Skip to content
ZINDORA
Insights

Earth Observation

The role of SAR in all-weather maritime monitoring

Radar sees through cloud and works at night, which on a monsoon coast is the difference between a monitoring capability and a seasonal one.

7 min read

Optical satellite imagery is intuitive: it looks like a photograph, and interpreting it needs no explanation. It also stops working whenever there is cloud, and stops working every night.

On the East African coast, where the monsoon brings extended cloudy periods, an optical-only approach is unavailable for a substantial part of the year. Synthetic aperture radar is what makes continuous observation possible.

Why radar does not care about weather

SAR is an active sensor. It transmits microwave pulses and measures what returns, so it supplies its own illumination and works in darkness. Sentinel-1 operates in C-band, around 5.405 GHz, at a wavelength of roughly 5.6 cm. Cloud droplets are far smaller than that wavelength and scatter the signal very weakly, so the pulse passes through cloud largely unaffected.

The word synthetic refers to how resolution is achieved. A real antenna long enough for fine resolution at orbital altitude could not be flown. SAR instead uses the satellite’s own motion, combining returns collected along the flight path to synthesise a much longer effective aperture. Sentinel-1’s Interferometric Wide swath mode, the default over most coastal waters, images a 250 km swath at roughly 5 by 20 m resolution.

Why a vessel appears at all

A radar image of the sea is an image of surface roughness. The ocean returns energy through Bragg scattering from small wind-driven capillary waves, and in the standard viewing geometry most of that energy scatters away from the sensor, so open water appears dark.

A vessel is different. A metal hull and superstructure present flat surfaces meeting at angles, forming corner reflectors that send energy directly back along the path it arrived from. Against a dark sea, a ship is a bright point.

Detection algorithms exploit exactly this contrast. The standard family is CFAR, or constant false alarm rate: rather than applying a fixed brightness threshold, the detector estimates the local statistics of the surrounding sea clutter and flags pixels that are improbably bright relative to their own neighbourhood. It adapts as conditions change across the scene.

What radar does not do

Presenting SAR as a solution without its limits would be the same error as trusting AIS uncritically.

Sea state cuts both ways. In very low wind the sea becomes specular and dark, which raises contrast but can also produce dark patches that confuse the clutter estimate. In high wind the surface roughens, the clutter floor rises, and smaller targets stop being distinguishable.

Material matters. The strong return comes largely from metal. Wooden and fibreglass craft return far more weakly, and the smallest vessels may not be reliably detectable at all. This is a real constraint for nearshore African waters, where much of the fleet is exactly that kind of vessel.

Artefacts exist. Azimuth ambiguities can place a faint replica of a bright target elsewhere in the image. An unvalidated detector will find things that are not there.

A detection is not an identification. SAR sees an object with a size and a position. It does not see a name, a flag or an intent.

Revisit, and why it constrains everything

How often a sensor returns sets the ceiling on what can be observed. Sentinel-1A launched in 2014 and Sentinel-1B in 2016, giving a two-satellite constellation until Sentinel-1B suffered a power system failure in December 2021. For the following three years, coverage relied on a single satellite. Sentinel-1C launched in December 2024 and restored the pair.

Even at its best this is a sampling of a continuous process. Radar gives repeated, weather-independent snapshots. Between passes, nothing is observed at all. Any honest maritime product has to be explicit that it reports on observed moments, not on continuous presence.

The practical conclusion

Radar and optical imagery are not competitors and should not be ranked. They answer different questions and fail under different conditions, which is precisely why using both is worth the additional complexity.

For a monitoring system on a monsoon coast, radar carries the detection task because it is the only modality that works every day. Optical corroborates and characterises when the sky allows.