Definition of variable "short names" and "definition" reference

[fmigneault]

Variables often have an official name and a "short name" variant.

For example, ECMWF land_sea_mask that also defines short-name: lsm.
https://codes.ecmwf.int/grib/param-db/172

It would be worth having a short_name property for describing these variables, since they are commonly used in scripts loading the data, and it might not be that obvious for non-expert users to resolve the mapping.

Alternatively (or additionally), there could also be directly a property (definition or similar) that contains the "https://codes.ecmwf.int/grib/param-db/172" link for explicit referencing.

[fmigneault]

Maybe can be addressed with stac-extensions/cf#8

[fmigneault]

Using CF extension below.

Note that https://codes.ecmwf.int/grib/param-db/172 provides (under the NetCDF tab) cfName: land_binary_mask which differs from the ECMWF land_sea_mask parameter, the readable name, and the alternative lsm short-name often used in code for data access.

Ideally, all of these "name" representations would be identified by distinct fields.

{
  "properties": {
    "cube:dimensions": {
      "time": {
        "type": "temporal",
        "extent": [
          "1940-01-01T00:00:00Z",
          "2100-12-31T23:59:59Z"
        ],
        "step": "P1H"
      },
      "lat": {
        "type": "spatial",
        "extent": [
          54.2,
          22.6
        ],
        "axis": "y"
      },
      "lon": {
        "type": "spatial",
        "extent": [
          233.6,
          297.5
        ],
        "axis": "x"
      }
    },
    "cube:variables": {
      "land_sea_mask": {
        "definition": "https://codes.ecmwf.int/grib/param-db/172"
        "cf:standard_name": "land_binary_mask",
        "title": "Land-sea mask",
        "common_name": "lsm",        
        "description": "Proportion of land-sea, where 1 indicates land and 0 indicates sea.",
        "type": "data",
        "data_type": "float32",
        "extent": [
          0,
          1
        ],
        "dimensions": [
          "lat",
          "lon"
        ]
      }
    }
  }
}

[m-mohr]

What is used for "data access", i.e. what can I use in code to load the thing? That's the key to be used.
"common name" for me implies a pre-defined list of values, similar to eo:common_name or the CF standard name.
What is this short name used for? Wondering whether themes or externalIds (see OGC API - Records) could be used.

[fmigneault]

In this specific case, lsm is the attribute that is available in the code.

I used "short name" to match the field indicated in ECMWF, but it could be anything else.

What I think matters most is having the https://codes.ecmwf.int/grib/param-db/172 reference.
Similarly to OGC API Common UMD (https://docs.ogc.org/DRAFTS/20-024.html#umd-collection-definition), where definition/unit/unitLang references can be provided, I think it would make sense to have a definition field in cube:dimension and cube:variables since unit is already available. It would help reference more complex/ambiguous definitions to ensure their correct interpretation. The same unitLang could also be considered to distinguish between UCUM, UDUNITS-2 or another like QUDT.

The "common name" seems somewhat arbitrary depending on whether you consider ECMWF or CF to be your reference, both of which offer pre-defined lists/catalogues.