Earth Data School/Burn severity — NBR & dNBR
Lesson 6.1 · 15 of 17

Burn severity — NBR & dNBR

After a wildfire, recovery needs a map of how badly each patch burned. Satellites read that from how charred ground reflects infrared light — the headline fire product on the NASA Disasters Portal.

In one lineHealthy plants reflect lots of near-infrared and little shortwave-infrared; char flips that. NBR measures the contrast, and the change in NBR before vs. after a fire (dNBR) tells you how severe the burn was.

From NDVI to NBR

If you've read the NDVI guide, this is its twin. NDVI contrasts near-infrared with red; the Normalized Burn Ratio swaps red for shortwave-infrared (SWIR), the band that lights up over dry, charred ground:

NBR = (NIR − SWIR) / (NIR + SWIR)

Living vegetation: high NIR, low SWIR → high NBR. After a fire: NIR collapses (no healthy leaves), SWIR rises (exposed char and soil) → low NBR.

dNBR — measuring the change

A single NBR doesn't tell you severity; a desert has low NBR without ever burning. The trick is the difference between a pre-fire and post-fire scene:

dNBR = NBR_before − NBR_after

A big positive dNBR = a big drop in greenness = a severe burn. The US Geological Survey bins it into standard classes — unburned, low, moderate, high — which is what fills a burn-severity map.

Play with it

The "before" is a healthy forest. Move the post-fire reflectances — drop the near-infrared (leaves gone) and raise the shortwave-infrared (exposed char) — and watch dNBR climb into the severe classes.

before
after
unburnedlowmoderatehigh
NBR before: NBR after:

Do it yourself

editable · runs in your browser

The honest caveats

  • Use clear scenes. Fresh smoke and cloud bias the bands — take the first clear post-fire image.
  • dNBR is a proxy. Field plots (the Composite Burn Index) calibrate it; thresholds shift by ecosystem.
  • Pre/post timing matters. Too long after, regrowth or rain changes NBR for reasons other than the fire.