You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Right now, we only use a cloud screen to toss out images that have too many clouds, but then we calculate features from all the pixels in the selected image. A better way would be to actually remove the cloud pixels. Sentinel 2's SCL band is too lossy for this (throws away too much data). Google's Cloud Score+ is supposed to be much more accurate and fine grained so we should look into using this algorithm to drop cloud pixels before calculating features from band statistics.
Right now, we only use a cloud screen to toss out images that have too many clouds, but then we calculate features from all the pixels in the selected image. A better way would be to actually remove the cloud pixels. Sentinel 2's SCL band is too lossy for this (throws away too much data). Google's Cloud Score+ is supposed to be much more accurate and fine grained so we should look into using this algorithm to drop cloud pixels before calculating features from band statistics.
https://developers.google.com/earth-engine/datasets/catalog/GOOGLE_CLOUD_SCORE_PLUS_V1_S2_HARMONIZED
https://medium.com/google-earth/all-clear-with-cloud-score-bd6ee2e2235e
The text was updated successfully, but these errors were encountered: