Closed bekozi closed 5 years ago
Are we dealing with geometry collections?
The coordinate index is implicitly a geometry collection in its VLen or CRA form. So, yes, I'd say we are supporting geometry collections. We could make this more explicitly linked to the OGC Geometry Collection if we think that's a good idea. The reference implementation uses a CFGeometryCollection
to manage indices/coordinates: https://github.com/bekozi/netCDF-CF-simple-geometry/blob/master/src/python/ncsg/geometry/base.py#L13.
Per the draft CF 1.8 spec, the geometry container may have a grid_mapping attribute provided that the data variable also has it.
This package currently doesn't deal explicitly with data variables and their other supporting variables, and one can easily add the grid_mapping attribute with the netcdf4 package; therefore, I think with support for grid_mapping in the CF spec and the tooling available to add the attribute, this issue is resolved.
Are we dealing with geometry collections?
...heh, I see a bunch of new comments and issues. Maybe I need to catch up!