A flood happens under thick cloud, which is exactly when a camera in orbit is useless. The satellites that map it do not use light at all

There is an awkward fact about watching the Earth from space. The events that most need watching — a flood, a cyclone landfall, a landslide during heavy rain — happen under exactly the conditions that blind an ordinary satellite. An optical satellite is a camera. It needs sunlight, and it needs a clear line of sight. During the monsoon it has neither, and the days when a flood map would save lives are precisely the days a camera in orbit returns a photograph of the top of a cloud.

The satellites that map floods solve this by not using light at all.

A camera that brings its own light, at the wrong frequency to be stopped

Synthetic aperture radar is an active instrument. Rather than collecting sunlight reflected off the ground, it transmits its own pulses of microwave energy downward and listens for the echo, measuring how long the return takes and how strong it is.

Two consequences follow, and both are decisive.

It does not need the sun, so it works at night as readily as at noon. And microwaves at the wavelengths used — centimetres, rather than the fraction of a micrometre that visible light occupies — pass through cloud, rain and smoke almost unimpeded. Cloud droplets are far too small to interact strongly with a wave that long. To a radar satellite, a monsoon storm system is very nearly transparent.

That single property is why flood mapping, cyclone monitoring and disaster response depend on radar rather than on photographs.

The trick in the name

Radar resolution normally depends on antenna size: the larger the antenna, the narrower the beam, the finer the detail. At orbital altitude, resolving features a few metres across along the direction of flight would demand an antenna kilometres long, which is not a thing anyone is launching.

The synthetic aperture is the solution, and it is elegant. The satellite is moving fast — several kilometres per second — so a point on the ground stays within the beam for a while, and the radar illuminates it repeatedly from a succession of positions along the orbit. If the returns from all those positions are recorded coherently, preserving phase as well as amplitude, they can afterwards be combined as though they had been collected simultaneously by one enormous antenna as long as the distance the satellite travelled.

The antenna is synthesised from motion and computation instead of being built from metal. Remarkably, the achievable along-track resolution then becomes independent of altitude — which is why this technique works from orbit at all.

Why water is black and a city glows

Reading a radar image is not like reading a photograph, and the difference trips up everybody at first. Brightness in a radar image is not colour or illumination. It is backscatter: how much of the transmitted energy came back towards the satellite.

That depends mostly on roughness, measured against the radar's own wavelength.

Calm water is smooth, so it behaves like a mirror. The pulse strikes it and reflects away at an equal and opposite angle, travelling onward away from the satellite rather than returning. Almost nothing comes back, and still water therefore appears black.

This is the property that makes radar flood mapping close to automatic. A field, a road or a village that was moderately bright last week and is pure black today is under water. No interpretation of subtle colour is required, which is also why the processing can be run quickly enough to matter during an emergency.

Rough surfaces scatter diffusely, sending energy in many directions including back to the sensor, so vegetation and bare rough ground appear mid-grey. Cities are brilliant. A vertical wall meeting a flat road forms a corner that reflects energy back along precisely the path it arrived — a double bounce — which returns a very strong signal, so built-up areas light up.

Radar images also carry speckle, a grainy salt-and-pepper texture that looks like noise and is not. It arises because the radar is coherent: returns from many scatterers within one pixel add together with random relative phases, sometimes reinforcing and sometimes cancelling. It is a real physical effect and it is suppressed by averaging several looks at the same scene, at some cost in resolution.

An optical satellite records how the ground was lit. A radar satellite records how the ground is shaped. That is why one of them stops working in the monsoon and the other does its most valuable work then.

Measuring the ground moving, in millimetres, from orbit

The most striking capability comes from the phase information that ordinary imaging discards.

If the same area is imaged on two passes from very nearly the same orbital position, the two records can be compared phase by phase. Any change in the distance from satellite to ground between the passes shows up as a phase difference — and because the wavelength is only a few centimetres, fractions of a wavelength are detectable. Interferometric SAR can therefore measure ground displacement at the level of millimetres, over whole landscapes, from several hundred kilometres up.

The applications are exactly what you would want. Slow subsidence of land where groundwater has been over-extracted, which is a serious and under-measured problem across northern India. Movement on a hillside before a landslide. Deformation around a fault after an earthquake. Settlement of embankments, dams and mining areas. Measurements of this kind have been used to study subsidence in Indian hill towns and in heavily pumped plains, and the value is that the technique is retrospective — archived imagery lets you ask what the ground was doing for years before anybody was worried about it.

Where it actually fails

Radar is not magic, and an honest account has to include the limits, because flood maps get used for compensation and relief.

Wind roughens water. A lake surface disturbed by a squall scatters energy back and stops being black, so flooding can be missed exactly during the storm that caused it. Dense vegetation hides water beneath it from shorter-wavelength radar, so flooding under a forest canopy or in tall standing crops may not register — longer wavelengths penetrate better, which is one reason L-band instruments are valuable. Urban flooding is genuinely hard: the geometry of buildings produces layover and radar shadow, and the bright double-bounce returns that make cities visible also confuse the interpretation of water between them. Dry smooth surfaces, such as a flat desert or an airport apron, also appear dark and can be mistaken for water by a naive threshold.

These are not reasons to distrust radar flood maps. They are reasons such maps should be produced with ground truth and stated uncertainty rather than treated as photographs of fact.

Why it matters for students and researchers

India has an unusually strong position here. ISRO has flown a series of radar imaging satellites, the national remote sensing infrastructure distributes flood and disaster products during the monsoon, and NISAR — the joint NASA–ISRO mission carrying both L-band and S-band radar, launched in 2025 — is designed precisely for repeat global measurement of deforming and changing land surfaces. Alongside that, freely available radar data from international missions means a student with a laptop and an internet connection can work with the same imagery a disaster response agency uses.

The open problems are well suited to that situation. Automated flood extent extraction that holds up in urban and vegetated terrain remains unsolved in the general case. Radar backscatter is sensitive to soil moisture and to crop structure, which makes it a promising and still imperfect route to agricultural monitoring that works during the growing season regardless of cloud. Subsidence mapping over Indian aquifers is enormously consequential and thinly published. And the operational question — how a raw radar product becomes a map that a district administration can act on within hours — is an engineering and institutional problem as much as a scientific one.

That span, across satellite communication and remote sensing, is the stated scope of the International Journal of Satellite Remote Sensing, a peer-reviewed journal launched in 2023. For electronics, communication and geoinformatics students, radar imaging is a satisfying subject precisely because so little of it is intuitive: the picture looks like a photograph, and almost every instinct you bring from photographs is wrong.

Frequently asked questions

How do satellites take images through clouds?

By using radar rather than light. Synthetic aperture radar transmits microwave pulses and measures the echo, and microwaves at these wavelengths pass through cloud, rain and smoke with very little attenuation. It also requires no sunlight, so it works at night.

What makes it a "synthetic" aperture?

Fine detail along the flight direction would need an antenna kilometres long. Instead, the moving satellite illuminates the same ground point from many positions and combines those recordings coherently afterwards, synthesising the effect of one very large antenna.

Why does water appear black in radar images?

Calm water is smooth relative to the radar wavelength, so it reflects the pulse away from the satellite like a mirror rather than scattering it back. Very little energy returns, so still water appears dark — which makes flood extent easy to detect.

Why do cities appear so bright?

Because a vertical wall meeting a horizontal road forms a corner reflector, sending energy back along the direction it came from. This double-bounce return is very strong, so built-up areas appear much brighter than vegetation or bare ground.

What is InSAR used for?

Comparing the phase of two radar images of the same area taken at different times reveals changes in distance to the ground at millimetre scale. It is used to measure land subsidence, landslide movement, earthquake deformation and settlement of dams and embankments.

When does radar flood mapping go wrong?

When wind roughens the water surface so it no longer appears dark, when flooding sits beneath dense vegetation that shorter-wavelength radar cannot penetrate, in urban areas where building geometry causes layover and shadow, and where dry smooth surfaces are mistaken for water.