Synthetic Aperture Radar — How a Moving Antenna Sees What No Camera Can

At any given moment, roughly 60 percent of Earth’s land surface is obscured by cloud cover. Add nighttime darkness, and the fraction of the planet inaccessible to optical satellites at any one instant is consistently above half. Optical sensors work exactly like cameras — they detect sunlight reflected from the surface. When there is no sunlight, or a cloud intercepts it first, the image is blank.

Synthetic aperture radar is the engineering response to that constraint. A SAR satellite does not wait for sunlight. It generates its own illumination — microwave pulses transmitted toward the surface at frequencies between roughly 1 and 12 gigahertz — and measures the energy that scatters back to the antenna. Microwaves at these frequencies pass through cloud water droplets, fog, smoke, and light rain with minimal attenuation. The sensor works at 2 a.m. over a monsoon-covered delta with the same fidelity as noon over the Sahara.

The physics behind how SAR achieves its resolution, however, is genuinely counterintuitive. The resolution of a real aperture radar improves as the antenna grows larger. SAR inverts this relationship entirely. The smaller the physical antenna, the finer the detail the system can resolve in one dimension. Understanding why requires following the signal from transmission to image.

Key parameters

Band	Wavelength	Frequency	Best commercial resolution	Penetration
X-band	~3.1 cm	8–12 GHz	16 cm (ICEYE Gen4)	Canopy surface only
C-band	~5.6 cm	4–8 GHz	5 m (Sentinel-1C)	Light vegetation, atmosphere
L-band	~23.5 cm	1–2 GHz	~3 m (ALOS-2)	Vegetation, dry soil
P-band	~69 cm	0.3–1 GHz	~6 m (BIOMASS, 2025)	Forest structure, ice sheets

---

What Passive Sensors Cannot See

Optical Earth observation satellites are passive instruments. They carry no illumination source. Landsat, Pléiades, WorldView — all of them record reflected sunlight. The moment clouds intervene, or the orbit crosses into Earth’s shadow, the acquisition is lost.

This matters operationally. The applications where timing is most critical — flood mapping after a cyclone, oil spill detection following a maritime accident, crop damage assessment, military movement monitoring — are precisely the applications where weather and time-of-day constraints are most likely to block an optical pass. A 50-cm optical satellite over cloud-covered coastline in a tropical storm produces nothing. A SAR satellite in the same geometry produces a usable image.

SAR’s all-weather capability is not a marginal advantage. It defines which problems radar-based Earth observation can solve that optical systems structurally cannot. For persistent maritime surveillance, where areas of interest are wide and revisit windows are brief, optical and SAR are not interchangeable tools — they are sensors with different operational envelopes.

The Counterintuitive Physics of the Synthetic Aperture

A real aperture radar resolves two closely spaced targets in the along-track direction if their angular separation exceeds $\theta = \lambda / D$, where $\lambda$ is wavelength and $D$ is antenna length. To achieve 1-metre along-track resolution at 500 km altitude with a 3-cm wavelength, the antenna would need to be approximately 15 kilometres long. That is not buildable.

SAR resolves the problem by exploiting the satellite’s own motion. As the satellite moves along its orbit, the antenna illuminates each ground target from a sequence of progressively changing angles over several seconds. The radar records the return signal — magnitude and phase — at each position. Signal processing then coherently combines all those returns as if they had been collected simultaneously by a single, physically large antenna. This is the synthetic aperture: a virtual antenna whose length equals the distance the satellite travelled while the target was illuminated.

The counterintuitive result: the achievable along-track (azimuth) resolution of a SAR is:

$\delta_{az} = \frac{D}{2}$

where $D$ is the length of the physical antenna. A shorter antenna illuminates each target over a wider range of angles, creating a longer synthetic aperture and finer resolution. Larger physical antenna, worse SAR resolution. The relationship inverts.

Range resolution — the ability to distinguish two targets at different distances along the radar line of sight — is governed by a separate formula and a separate variable:

$\delta_r = \frac{c}{2B}$

where $c$ is the speed of light and $B$ is the transmitted pulse bandwidth. For 600 MHz bandwidth — the figure Capella Space’s Acadia satellites achieve — range resolution reaches 25 centimetres. Both azimuth and range resolution are independent of range to target. A SAR at 500 km altitude produces the same resolution as one at 600 km, all else equal. Optical systems, limited by diffraction, cannot make that claim.

What Frequency Determines

Shorter wavelength means finer potential resolution and weaker surface penetration. Longer wavelength means coarser resolution and deeper penetration into vegetation and soil. Every SAR system trades between these.

X-band, at 3.1 centimetres, sits at the high-frequency end of operational SAR. The short wavelength allows fine azimuth and range resolution — ICEYE’s Gen4 satellites, launched in March 2025, achieve 16 centimetres in spotlight mode, and Capella Space’s Acadia class reaches 25 centimetres. X-band interacts with the top of the vegetation canopy and produces strong returns from metallic surfaces and building edges. It cannot detect what is below the canopy.

C-band, at 5.6 centimetres, is the band of ESA’s Sentinel-1 constellation — the most widely used SAR system in the world. Operating at 5.405 GHz, Sentinel-1C achieves 5-metre resolution in strip-map mode across an 80-kilometre swath, and 20-metre resolution across 400 kilometres in interferometric wide-swath mode. C-band penetrates light vegetation and thin atmospheric layers but does not reach the soil beneath dense forest. For the ESA Copernicus programme, C-band represents the operational balance between resolution and penetration that supports the widest range of civilian applications.

L-band, at 23.5 centimetres, penetrates dense forest canopies and dry soils to depths that X and C cannot reach. JAXA’s ALOS-2 satellite uses L-band for forest structure mapping, soil moisture retrieval, and subsurface geology. ESA’s BIOMASS satellite, launched in April 2025, uses P-band — 69 centimetres — specifically to measure forest above-ground biomass through the entire forest structure, a measurement that requires penetrating to the forest floor.

What the Backscatter Reveals

A SAR image does not look like a photograph. It is a map of radar backscatter — the fraction of transmitted microwave energy that returns to the antenna from each resolution cell on the ground.

Rough surfaces scatter energy in many directions, returning a significant fraction to the sensor. Vegetation canopies, urban surfaces, and rocky terrain appear bright. Smooth surfaces — calm water, asphalt roads, airport runways — act as specular reflectors, directing energy away from the antenna at the angle of incidence. They appear dark. Urban structures produce characteristic “double bounce” returns when the radar pulse reflects off a vertical wall and then off the adjacent ground — or vice versa — forming a dihedral corner reflector geometry that redirects most of the incident energy back toward the sensor with particularly high intensity.

These response patterns encode information that optical imagery cannot capture. An oil slick damps the small-scale roughness of the sea surface, appearing as a dark patch against the surrounding ocean backscatter. Ship hulls produce strong point-like returns. Flooded areas under forest canopy — invisible to optical sensors — produce distinctive double-bounce signals as microwave pulses reflect off the water surface and tree trunks. These are measurements, not images in the photographic sense.

InSAR — When Phase Difference Becomes a Measurement

Backscatter intensity is one dimension of the SAR signal. Phase is another. Every return carries a phase value determined by the total path length the microwave pulse travelled from antenna to target and back. When two SAR passes image the same area from nearly identical geometries, the phase difference between the two acquisitions encodes any change in that path length — including surface displacement.

The relationship is:

$\phi = \frac{4\pi}{\lambda} \cdot \Delta r$

where $\phi$ is the phase difference in radians, $\lambda$ is wavelength, and $\Delta r$ is the change in range (line-of-sight displacement) between passes. For Sentinel-1C at 5.6-centimetre wavelength, a displacement of 2.8 centimetres along the radar line of sight produces a phase shift of one full cycle ($2\pi$ radians). This is the system’s native sensitivity unit.

Persistent Scatterer InSAR (PS-InSAR) extends this by identifying stable radar reflectors — individual buildings, exposed rock faces, infrastructure — that maintain consistent backscatter characteristics across dozens or hundreds of acquisitions spanning years. By tracking phase evolution at these persistent scatterers through a time series of interferograms, displacement rates of 1 to 2 millimetres per year become measurable from orbit. A satellite at 700 km altitude, moving at 7.5 km/s, detecting subsidence at the scale of a coat of paint per year.

Applications span earthquake co-seismic deformation, pre-eruptive volcanic inflation, dam and bridge structural monitoring, groundwater extraction-driven subsidence, and geological carbon sequestration reservoir monitoring — any phenomenon that displaces the ground surface at rates above the millimetre-per-year threshold.

The Commercial Resolution Race

For three decades, high-resolution SAR was the domain of government programmes. ESA’s ERS-1 launched in 1991. Sentinel-1A launched in 2014. Access was free but revisit was limited — with a two-satellite Sentinel-1 constellation, the equatorial revisit period is six days.

Commercial SAR changed the architecture of access. ICEYE, founded in Finland in 2014, launched its first satellite in 2018 and operates more than 30 X-band SAR satellites as of 2025. Its Gen4 satellites, launched in March 2025, produce 16-centimetre resolution images and can capture up to 500 scenes per day per satellite. Capella Space’s Acadia-class satellites deliver 25-centimetre resolution with 600 MHz bandwidth. Both constellations achieve sub-hourly revisit at specific latitudes through orbital phasing.

The SAR satellite market was valued at 4.6 billion dollars in 2025 and is projected to grow to 9.8 billion by 2034. The driver is not resolution for its own sake. It is the convergence of sub-metre resolution, all-weather availability, and commercially accessible revisit rates — a combination that makes SAR viable for applications that previously required classified government systems or were simply not possible.

The physics has not changed since SEASAT first demonstrated spaceborne SAR imaging in 1978. What changed is who can buy it.