Estimate properly the size of ocean 3D variables #2
Description
The ocean component has quite a few 3D variables which are provided in a single file. This differs from the atmosphere, where a set of pressure levels have been selected and variables are defined as 2D variables per level (e.g. ta850). Programmatically, the simplest way to know if a variable is 3D is by checking the dimensions in the main data request file or in the CMOR tables themselves. The dimension olevel indicates the vertical ocean dimension:
This, however, does not provide the size of this dimension, which is model-dependent. For the estimate (#1), we could use an average value across models, which will be better than the current assumption of a 2D field. For Med-CORDEX, a finer estimate could be achieved by retrieving the vertical levels from the model documentation, e.g.:
https://github.com/Med-CORDEX/model-documentation/blob/2cc1366eadce926f8397a874ba35abee8e9df52e/_models/CNRM-RCSM6B.md?plain=1#L254
but I'm not sure that the levels requested in the ocean data request (depth levels in meters) are the native levels in the ocean models or these need to be interpolated (to a predefined set of depth levels?). @sam-somot?
The same problem affects soil variables (sdepth dimension), which is relevant for WCRP-CORDEX/simulation-status#40