QC dimensions don't match data variables

[naught101]

Here's some output form FluxnetLSM

                time:long_name = "time" ;
        float Tair(time, z, y, x) ;
                Tair:units = "K" ;
                Tair:_FillValue = -9999.f ;
                Tair:long_name = "Near surface air temperature" ;
                Tair:Fluxnet_name = "TA_F_MDS" ;
                Tair:Standard_name = "air_temperature" ;
                Tair:ERA-Interim\ variable\ used\ in\ gapfilling = "TA_ERA" ;
                Tair:Missing_\% = 0. ;
                Tair:Gap-filled_\% = 58.6 ;
                Tair:Gapfilling_method = "ERA-Interim" ;
        float Tair_qc(time, y, x) ;
                Tair_qc:units = "-" ;
                Tair_qc:_FillValue = -9999.f ;
                Tair_qc:long_name = "Tair quality control flag" ;
                Tair_qc:Fluxnet_name = "TA_F_MDS_QC" ;
                Tair_qc:Standard_name = "NULL" ;
                Tair_qc:Missing_\% = 0. ;
                Tair_qc:Gap-filled_\% = 0. ;
        float SWdown(time, y, x) ;

The Tair_qc flag has different dimensions to the Tair variable. I don't know how I've managed to avoid noticing thing until now, but some of my existing code just started not working after upgrading xarray, and I think the previous version was flattening dimensions on the array that is now not working..

[aukkola]

Do you mean the z-dimension? I think that was added because CABLE uses the extra z-dim for air temperature and a couple of other variables (air pressure?). It should always be set to 1. It is somewhat redundant so can be removed from the package.

[naught101]

I guess it should be optional, but either way, the QC flag dimensions should match the variable dimensions. I'm not likely to find time to deal with this any time soon, as I can work around it fairly easily right now.

[aukkola]

Fixed. Some variables left as 3D following previous PLUMBER experiment