Multispectral imagery is remote-sensing data recorded in several discrete spectral bands—typically a handful, from visible through near- and shortwave-infrared—rather than a single panchromatic channel. Each band measures reflected energy in a defined wavelength range, letting analysts detect materials and conditions invisible to the naked eye.
Why it matters
Different surface materials reflect and absorb light distinctively across the spectrum. By combining bands, GIS and remote-sensing workflows can map vegetation vigour, soil and rock types, water, iron oxides, and clay minerals. This makes multispectral data a workhorse for geological reconnaissance, environmental baselines, and land-cover classification, often at no cost from public satellites.
Concrete example
Sentinel-2 carries 13 bands at 10–60 m resolution. Common combinations:
- Natural colour: B4 (red), B3 (green), B2 (blue).
- Color infrared (vegetation): B8 (NIR), B4, B3.
- NDVI:
(B8 − B4) / (B8 + B4), a normalised vegetation index. - Geological / mineral hints: shortwave-infrared bands B11 and B12 help highlight clays and carbonates; ratios such as B4/B2 flag iron oxides.
Landsat 8/9 OLI offers a comparable band set at 30 m, with a long historical archive.
Multispectral vs hyperspectral
Multispectral sensors record a few broad bands; hyperspectral sensors (such as EnMAP or PRISMA) record hundreds of narrow contiguous bands, enabling finer mineral discrimination at the cost of data volume and complexity.
Common pitfall
Spectral indices are sensitive to atmosphere, illumination, and ground cover. Apply atmospheric correction (surface reflectance products) and remember that NDVI maps vegetation, not rock—dense canopy can mask the geology you are trying to see. Always validate spectral interpretations against ground truth.