I just faced almost the same problem.
I have a computed layer with x,y dimension, and i want to calculate a global statistic (i.e. the 35° percentile ) for the whole extent of my area
Up to now i don’t consider the temporal dimension but in future i will.
The calculated value(s) will be than used to mask another layer.
My problem connects well also with the percentile/quantile topic How to use quantiles on a 2d data cube (x,y) - openEO Platform - EODC Discussions
I tried with both aggregate_spatial using a polygon as big as the image and with reduce_spatial ( that as pointed in the posts above con be done only one dimension by time that is not so correct )
Are there any updates on the implementation of reduce_spatial in 2 dimension at the same time ?
Any updates on filter_labels ?
@stefaan.lippens @jeroen.dries