Sea-ice freeboard and thickness from radar and laser altimetry
Radar and laser altimeters measure the sliver of sea ice above the waterline; converting that freeboard to thickness requires assumed densities and carries real uncertainty. This page explains the physics, the sensors, and the honest limits of each approach.
Sensors
- SIRAL / CryoSat-2 (ESA): Ku-band (13.575 GHz) radar altimeter operating in three modes: conventional pulse-limited, SAR (SARIn) and interferometric SAR. Along-track resolution in SAR mode is approximately 300 m; across-track remains several kilometres. Revisit at the poles is roughly 30 days for an exact repeat but the drifting orbit provides denser polar sampling. Primary source of Arctic and Antarctic freeboard records since 2010.
- ATLAS / ICESat-2 (NASA): Photon-counting lidar at 532 nm, six beams in three pairs, 17 m along-track footprint spacing at 10 kHz. Measures ice surface elevation with centimetre-level precision on individual segments, though cloud cover causes complete data loss. Repeat cycle is 91 days; the polar orbit reaches 88° latitude. Complements radar by measuring snow surface rather than penetrating into it.
- SRAL / Sentinel-3A and 3B (ESA/EUMETSAT): Dual-frequency (Ku and C-band) SAR altimeter. Along-track resolution in SAR mode is approximately 300 m. Designed primarily for ocean and ice-sheet elevation, it provides Arctic freeboard data with a 27-day exact repeat and good operational continuity, though its sea-ice freeboard products are considered secondary to CryoSat-2's in scientific literature.
- AltiKa / SARAL (CNES/ISRO): Ka-band (35.75 GHz) altimeter. The shorter wavelength reduces penetration into dry snow relative to Ku-band, making the radar return closer to the true snow surface. Useful for cross-calibration studies and for estimating snow depth on ice when differenced against Ku-band measurements. Orbit inclination limits polar coverage to approximately 81.5° latitude.
The freeboard problem: a small signal carrying large consequences
Sea ice floats with only a fraction of its thickness above the waterline. For typical first-year Arctic ice with a density around 900 kg/m³, freeboard is roughly one-tenth of total thickness. A satellite altimeter measuring 20 cm of freeboard is inferring perhaps 2 metres of ice below, with every centimetre of measurement error amplified by that same factor of ten. This is not a flaw in the method; it is the physics of buoyancy, and it is why altimetry-derived thickness products carry uncertainty ranges that honest users should never ignore.
The conversion from freeboard to thickness requires assuming the densities of both the ice and the snow layer sitting on top of it. Snow density on Arctic sea ice varies from around 200 kg/m³ for fresh snow to over 350 kg/m³ for wind-packed or metamorphosed layers. The snow depth itself is poorly constrained from space. Most operational products use climatological snow-depth and snow-density fields derived from historical in-situ records, such as the Warren et al. climatology, which was built largely from Soviet drifting-station data and may not represent the thinner, wetter snow conditions now common in a warming Arctic. The resulting thickness uncertainty for a single CryoSat-2 grid cell is typically quoted at 0.5 to 1.0 m, even after spatial averaging.
What Ku-band radar actually sees inside snow
CryoSat-2's SIRAL transmits at Ku-band, a frequency at which dry snow is largely transparent. The radar return is expected to come from the snow-ice interface rather than the snow surface, which would be convenient: it would mean the instrument measures ice freeboard directly, without needing to know snow depth. The reality is messier. Wet snow, layered ice lenses within the snowpack, and surface roughness all scatter energy at shallower depths, causing the effective scattering horizon to sit somewhere between the snow surface and the snow-ice interface. This penetration uncertainty shifts estimated freeboard, and therefore thickness, by amounts that are difficult to correct without coincident snow measurements.
Ka-band altimetry, as flown on SARAL/AltiKa, penetrates dry snow much less, returning a signal closer to the true snow surface. Differencing Ka-band and Ku-band measurements over the same ice offers a way to estimate snow depth, a technique explored in published literature and operationally promising but sensitive to the spatial and temporal mismatch between two satellites on different orbits. It is not a solved problem.
ICESat-2: photons do not penetrate snow, but clouds stop them entirely
ICESat-2's ATLAS fires 532 nm green laser pulses and counts individual returning photons. At this wavelength, even a thin snow surface scatters almost all energy back immediately, so the instrument measures the snow surface elevation with no penetration ambiguity. Combine ICESat-2 snow-surface elevation with a Ku-band radar measurement of the snow-ice interface and you get a direct snow-depth estimate, which in turn sharpens the freeboard-to-thickness conversion. Several research groups have demonstrated this fusion approach over the Arctic, and it represents the most physically grounded path to reduced thickness uncertainty currently available from space.
The constraint is cloud cover. Arctic clouds are frequent, particularly in summer, and a single opaque cloud layer returns zero photons from the ice below. CryoSat-2's microwave pulses pass through cloud without loss. The two sensors are therefore complementary in a practical operational sense: radar provides all-weather continuity, lidar provides precision when skies permit. Neither alone gives the full picture, and no fusion product should be presented without flagging the cloud-gap fraction in the lidar input.
Resolution floors and the thin-ice problem
Neither CryoSat-2 nor ICESat-2 reliably resolves ice thinner than roughly 0.5 m. This is not primarily a sensitivity limit; it reflects the combination of freeboard measurement noise, the amplification of that noise through the buoyancy conversion, and the spatial averaging required to reduce random error to acceptable levels. Thin ice, typically new or young ice forming in leads and polynyas, is precisely the ice most relevant to heat exchange between ocean and atmosphere: a 10 cm layer of new ice reduces ocean heat loss by an order of magnitude compared to open water, yet it sits below the practical detection floor of current altimeters.
Spatial resolution is a separate constraint. CryoSat-2 in SAR mode achieves roughly 300 m along-track, but the across-track footprint remains several kilometres. Narrow leads, pressure ridges and the ice-edge zone are all undersampled. Products are typically delivered on grids of 25 km or coarser after averaging to reduce noise, which means the thickness of individual floes is not what is being measured. It is a regional mean, and buyers planning route-optimisation for individual vessels should treat gridded thickness products as background context rather than ice-by-ice navigation data.
Operational uses: shipping, climate monitoring and defence awareness
For Arctic route planning, thickness products from CryoSat-2 and ICESat-2 inform decisions about whether a given corridor is likely to be navigable by a vessel of a given ice class. The Northern Sea Route and Northwest Passage both see seasonal variation in multi-year ice fraction that thickness records capture better than extent maps alone. Multi-year ice, which has survived at least one melt season, tends to be thicker and harder than first-year ice of the same extent; a passive microwave extent map cannot distinguish them reliably, but an altimeter-derived thickness product can, at the spatial scales described above.
For climate monitoring, the Arctic sea-ice volume record derived from CryoSat-2 (produced operationally by institutions including the Alfred Wegener Institute and University College London's Centre for Polar Observation and Modelling) provides a physically meaningful metric of ice-system change that extent alone cannot supply. Volume declined significantly over the CryoSat-2 record, though year-to-year variability remains large. Defence and sovereignty applications, including monitoring of exclusive economic zones and contested Arctic passages, benefit from the same thickness context: ice that appears continuous on optical or SAR imagery may be thin enough to be transited by surface vessels or submarines without the signatures associated with breaking heavy multi-year ice.
Satellize can ingest CryoSat-2 and ICESat-2 open-access products, fuse them with Sentinel-3 SRAL continuity data, and deliver thickness and freeboard time series as GIS layers or structured feeds for clients with Arctic operational or policy mandates. The analytical pipeline is the same class of sensor-fusion work that underpins the Tonga crop-estimation programme, applied to a very different physical regime.
What the record does not yet answer
The CryoSat-2 archive begins in 2010. ICESat-2 launched in 2018. The earlier ICESat mission (2003 to 2009) used a different lidar technology with a larger footprint, and bridging the two laser records requires careful cross-calibration. There is no continuous, high-confidence satellite thickness record extending back to the 1980s, unlike the passive microwave extent record. Thickness trends over the period of maximum Arctic change, the 1980s and 1990s, rest primarily on submarine sonar data from declassified US and UK naval transects, which are spatially sparse and not publicly available in full.
Snow depth on sea ice remains the largest single source of uncertainty in current thickness retrievals. Until a dedicated snow-depth mission flies, or until the Ka-minus-Ku differencing approach matures into an operational product with quantified uncertainty, thickness estimates will carry error bars that are uncomfortable for applications requiring precision. Quoting a thickness value without its uncertainty range is misleading. Any product or service that does not surface that uncertainty should be treated with scepticism.
Typical figures
| Along-track resolution (SAR mode) | ~300 m (CryoSat-2 SIRAL, Sentinel-3 SRAL) |
| Along-track footprint (ICESat-2 ATLAS) | ~17 m photon-counting segments; 70 cm beam diameter at surface |
| Across-track resolution | Several km (radar altimeters); ~90 m beam separation within ICESat-2 pairs |
| Typical gridded product resolution | 25 km (CryoSat-2 operational products after spatial averaging) |
| Revisit / repeat cycle | CryoSat-2: ~30-day exact repeat, dense polar sampling; ICESat-2: 91 days; Sentinel-3: 27 days |
| Radar frequency (CryoSat-2 / Sentinel-3) | Ku-band 13.575 GHz; Sentinel-3 adds C-band 5.41 GHz |
| Radar frequency (SARAL/AltiKa) | Ka-band 35.75 GHz; reduced snow penetration relative to Ku-band |
| Lidar wavelength (ICESat-2) | 532 nm (green); no snow penetration; cloud-opaque |
| Minimum reliably detectable thickness | ~0.5 m (practical floor after buoyancy conversion and noise averaging) |
| Archive depth | CryoSat-2 from 2010; ICESat-2 from 2018; SARAL/AltiKa from 2013 |
Analytics Satellize can run
| Arctic freeboard time series | Retracking of CryoSat-2 SIRAL Level-1b waveforms using TFMRA or threshold-first-maximum-retracker-algorithm approaches, following published ESA and CPOM methods | Monthly gridded freeboard GIS layer (NetCDF or GeoTIFF) with per-cell uncertainty estimates |
| Ice thickness with uncertainty bounds | Freeboard-to-thickness conversion applying isostatic balance with Warren climatological snow depth and published density assumptions; uncertainty propagated from snow-depth and density priors | Seasonal thickness maps with explicit uncertainty ranges, delivered as GIS layers or structured JSON feed |
| Snow depth on sea ice | Ka-minus-Ku differencing of coincident SARAL/AltiKa and CryoSat-2 passes over the same ice, following published dual-frequency retrieval approaches | Experimental snow-depth grids with quality flags indicating temporal mismatch and coverage gaps |
| Multi-year versus first-year ice fraction | Thickness threshold classification combined with backscatter texture from Sentinel-1 SAR, cross-referenced against CryoSat-2 thickness products | Weekly ice-type classification layer for specified Arctic sub-regions, GeoTIFF format |
| Route-corridor thickness profile | Along-track extraction of CryoSat-2 and ICESat-2 freeboard/thickness values within a user-defined corridor polygon, with cloud-gap flagging on lidar segments | PDF report and CSV time series covering the corridor for a specified season or voyage window |
| Ice volume anomaly monitoring | Spatial integration of gridded thickness over sea-ice extent mask derived from passive microwave concentration, following CPOM and AWI published volume methods | Monthly volume anomaly relative to the CryoSat-2 baseline period, delivered as a structured data feed or dashboard update |
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.