Output data needs to indicate whether a node is a boundary one or not

[pmav99]

This is needed for GDACS

[brey]

I have second thoughts about adding it to the post-processing. The code is

import pyposeidon.mesh as pm
    #add boundary nodes
    mesh = pm.set(type='tri2d', mesh_file=os.path.join(b.rpath + 'hgrid.gr3'))
    
    out_total = xr.merge([out,mesh.Dataset[['type','node','id']]])

but it would take long for a big global mesh and the data will be in every output file. I suggest we save mesh.Dataset[['type','node','id']] to a single file once and this can be used to identify boundaries for all runs.

[pmav99]

Sure, it can be part of the model definition.

[pmav99]

That being said, do we need to call xr.merge()? Wouldn't that be faster?

out_total = out.assign(
    btype=mesh.Dataset.type,
    bnode=mesh.Dataset.node,
    bid=mesh.Dataset.id
)

[brey]

I agree if we needed to merge them. We will not do that. We are just going to save to a netcdf file that relates to the model and is unique.

import pyposeidon.mesh as pm
    #add boundary nodes
    mesh = pm.set(type='tri2d', mesh_file=os.path.join(b.rpath + 'hgrid.gr3'))
    
   mesh.Dataset[['type','node','id']]].to_netcdf("boundary_connectivity.nc")

This will be done once and the file can be used to identify boundary nodes for all outputs, if need be. It doesn't need to be part of pyposeidon.

[brey]

I think we can close this issue.

[pmav99]

Should we export this file by default? Should we also be able to read it? And/or to add a reference it in the *model.json ? Or maybe just document, that extracting this information can be done with the snippet above?