I am currently in a remote sensing course at my university and am working on a change detection project. I am having some trouble creating an ndwi raster from landsat 8 imagery. I am using arcgis 10.5.

I have downloaded the imagery from USGS earth explorer and am running the raster calculator tool to get my output. The formula I am using is:

float(band 3 - band 5)/float(band 3 + band 5)

My output contains low value of -20.4 and high values of 31.3 which definitely seems off to me. I ran a similar calculation on landsat 7 imagery from 2001 and got a result that ranges from -1 to 1. I am very new to the remote sensing field and am stumped on why my results for the landsat 8 imagery are so extreme.

Can anyone provide insight into why this is happening and what I can do to correct my calculations?

1 Answer 1


I always convert the raw data to top-of-atmosphere reflectance first, and then derive the index. I do this to address issues of variations of earth-sun geometry during the year, and also atmospheric path length. If this is done, then the index should range from -1 to +1, which would assume a theoretically perfect response of vegetation, and water, respectively. You should be able to find the procedure somewhere on the Lands at website.

Good luck!

Stuart McFeeters, Ph.D., GISP CSU-Fresno

  • Thank you I appreciate the great answer. Looks like there is some additional processing I need to do using numbers from the metadata. Thanks again! Nov 12, 2017 at 8:02
  • Holy...I just realized who answered my question. Legendary. Thank you! I ended up getting an A on my project. Cheers! Jun 19, 2018 at 20:55

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.