NetworkValidator

[DanielTollenaar]

We see some network-issues in regional models (see #102). We want to get an overview within Python so we can (1) see all issues at once and (2) correct them on-the-fly.

Functions build:

• node_overlapping: Check if the node-geometry overlaps another node within tolerance (default=1m)
• node_duplicated: Check if node_id is duplicated
• node_internal_basin: Check if a Node with node_type Basin is not connected to another node
• edge_duplicated: Check if the from_node_id and to_node_id in the edge-table is duplicated
• edge_missing_nodes: Check if the from_node_id and to_node_id in the edge-table are both as node-id in the node-table
• edge_incorrect_from_node: Check if the from_node_type in edge-table in matches the node_type of the corresponding node in the node-table
• edge_incorrect_to_node: Check if the to_node_type in edge-table in matches the node_type of the corresponding node in the node-table
• edge_incorrect_connectivity Check if the geometries of the from_node_id and to_node_id are on the start and end vertices of the edge-geometry within tolerance (default=1m)

[Huite]

I noticed:
#102 (comment)

Maybe worthwhile to check connected components:

https://docs.scipy.org/doc/scipy/reference/generated/scipy.sparse.csgraph.connected_components.html

I guess multiple separate networks should be valid. However, if a component is 1- or 2-sized, it's probably wrong?

n_labels, labels = scipy.sparse.csgraph.connected_components(adj_matrix)
if (np.bincounts(labels) == 1).any():
    raise ValueError("...")

Easiest to setup as a COO sparse matrix using from_node_id and to_node_id. If a Ugrid1d topology is available, the node_node_connectivity can be directly used:

https://deltares.github.io/xugrid/api/xugrid.Ugrid1d.node_node_connectivity.html#xugrid-ugrid1d-node-node-connectivity

[DanielTollenaar]

Fixed by #129