Significant wave height and swell propagation in polar seas
Shrinking sea ice opens vast new ocean fetches in the Arctic and Southern Ocean, driving wave heights that infrastructure and shipping routes were never designed for. Radar altimeters and SAR wave-mode imagery now give us the numbers.
Sensors
- Sentinel-6 Michael Freilich (Poseidon-4 altimeter): Dual-frequency (Ku and C band) radar altimeter measuring significant wave height (SWH) along a 1,336 km repeat ground track with 10-day exact-repeat cycle. SWH precision is approximately 0.25 m at open-ocean wave heights above 1 m; degrades near ice edges where return waveforms are contaminated by mixed surfaces. Provides the primary operational SWH record from 2020 onward.
- Jason-3 (Poseidon-3B altimeter): Ku/C-band altimeter on the same ground track family as Sentinel-6, providing continuity with the TOPEX/Poseidon record back to 1992. SWH accuracy comparable to Sentinel-6 in open water. Jason-3 remains operational as a tandem asset, giving denser along-track sampling when both satellites are active.
- Sentinel-1 (wave mode, IW and WV sub-modes): C-band SAR operating in wave mode acquires 20 × 20 km imagettes at 5 × 5 m resolution every 100 km along track. Cross-spectral analysis of these imagettes yields directional ocean wave spectra: dominant wavelength (typically resolved for swells above roughly 80 m wavelength), propagation direction, and period. This directional information is fundamentally unavailable from nadir altimetry.
- CFOSAT SWIM (Surface Waves Investigation and Monitoring): A rotating Ku-band scatterometer-altimeter hybrid that measures directional wave spectra at incidence angles from 0° to 10°, filling the gap between altimeter nadir and SAR off-nadir geometry. SWIM resolves dominant swell directions with a 70 km footprint and a roughly 13-day repeat. Coverage extends into polar seas but degrades poleward of about 80°N due to orbital inclination.
Why polar wave climates are changing faster than the models predicted
Significant wave height in the Arctic Ocean has been rising measurably over the past three decades, and the mechanism is straightforward: wave height scales with fetch, and fetch scales with open water. As September sea-ice extent has declined, the Beaufort and Chukchi Seas now expose thousands of kilometres of open water to prevailing winds that previously encountered ice within a few hundred kilometres. A 2014 study published in Geophysical Research Letters, drawing on the Jason altimeter record, found that Arctic SWH increased by roughly 0.5 m over the period 1992 to 2012, with the largest trends in the Beaufort Sea. The Southern Ocean, already the world's most energetic wave environment, is also seeing shifts in swell propagation patterns as the Antarctic sea-ice margin shifts seasonally.
The engineering consequence is serious. Offshore structures, subsea cables, and coastal communities in places like Alaska and Svalbard were sited under wave-climate assumptions that no longer hold. Insurers, port authorities, and Arctic-route planners need quantitative SWH statistics, not qualitative assessments.
What a radar altimeter actually measures, and where it fails
A nadir-pointing radar altimeter fires pulses straight down and measures the shape of the return waveform. In open ocean, the leading edge of that waveform broadens in proportion to the variance of the sea surface, and from that variance the instrument derives SWH. Sentinel-6 Poseidon-4 operates in both low-resolution mode and synthetic-aperture (high-resolution) mode, the latter reducing along-track averaging to roughly 300 m and improving performance in coastal and ice-marginal zones.
The honest limits are worth stating plainly. Altimeters sample a single ground track. Between tracks, which can be 150 km apart at mid-latitudes and converge only slowly toward the poles, there is no direct measurement. Waveform contamination begins when sea ice occupies even a fraction of the altimeter footprint, which is typically 2 to 10 km in diameter depending on wave height and mode. Near the ice edge, SWH retrievals become unreliable and require careful flagging. The 10-day repeat cycle of Sentinel-6 means that a fast-moving storm system may pass through an Arctic basin between overpasses entirely unobserved.
Directional spectra: what SAR wave mode adds to the picture
Knowing that waves are 3 m high is useful. Knowing that a 14-second swell is propagating from 240° and will reach a particular ice edge in 18 hours is operationally decisive. Sentinel-1 wave mode provides that second layer. The instrument acquires small SAR imagettes and, from the spatial autocorrelation of surface roughness patterns within each imagette, produces a two-dimensional image spectrum. After resolving the 180° directional ambiguity inherent in SAR (which cannot distinguish a wave travelling north from one travelling south without additional information), the spectrum gives dominant wavelength, direction, and spectral width.
In polar seas, swell propagation is complicated by sea ice acting as a low-pass filter: short-period wind waves are dissipated rapidly at the ice edge, while long-period swells can penetrate tens to hundreds of kilometres into the marginal ice zone before their energy is absorbed. SAR wave mode imagettes acquired over the marginal ice zone can, under favourable conditions, detect this penetrating swell signal, though ice surface texture can introduce artefacts that require careful quality control. CFOSAT SWIM adds a complementary directional measurement with a wider swath and without the ambiguity problem, at the cost of coarser spatial resolution.
Fusing altimeter tracks into a wave-climate product
A single altimeter gives a one-dimensional SWH transect. Useful statistics require combining multiple satellites and multiple repeat cycles. The standard approach is to bin along-track SWH observations into spatial grids (commonly 1° × 1° or 0.5° × 0.5°) and compute monthly or seasonal means, percentiles, and trends. When Jason-3 and Sentinel-6 operate in tandem on interleaved ground tracks, the combined sampling interval halves to roughly five days, which helps capture seasonal variability in rapidly changing Arctic conditions.
Wave model reanalysis products such as ERA5 from ECMWF provide a continuous gridded background, but their accuracy in polar seas is limited by sparse in-situ validation and by the challenge of representing the ice-edge boundary condition. Altimeter observations are the primary constraint used to correct these models. The practical output for an operator is a gridded SWH climatology with uncertainty bounds, a trend estimate per grid cell, and, where SAR wave mode data are available, a directional swell rose.
Operational limits that buyers should understand before commissioning work
Cloud cover is irrelevant for radar instruments, which is one genuine advantage over optical sensors. But several other constraints matter. First, altimeter coverage above 82°N is sparse: Jason-3 reaches only 66°N, so the high Arctic depends on Sentinel-6 and, for some periods, CryoSat-2 altimeter data repurposed from its primary ice mission. Second, SWH below roughly 0.5 m is at the noise floor of current altimeters in high-resolution mode; this matters less in polar seas, where waves of interest are typically larger, but affects calm-water marginal-ice-zone studies. Third, the SAR wave-mode imagette spacing of 100 km means that spatial gradients in the wave field, particularly the sharp attenuation at the ice edge, are resolved only statistically across many passes rather than in a single overpass.
Satellize processes Sentinel-1 and Sentinel-6 data on open-constellation pipelines and can add commercial SAR tasking where higher-frequency revisit is needed for a specific area of operational interest.
From wave spectra to decisions: what the analytic chain produces
The end product for most clients is not a satellite image. It is a number, or a distribution of numbers, attached to a location and a time. For an offshore operator assessing structural fatigue loads on a platform in the Barents Sea, that means a long-term SWH exceedance curve: the probability that SWH exceeds 8 m, 10 m, or 12 m in any given month. For a shipping company planning Arctic transits, it means a seasonal swell-propagation forecast constrained by the satellite-derived wave climatology. For a coastal engineer in Alaska, it means a trend line showing how the 99th-percentile wave height at a particular shoreline has shifted over the satellite record.
Each of these products rests on the same underlying data pipeline: altimeter SWH retrievals quality-controlled against ice-mask data, gridded and aggregated, then combined with directional spectral information from SAR wave mode where the spatial coverage is sufficient. The honest answer about uncertainty is that polar wave climatology from satellites carries larger error bars than mid-latitude climatology, because the record is shorter, the ice-edge contamination problem is real, and in-situ buoy validation is sparse. Quoting a trend without quoting its confidence interval is not analysis; it is a guess with a satellite attached.
Typical figures
| SWH precision (open ocean) | ~0.25 m (Sentinel-6 Poseidon-4, HR mode); degrades within ~50 km of ice edge |
| Altimeter along-track sampling | ~1 Hz (~7 km spacing) standard; 20 Hz (~350 m) in high-resolution mode |
| Altimeter repeat cycle | 10 days (Sentinel-6 and Jason-3); combined tandem sampling ~5 days |
| SAR wave-mode imagette size | 20 × 20 km at 5 × 5 m resolution; acquired every ~100 km along track |
| Minimum resolved swell wavelength (SAR wave mode) | ~80 m (limited by SAR image spectrum aliasing at short wavelengths) |
| CFOSAT SWIM footprint | ~70 km diameter; directional spectra at 0°–10° incidence angles |
| Latitudinal coverage | Sentinel-6: ±66°; Jason-3: ±66°; Sentinel-1: ±87.5°; CFOSAT: ±80° |
| Archive depth | TOPEX/Poseidon–Jason altimeter SWH record from 1992; Sentinel-1 wave mode from 2014 |
| Delivery formats | NetCDF (altimeter L2/L3), GeoTIFF (gridded climatology), GeoJSON (track-based SWH), CSV exceedance tables |
| Cloud sensitivity | None: all instruments are active microwave; unaffected by cloud or polar darkness |
Analytics Satellize can run
| Gridded SWH climatology with trend and percentile layers | Along-track altimeter retrieval aggregation; least-squares trend fitting per grid cell; bootstrap uncertainty estimation | NetCDF or GeoTIFF climatology grid; PDF summary report with trend maps and confidence intervals by season |
| SWH exceedance probability curves for a named location | Extreme-value analysis (GEV or GPD fitting) applied to multi-mission altimeter SWH time series within a defined radius | CSV table of return-period SWH estimates (1-year, 10-year, 50-year); accompanying uncertainty bounds |
| Directional swell-propagation analysis for a target area | Sentinel-1 wave-mode cross-spectral analysis; 180° ambiguity removal using ERA5 wind-wave model background; swell-rose construction | Directional swell-rose GIS layer; dominant swell period and direction statistics by month |
| Ice-edge wave-attenuation profile | Altimeter SWH transects perpendicular to the ice edge co-registered with sea-ice concentration maps; exponential decay fitting to estimate attenuation coefficient | Attenuation-coefficient raster per season; time-series chart of ice-edge SWH versus distance into marginal ice zone |
| Fetch-change attribution analysis | Open-water fetch computed from passive-microwave ice-concentration maps; correlation with altimeter SWH anomalies at matching locations and lags | Annual report linking fetch increase to SWH trend; suitable for engineering design-basis update or insurance review |
| Seasonal wave-climate monitoring feed | Near-real-time Sentinel-6 L2 SWH ingestion; automated flagging of SWH anomalies exceeding a client-defined threshold within a bounding box | Weekly GeoJSON feed of flagged high-SWH events; optional email alert when threshold is breached |
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.