I am working with Landsat 8 data. I have performed the necessary pre-processes that I want (atmospheric and topographic correction and pan-sharpening, using QGIS and ERDAS). I am now working on removing clouds and cloud shadow. I used the "Iso Cluster Unsupervised Classification" tool first. I then manually color coated vegetation vs. cloud/shadow coverage (red and green). I used the "reclassfy" tool to combine the two classes.

I am stuck with: Now I want to make the cloud/shadow coverage value 0 and the vegetation value 1. Then multiple that by my original pre-processed Landsat 8 image. This way all the pixels with the cloud/shadow coverage will disappear, but make sure I keep the RGB bands from the original. Any suggestions?

I have tried multiplying the cloud/shadow and vegetation data by the original in the raster calculator and I get an output with no clouds, but the RGB bands are no longer there.

  • Have you tried something like this in the raster calculator (not tested) Con(binary == 0, 0, original_raster), where binary is your 0-1 cloud/shadow - vegetation raster? – umbe1987 Jul 9 '19 at 7:57
  • I have not tried that. I will try it soon. I just got it to work in ERDAS by assigning the cloud/shadow vegetation raster as a float single. – Zman3 Jul 9 '19 at 8:11
  • 2
    If using the ArcGIS raster calculator, you need to multiply each band separately and then recombine them. – user2856 Jul 9 '19 at 10:45
  • @user2856 Thank you that works, but its a bit time consuming since I have to take each band apart, multiply them individually and then recombine them. I found an alternate solution with ERDAS that uses the Model Maker. Much faster. – Zman3 Jul 10 '19 at 1:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.