IBI: Built-Up Mapping That Cancels the Soil and Water Noise
The reason NDBI over-maps cities is that soil and water sneak into the signal. IBI's insight is to build a new index out of three existing ones so those confounders cancel — leaving a cleaner built-up layer.
How it works
IBI combines NDBI (built-up), SAVI (vegetation) and MNDWI (water) into a single normalized index. By setting built-up against the average of the vegetation and water signals, it suppresses the two classes most often confused with urban land, sharpening the boundary NDBI blurs.
The formula
where SWIR1 is shortwave-infrared 1 reflectance (~1.6 µm), NIR is near-infrared reflectance, Red is red reflectance, Green is green reflectance.
Reading the values
> 0 built-up · < 0 non-built-up, with vegetation and water pushed strongly negative — cleaner separation than NDBI alone.
Which bands to use
| Sentinel-2 | SWIR1 B11 · NIR B8 · Red B4 · Green B3 |
| Landsat 8/9 | SWIR1 B6 · NIR B5 · Red B4 · Green B3 |
In practice
It is used where accurate built-up extraction matters — impervious-surface mapping, urban-growth quantification, and land-cover classification across mixed landscapes.
Where it struggles
Being a composite, IBI is only as good as its three inputs; an error in NDBI, SAVI or MNDWI propagates through. It still needs a threshold, so validate against reference points rather than assuming zero is the cut.
Compute IBI on your own study area
Skip the code. Draw or upload a boundary and Spatial Research Suite runs IBI on live Sentinel-2 or Landsat imagery — with cloud masking, exports and citations built in.
Run this analysis in GISforus →Frequently asked
Why is IBI more accurate than NDBI for cities?
It combines built-up, vegetation and water indices so that soil and water — NDBI's main confusers — are cancelled out.
Is IBI harder to compute than NDBI?
Slightly — you first compute NDBI, SAVI and MNDWI, then combine them, so it inherits their band requirements and any errors.