I am trying to get landsat reflectance values corresponding to field plots spread all over Finland using the Google Earth Engine. As you know, Finland is up North, between latitudes of ~60 and 70 degree N. I am extracting from this landsat collection. Upon examining the extracted values, I noticed that most (99.9%) of the pixels have a 'pixel_qa' value of 322. From my understanding of the bitmask, I gathered that the value of '322' means that the 'cloud confidence' and the 'cirrus confidence' was low. Why is this? Can I still assume that the cloud/noncloud classification is good enough?


I got a reply from someone at USGS: "You want low confidence in clouds, that means that there is very little chance that there is a cloud there when there shouldn't be. A value of 322 means that the pixels are clear and there are not clouds present."

So, in conclusion, a value of 322 is okay to work with.

  • Is more than that, you should decode pixel_qa band in the bits you need. Check this question, it could help you gis.stackexchange.com/questions/271322/… – aldo_tapia Jan 27 at 12:23
  • Thanks for that link! Actually, I was confused by what they meant in bits 6-7 and 8-9. For example, a "cloud confidence" value of "none" seems to say that the algorithm has very low confidence on its output (clouds vs no-cloud)! I have given the feedback to the USGS person that they may want to word that better. Something like "probability of cloud contamination in the pixel"... – user7831861 Jan 28 at 16:15
  • As far I know, none is non-determine status. Check this page: usgs.gov/core-science-systems/nli/landsat/… – aldo_tapia Jan 28 at 19:38

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.