Description
I've found that the QC checks in pvanalytics flag values outside the testing domain as erroneously, which I strongly believe is against the purpose of the testing domain.
As an example, the BSRN comparison and diffuse ratio tests are stated to only be performed for GHI>50 W/m^2. The reason for this threshold is that the uncertainty at lower irradiance is too high to perform the test reliably. This is NOT saying that values lower than 50 W/m^2 should be flagged.
If values lower than 50 W/m^2 should be flagged, it certainly does not make sense for the BSRN lower limits on the individual components to be -4 W/m^2. For example, the below data point which matches the closure equation perfectly (i.e., GHI = DHI+DNI*cos(Z)) is flagged as erroneously because it's below 50 W/m^2.
qcrad_consistency_flag = check_irradiance_consistency_qcrad(
solar_zenith=pd.Series(80),
ghi=40,
dhi=40,
dni=0)
I have a series of QC checks that I would like to add, but this issue is a blocker for now.
This issue is related to #188 (comment) and #167 (comment)