SWOT: A Small Height Difference and a Demanding Reference

A fine feature in a satellite map of sea surface height raises a difficult question: is the feature part of the ocean, or part of the measurement error? Putting another instrument in the sea does not automatically settle it. The independent instrument also has uncertainties. A 2025 study validating SWOT's KaRIn measurements confronted this problem directly. It compared observations from orbit with an ocean-based reference while retaining the possibility that the reference contributed to their disagreement. That is a more informative test than treating a detailed map as proof of its own accuracy. [1]

SWOT launched on December 16, 2022. Its Ka-band radar interferometer provides two-dimensional water-surface altimetry on two 50-kilometer swaths separated by a 20-kilometer central gap. With an information cutoff of October 13, 2026, this article examines the technology and the open-access validation study by Jinbo Wang and colleagues. The relevant quantity is surface elevation and its variability at specified spatial scales, not seabed depth or water chemistry. Keeping that measurement target explicit helps explain what the independent comparison was designed to establish. [1]

Geometry is part of the sensor

KaRIn's antennas are separated by approximately 10 meters. Echoes from the same water location arrive with a slight phase difference; that difference, the antenna separation and radar wavelength contribute to the height calculation. The measurement therefore depends on known geometry, not radar power alone. Spacecraft orientation matters as well. A change in the instrument's configuration must not be mistaken for a real slope in the water surface. [2]

NASA's technology-development account shows why that geometry required a substantial engineering history. The deployable antennas needed structural flatness and thermal stability. A full-scale reflectarray prototype developed in 2008 became a basis for the later design. Another project in 2010 addressed deployable masts, hinges, latches and pointing adjustment. The baseline is consequently more than a number in an equation: it is a relationship that hardware must establish after deployment and maintain during measurement. [3]

This illustrates a general challenge of precision sensing. When the target signal is small, an instrument's own changes can compete with the changes it is intended to observe. Mechanical stability is not merely protection against damage. It supports the physical model through which the signal receives its meaning. The antenna-development work made the intended measurement geometry practical, connecting a mathematical relationship to a structure that could operate in space. [2][3]

Receiving, processing and transmitting information

The development history also includes microwave integrated-circuit electronics and an FPGA-based onboard interferometric radar processor. One objective was matching radar output to available downlink capacity. A sensor producing more information does not imply that the complete mission can transmit all that information automatically. Processing and communication belong to the route from electromagnetic echo to usable science data. [3]

Receiver development addressed low power consumption alongside precision over a wide bandwidth. These are different design concerns, rather than alternative names for sensitivity. The mission must receive a signal, process it and deliver data for analysis within its resource limits. Judging the technology solely by an antenna or detector would leave out the steps that carry the measurement to users. The engineering chain explains why a remote-sensing capability can take years of complementary developments to realize. [3]

An orbit chosen for comparison

The validation paper distinguishes two sampling phases. From March 30 to July 10, 2023, the calibration and validation phase used a one-day-repeat orbit. Selected crossover locations could be sampled twice daily, accumulating comparison data relatively quickly. This did not mean that every place on Earth was observed twice a day. Increased temporal sampling came at the expense of broad spatial coverage. [1]

The science orbit subsequently provided near-global coverage between approximately 78 degrees north and south over a 21-day repeat cycle. This change in the balance of time and area affects interpretation. An individual pass concerns a particular observation time; a map assembled over a cycle combines passes obtained at different times. Fine spatial structure is not the same as a simultaneous picture of every structure around the world. Nor does seeing detail establish that it remained unchanged between observations. [1]

A surface reference built from water below it

The field array was located about 300 kilometers west of Monterey, California, in the open ocean. Eleven profiling moorings were spaced approximately 10 kilometers apart. Wave-powered profilers measured temperature and salinity in the upper 500 meters. Total water depth in the region was about 4,500 meters. The reference thus did not directly sample the entire water column down to the seabed. [1]

Temperature and salinity were used to derive density changes. Integrating density anomalies over depth produced a steric height anomaly, which served as a proxy for the dynamic sea surface height anomaly. The independent system was valuable because it constructed a comparable quantity from different physical observations, rather than repeating the radar measurement. Independence, however, does not make a proxy universally applicable or error-free. [1]

The authors explain that upper-ocean steric height may be insufficient in western boundary currents or shallow shelf regions with strong deep barotropic signals. Full-depth temperature and salinity measurements, together with bottom pressure, may be needed there. Success in the California Current validation region therefore cannot be silently extended to every coast and ocean environment. The comparison depends on why this reference quantity is suitable for this region. [1]

Making the two records comparable

Satellite products and profiling observations do not initially provide identical quantities at identical times. The study applied relevant geophysical corrections to satellite height data and retained signals present in the in situ record for a consistent comparison. Temporal means and spatial and temporal linear trends were addressed. Profiler data were matched to satellite pass times, and spatial sampling aligned the satellite observations with mooring locations. These steps defined the variability being tested. [1]

The spectral analysis used 851 paired observations across 85 snapshots meeting the validity criteria. Those numbers describe different parts of the dataset: a snapshot contains multiple pairs. Mooring movement and missing observations also meant the spatial sampling was not a perfectly regular line. The analysis accounted for irregular sampling. Counting each pair as an independent satellite pass would inflate the apparent scope of the evidence and obscure the structure of the comparison. [1]

What a 0.56-centimeter difference means

After spatial and temporal detrending, the reported root-mean-square difference between the two anomaly measurements was 0.56 centimeters. The result concerned spatial scales below 100 kilometers and temporal scales below 90 days. It is not a statement that all raw, absolute satellite heights everywhere have that error. The principal validated spatial wavelength range was 20 to 100 kilometers. A product's grid spacing and the physical scales tested by validation are not interchangeable. [1]

The difference also contains uncertainties from both observing systems. Leaving out contributions below 500 meters, finite profiling speed, gaps, interpolation and time mismatches can affect the reference. Satellite corrections and measurement noise can contribute as well. The researchers did not separately quantify every source's share of the disagreement. Allowing reference error does not undermine the comparison; it prevents a joint difference from being mislabeled as a single instrument's complete error budget. [1]

The evidence supports KaRIn's ability, under this regional comparison and processing framework, to carry small-scale ocean information consistent with the mission requirements. Its technological basis includes maintained two-antenna geometry and a usable data-processing chain. Confidence in the scientific interpretation additionally depends on a carefully matched, physically justified independent reference. The important result is not a claim of perfect measurement. It is strong agreement in a comparison whose scope, transformations and shared uncertainties remain visible. That makes the small structures more than attractive detail on a map: they become observations with a defensible connection to the ocean being studied. [1][3]

Sources

[1] Wang, J. and colleagues (2025). SWOT Mission Validation of Sea Surface Height Measurements at Sub-100 km Scales. Geophysical Research Letters, 52, e2025GL114936; doi:10.1029/2025GL114936. Open access, Creative Commons Attribution. Independently written explanation without quotations; supplementary data and code were not rerun. CC BY 4.0.

[2] NASA/JPL, December 12, 2022, Latest International Water Satellite Packs an Engineering Punch. Prelaunch technical account; the planned launch date is not treated as the actual date.

[3] NASA Earth Science and Technology Office, Early Technology Developments for the SWOT Mission.