Skip to content

IO Utilities

diffroute can ingest RAPID routing configurations and convert them into ready-to-route RivTree or RivTreeCluster instances. The helpers live in diffroute.io and handle both single-VPU and multi-VPU projects.

Reading a single RAPID project

from pathlib import Path

from diffroute.io import read_rapid_graph

vpu_root = Path("/path/to/rapid/VPU1234")

# Without segmentation: returns a single RivTree
g = read_rapid_graph(vpu_root)

# With segmentation: returns a RivTreeCluster
clustered = read_rapid_graph(
    vpu_root,
    plength_thr=50_000,
    node_thr=1_000
)

print(type(g), len(g))
print(type(clustered), len(clustered))
  • RAPID parameters (k, x) are converted to Muskingum IRF coefficients and stored in the resulting data structure.
  • When plength_thr and node_thr are provided, the function runs the graph segmentation pipeline (define_schedule) before building the RivTreeCluster.
  • When plength_thr and node_thr are set as None, staging is omitted and a RivTree instance is returned.

Reading multiple RAPID projects at once

Large problems are often split into sub-clusters in RAPID. diffroute provides a helper function to load arbitrarily many such directories and schedule across these clusters as a single river network. 

from pathlib import Path

from diffroute.io import read_multiple_rapid_graphs

vpu_roots = [
    Path("/path/to/rapid/VPU0101"),
    Path("/path/to/rapid/VPU0102"),
]

gs = read_multiple_rapid_graphs(
    vpu_roots,
    plength_thr=40_000,
    node_thr=800
)

print(f"Global reaches: {len(gs.nodes_idx)}")

read_multiple_rapid_graphs merges the VPU graphs into a single NetworkX DAG, combines their Muskingum parameters, and applies the same optional segmentation logic as the single-project loader.