WEPPcloud

← Back to help

ADR-0113: Native accepted-boundary water and current-event sediment accounting

  • Status: Accepted for U2c current-event repair in 261029 artifact revision 2; C3 water formulation rejected.
  • Date: 2026-10-30
  • Decision venue: WEPPcloud Roads agent session, America/Los_Angeles.
  • Participants: project owner and Codex; independent implementation/scientific agents review evidence.
  • Decision owner: project owner authorizes ablation and the undeployed 261029 reissue.
  • Implementer: Codex.

Context

The repaired wepp_261029 rejects two generated-soil Roads fixtures because a daily runoff calculation asks the hourly allocator for more water than exists. The allocator's refusal is correct. Equivalent-plane calculation can recompute upstream infiltration after that upstream soil's water balance has committed. Later O2 traces establish melt-support compression as the causal defect in the two retained failures; upstream GRNA already uses pre-event STOLD. See ADR-0114 for the narrower corrective formulation. A separate late-return path enables sediment routing after current-event setup was skipped, reusing prior-event coefficients and duration.

Decision for this ablation

Retain the independently tested U2c current-event repair in revision 2. Reject and roll back C3 water processing: output accounting passed, but connected hydraulic support, peak-area frames and incoming-sediment deposition did not. No Roads parameter default is accepted. Define separate local water and connected hydraulic geometry contracts before another water mutation.

Evaluated C3 water proposal (rejected)

The experiment tested the following formulation. For continuous hourly UI hillslopes with no contour or irrigation system, use the same sequential boundary calculation for every receiving OFE, independent of whether return is positive. Use local melt, soil capacity and length, plus accepted upstream Hortonian input. Normalize the upstream excess shape captured before return is added to its accepted post-recession volume, using the existing effective-length conversion. Merge that input after local interception; keep return and lateral water on the existing hourly stock path. Keep local-only rainfall intensity and energy for raindrop detachment. Preserve the existing calculation outside the explicitly eligible model domain. Hortonian-only events within that domain intentionally use sequential accounting.

Retained U2c event-state proposal

Require current-event erosion preparation before ROUTE/SLOSS. Preserve late water and its source-time metadata, but do not route stale erosion state. Publish zero meteorological duration for a confirmed late return source using a local value; leave the shared global duration unchanged for legacy consumers such as furrow irrigation. No parameter values, guard thresholds or acceptance bounds change.

Rationale and alternatives

Clipping negative FIN, deleting incoming return, relaxing the inventory guard, or adjusting soils would hide the native request defect. Pure-return separation alone completes Birmingham but does not resolve mixed-source Charleston. The original return-only scope proved discontinuous; the broader consistent scope required its own scientific assessment. Clearing global duration fixes the retained event but risks a later irrigation consumer; the local publication value avoids that new dependency. A return-triggered formulation was rejected: a 45-case native subproblem found up to 10.682542 mm finite change at zero-to-positive return with unchanged Hortonian/local forcing. Consistent receiving-element accounting removes the method switch; an arbitrary epsilon would merely move it. Independent scientific review supports testing this broader scope, not promotion without validation.

Evidence and acceptance

Native canonical checkout: /workdir/wepp-forest-release-20261009. Work package: docs/work-packages/20261030-roads-native-accounting/ in that repo; committed four-case reproduction: 755a15fbe. Active incident: docs/ablation/20261030_roads_daily_coupling_and_event_state/. Roads continuation records the original 840-case failure panel. The native plan requires frozen-input replays, source accounting controls, native regression/watchlist gates and independent review. Existing output audits are not complete water closure.

Risks and rollback

The water candidate changes shared multi-OFE process behavior beyond Roads; peak timing, deposition, storage coupling and legacy shared state require review. An excess-flow shape is not a newly validated travel-time hydrograph. Existing Hortonian transfer uses length scaling; variable-width closure is not established. Reject any lane that fails stock/source conservation, generates rainfall energy from runon, or changes excluded paths. Finalize each upstream segment before starting the next receiver; preserve RDAT interval-versus-endpoint semantics. Independent review caught and rejected a candidate that otherwise suppressed valid upstream erosion. Keep rejected lanes as evidence and restore accepted source. Release promotion and UI implementation are separate decisions; the UI specification/layout remains human-gated.

Initial C3 ablation disposition — 2026-10-30 (historical)

Canonical native package: /workdir/wepp-forest-release-20261009/docs/work-packages/20261030-roads-native-accounting/. The separate U2c stale-event repair is retained for validation. C3 water changes are rejected and rolled back despite 840/840 output-accounting passes: independent controls demonstrate lost connected hydraulic support, inconsistent peak-area normalization and skipped deposition when runon fully infiltrates. Cedar peak +42.64% and sediment +35.05% with water +0.00212% require this hold. C3 magnitude is 719/840, while literal-legacy attribution is 96/96. Neither is a Roads default acceptance. Frozen inputs, candidates, review probes and Cedar figures are retained in the native package. The water mechanism remains unresolved; C4 diagnostic controls define the next geometry/accounting contract before another behavioral patch. Keep wepp_261029 pinned, withheld materials unused, and parameter tuning paused. The human gate on configuration specification/layout still applies.

Revision 2 acceptance

The final matched source ba9f2ad91b5165d5cd2954bd64b5df608798541e is scientifically accepted for the bounded repairs above and vendored under the owner-authorized undeployed wepp_261029 identity, artifact revision 2. Validation includes 1,059 native tests, 840 Roads physical cases, 99 Disturbed tests, 18 watchlist cases and fresh matched Cedar outputs. The full 24-year channel audit has zero negative/nonfinite class transfers, with all nine standard outputs unchanged by observation. Operational release evidence and the preserved original identities are in the reissue package.

Cedar outlet sediment is 517,815.77 kg, about 187.456% above the original release, principally from recovered channel transport. This is not field calibration or complete physical budget closure. Regenerate all contributing hillslope, PASS and watershed outputs with the exact matched pair. Roads magnitude acceptance remains unmet (719/766 meaningful-reference and 30/74 small-reference cases); defaults and the human UI specification/layout gate remain unchanged. Earlier experimental dispositions above are retained as history and do not authorize the rejected C3 formulation.