Repository navigation
Adopt conservative bifurcation FACC budgets with fixed full-river area #236
Description
Adopt the validated global conservative bifurcation correction as the local working FACC authority. User approved adoption on 26 October 2026, with facc_total and SWORD widths unchanged and mapping/direction work tracked independently.
Every allocated side share now debits the main edge, including non-rejoining distributaries. The main edge receives 1 minus actual side allocations. The dynamic whole-area guard is removed; whole totals are fixed to the frozen 21 October parent. Existing SWORD characteristic widths, topology and parent source/credit rules are retained.
Validated population: 60 basins, 248,673 reaches, 11,112,454 nodes. Overbudget forks above 1 km² decrease 903 -> 0 across 2,893 forks. Source-ID terminal budget violations decrease 16,671 -> 0. All full-area endpoints and nodes match the parent; 38 focused tests and global ledger/node checks pass.
The 95th-percentile absolute source-union junction residual decreases 1,118.93 -> 892.34 km²; deficit screens decrease 1,913 -> 1,671, while positive screens increase 4,304 -> 4,328. Thus internal accounting improves, but not every discrepancy statistic improves. HydroBASINS facc_total predictions and scores remain unchanged: 5,312/6,262 within 20% (84.83%), median absolute relative error 0.961%. Improvements against original v17b in paired NA/EU/OC references precede this split-only change.
Implementation: scripts/maintenance/adopt_facc_bifurcations.py materializes the working product and config/facc_adopted.json selects it. Product: data/derived/facc/bifurcation_20261026/, with per-region reach/node Parquets and a hashed manifest. Frozen scientific source: output/facc_detection/variants/rule0021_conservative_forks_fixed_total_20261026/runs/global_candidate/.
Scope: local working product and analysis authority. Published exports, original shipped source, databases, widths, and topology remain untouched. Preserve AS edge 43308000201 -> 43309000191 as a deferred investigation. Reference mapping, native/interpolated placement and direction cases have separate follow-ups. Related validation context: #232.
Completed and verified locally. Authority: config/facc_adopted.json. Twelve materialized regional files passed manifest/hash/count readback; all 1,580 frozen candidate files matched, whole reach/node totals are unchanged, and SWORD widths/topology remain unchanged. Separate follow-ups: #233 (reference units), #234 (source placement), #235 (deferred direction case). The local adoption supersedes the earlier audit hold; published data and databases are untouched.
Activity