Fire & Snow

BAI: Highlighting Char and Ash by Spectral Distance

Concept article · Updated · by Dr. Anant Kumar Pathak

Instead of a normalized ratio, BAI asks a different question: how spectrally close is this pixel to freshly burned ground? It measures the distance to a charcoal reference point, so recently burned area lights up.

How it reads the surface

The constants 0.1 and 0.06 are the red and near-infrared reflectances of charred surfaces; pixels near that point produce a small denominator and therefore a large BAI. This emphasises the ash and char signal that a normalized index can dilute.

How it is calculated

BAI = 1 / ((0.1 − Red)² + (0.06 − NIR)²)

where Red is red reflectance, NIR is near-infrared reflectance.

What the numbers mean

Typical range: 0 upward (higher = more likely burned)

High values highlight charcoal / ash; near-zero = unburned. It measures spectral distance from a reference "burned" point.

Advertisement

Band configuration

Sentinel-2Red B4 · NIR B8
Landsat 8/9Red B4 · NIR B5

Where it is used

BAI is used for burned-area discrimination shortly after fire, especially in Mediterranean and savanna systems where the char signal is strong and fresh.

Limitations to know

Sensitive to bright soils and less transferable across biomes than NBR.

Because it keys on absolute reflectance values, BAI is sensitive to bright soils and loses skill as ash weathers away — it is less transferable across biomes than NBR, so calibrate expectations per landscape.

Compute BAI on your own study area

Skip the code. Draw or upload a boundary and Spatial Research Suite runs BAI on live Sentinel-2 or Landsat imagery — with cloud masking, exports and citations built in.

Run this analysis in GISforus →

Frequently asked

How is BAI different from NBR?

BAI measures spectral distance to a fixed "charred" reference in red–NIR space rather than a normalized band ratio, which sharpens fresh burn signal but is less transferable.

When does BAI work best?

Shortly after fire, before ash weathers, in landscapes with a strong char contrast such as Mediterranean shrubland and savanna.

Primary reference: Chuvieco, E., Martín, M.P. & Palacios, A. (2002). Assessment of different spectral indices in the red–near-infrared spectral domain for burned land discrimination. International Journal of Remote Sensing 23(23), 5103–5110.