Common Misconceptions Regarding UAV-Collected NDVI Imagery

If you are considering spending several thousand dollars on a proclaimed “Infrared” or “NDVI” camera for your drone, this is a good article to read and consider before doing so.  As you’ll read, not all NDVI images are create equal and the Normalized Difference Vegetation Index was actually created some 40-years-ago to work with images captured by satellites, not within our atmosphere by less precise cameras on drones.

Also note that Near Infrared, the most predominate light wave emitted by healthy plant tissue, images are actually black and white.  The NDVI images you’re accustomed to seeing online are actually created using software after the fact.

The raw image take with a blue notch filter gives a general idea about the scene. Feel free to download and play with all of the vegetation indices you see thrown around.The NIR-VIS index returns a very reproducible vegetation map. Here we've applied a false coloring that scales from green (dense vegetation) to yellow to red to grey (no vegetation). Notice the tractors, houses, and roads have a low signal, while the fields have a higher signal.

The NIR-VIS image is in sharp contrast to the NDVI image. Here, the tractors looks like vegetables! These normalization problems are an artifact of the varied light conditions we experience on Earth that the Landsat simply doesn't have to deal with.

 

 

 

 

 

 

 

The first image shows the raw photo taken by their camera with a blue notch filter, the center shown after it has been altered using Agribotix software using a NIR-VIS index, notice the tractor is black and the vegetation is green. The image on the right was altered using a standard NDVI, you’ll notice the tractor lights up in green as if it were a plant, a common issue found when using NDVI.

 

Content and images credit Agribotix article “Misconceptions about UAV-collected NDVI imagery and the Agribotix experience in ground truthing these images for agriculture” posted June 10, 2014.