Issues in auxiliary file conversion output

[niallmcc]

n EOCIS-AUXILARY-L4-SOCIOECONOMIC_ITL3-MERGED-*-fv1.0.nc, int64 is used as the data type (should be int) and I guess we should call it "time" rather than "t"

int64 t(t) ;             t:units = "days since 1999-01-07 00:00:00" ;             t:calendar = "proleptic_gregorian" ;

In EOCIS-AUXILARY-L4-LANDCLASS-MERGED-*-fv1.1.nc there is no time or t attribute, we should probably make it consistent with what we do for the above files

[niallmcc]

Some further corrections mentioned by Alison

Flagged by the NCAS cf checker ( https://cfchecker.ncas.ac.uk/)

• In many of the files, they’re specifying non-existent bounds attributes (x_bnds, y_bnds)
• In some of the files, there are a lot of ‘standard_names’ that don’t exist in the standard name table (e.g. in the elevation file and some others). I would just remove the standard names in those cases, and make sure the information is in the long_name attribute.
• Similarly some of the files use the standard name ‘region’ - but then the regions are non-standard, so I’m not sure we can use that here.
• It’s also showing an error with flag_values in some of the files e.g. landcover (ERROR: Attribute flag_values of incorrect type (expecting 'Data Variable' type, got 'Numeric' type)) - I think that’s because the variable is a short but the flag values are unsigned bytes, when they should be the same.
• Finally it’s also showing some errors with the grid mappings – again I think it’s trying to reference some non existent grid mapping variables e.g. in the builtarea file, the variables have e.g. ‘urban_area:grid_mapping = "crsOSGB: x y crsWGS84: lat lon" ‘, but only the crsOSGB is actually defined in the file.

[oembury]

It’s also showing an error with flag_values in some of the files e.g. landcover (ERROR: Attribute flag_values of incorrect type (expecting 'Data Variable' type, got 'Numeric' type)) - I think that’s because the variable is a short but the flag values are unsigned bytes, when they should be the same.

_FillValue, missing_value, valid_min, valid_max, valid_range, flag_values, and flag_mask attributes must have the same type as the variable they are attached to.

actual_range must have the same type as add_offset/scale_factor if used, otherwise the same type as the variable.

[niallmcc]

@oembury what tool do you use to check for compliance with CF?

[oembury]

From https://cfconventions.org/software.html

• cf-checker: web tool download: https://github.com/cedadev/cf-checker
• IOOS Compliance Checker: web tool download: https://github.com/ioos/compliance-checker