Relative humidity min/max, clipping and OOL

[ladsmund]

The clip_values function currently restricts rh_[uli]_cor to the range of 0 to 100.

Is this approach valid, given that the allowable range for these values was initially set between 0 and 150%, accommodating supersaturated states?

Would it be more appropriate to validate and filter the inferred values before computing the corrected humidity variables? Clipping might be overly restrictive in cases where supersaturation is expected or meaningful.

#307 (review)

[BaptisteVandecrux]

To me there is no doubt that

• rh_[uli] should have a range of 0-100
• rh_[uli]_cor should have a range of 0-150 to allow supersaturation
• the following lines have been identified as dysfunctional (found bug it clip_value for rh_u_cor and rh_l_cor, adjusting limit for rh_u and rh_l and problem with the specific humidity's unit in variable.csv #220) and shouldn't be brought back into Fix OOL Dependency Handling, RH Clipping, and Bufr Export Rounding #307:

pypromice/src/pypromice/process/value_clipping.py

Lines 45 to 52 in a46e96e

	if var in ["rh_u_cor", "rh_l_cor"]:
	# Nan inputs should stay nan
	was_nan = ds[var].isnull()
	if ~np.isnan(row.lo):
	ds[var] = ds[var].where(ds[var] >= row.lo, other=0)
	if ~np.isnan(row.hi):
	ds[var] = ds[var].where( ds[var] <= row.hi, other=100)
	ds[var] = ds[var].where(~was_nan)

Would it be more appropriate to validate and filter the inferred values before computing the corrected humidity variables?

I am not sure what you are suggesting here.