Problem when using UGRID aggregated in THREDDS

[marceloandrioni]

Hello, following @Unidata/thredds#526 (comment) by @drf5n I managed to aggregate unstructured grid results in TDS 4.6.14 as a FMRC (feature model run collection). The solution was basically to add a record dimension to aggregate everything along.
TDS shows the new aggregated dataset just fine using the OPENDAP protocol. All other protocols (NCSS, WMS, WCS) fail, but I think that is expected since only TDS 5 will have true UGRID support.

Back to the main issue, when trying to open the opendap link using ncWMS2/EDAL (version 1.4.1, not the one inside TDS) it fails with the message:
uk.ac.rdg.resc.edal.exceptions.DataReadingException: Coordinate variables listed in mesh_topology:node_coordinates must share the same single dimension to be UGRID compliant

Checking the code my extremely limited java knowledge tells me that the problem is because now latitude and longitude have one extra dimension (record) besides the main one (node).

Here some examples of the files I am testing with.

Thank you very much.

Please check your files. Maybe they follow the UGRID convention, but they are not CF-compliant:

------------------
Checking variable: element
------------------
ERROR: (3.3): Invalid standard_name: face_node_connectivity
ERROR: (3.1): Invalid units: nondimensional

------------------
Checking variable: latitude
------------------
ERROR: (4.3): Invalid value for positive attribute

------------------
Checking variable: longitude
------------------
ERROR: (4.3): Invalid value for positive attribute

------------------
Checking variable: mesh_topology
------------------
ERROR: (3.3): Invalid standard_name: mesh_topology
INFO: attribute cf_role is being used in a non-standard way
INFO: (3.1): No units attribute set.  Please consider adding a units attribute for completeness.
ERROR: (9.5): Invalid value for cf_role attribute

[marceloandrioni]

Hi @JaviRG , thank you for your reply. Those error messages came from EDAL/ncWMS log?

No, I just checked your files using https://pumatest.nerc.ac.uk/cgi-bin/cf-checker.pl

[marceloandrioni]

Hi @JaviRG, thanks for pointing me out to cf-checker. I never tried before but it seems a great tool. I will try to fix the errors listed by it to see if then the file can be read/displayed using EDAL.

Thank you.