Hyperspectral imaging is a way for a satellite to see hundreds of narrow bands of colour, so a diseased crop, a pollutant or a mineral each shows its own signature. It is one of the sensors on an Earth observation satellite.
An ordinary camera records three bands of colour: red, green and blue. A hyperspectral sensor splits the light into hundreds of narrow bands, including colours the eye cannot see. Every material reflects light differently across those bands. A plot of how much light a material reflects at each band is its signature. Healthy leaves reflect much near-infrared light. A stressed or diseased plant reflects less of it and shows a different pattern in other bands.
The signature is the basis of every use. Software holds a library of known signatures, such as healthy wheat, rusted wheat, a certain clay mineral or algae in water. It compares each patch of ground with the library and names the closest match.
Say a water board in Bhopal wants to know which of its lakes have a heavy growth of algae, which makes the water unsafe. A hyperspectral satellite records each lake. Algae has its own signature, and the software reports which lakes match it and how strongly. Sending a boat to take samples from every lake would take weeks.
The same method can find a mineral that a mining company is looking for. It can also measure the carbon in the soil, which helps with soil carbon credits. On a farm, as in the box, it can find disease early, and precision tools such as precision spraying can then treat only that patch.
The pictures are large and need geospatial AI to read. A hyperspectral sensor shows more detail in colour than a normal camera, but its pictures are less sharp, so the two are often used together.