I've got the following problems with my project. My task is complete the NDSI index calculation in the Antartic region. A problem is some pixel values are much lower than should be.

I'm using ArcMap 10.4.1 verison. My data came from Landsat OLI 8 satellite (https://earthexplorer.usgs.gov/) and the DEM data from REMA database (https://www.pgc.umn.edu/data/rema/)

I've read about Topographic correction, but It didn't work succesfully. I didn't get higher pixel values behind the mountainous area. Luckily I found in QGIS the Topographic correction and I tried all of the topographic correction methods. In the final results I got the exact same NDSI values everywhere. (The input values was different after the topographic correction)

I also tried this toolbox, but the ArcMap crashed while the last correction was running. That happend 3 times in the same point of the progress.( https://hub.arcgis.com/datasets/fd6ed990984945d5a9827418e67f26df )

1 Answer 1


USGS provide terrain corrected Landsat data. (Landsat 8 OLI) Not need doing it again. Your problem can related with NoData value (first need to set it) or with data type (better practice is transform data to Float type before any processing).

  • I transformed my data to Float type but there is no changing in the NDSI values. I've set the NoData values yet, but the main problem is I still lower NDSI values behind the mountains. For example I have lower values then the water. Thats why I cant separate the values between the snow and the water or ice. But these problems are only behind the mountains.
    – GeoKapor
    Mar 20, 2021 at 11:09
  • Are you mean in topographical shadow zones? Then yes, it's possible. Deep shadows can to provide that effect. Mar 21, 2021 at 19:50
  • And is there any way to remove these cast shadows?. I mean to get a higher values, and then I can separate the NDSI values between the water and the snow.
    – GeoKapor
    Mar 21, 2021 at 22:36
  • I'm afraid isn't. In general case it's really not a trivial task. Mar 23, 2021 at 20:23
  • Please, look at that researchgate.net/publication/… Mar 23, 2021 at 20:33

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.