Since the beginning of COMT, we've been imagining that providers were capable of editing an NcML template for each new simulation. But with increasing reliance on standard_names, the introduction of SGRID conventions, and the requirement that each non-coordinate variable have an attribute content_coverage_type: modelResult, it was getting too complicated.
To make it easier for providers, we created a 'yaml2ncml.py' script, which currently works only for ROMS, but could be modified to work with other models. It contains a ROMS lookup table for standard_name mapping from ROMS variable names, detection of SGRID information (e.g. cell centers or cell walls). So the only thing that the provider does is to edit a YAML file to provide title, abstract, etc, which variables to include or exclude, and the aggregation scan pattern.
Since the beginning of COMT, we've been imagining that providers were capable of editing an NcML template for each new simulation. But with increasing reliance on standard_names, the introduction of SGRID conventions, and the requirement that each non-coordinate variable have an attribute
content_coverage_type: modelResult, it was getting too complicated.To make it easier for providers, we created a 'yaml2ncml.py' script, which currently works only for ROMS, but could be modified to work with other models. It contains a ROMS lookup table for
standard_namemapping from ROMS variable names, detection of SGRID information (e.g. cell centers or cell walls). So the only thing that the provider does is to edit a YAML file to provide title, abstract, etc, which variables to include or exclude, and the aggregation scan pattern.The code is at http://github.com/usgs-cmg/yaml2ncml
I have used it to create NcML for 2 of Marjy's datasets: https://github.com/ioos/comt_2/tree/master/cb_hypoxia