Often we encounter multi-spectral and hyper-spectral sensors offering multiple NIR bands. For instance, the WorldView 3 sensor has 8 MS bands, out of them, two bands fall within NIR region.

Near-IR1: 770 - 895 nm

Near-IR2: 860 - 1040 nm

Reference (https://www.satimagingcorp.com/satellite-sensors/worldview-3/)

In case of hyper-spectral imagery, there are number of NIR bands with narrower bandwidth. So, in terms of NIR spectral range, there are many choices which apparently makes it hard to select suitable NIR band to compute spectral indices like NDVI.

What are the conceptual and theoretical basis to select optimal NIR spectral range to compute NDVI?

1 Answer 1


Between 770 and 1040 nm, you are nearly on a plateau in term of reflectance from the green vegetation (in fact, it decreases a little bit when the wavelength increases), so your green vegetation NDVI will be comparable. However, dry vegetation and bare soil usually have an increasing reflectance when the wavelength increases. So, if you want to optimise the discrimination of green vegetation from other land covers, you should use the first NIR (770-895).

In case of hyperspectral data, I would also consider the signal to noise ratio of the bands and make sure to avoid the spectral bands that are strongly affected by the atmospheric gasses (e.g. 1043 nm for ozone). Also note that, with hyperspectral data, there are several indices derived from the "red edge" band.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.