1

I am a new ENVI user and I am trying to perform a supervised classification of a Landsat 8 image with the maximum likelihood algorithm.

In order to run the classification, I have to insert a Scale Factor Unit that, for uncalibrated integer data, is equal to the maximum resolution of the sensor (that, for a 16-bit instrument, is 65535).

In this case it works fine, but if I try to perform the classification on an image with an atmospheric correction (Dark Object subtraction), it returns an unclassified image.

The atmospheric correction rescaled the image on lower values, but I can't understand which scale factor I have to insert.

  • Unless it has specifically been resealed, data that has been atmospherically corrected to at-sensor-reflectance is floating point [0-1] and not 16-bit integer. – Jeffrey Evans Feb 17 '16 at 14:35
1

That is probably not working because atmospherically corrected images have reflectance values instead of Digital Numbers, which is what your 65535 refers to. Since you're not dealing with pixel values anymore, it's not gonna work. What version of ENVI are you using? Version 5.0 sets a default value of 1.0 for Max Likehood classification. I've always used it and never had problems.

Your Answer

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

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