Question: Can we create a guide or set of guides to interpreting infrared or NDVI images?

warren is asking a question about infragram-challenges
Follow this topic

by warren | November 16, 2017 01:23 | #15196

Folks often ask for help in interpreting multispectral imagery (like NDVI images) such as collected using DIY Infragram cameras, and this is just as relevant for satellite-based infrared imagery.

Maybe we could develop guides for things like:

  • comparing agricultural practices
  • identifying groundwater patterns

There are already some good notes on NDVI in agriculture on the Public Lab site but they're not necessarily in step-by-step guide format. And there's tutorials on creating imagery composites, but not on what exactly they mean!

We'd love help in building out this kind of resource!


This would be really helpful to me. I've struggled a lot with the infragram trying to make sense of the processed images, which has as much to do with what you're looking for as it does with what you're looking at.

Katie, it sounds like what would be helpful for you is a better NDVI wiki. That would explain what it can mean when NDVI is high, and why it might be low. I think many people must struggle to understand the NDVI images that are posted at publiclab because most of those images are uninterpretable for several reasons (poor photography, poor processing, presented without a color key, limited context).

Interpreting what is going on in a well made NDVI image is not that hard, although really understanding the biological patterns represented is harder (which is why there are graduate degrees in remote sensing). Because all the DIY NDVI images posted at publiclab have unique stories of place, vegetation, technique, and presentation, each one requires expert scrutiny to extract the meager meaning present. It's a big task to summarize that process. It is a smaller task to summarize the process of interpretation for really good NDVI images. As Jeff suggests, that process is different for agriculture, hydrology, and other investigations (forests, wetlands, turf, soils, etc). So even that is a big job and the result might not be all that applicable to the quality of work being presented at publiclab.


What if we could start by compiling some examples of interpretation (Chris has posted some great ones), and try to note what were the "givens" that enabled the analysis to work. That way, people could look at an example similar to their own site of interest, and at least know what they have to do before they'll be able to do an analysis (i.e. troubleshooting or controlling specific variables like lighting conditions).

I love Chris's list here:

  • poor photography
  • poor processing
  • presented without a color key
  • limited context

Can we expand on these a bit to have a list of things to look out for? There are plenty of examples of people working through these issues through Q&A from the past few years, so maybe we can do better to list out and thoroughly answer peoples questions (if possible) on these subjects while providing more/better guidance on doing the analysis once they've gotten past these issues.

Is this a question? Click here to post it to the Questions page.

Reply to this comment...

Log in to comment