Clouds & Cloud masking
today we used cloud-masked satellite imagery for the NDVI task. Insa also said that clouds are a substantial problem for much Sentinel-2 data. How do I decide which satellite imagery are still "usable" although they have clouds? Are their any rules of thumb? For example, anything <20% cloud cover is ok when I would like to use large areas of a satellite image, e.g. for forest monitoring?
The fewer clouds the better, of course. It always depends a little on your research question. If you are flexible in terms of time (temporal resolution), you can search until you find a scene with the fewest clouds. If you want to perform less complex analyses, it is of course fine to use images with more clouds than if you want to perform highly accurate analyses in which you are heavily dependent on the extracted values. There is no rule of thumb, so to speak.
Thanks for asking.