I have NDVI raster data and vector line data for the St George Queensland irrigation channels.

I am trying to detect irrigation channel leakage using high NDVI values detected on the immediate bank.

How would I identify high NDVI values close to the channel lines in ArcGIS Desktop?

  • 1
    Can you please refine your question into specific questions? What program are you using? What was your approach before asking here? These are very basic tasks that should be answered in most help pages of common software or tutorials and literature.
    – Matte
    Aug 20, 2018 at 6:49
  • 1
    Welcome to GIS SE. As a new user, please take the Tour, which explains how our Focused Q&A model operates. Each question needs to include one task, with complete details of the software in use, steps taken, and problem encountered.
    – Vince
    Aug 20, 2018 at 7:27
  • Good afternoon Luke, I'll answer the easy part. I have ERDAS Imagine 2015, ARCGIS 10.6 and QGIS 3.2. I can also get IDRISI Selva if necessary.Irrigation Channel line data. Not sure what you mean here but I have 3m pixel images of the St George Queensland irrigation area. Neville Brisbane
    – Nifty
    Aug 25, 2018 at 5:43
  • I've made some edits that should allow your Q. to be reopened. Roll back if you disagree. You can ask about exporting pixel coordinates in a separate Q. (or search, pretty sure it's been asked before)
    – user2856
    Aug 28, 2018 at 4:19

1 Answer 1


Potential solution for ArcGIS, assuming that you have the spatial analyst extension. You will need to define two thresholds: one for the distance that you consider as "near" and one for the NDVI that you consider as "high".

1) Compute the Euclidian distance between to your lines. Make sure that the resolution is the same as your NDVI raster + use snapping in the environment so that the two layers nicely match each other.

2) With the raster calculator, identify areas of large NDVI (e.g. > 0.6) within a given distance (e.g < 10 m)

Con( ( "distance" < 10 ) & ( "NDVI" > 0.6 ) , 1, 0 )

  • I would suggest Euclidian Distance rather than buffer + vector to raster. Mainly because it's easier to experiment with different distance thresholds.
    – user2856
    Aug 28, 2018 at 6:55
  • good idea, I've updated my answer
    – radouxju
    Aug 28, 2018 at 7:05

Your Answer

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

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