ee_extract_tidy/ee_extract handling of masked/NA values #33
Description
When a feature lands on a masked image value that feature is omitted from the extracted output data.frame/sf object. This is usually not a significant problem - say I have 10 points and 2 land in masked areas - my output data.frame will have 8 records. Not a big deal, I probably don't want to include these 2 masked points in most subsequent analysis anyways, but if I do I can always figure it out as long as the records had uids of some sort. However, it becomes a problem if all of your points fall in a masked area: ee_extract_tidy throws a cryptic error triggered by tidyr::pivot_longer because there is nothing to pivot on.
I've thought of 2 options so far:
1.) set up a way to return NA values for masked features rather than omitting them - i think this would be advantageous. Would allow you to quickly calculate % of locations covered by cloud and get rid of the pivot error. I started writing this, but then realized the path i was following would end up being more work than expected especially in the case of multiple band extractions.... perhaps a simpler, better solution will come soon.
2.) put in a more informative warning (i.e "all feature fall in masked/NULL area so no data was extracted") - this might be a nice interim solution
As we can see the problem was flagged in the issue below and was not immediately obvious due to the cryptic message - either of the above would be an improvement i think