After a wildfire, land managers need to know which areas burned and how badly, often across thousands of hectares. The Normalized Burn Ratio (NBR) does this from satellite imagery. Comparing NBR before and after a fire, called dNBR, gives a burn severity map used for recovery planning, erosion risk and habitat assessment.

How NBR works

Healthy vegetation reflects strongly in near-infrared (NIR) and weakly in the second shortwave infrared band (SWIR2). Burning flips this: charred, bare ground reflects less NIR and more SWIR2.

NBR = (NIR − SWIR2) ÷ (NIR + SWIR2)

High NBR means healthy vegetation; low or negative NBR means burned or bare land.

dNBR: before minus after

A single NBR image can't tell burned land from naturally bare ground. Subtracting the post-fire NBR from the pre-fire NBR shows change:

dNBR = NBRbefore − NBRafter

Worked examples
SurfaceNIRSWIR2NBRdNBR
Forest before the fire0.450.100.64
Same forest after0.120.22−0.290.93
Woodland before0.400.120.54
Same woodland after a light burn0.300.150.330.21

For the forest: NBR before = (0.45 − 0.10) ÷ (0.45 + 0.10) = 0.64; after = (0.12 − 0.22) ÷ (0.12 + 0.22) = −0.29; dNBR = 0.64 − (−0.29) = 0.93. Values use typical surface reflectance, the share of light reflected from 0 to 1. They're illustrative; real values vary with sensor, season and site.

Reading burn severity

The USGS FIREMON program (Key and Benson) published widely used dNBR classes:

dNBR burn severity classesA scale of dNBR values from minus 0.5 to 1.3 divided into USGS classes: enhanced regrowth below minus 0.1, unburned to 0.1, low to 0.27, moderate-low to 0.44, moderate-high to 0.66, and high above. The forest example at 0.93 falls in high severity; the light burn example at 0.21 is low severity.Regrowth: -0.50 to -0.10RegrowthUnburned: -0.10 to +0.10UnburnedLow: +0.10 to +0.27LowModerate-low: +0.27 to +0.44Moderate-lowModerate-high: +0.44 to +0.66Moderate-highHigh: +0.66 to +1.30High-0.5-0.10.10.270.440.661.3Forest example: dNBR 0.93Light burn example: 0.21
dNBR severity classes after Key and Benson (USGS FIREMON). Negative values mean vegetation is greener than before. Hover a class for its range.
dNBR burn severity classes
dNBRSeverity
Below −0.25High post-fire regrowth
−0.25 to −0.10Low post-fire regrowth
−0.10 to +0.10Unburned
+0.10 to +0.27Low severity
+0.27 to +0.44Moderate-low severity
+0.44 to +0.66Moderate-high severity
Above +0.66High severity

These breaks are a starting point. Severity looks different in grassland, shrubland and forest, so field plots are used to calibrate them for each region.

Which bands to use

Bands for NBR
SatelliteNIRSWIR2Resolution
Sentinel-2B8 (or B8A)B1220 m for SWIR2
Landsat 8 and 9B5B730 m

Pick pre- and post-fire images from the same season and with similar sun angles, so the change you measure is the fire, not the calendar. Many studies use an image soon after the fire and another a year later to see which areas recover.

Limits

  • Water gives misleading NBR values because both bands are close to zero. Mask water first, for example with NDWI.
  • Clouds and smoke must be removed from both images.
  • Sparse vegetation has little to lose, so even severe fires give small dNBR. Relative dNBR (RdNBR) adjusts for this.
  • Severity is not the same as ecological impact. Field checks are still needed.

Frequently asked questions

What's the difference between NBR and NDVI?

Both use NIR. NDVI pairs it with red light and tracks greenness; NBR pairs it with SWIR2 and is more sensitive to charred ground and lost moisture.

How soon after a fire can I map it?

As soon as a clear image is available, often within days. A second assessment a year later shows lasting effects.

Is there free burn severity data?

In the US, the Monitoring Trends in Burn Severity (MTBS) program publishes maps for large fires.

Mapping a fire or its recovery?

We'll map burn severity and track regrowth.

Before-and-after imagery, severity classes and area totals for your land.

Talk to usSee change maps