Ice sheet grounding line migration and basal melt detection
The grounding line separating grounded ice from floating ice shelf shifts as ocean heat melts ice from below. Differential InSAR captures the tidal flexure zone that marks this boundary, revealing retreat rates on glaciers like Thwaites and Pine Island.
Sensors
- Sentinel-1 A/B (C-band SAR, ESA): 6-day repeat at high latitudes (12-day single-satellite), IW mode at 5 x 20 m range-azimuth resolution. C-band phase coherence is maintained over cold, dry ice surfaces for months, making it the primary workhorse for current grounding-line InSAR campaigns.
- ERS-1/2 Tandem Mission archive (C-band SAR, ESA): 1-day repeat during 1995-1996 tandem phase, 25 m resolution. The short baseline almost eliminates topographic phase, which is why the original Rignot grounding-line detections on Pine Island and Thwaites used this archive. Irreplaceable for historical baseline.
- Envisat ASAR (C-band SAR, ESA): 35-day repeat, 30 m resolution in Image mode. Bridges the gap between the ERS archive and Sentinel-1, covering 2002 to 2012. Longer repeat introduces tidal aliasing risk that requires careful tidal model correction.
- ALOS-2 PALSAR-2 (L-band SAR, JAXA): L-band (1.27 GHz) penetrates firn and maintains coherence over longer time intervals than C-band on wet or disturbed surfaces. 14-day repeat, stripmap mode at 3 x 3 m (fine beam). Useful for areas where C-band loses coherence near the grounding zone due to surface melt.
What the grounding line is, and why it moves
Where an ice sheet rests on bedrock, it is coupled to the bed by friction. Where the ice becomes buoyant and lifts off, it floats as an ice shelf. The boundary between these two states is the grounding line. It is not a sharp edge but a zone, typically kilometres wide, across which the ice transitions from rigid contact with the bed to elastic flexure driven by ocean tides.
Ocean warming drives basal melt beneath the ice shelf, thinning it. As the shelf thins, buoyancy increases and the grounding line migrates inland, often along a retrograde bed slope where retreat is self-reinforcing. On Thwaites Glacier, published InSAR studies using ERS and Envisat data documented retreat of roughly 1 km per year over parts of the grounding zone between the 1990s and 2010s, with the total grounded area loss running to hundreds of square kilometres. These figures come from peer-reviewed work by Rignot and colleagues published in the open literature; site-specific rates vary and continue to be revised as new data arrives.
The elastic beam and why tides are the signal, not the noise
Tidal loading is usually treated as a nuisance in InSAR. At the grounding line it is the measurement. The floating ice shelf rises and falls with the tide; the grounded ice does not. In between, the ice bends elastically, and the width and shape of that flexure zone encode the ice thickness, the elastic modulus of the ice, and the position of the grounding line itself.
The standard model treats the ice as a thin elastic beam clamped at the grounding line. The characteristic flexure length, sometimes called the elastic length scale, is proportional to the fourth root of ice thickness cubed divided by the elastic modulus and the water density. For typical Antarctic ice shelf thicknesses of 300 to 800 m, this produces flexure widths of roughly 5 to 20 km. A double-difference interferogram, formed by subtracting two interferograms acquired at different tidal phases, isolates the differential vertical displacement across this zone. The landward limit of measurable displacement defines the grounding line position to within roughly one SAR resolution cell, often better than 100 m in practice.
The honest caveat: the method requires that the two interferometric pairs capture a meaningful tidal amplitude difference. If both pairs happen to be acquired near the same tidal phase, the double-difference signal is small and the grounding line position becomes ambiguous. This is the tidal aliasing problem.
Tidal aliasing: the repeat-pass trap
SAR satellites acquire on fixed repeat cycles that have no relationship to the 12.4-hour semidiurnal tidal period. Sentinel-1's 6-day repeat at Antarctic latitudes aliases into a tidal phase offset that is nearly constant for long stretches of time, meaning successive interferograms may consistently sample similar tidal states. When that happens, the differential tidal signal in the double-difference is small, and the flexure zone appears narrow or disappears entirely.
The practical solution is to combine passes from different orbital tracks, different satellites, or different epochs separated by enough time that the aliased phase offset changes. Tidal models such as CATS2008 (a circum-Antarctic barotropic model) are used to predict the tidal height at each acquisition epoch, allowing analysts to select pairs that maximise the differential tidal amplitude and to correct residual tidal phase from the interferogram. Even with good pair selection, tidal model errors of a few centimetres propagate into grounding line position uncertainty of tens to hundreds of metres, depending on local ice geometry. That is a real limit, not a solvable one with current models.
What the archive reveals that a single campaign cannot
Grounding line retreat is slow by human standards but fast by glaciological ones. Detecting it reliably requires comparing positions separated by years to decades. The ERS tandem archive from 1995 to 1996, Envisat ASAR from 2002 to 2012, and Sentinel-1 from 2014 onwards together provide nearly three decades of potential coverage over West Antarctica. Each sensor era has different coherence properties, resolution, and tidal aliasing characteristics, so cross-era comparisons require careful phase calibration and consistent tidal correction.
The archive also captures episodic events. Rapid grounding line retreat associated with ice shelf thinning pulses, or temporary re-advance during periods of reduced ocean heat delivery, appears as a non-monotonic signal in the time series. A single campaign snapshot would misread these episodes as either faster or slower long-term change than is actually occurring. Time-series analysis across the full archive is therefore not a luxury but a methodological requirement.
Resolution floors, cloud and what InSAR cannot tell you
InSAR measures surface displacement in the radar line-of-sight direction. Converting this to vertical displacement requires knowledge of the local incidence angle and an assumption that horizontal motion is either negligible or independently constrained. Near the grounding line, ice is flowing seaward at speeds that can exceed 1 km per year on fast-moving glaciers. That horizontal flow contributes a line-of-sight phase that must be separated from the tidal flexure signal. Ascending and descending pass combinations help decompose the two components, but full decomposition requires at least two independent look directions and is not always geometrically achievable at high latitudes where orbital tracks converge.
InSAR cannot directly measure basal melt rate. It constrains grounding line position and its change over time. Basal melt is inferred by combining grounding line retreat with ice thickness data (from airborne radar such as CReSIS or Operation IceBridge) and a mass-balance model. The InSAR result is one input into a larger chain, not a standalone melt estimate. Cloud cover is irrelevant for SAR, which is one of the few genuine advantages of microwave over optical methods in the Antarctic.
Satellize runs grounding-line InSAR analytics on open Sentinel-1 acquisitions and the ESA archive, applying double-difference processing with CATS2008 tidal correction. The workflow is the same class of analysis used in the peer-reviewed literature, not a proprietary black box. Clients receive georeferenced flexure-zone maps and grounding line position shapefiles with uncertainty envelopes, comparable in structure to the outputs Satellize delivers for its Tonga crop-estimation programme, adapted for polar geodesy.
What a grounding line product actually looks like in practice
The primary deliverable is a grounding line position vector, updated at each available interferometric epoch, with an uncertainty band derived from tidal model residuals and phase noise. Secondary outputs include the width of the flexure zone (which constrains ice thickness when elastic modulus is assumed) and a displacement amplitude map showing peak differential vertical motion across the zone.
For change detection, the time series of grounding line positions is the key product. Retreat rates are expressed in metres per year with confidence intervals. Where the bed topography is known from airborne surveys, retreat can be converted to grounded ice area loss. Where it is not, the linear retreat distance is the honest reportable quantity. Any client commissioning this analysis should expect to supply, or jointly source, co-registered ice thickness data to make the step from geometric change to volumetric change.
Typical figures
| Primary SAR frequency | C-band (5.405 GHz, Sentinel-1); L-band (1.27 GHz, ALOS-2 PALSAR-2) for coherence-challenged surfaces |
| Spatial resolution (processed interferogram) | 20–100 m after multi-looking; grounding line position accuracy typically 50–500 m depending on tidal amplitude and coherence |
| Revisit (Sentinel-1, Antarctic latitudes) | 6 days (dual satellite); single-satellite 12 days. Tidal aliasing means not every repeat is analytically useful |
| Minimum detectable vertical displacement | ~5 mm in a single interferogram under good coherence; double-difference noise floor ~1–2 cm over the flexure zone |
| Archive depth | ERS tandem from 1995; Envisat 2002–2012; Sentinel-1 from 2014. Effective three-decade record for West Antarctica |
| Tidal correction model | CATS2008 circum-Antarctic barotropic model; residual tidal errors typically 2–5 cm, propagating to tens to hundreds of metres in grounding line position |
| Cloud sensitivity | None. SAR is unaffected by cloud or polar darkness |
| Delivery formats | GeoTIFF displacement maps, Shapefile/GeoPackage grounding line vectors, CSV time-series with uncertainty columns |
| Ancillary data required | Precise orbital state vectors (ESA DORIS/POD), DEM for topographic phase removal (TanDEM-X or REMA), ice thickness for melt-rate inference |
Analytics Satellize can run
| Grounding line position map | Double-difference InSAR with elastic beam flexure model; tidal phase correction via CATS2008 | Georeferenced Shapefile with position uncertainty envelope, delivered per interferometric epoch |
| Grounding line retreat time series | Multi-epoch InSAR stack comparison across ERS, Envisat and Sentinel-1 archives | CSV and chart showing retreat distance (m/yr) with confidence intervals per glacier sector |
| Tidal flexure zone width map | Differential displacement profile fitting to elastic beam model; width extracted at half-amplitude | GeoTIFF raster of flexure zone width, usable as proxy for local ice thickness |
| Grounded ice area change | Polygon differencing of grounding line positions across epochs, intersected with bed topography where available | Area-change table (km² per epoch) with source interferogram metadata |
| Tidal aliasing risk assessment | CATS2008 tidal prediction at candidate acquisition epochs; differential amplitude scoring for pair selection | Acquisition planning report identifying optimal pass combinations for a given glacier and time window |
| Line-of-sight displacement decomposition | Ascending/descending pass combination to separate vertical tidal flexure from horizontal ice flow contribution | Separate vertical and horizontal displacement GeoTIFF layers with decomposition uncertainty |
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.