The current behaviour is to use np.nan but this makes the MCA to exclude area where any of the rasters has np.nan values. To avoid this we can set the nodata value to 0 instead. However, this need testing to ensure that the current applications don't get affected.
The current behaviour is to use
np.nanbut this makes the MCA to exclude area where any of the rasters hasnp.nanvalues. To avoid this we can set thenodatavalue to 0 instead. However, this need testing to ensure that the current applications don't get affected.