3 year composite for 2013/2015 nir looks like this:
Those are clouds coming through in the northwest of the image.
First, I limited items to tier one only. That helped some (but may have effects on available data).
In this case I looked at adjacent masks for a single date, here 2013-07-18 (this is pathrow 87 56 btw).
First image, identified clouds are blue, orange is clear
Second image, taken half a second later:
Obviously missing some clouds, as this is nir after masking
I created a new method to combine cloud masks for the same day with bitwise-or,here is new output
Issues with clouds are causing noise.
In this case, here is the effect:
3 year composite for 2013/2015 nir looks like this:
Those are clouds coming through in the northwest of the image.
First, I limited items to tier one only. That helped some (but may have effects on available data).
In this case I looked at adjacent masks for a single date, here 2013-07-18 (this is pathrow 87 56 btw). First image, identified clouds are blue, orange is clear
Second image, taken half a second later:
Obviously missing some clouds, as this is nir after masking
I created a new method to combine cloud masks for the same day with bitwise-or,here is new output
We're getting there.