Sea-level rise measurement from radar altimetry
Nadir-pointing Ku-band radar altimeters on a continuous satellite series have measured global mean sea-level change since 1992 at millimetre-per-year precision, revealing thermal expansion, ice-melt contributions, and regional divergence from the global mean.
Sensors
- Sentinel-6 Michael Freilich (ESA/EUMETSAT/NASA/NOAA, launched 2020): Carries the Advanced Microwave Radiometer-C and a Poseidon-4 dual-frequency (Ku and C band) altimeter in open-loop tracking mode. Flies the same 1,336 km, 66-degree inclination reference orbit as the Jason series with a 9.9156-day exact-repeat cycle, delivering sea-surface height (SSH) precision around 2.5 cm per individual measurement, improving to roughly 1.5 mm/year on the global mean trend when averaged over the full mission.
- Jason-3 (OSTM, CNES/EUMETSAT/NASA/NOAA, launched 2016): Operational partner to Sentinel-6, maintaining the reference groundtrack. Poseidon-3B altimeter at Ku and C band; same 9.9156-day repeat. Jason-3 data extend the unbroken TOPEX/Poseidon heritage record through the Sentinel-6 overlap period, enabling rigorous inter-mission bias correction essential for trend continuity.
- SARAL/AltiKa (ISRO/CNES, launched 2013): Ka-band (35.75 GHz) altimeter, the only operational Ka-band altimeter on the reference orbit. Shorter wavelength reduces ionospheric delay and improves coastal and ice-edge performance. Pulse-limited footprint approximately 8 km diameter, smaller than Ku-band instruments, which helps in marginal seas. Flies a 35-day repeat drifting orbit since 2016 after gyroscope failure, reducing systematic sampling bias.
- CryoSat-2 SIRAL (ESA, launched 2010): Synthetic Aperture Radar Interferometric Altimeter operating in Ku band. SAR mode reduces the along-track footprint to roughly 300 m, making it the primary instrument for sea-ice freeboard and polar ocean SSH where conventional pulse-limited altimeters struggle. Its 369-day long-repeat orbit provides dense cross-track sampling useful for mapping regional sea-level gradients, though the long repeat limits temporal resolution for trend monitoring.
What a radar pulse actually measures, and what it does not
A nadir-pointing altimeter fires roughly 1,800 Ku-band pulses per second toward the ocean surface and times the two-way travel of the returned waveform to a precision of a few centimetres. That range measurement, subtracted from the precisely known orbital altitude, gives the sea-surface height above the reference ellipsoid. The ellipsoid is not the ocean's natural resting surface. To recover meaningful oceanography you must subtract the geoid, the gravitational equipotential that defines where a motionless ocean would sit. What remains is dynamic topography: the real signal of ocean circulation, heat content, and long-term sea-level change.
Three corrections dominate the error budget. The ionosphere advances the pulse; Ku and C dual-frequency measurements allow the delay to be estimated and removed. The wet troposphere slows the pulse; an on-board microwave radiometer measures column water vapour to apply a correction that is accurate to roughly 1 cm in the open ocean but degrades near land. The dry troposphere correction comes from meteorological models. After all corrections, a single-pass SSH measurement carries roughly 2 to 3 cm of random noise. The scientific power comes from averaging millions of passes over years, which drives the uncertainty on the global mean trend down to around 0.3 to 0.4 mm/year.
Thirty years of continuity: why the handover between missions matters more than any single satellite
TOPEX/Poseidon flew from 1992 to 2006. Jason-1 overlapped it from 2001. Jason-2 overlapped Jason-1 from 2008. Jason-3 overlapped Jason-2 from 2016. Sentinel-6 Michael Freilich overlapped Jason-3 from 2020. Each overlap period, typically six months to a year on the same groundtrack, allows scientists to measure the inter-mission bias directly from simultaneous observations rather than inferring it from models. Without those overlaps, a bias of even 5 mm between successive missions would corrupt the trend estimate by roughly 0.5 mm/year over a decade, which is larger than the signal of interest in some regional analyses.
The current best estimate of global mean sea-level rise from this merged record is approximately 3.3 to 3.7 mm/year averaged over the satellite altimetry era, with the rate accelerating to around 4.5 mm/year in the most recent decade. Separating this into contributions requires external data: Argo float profiles and GRACE/GRACE-FO gravimetry attribute roughly half the rise to thermal expansion of warming water and the remainder to mass addition from melting ice sheets and glaciers. Altimetry alone cannot make that separation; it measures the sum.
Where the measurement breaks down: the coastal zone problem
Within roughly 10 to 20 km of the shoreline, the standard ocean-mode waveform becomes contaminated. The altimeter footprint, typically 2 to 10 km in diameter depending on sea state and instrument mode, begins to include land returns, which are far stronger reflectors than water. The leading edge of the returned waveform, from which range is extracted, is distorted. Standard Brown-model retracking algorithms, designed for open-ocean waveforms, produce biased SSH estimates in this zone.
Specialised coastal retracking algorithms (ALES, PISTACH, and similar published methods) can recover usable data to within 5 to 10 km of shore in many cases, but accuracy degrades and data gaps increase. CryoSat-2 in SAR mode and Sentinel-6 in high-resolution mode help by reducing the along-track footprint, but they do not eliminate the problem. For small island states, narrow continental shelves, or estuarine environments, tide-gauge records remain essential ground truth. The altimetry record and the tide-gauge record must be reconciled through a common terrestrial reference frame, specifically the International Terrestrial Reference Frame (ITRF), which itself carries uncertainties of a few tenths of a millimetre per year in its vertical component.
Regional sea level diverges sharply from the global mean
The global mean is a useful headline but a poor planning tool. Regional sea-level change can differ from the global mean by a factor of two or more in either direction. The western tropical Pacific has risen two to three times faster than the global mean since 1993, driven by trade-wind intensification and associated heat redistribution. Parts of the northern Atlantic have risen more slowly, or even fallen temporarily, due to weakening of the Atlantic Meridional Overturning Circulation.
Vertical land motion complicates every regional estimate. A coast subsiding due to groundwater extraction or sediment compaction experiences relative sea-level rise that can far exceed the absolute ocean signal. Altimetry measures absolute sea-surface height; it says nothing about what the land beneath a tide gauge is doing. Combining altimetry with GNSS-measured land motion at tide-gauge sites is the standard approach for separating the two contributions, but GNSS station density remains uneven globally.
What analytical products the data support, and what they do not
The merged altimetry record supports trend maps at roughly 1/4-degree spatial resolution, anomaly time series at any point in the open ocean, and detection of interannual variability driven by ENSO, the Pacific Decadal Oscillation, and similar modes. It does not support sub-kilometre coastal inundation modelling directly; that requires the altimetry trend as a boundary condition fed into a hydrodynamic model with high-resolution bathymetry and topography.
For governments assessing infrastructure exposure, the most useful deliverable is a regional trend map with uncertainty bounds, disaggregated by the altimetry-measured absolute signal and the land-motion component from GNSS. Satellize runs this type of analysis against open Copernicus and NASA altimetry products for coastal planning clients, including small island states where the gap between global-mean projections and local reality can be significant. The Kingdom of Tonga crop-estimation programme is a different engagement, but it illustrates the same principle: regional specificity matters more than global averages when decisions are local.
One honest caution: altimetry-derived trends require at least a decade of data to separate a real acceleration from interannual noise. A five-year regional trend extracted from any single altimeter mission should be treated with scepticism unless it is consistent with the longer merged record.
Typical figures
| Primary frequency | Ku band (~13.6 GHz) on TOPEX/Jason/Sentinel-6; Ka band (35.75 GHz) on SARAL/AltiKa |
| Orbital repeat cycle | 9.9156 days (Jason/Sentinel-6 reference orbit); 35 days (SARAL drifting orbit); 369 days (CryoSat-2) |
| Orbital altitude | ~1,336 km (Jason/Sentinel-6); ~800 km (SARAL, CryoSat-2) |
| Along-track measurement spacing | ~580 m at 1 Hz; ~58 m at 20 Hz (Sentinel-6 open-loop mode) |
| Single-pass SSH precision (open ocean) | ~2–3 cm RMS after corrections |
| Global mean sea-level trend precision | ~0.3–0.4 mm/year (merged multi-mission record) |
| Coastal exclusion zone (standard retracking) | ~10–20 km from shore; reducible to ~5 km with coastal retracking algorithms |
| Continuous archive depth | 1992–present (TOPEX/Poseidon through Sentinel-6 Michael Freilich) |
| Standard data products | Geophysical Data Records (GDR) in NetCDF; merged gridded products (CMEMS DUACS) at 1/4-degree, daily |
| Latency (near-real-time products) | ~3–5 hours for OGDR (operational); ~1 day for IGDR; ~60 days for final GDR |
Analytics Satellize can run
| Regional absolute sea-level trend map | Linear trend estimation on merged multi-mission gridded SSH anomaly (DUACS/CMEMS), with seasonal signal removed by harmonic fitting | GeoTIFF trend map with per-pixel uncertainty bounds; PDF summary report for planning authorities |
| Point-location SSH anomaly time series | Extraction from gridded DUACS product at client-specified coordinates; ENSO and PDO index regression to separate forced variability from trend | CSV time series with annotated anomaly events; annual trend update feed |
| Relative sea-level estimate at tide-gauge sites | Differencing of altimetry-derived absolute SSH trend and GNSS-measured vertical land motion at co-located stations, following published IGS/SONEL methodology | Site-specific relative sea-level report with confidence interval; comparison against IPCC AR6 regional projections |
| Coastal infrastructure exposure ranking | Altimetry trend combined with 1-arc-second coastal DEM (SRTM or CoastalDEM) to estimate land area below projected sea-level contours at 2050 and 2100 under IPCC scenarios | GIS polygon layer of exposure zones by scenario; ranked asset table for port and road infrastructure |
| Interannual variability attribution | Regression of regional SSH anomaly time series against ENSO indices (MEI, ONI) and PDO; residual trend isolation following published AVISO methods | Annotated time-series chart distinguishing climate-mode variability from secular trend; briefing note |
| Thermal expansion vs. mass contribution decomposition | Steric sea-level change estimated from Argo float climatology (ISAS or EN4); mass contribution as residual, cross-checked against GRACE-FO mascon products | Stacked bar chart of annual contributions; data table for climate-finance reporting |
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.