Sensors
- Sentinel-1 SAR (C-band, 5.405 GHz): Extra-Wide Swath mode gives 400 km coverage at 20 m resolution; Interferometric Wide Swath gives 250 km at 5x20 m. Backscatter anisotropy and GLCM texture metrics (contrast, homogeneity, entropy) distinguish deformed from level ice. Repeat pass every 6 days over the Arctic with both satellites combined, though actual revisit over a fixed point depends on latitude and orbit geometry.
- RADARSAT-2 SAR (C-band, 5.405 GHz): Fine-Quad polarisation mode at 8 m resolution enables polarimetric decomposition (entropy, anisotropy, alpha angle) that separves volume scattering in ridged ice from surface scattering in level ice. Commercial tasking allows targeted acquisition over specific shipping corridors or operational areas.
- ICESat-2 ATL07 (photon-counting lidar, 532 nm): Along-track sea-ice freeboard at roughly 17 m along-track posting, with a 91-day exact-repeat orbit. ATL07 provides freeboard height profiles that cross-check SAR-derived sail height estimates and constrain the freeboard-to-thickness conversion for ridged ice, where the isostatic relationship differs significantly from level ice assumptions.
- Operation IceBridge airborne laser altimeter (ATM and Riegl systems): Airborne laser scanning at sub-metre resolution over transects up to roughly 250 km long. Provides the highest-fidelity sail-height and ridge-spacing ground truth available, used to calibrate satellite-derived ridge statistics. Coverage is campaign-based and not operationally continuous.
- RADARSAT Constellation Mission (RCM, C-band): Three-satellite constellation achieving daily Arctic revisit above 70°N. Low-Noise and Medium-Resolution modes at 16–30 m resolution support near-daily deformation monitoring during dynamic ice seasons when ridge formation is most rapid.
What a pressure ridge gives away to a radar
A pressure ridge is not simply a pile of ice. It is a mechanically fractured zone where convergent motion has buckled and rafted ice slabs into a sail above the waterline and a keel below it. The keel-to-sail ratio averages around 4:1 for first-year ridges and can exceed 4.5:1 for multi-year ridged ice, meaning a 2-metre sail implies a keel penetrating 8 metres or more. That geometry is invisible to optical sensors and largely invisible to passive microwave, but it is legible to radar.
C-band SAR responds to ridge topography through two mechanisms. First, the rough, faceted surface of a ridge sail produces stronger and more isotropic backscatter than the smoother, wind-glazed surface of level ice. Second, the internal void structure and brine drainage within aged ridged ice introduces volume scattering that alters the polarimetric signature. Analysts use Grey-Level Co-occurrence Matrix (GLCM) texture features, particularly contrast and entropy computed over 7x7 or 11x11 pixel windows, to separate deformed ice from level ice at Sentinel-1 IW resolution. The separation is reliable for ridges with sails above roughly 0.5 metres; shallower deformation blends into the background noise.
Backscatter anisotropy and what it cannot resolve alone
Ridge orientation relative to the radar look direction matters considerably. A ridge running perpendicular to the azimuth produces a strong specular return from its windward face; one running parallel to the look direction presents a much weaker signal. This anisotropy means that a single-pass SAR image can undercount ridges aligned with the flight path by 20 to 40 percent in some geometries, a figure documented in studies using Operation IceBridge laser transects as reference. Ascending and descending pass composites from Sentinel-1 partially correct this, since the two look directions are roughly orthogonal over the Arctic.
Polarimetric data from RADARSAT-2 Fine-Quad mode adds a further discriminant. The Cloude-Pottier entropy (H) and anisotropy (A) parameters separate volume-scattering ridged ice from surface-scattering level ice with fewer look-direction artefacts. In practice, analysts combine single-pass texture metrics with multi-pass coherence: low interferometric coherence between passes separated by a day or two flags areas of rapid deformation, which are also the areas most likely to be generating new ridges. The coherence method is sensitive to temporal decorrelation from surface motion, so it works best during the dynamic autumn freeze-up and spring break-up seasons rather than in the stable mid-winter pack.
ICESat-2 as the height arbiter
SAR texture tells you where deformed ice is. ICESat-2 ATL07 tells you how high the sails are. The ATLAS photon-counting lidar resolves freeboard at roughly 17-metre along-track intervals with a vertical precision of a few centimetres over smooth ice, degrading to 10–20 cm over rough ridged surfaces where the photon return is spread across multiple height bins. That precision is sufficient to distinguish ridge sails statistically: a population of ATL07 height profiles over a SAR-classified deformed-ice zone allows analysts to derive the sail-height distribution and estimate the 90th-percentile sail height, which is the operationally relevant figure for under-keel clearance calculations.
The 91-day repeat cycle is the main constraint. ICESat-2 does not provide daily coverage of any given area, and its six beam pairs cover a narrow swath. Over a specific shipping corridor or operational polygon, a usable ATL07 overpass may occur only once or twice per month. The practical workflow fuses infrequent ICESat-2 height calibration with more frequent SAR texture classification, using the lidar data to anchor the SAR-derived deformation index to physical height units rather than treating it as a dimensionless relative score.
Ridge density as a model input, not just a map product
The operational value of ridge mapping extends beyond navigation hazard assessment. Under-ice drag, which governs the momentum exchange between ocean currents and the ice pack, depends on the form drag from keels. Level-ice drag parameterisations underestimate total drag by a factor of two to five in heavily ridged regions, according to published coupled ocean-ice model sensitivity studies. Accurate ridge density, expressed as ridges per kilometre along a transect or as the fraction of grid cells exceeding a deformation threshold, is a direct input to the keel-drag term in models such as CICE and LIM.
SAR-derived ridge density maps at 100-metre to 500-metre posting are compatible with the grid resolution of operational Arctic ocean models. The conversion from SAR texture score to ridge-per-kilometre density requires calibration against airborne or in-situ observations. Operation IceBridge laser altimeter transects, now partially superseded by ICESat-2 but still the densest historical record, provide the calibration dataset for several Arctic sub-regions. Outside those calibrated areas, the density estimate carries an uncertainty of roughly ±30 percent, which is honest to state and still useful for model sensitivity testing.
Limits the data cannot overcome
Three constraints bound what satellite remote sensing can deliver here. First, keel depth is never directly observed from above. It is always inferred from sail height via an assumed isostatic ratio, and that ratio varies with ridge age, snow loading, and brine content. The uncertainty in keel depth from a satellite-derived sail height is typically ±1 to 2 metres for individual ridges, which matters for vessels with under-keel clearance margins in that range.
Second, C-band SAR cannot penetrate snow cover to sense the ice surface directly. A wet snow layer in spring attenuates the signal and suppresses backscatter from ridges, potentially causing deformed ice to be misclassified as smooth. This is a known seasonal bias that analysts must flag explicitly in spring products. Third, the spatial resolution of Sentinel-1 IW mode (5x20 m in range and azimuth) means that isolated ridges narrower than about 20 metres may not be resolved as discrete features, though their aggregate effect on texture metrics is still detectable at the scene level.
Satellize's analytics pipeline for polar clients combines open Sentinel-1 acquisitions with commercial RADARSAT-2 tasking where resolution or polarimetry requirements exceed what Sentinel-1 can provide, following the same open-plus-commercial fusion approach used in the Tonga crop-estimation programme.
Typical figures
| SAR spatial resolution (Sentinel-1 IW) | 5 m (range) x 20 m (azimuth), multi-looked to ~20 m for texture analysis |
| SAR spatial resolution (RADARSAT-2 Fine-Quad) | ~8 m, single-look complex |
| ICESat-2 ATL07 along-track posting | ~17 m; vertical precision ~2–5 cm over smooth ice, ~10–20 cm over ridged surfaces |
| Sentinel-1 Arctic revisit (dual satellite) | 6-day exact repeat; effective revisit 1–3 days at high latitudes due to orbit overlap |
| ICESat-2 repeat cycle | 91 days; sub-cycle passes provide partial coverage at shorter intervals |
| Minimum detectable sail height (SAR texture) | ~0.5 m sail height; below this, deformed ice is not reliably separated from level ice |
| Ridge density product posting | 100 m to 500 m grid, depending on SAR mode and averaging window |
| SAR frequency / polarisation | C-band 5.405 GHz; VV, VH (Sentinel-1); HH, HV, VH, VV (RADARSAT-2 Fine-Quad) |
| Archive depth (Sentinel-1) | From October 2014 (Sentinel-1A launch); Sentinel-1B data 2016–2021 |
| Delivery formats | GeoTIFF deformation-intensity raster, GeoPackage ridge-line vectors, NetCDF for model ingestion, PDF summary report |
Analytics Satellize can run
| Deformed-ice fraction map | GLCM texture classification (contrast, entropy, homogeneity) on Sentinel-1 IW backscatter, threshold-trained against Operation IceBridge or ICESat-2 reference transects | GeoTIFF raster, 100 m posting, updated per available SAR acquisition |
| Ridge density index (ridges per km) | Conversion of SAR texture score to linear ridge density using published calibration relationships from airborne laser altimeter transects | GeoPackage polygon layer with per-cell density attribute, uncertainty band included |
| Sail-height distribution (90th percentile) | ICESat-2 ATL07 freeboard profile statistics extracted over SAR-classified deformed-ice zones | CSV summary table per analysis polygon, with histogram plots in PDF report |
| Polarimetric deformation index | Cloude-Pottier H/A/alpha decomposition on RADARSAT-2 Fine-Quad data, separating volume-scattering ridged ice from surface-scattering level ice | GeoTIFF three-band decomposition layer and classified deformation raster |
| Rapid deformation alert (new ridging events) | Short-baseline InSAR coherence loss between consecutive Sentinel-1 or RCM passes flagging zones of active ice convergence | Alert polygon GeoPackage, delivered within 12 hours of SAR acquisition processing |
| Keel-drag parameter field for ocean-ice models | Ridge density and estimated keel depth converted to form-drag coefficient following published parameterisations (e.g. Tsamados et al. 2014 framework) | NetCDF field on user-specified model grid, with documented uncertainty bounds |
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.