netCDF4 groups, dimensions & GenerateDatasetsXml:EDDTableFromMultidimNcFiles

[Rob Cermak]

General question about dimensions handled by EDDTableFromMultidimNcFiles?

Typical glider profiles are aligned with a 1D dimension (time).  I have managed to combine the 2D echogram within a 1D profile of temperature and salinity.  See: https://hoodfish.fish.washington.edu/erddap/files/unit_507_20240304T0000_CF2/

After using GenerateDatasetsXml, the resultant dataset.xml needed a little adjustment.  But it works.  I can access temperature and salinity with its usual dimensions and access the echogram using its dimensions.

Let me attach the dataset.xml for inspection.  Just the <dataset> and sanitized.

Now, here comes consideration for netCDF4 groups.  Theoretically, I could move the echogram into an [acoustics] group.  I could create time, depth, latitude, longitude dimensions within that group used in a different structure... depth(time, n_bins).

The question is will the discovery of LLAT from the [acoustic] group conflict with the discovery of LLAT from the [root] group using EDDTableFromMultidimNcFiles?  

The operation of netCDF4 says the groups should be isolated from other groups.  That doesn't mean EDDTableFromMultidimNcFiles necessarily operates that way.

Or do we still need to use unique dimensional labels throughout the netCDF file?  That would be a general consideration? Any variable or dimension name needs to be unique throughout?  Like...

[acoustics]:

acoustics_depth(acoustics_time, acoustics_bin)

echogram_sv(acoustics_time, acoustics_bin)

Thank you!

Rob