Light reflected from a rock carries its composition. Just split that light into colors, like a rainbow, and look carefully for missing shades. That's how hyperspectral imaging works: the camera takes hundreds of pictures of the same spot, each in a very narrow wavelength range. Minerals absorb light differently, and these 'dips' in the spectrum reveal what we're looking at.

Scientists took a thin slice, just 3 millimeters, of the lunar meteorite Bechar010, found in Algeria, and imaged it under a microscope with such a camera. The result was a data cube: 791 by 1024 pixels and 224 color layers. A computer trained on samples with known compositions learned to distinguish olivine from pyroxene with 93.7% accuracy. Olivine is a mineral rich in magnesium and iron, common in the lunar highlands. Pyroxene is more often found in the dark 'maria'—frozen lava plains.

Then the same method was applied to images of the Moon itself, taken from Earth through an amateur telescope, a Celestron 8SE. One pixel of such an image covers 3 kilometers of the lunar surface. After processing, the mineral map matched data from India's Chandrayaan-1 orbiter by 88%. This means we can map the Moon's composition from Earth without waiting for expensive space missions.

The lunar meteorite Bechar010 became a bridge between the lab and the sky. What the camera saw in a rock slice, the telescope confirmed on the whole Moon. Now this approach can be directed at other bodies in the solar system, and we can produce detailed mineral maps without leaving Earth.