ADR-0116: Anchor channel lateral sediment to its assigned source mass
- Status: Accepted for channel sediment-source accounting in 261029 artifact revision 2.
- Date: 2026-10-30.
- Decision venue: WEPPcloud Roads development conversation, America/Los_Angeles.
- Participants: project owner and Codex, with independent scientific and implementation reviewers.
- Decision owner: project owner, authorizing defect investigation and remediation.
- Implementer: Codex.
Context
Independent watershed tracing during ADR-0114/0115 validation reproduces all nine
released and candidate output files exactly. On Cedar Creek 1985-11-06, one
channel's lateral hill supplies about 1.845 kg, but CHNRT injects about 136,060 kg.
The source-prefix reproduction and ten controls were committed in the native
repository as b0f33584f before repair.
CHNRT multiplies a lateral sediment concentration by an effective hydraulic flow. Its concentration weights use actual lateral water volume, while the hydraulic flow can represent a different volume. CHNERO's existing large-top-flow guard makes this extreme: it truncates virtual top length to zero, which CHNRT interprets as all-lateral flow. Clean channel water can amplify the source even without that guard; negative effective lateral volume can suppress a real source. This is a legacy shared channel defect, not a Roads sediment coefficient issue.
Proposed decision and isolated scope
In U7, compute each class's actual lateral input mass as the sum of
tmpvol(element) * sedcon(class, element) for the four assigned left/right
hillslope and impoundment slots. For a valid routed event, distribute that mass
uniformly over the actual computational span x(nptsc)-x(1) and rundur.
This makes integration by the existing segment DX values recover the assigned
source, independent of peak, effective hydraulic length, or clean-water dilution.
Retain the existing lateral conversion constants 35.31984 * 0.06241818.
Their SI roundtrip with 0.453592 is approximately 0.99998898248, so validation
must state the small existing unit-rounding discrepancy rather than claim exact
SI identity. The separate top-load factor 2.2064 has a different discrepancy;
changing it is outside this isolated lane.
Preserve positive rvolat eligibility. Zero duration or nonpositive computational
span returns no flux and avoids division; such positive-source inputs are invalid
routing domains, not successful conservation cases. Normalize against the actual
integrated span rather than nominal length because large virtual coordinates
round in single precision.
U7 does not modify water routing, channel hydraulics, the 1e5 guard, sediment transport/deposition formulas, soil properties, erosion parameters or output formats. The guard's incorrect all-top hydraulic limit is a separate confirmed investigation and cannot be declared repaired by mass normalization.
Alternatives and rationale
Clipping final sediment or fitting a coefficient would conceal created source mass. Replacing hydraulic geometry at the same time would prevent attribution. Scaling by nominal physical length misses computational-coordinate rounding. Independent source summation also avoids cross-weighting a hill and impoundment when both occupy a side; the input parser does not prohibit that combination.
Validation and risks
Require actual-source controls across both sides of the guard, ordinary support mismatch, each contributing slot, mixtures, zero source, negative effective hydraulic lateral volume, and defensive invalid domains. A newly reachable positive sediment source with zero effective lateral hydraulic flow also requires full consumer assessment: CASE12/CASE34 contain no-lateral-water assumptions. A producer-only conservation result cannot establish complete routing correctness.
After bounded controls, retain isolated Cedar results and paired broader native,
Roads, Disturbed and watchlist evidence for the final selected repair. Keep U6
source 97b2b1e3e and installed wepp_261029 as immutable comparators. No withheld
Roads materials, parameter retuning, release promotion or UI implementation is
implied. Roll back an invalid experimental lane; retain its evidence.
Evidence
Native work package:
/workdir/wepp-forest-release-20261009/docs/work-packages/20261030-roads-native-accounting/.
See artifacts/U6-Cedar-channel-independent-scientific-review.md,
artifacts/U7-independent-channel-lateral-mass-review.md, and incident reproduction
review_channel_lateral_mass.py with review-channel-lateral-mass-red.json.
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.