The Normalized Difference Vegetation Index, or NDVI, turns two satellite bands into a single number that describes how green and vigorous vegetation looks from space. Healthy leaves absorb red light for photosynthesis and reflect near-infrared light strongly, so the contrast between those two bands reveals plant condition far better than a normal colour photograph. The formula is simple: NDVI equals near-infrared minus red, divided by near-infrared plus red, giving values from minus one to plus one.

With Sentinel-2, the bands you need are B04 for red and B08 for near-infrared, both at 10-metre resolution. With Landsat 8 or 9, the equivalents are Band 4 for red and Band 5 for near-infrared, at 30 metres. Download Level-2 surface reflectance products where possible, because they have already been corrected for atmospheric effects, and check the cloud mask before you trust any pixel. A scene with 12 percent cloud cover may still be usable if the clouds sit outside your study fields, but a thin haze over the whole scene will depress every NDVI value.

Cloud and archive discipline also changes NDVI interpretation more than most beginners expect. Save every NDVI raster with a filename that records sensor, date and product level, such as S2_NDVI_2026-04-18_L2A.tif, and keep a short log of cloud percentage, solar angle and any masking you applied. When a colleague revisits the project six months later, that log is the difference between a reproducible result and a colourful picture nobody can defend. If you compare NDVI across years, use anniversary dates within ten days of each other, because a wheat field in March and the same field in May are different crops in practical terms even though the boundary has not moved.