crevassing and cloud.
Sensors
- Sentinel-1 C-band SAR (ESA): 5 m × 20 m ground range resolution in Interferometric Wide Swath mode, 250 km swath, 6-day exact repeat at mid-latitudes (12-day at polar gap edges). Provides interferometric phase for slow-flow ice (mm to ~10 cm per repeat cycle) and amplitude offset tracking for fast outlet glaciers where phase decorrelates. All-weather, day-night.
- ALOS-2 PALSAR-2 L-band SAR (JAXA): L-band (1.27 GHz) penetrates a few metres into dry snow, maintaining coherence over longer time intervals than C-band. Stripmap mode delivers 3 m × 3 m resolution; ScanSAR covers 350 km at 60 m. Repeat cycle is 14 days. Particularly valuable for cold, dry ice sheets where C-band loses coherence.
- Landsat 8/9 OLI (USGS/NASA): 15 m panchromatic band, 30 m multispectral, 185 km swath, 16-day repeat. The primary input to the NASA ITS_LIVE global velocity mosaic via normalised cross-correlation of surface features. Useless under cloud or in polar night; best for ablation-zone feature tracking on temperate glaciers.
- Sentinel-2 MSI (ESA): 10 m visible and near-infrared bands, 290 km swath, 5-day repeat at the equator (more frequent at high latitudes due to orbit overlap). Supplements Landsat in ITS_LIVE processing and enables shorter-baseline optical feature tracking. Cloud cover remains the principal operational constraint.
Two speed regimes, two completely different measurement strategies
Glacier ice does not move at one speed. Slow interior flow on ice sheets or high-mountain glaciers may advance only a few centimetres per day. Fast outlet glaciers, including Jakobshavn Isbræ in Greenland and Pine Island Glacier in Antarctica, can exceed 10 km per year, which is roughly 27 metres per day. That three-order-of-magnitude range means no single technique covers the full spectrum.
Interferometric SAR (InSAR) exploits the phase difference between two SAR acquisitions of the same scene. A displacement of half the radar wavelength (about 2.8 cm for Sentinel-1 C-band) shifts the phase by a full fringe. This is extraordinarily sensitive for slow creep, but it requires the surface to remain coherent between passes. Wet snow, heavy crevassing and rapid surface change all destroy coherence, producing noise rather than signal. When coherence collapses, analysts switch to amplitude offset tracking: cross-correlating patches of SAR intensity imagery to find the shift that maximises correlation. Offset tracking is less precise, typically resolving displacements of around one-tenth of a pixel, but it works on surfaces that would defeat InSAR entirely.
What a floating roof gives away: the physics of SAR coherence
Coherence is the correlation coefficient between the complex SAR signals from two passes. Over stable bare rock it approaches 1.0. Over a fast-moving, heavily crevassed glacier snout it can fall below 0.2 within six days, making InSAR phase uninterpretable. L-band radar (ALOS-2, ~24 cm wavelength) maintains coherence longer than C-band (~5.6 cm) because its longer wavelength is less sensitive to small surface changes. On cold, dry Antarctic ice, L-band 14-day coherence can remain high enough for phase unwrapping. On a temperate Alpine glacier in summer melt, even a 6-day C-band pair may be incoherent in the ablation zone.
The practical consequence: analysts must choose the technique to match the target. A slow-moving ice divide on the Greenland Ice Sheet is a good InSAR candidate. The calving front of Helheim Glacier is not. Getting this wrong wastes processing time and produces misleading outputs, which is why method selection is the first, not the last, decision in any glacier-velocity workflow.
ITS_LIVE: a free global baseline that changes what a custom analysis needs to justify
NASA's Inter-Mission Time Series of Land Ice Velocity and Elevation (ITS_LIVE) project, hosted at the National Snow and Ice Data Center, produces a publicly accessible global mosaic of glacier surface velocities derived from Landsat 4 through 9 and Sentinel-2 imagery using optical feature tracking. The archive extends back to 1985. Annual and seasonal composites are available at 120 m posting for most glacierised regions. This is a genuinely useful baseline: a government or operator asking 'is this glacier accelerating?' can check ITS_LIVE before commissioning anything.
ITS_LIVE has real limits. Cloud cover creates gaps, particularly in maritime climates like Patagonia or Iceland. The 120 m posting smooths out sub-pixel detail. Temporal compositing can obscure short-lived acceleration events, such as the surge onset of a polythermal glacier. And it lags real time by weeks to months depending on the region. Where those gaps matter, higher-resolution SAR-based processing or shorter-baseline optical pairs fill in.
Surge glaciers and calving fronts: where the interesting events happen fast
Glacier surges are episodic instabilities in which a glacier accelerates by one to two orders of magnitude over months to years, then stagnates. Karakoram and Svalbard host a disproportionate fraction of the world's known surge-type glaciers. Detecting surge onset requires short revisit and sensitivity to velocity changes of tens of metres per day. Sentinel-1's 6-day repeat is well matched to this; a fortnightly optical pass often misses the critical early acceleration phase entirely.
Calving-front position is a related but distinct measurement. The terminus of a marine-terminating glacier can retreat or advance by hundreds of metres between satellite passes. Optical imagery at 10 m (Sentinel-2) resolves terminus position to roughly that order, but cloud frequently obscures the scene at exactly the wrong moment. SAR has no such problem. Combining SAR-derived velocity fields with terminus position time series gives the most complete picture of a calving glacier's state.
Honest limits: what satellite velocity data cannot tell you
Surface velocity is not the same as ice discharge. Converting velocity to volumetric flux requires ice thickness, which satellites do not measure directly. Airborne ice-penetrating radar surveys (such as those archived under NASA's Operation IceBridge) provide thickness transects, but coverage is uneven and the data may be years old. Bed topography beneath fast-flowing glaciers is often poorly constrained, introducing the largest uncertainty in any discharge calculation.
Temporal decorrelation over wet surfaces is a hard physical limit, not a processing shortcoming. No amount of algorithmic sophistication recovers InSAR phase that was never coherent. Similarly, optical feature tracking fails in polar night and under persistent cloud. Analysts who present glacier velocity products without stating the coherence mask or the cloud-gap fraction are omitting information a decision-maker needs.
Vertical motion, including dynamic thinning, is largely invisible to along-track velocity measurements. It requires either repeat altimetry (covered separately in the glacier mass balance page in this library) or differential InSAR in ascending and descending geometries, which is achievable with Sentinel-1 but requires careful orbit selection and processing.
From archive to operational monitoring
Most glacier-velocity work today is retrospective: researchers process long archives to understand decadal trends. Operational near-real-time monitoring is rarer but increasingly requested by hydropower operators, national hydrological services and polar logistics planners. A 6-day Sentinel-1 repeat cycle means a fresh velocity estimate is theoretically available every week for most glacierised regions outside the SAR coverage gaps near the poles.
Satellize processes open SAR and optical constellations for analytics clients and can configure glacier-velocity pipelines on the same infrastructure used for its Kingdom of Tonga crop-estimation programme. The methods are published and the input data are free; what varies is the processing chain, the validation approach and the format in which results reach the people who need to act on them. A national water authority wanting monthly velocity anomaly alerts for a set of glaciers feeding a reservoir system is a different brief from a research team wanting a 30-year Landsat time series. Both are solvable with the same underlying data.
Typical figures
| SAR spatial resolution (InSAR mode) | Sentinel-1 IW: ~5 × 20 m ground range; ALOS-2 Stripmap: ~3 × 3 m |
| Optical spatial resolution (feature tracking) | Landsat 8/9 panchromatic: 15 m; Sentinel-2 visible/NIR: 10 m |
| Repeat pass / temporal baseline | Sentinel-1: 6 days (mid-latitude); ALOS-2: 14 days; Landsat: 16 days; Sentinel-2: 5 days |
| Minimum detectable velocity (InSAR phase) | ~1 cm per repeat cycle (C-band, coherent surface); degrades to noise below coherence threshold |
| Minimum detectable velocity (amplitude offset / optical tracking) | ~0.1 pixel per baseline; roughly 0.5–1.5 m per repeat cycle depending on sensor and baseline |
| ITS_LIVE mosaic posting | 120 m; annual and seasonal composites; archive from 1985 (Landsat 4) |
| Swath width | Sentinel-1 IW: 250 km; ALOS-2 ScanSAR: 350 km; Sentinel-2: 290 km; Landsat: 185 km |
| Archive depth | Sentinel-1: from 2014; Landsat: from 1972 (Landsat 1); ALOS-2: from 2014 |
| Principal constraints | Temporal decorrelation (wet/crevassed surfaces, SAR); cloud cover and polar night (optical); ice thickness unknown (limits discharge conversion) |
| Delivery formats | GeoTIFF velocity fields, NetCDF time series, GIS vector terminus positions, tabular anomaly reports |
Analytics Satellize can run
| Surface velocity field | InSAR phase unwrapping (slow flow) or SAR amplitude offset tracking (fast flow), Sentinel-1 or ALOS-2 | GeoTIFF raster of horizontal velocity magnitude and direction, per repeat cycle |
| Velocity anomaly alert | Comparison of current velocity field against rolling baseline; threshold exceedance flagged | Automated alert with anomaly magnitude, affected area polygon and date, delivered by API or email |
| Surge onset detection | Multi-temporal amplitude offset tracking on 6-day Sentinel-1 pairs; acceleration rate computed | Time-series chart of velocity at user-defined centreline points; surge onset date estimate |
| Calving-front position time series | SAR backscatter edge detection or Sentinel-2 NDSI-based terminus delineation | Vector polyline per acquisition date; advance/retreat distance tabulated against baseline position |
| Coherence map | Complex coherence estimation from repeat-pass SAR pair | GeoTIFF coherence layer (0–1) indicating where InSAR phase is reliable; input to method-selection workflow |
| Long-term velocity trend | Optical feature tracking on Landsat archive (consistent with ITS_LIVE methodology); linear and breakpoint trend fitting | Decadal trend report with acceleration/deceleration rates per glacier, GIS layer, annual velocity mosaic stack |
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.