SAR offset tracking and InSAR glacier surface velocity
SAR interferometry and intensity offset tracking together measure glacier surface velocity across the full speed range, from millimetres per day on alpine glaciers to tens of metres per day on Greenland outlet glaciers, feeding directly into mass-balance and sea-level budgets.
Sensors
- Sentinel-1 A/B/C (C-band, 5.6 cm): 6- or 12-day repeat in standard IW mode at 20 m range resolution; 6-day repeat restored after Sentinel-1C launch in 2023. Short baselines favour coherence on slow glaciers; offset tracking applied where phase wraps or decorrelates on fast-moving ice.
- ICEYE X-series (C-band): Stripmap mode at approximately 3 m resolution; constellation enables same-day or next-day repeat tasking over selected glaciers, giving sub-daily velocity snapshots impossible with single-satellite missions.
- Capella Space (X-band, 3.1 cm): Spotlight mode down to approximately 0.5 m resolution; very short wavelength increases sensitivity to small displacements but also increases decorrelation risk on wet or rough ice surfaces.
- ALOS-2 PALSAR-2 (L-band, 23.6 cm): 14-day repeat; longer wavelength maintains phase coherence through moderate surface change, making it the preferred sensor for temperate glaciers during melt season when C-band loses coherence within days.
Two methods, one velocity field
Measuring how fast a glacier moves from orbit requires choosing between two fundamentally different SAR techniques, and the choice depends almost entirely on how fast the ice is moving. Differential InSAR (DInSAR) exploits the phase difference between two acquisitions of the same scene. One full phase cycle corresponds to half the sensor wavelength of displacement in the line-of-sight direction: roughly 2.8 cm for C-band Sentinel-1, about 1.6 cm for X-band. That extraordinary sensitivity is also the method's weakness. If the glacier moves more than a few centimetres between passes, the phase fringes become too dense to unwrap reliably. On fast outlet glaciers moving metres per day, InSAR phase is essentially uninterpretable.
Intensity offset tracking sidesteps phase entirely. It cross-correlates patches of SAR amplitude between two images and measures how far surface features, crevasse patterns, seracs, medial moraines, have shifted in both the azimuth (along-track) and range (across-track) directions. The two components together give the full two-dimensional horizontal velocity vector projected onto the surface. Precision is limited by image resolution and patch size rather than wavelength: typical accuracy is one-tenth to one-twentieth of the resolution cell, so roughly 1 to 2 m on a 20 m Sentinel-1 image. That is far coarser than InSAR phase, but it works at any speed. In practice, most operational glacier velocity products apply both methods and blend the results: InSAR where coherence holds, offset tracking everywhere else.
What the geometry actually measures, and what it misses
SAR looks sideways, not straight down. Every displacement measurement is a projection onto the satellite's line-of-sight vector, which combines range (horizontal, across-track) and a vertical component determined by the incidence angle, typically 30 to 45 degrees for Sentinel-1 IW mode. Azimuth offset tracking captures along-track motion. Range offset tracking captures across-track plus a vertical component. Combining ascending and descending passes from the same sensor, or fusing two sensors with different look geometries, allows decomposition into east-west and vertical components. North-south motion remains poorly constrained by any polar-orbiting SAR because the satellite flies nearly north-south and has almost no sensitivity to motion in that direction. For glaciers flowing roughly east-west this is manageable; for glaciers flowing north-south it introduces real ambiguity in the derived velocity magnitude.
Vertical motion is a separate complication. A glacier surface can lower due to dynamic thinning, surface melt, or both simultaneously. Offset tracking cannot separate horizontal flow from surface lowering without independent elevation data. For mass-balance work, velocity fields are combined with ice thickness estimates from airborne radar (such as NASA's Operation IceBridge data) and surface elevation change from ICESat-2 or CryoSat-2 altimetry. The velocity field alone is not a mass-balance measurement; it is one input among several.
The melt problem: when the surface decorrelates
Phase coherence requires that the scattering pattern of the ice surface remains stable between passes. Liquid water destroys this. During summer melt, C-band coherence on temperate glaciers can drop to noise levels within 24 to 48 hours. This is not a minor inconvenience; it eliminates InSAR as a usable tool across large parts of the Greenland ablation zone and virtually all of the Alps and Patagonia during the melt season. L-band PALSAR-2 is more tolerant because its longer wavelength penetrates a few centimetres into the ice and samples a more stable sub-surface layer, but it too loses coherence under heavy melt.
Short repeat intervals are the partial answer. An ICEYE or Capella pair separated by hours rather than days can maintain coherence that a 6-day Sentinel-1 pair would lose entirely. Commercial constellations are therefore not just a resolution upgrade; they change which glaciers are measurable at all during summer. The trade-off is cost and spatial coverage. A commercial spotlight acquisition covers perhaps 25 by 25 km. Sentinel-1 IW swaths cover 250 km. Operational glacier monitoring programmes, including the ESA Glaciers CCI and the NASA MEaSUREs ITS_LIVE project, rely on Sentinel-1 for continental coverage and accept that summer velocity maps in temperate regions will have gaps.
From velocity field to mass-balance number
A glacier velocity map is a snapshot of ice flux. To convert it into a mass-balance term, you need ice thickness at a flux gate, usually a cross-section near the terminus or grounding line. Multiply depth-averaged velocity by thickness by ice density and you have a mass flux in gigatonnes per year. The uncertainty chain is long. Velocity measurement errors of 1 to 5 per cent are achievable with good SAR pairs. Ice thickness from airborne radar carries uncertainties of 10 to 30 per cent in many regions because coverage is sparse. Depth-averaging corrections depend on assumptions about the flow regime. The resulting mass-flux estimates for individual glaciers often carry total uncertainties of 20 per cent or more, even when the velocity field itself looks clean.
For sea-level-rise assessments, the numbers that matter most are the large outlet glaciers of Greenland and West Antarctica: Jakobshavn Isbrae, Pine Island, Thwaites. These are also the glaciers where velocity changes are most dramatic and most frequently measured. Jakobshavn's acceleration from roughly 20 m per day in the 1990s to over 40 m per day in the early 2010s, documented using ERS, RADARSAT and ALOS data, is one of the clearest signals of dynamic ice loss in the observational record. Monitoring that kind of change requires consistent, long-term SAR archives, which is why Sentinel-1's open data policy and the continuity provided by Sentinel-1C matter operationally, not just scientifically.
Operational realities for a government buyer
A country with glaciated territory, whether for water-resource planning, hazard assessment, or treaty reporting on climate commitments, needs velocity products that are timely, spatially complete, and traceable to documented methods. The open Sentinel-1 archive, processed through standard offset-tracking pipelines such as GAMMA or ISCE, can deliver monthly velocity mosaics at 200 m to 500 m posting across an entire mountain range for essentially zero data cost. The cost is in processing, validation and interpretation. Commercial SAR adds targeted high-resolution acquisitions over specific glaciers of interest, particularly during events: a surge, a calving episode, a rapid acceleration that warrants emergency characterisation.
Satellize applies offset-tracking and InSAR processing pipelines to both open and commercially tasked SAR data, producing velocity fields, anomaly alerts and trend reports for government clients. The analytic approach is the same one underlying the Tonga crop-estimation programme: open data where it is sufficient, commercial tasking where the science demands it. Buyers should be clear-eyed about what a velocity product is and is not. It is a well-validated, physically grounded measurement. It is not, by itself, a complete glacier health assessment. That requires pairing it with elevation change, terminus position, and where possible, surface mass-balance modelling.
Typical figures
| Spatial resolution (Sentinel-1 IW offset tracking) | 20 m pixel spacing; velocity precision approximately 1–2 m per displacement measurement |
| Spatial resolution (ICEYE / Capella spotlight) | 0.5–3 m; enables sub-metre feature tracking on small or heavily crevassed glaciers |
| Repeat interval (Sentinel-1 constellation) | 6 days at mid-to-high latitudes with A/B/C; 12 days with single satellite |
| Repeat interval (commercial SAR) | Hours to 1 day with tasked ICEYE constellation; days with ALOS-2 PALSAR-2 (14-day standard) |
| Minimum detectable velocity (InSAR phase) | Approximately 1–5 mm per day at C-band; coherence typically lost above ~0.5 m per day at 6-day repeat |
| Minimum detectable velocity (offset tracking) | Approximately 0.1–0.5 m per day at Sentinel-1 resolution; faster glaciers tracked reliably |
| Frequency / wavelength | C-band 5.6 cm (Sentinel-1, ICEYE); X-band 3.1 cm (Capella); L-band 23.6 cm (ALOS-2 PALSAR-2) |
| Archive depth | Sentinel-1 from 2014; ERS/Envisat usable back to 1991 for historical baselines via ESA archive |
| Coverage per acquisition | Sentinel-1 IW swath 250 km wide; ICEYE stripmap ~30 km wide; Capella spotlight ~5–25 km scene |
| Delivery format | GeoTIFF velocity grids (easting/northing components), NetCDF time-series stacks, vector shapefiles for flux gates |
Analytics Satellize can run
| Monthly glacier velocity mosaic | SAR intensity offset tracking (azimuth and range) applied to Sentinel-1 IW SLC pairs; GAMMA or ISCE processing chain | GeoTIFF velocity magnitude and direction grids at 200 m posting, delivered monthly per glacier basin |
| Surge and acceleration alert | Automated change detection on rolling 30-day velocity baseline; threshold exceedance triggers alert | Email or API alert with velocity anomaly map and magnitude estimate within 48 hours of SAR acquisition |
| Seasonal velocity time series | Multi-year stack of 6-day Sentinel-1 offset-tracking pairs; seasonal decomposition to separate winter flow from summer acceleration | Annual report with time-series plots and trend statistics per named glacier or flux gate |
| High-resolution event characterisation | Commercially tasked ICEYE or Capella spotlight pairs at 1–3 m; sub-pixel offset tracking on crevasse patterns | GeoTIFF velocity map at 10–20 m posting, delivered within 24 hours of acquisition pair completion |
| Ice flux estimate at terminus or grounding line | Velocity field combined with client-supplied or published ice-thickness data (e.g. BedMachine); flux-gate integration | Tabular flux estimate in Gt per year with documented uncertainty range; suitable for treaty or IPCC reporting |
| Long-term velocity trend (decadal) | Historical baseline from ERS/Envisat/ALOS archive merged with current Sentinel-1 time series; linear and non-linear trend fitting | Trend report with velocity anomaly maps for selected epochs and statistical significance assessment |
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.