Nemo 3d #463
Nemo 3d #463
Conversation
parcels/field.py
Outdated
| indices = filebuffer.indices | ||
| # Check if parcels_mesh has been explicitly set in file | ||
| if 'parcels_mesh' in filebuffer.dataset.attrs: | ||
| mesh = filebuffer.dataset.attrs['parcels_mesh'] | ||
|
|
||
| dimension_filename = filenames[0] | ||
| with NetcdfFileBuffer(dimension_filename, dimensions, indices) as filebuffer: | ||
| depth = filebuffer.read_depth |
There was a problem hiding this comment.
This isn't a very pretty solution, I'd say. This only works for nemo, but may leave users of other datasets out. Even if they would put depth in their dimension_filename, that would be overloaded here. Should we implement this in a way that is more flexible for different use cases, not only nemo?
There was a problem hiding this comment.
Why don't we replace the mesh_mask file which is not super explicit by a dimesion_file, which can be a file (the mesh_mask) or a dict of files, per dimension.
This follows the logic as everywhere else, no?
There was a problem hiding this comment.
yes, that is a good idea. Although we need to support also that dimensions can also be a dictionary of dictionaries, for each variable. So then we could have two dictionaries: dimension_names and dimension_files.
The former is what we now have as dimensions, and the latter would default to filenames[0] (i.e. assume that all dimensions information is simply in the variable files) but can be overloaded as either a single dictionary (dimension_files['lon']) or a dictionary-of-dictionaries (dimension_files['U']['lon']).
Makes sense?
field.from_netcdf()
in fset.from_netcdf
| " 'data': data_path + 'U_purely_zonal-ORCA025_grid_U.nc4'},\n", | ||
| " 'V': {'lon': data_path + 'mesh_mask.nc4',\n", | ||
| " 'lat': data_path + 'mesh_mask.nc4',\n", | ||
| " 'data': data_path + 'V_purely_zonal-ORCA025_grid_V.nc4'}}\n", |
There was a problem hiding this comment.
Does this new API for filenames also need to be documented? Perhaps explain here in the nemo tutorial, and also add in the docstring of FieldSet.from_netcdf etc?
parcels/field.py
Outdated
| data_filenames = filenames['data'] if type(filenames) is dict else filenames | ||
| if type(filenames) == dict: | ||
| assert len(filenames['lon']) == 1 | ||
| assert filenames['lon'] == filenames['lat'] |
There was a problem hiding this comment.
why can't 'lon' and 'lat' not be in different files? Seems too strict to me? Or, if we need to be this strict, then at least add an error message
parcels/fieldset.py
Outdated
| raise IOError("FieldSet files not found: %s" % str(notfound_paths)) | ||
| for fp in paths[dim]: | ||
| if not path.exists(fp): | ||
| raise IOError("FieldSet file not found: %s" % str(fp)) |
There was a problem hiding this comment.
Here, we have two copies of almost the same code. Create a function instead, for conciseness an ease of maintenance?
for convenience (explicit not implemented error if not in same file)
| @@ -33,7 +33,17 @@ | |||
| "cell_type": "markdown", | |||
| "metadata": {}, | |||
| "source": [ | |||
| "We can create a `FieldSet` just like we do for normal grids. Note that we need to provide an extra filename for the `mesh_mask` file that holds information on the Curvilinear grid." | |||
| "We can create a `FieldSet` just like we do for normal grids.\n", | |||
| "Note that NEMO is discretised on a C-grid. U and V velocities are not located on the same grid (see https://www.nemo-ocean.eu/doc/node19.html).\n", | |||
There was a problem hiding this comment.
Is 'grid' the right word here, is 'not located on the same grid'? Perhaps 'location'?
| "|__V0__|\n", | ||
| "```\n", | ||
| "\n", | ||
| "To interpolate U, V velocities on the C-grid, Parcels needs to read the f-nodes, which are located on the corners of the cells." |
There was a problem hiding this comment.
change to 'lower-left corners' to be more explicit which corner?
parcels/fieldset.py
Outdated
| @@ -160,6 +160,9 @@ def from_netcdf(cls, filenames, variables, dimensions, indices=None, | |||
| :param filenames: Dictionary mapping variables to file(s). The | |||
| filepath may contain wildcards to indicate multiple files, | |||
| or be a list of file. | |||
| filenames can be a list [files], a dictionary {var:[files]}, | |||
| a dictionary {dim:[files]} (if lon, lat, depth not stored in same files as data), | |||
| a dictionary of dictionaries {var:{dim:[files]}} | |||
There was a problem hiding this comment.
add 'or' before the dictionary-of-dictionaries?
parcels/fieldset.py
Outdated
| or be a list of file. | ||
| filenames can be a list [files], a dictionary {var:[files]}, | ||
| a dictionary {dim:[files]} (if lon, lat, depth not stored in same files as data), | ||
| a dictionary of dictionaries {var:{dim:[files]}} |
| U0 U1 | ||
| |__V0__| | ||
| To interpolate U, V velocities on the C-grid, Parcels needs to read the f-nodes, | ||
| which are located on the corners of the cells. |
There was a problem hiding this comment.
see minor comments on the jupyter notebook
No description provided.