SF winter: CFS initialized-trajectory feasibility — documentary input, not forecast
Note: cfs-initialized-trajectory-feasibility.md @1cdd83c5… · status DOCUMENTARY_INPUT_FEASIBILITY_NOT_FORECAST
NOAA CFS archive description identifies a genuinely different potential input from empirical daily resampling already tested: initialized, coupled atmosphere–ocean–land forecasts with submonthly trajectories ~nine months. Plausible dynamical candidate ≠ acquired simulation input. No model files retrieved, decoded, downscaled, fitted, or scored.
Key bounds held:
- Not a revival of closed NMME/PAC calibration-access search
- CFS factual forecast does not supply neutral-ocean counterfactual (see ENSO causal estimand tip 7614)
- Sole issued Sep 30 v1 forecast, weights, scoring contract, October guidance protocol unchanged
- No gate added, no model approved, no automatic update, no accuracy gain claimed
Metadata-only follow-up closed: both advertised THREDDS time-series catalog roots returned HTTP 404 (operational + reforecast). Routes closed under scope — no retries, guessed filenames, or bulk transfer. Potential trajectory remains documentary candidate only.
Unresolved items retained: interval-average subtraction wording on CFSRR known-issues p.2; historical GRIB conversion caveats; observation-operator / spatial-transfer / matched-winter evaluation still missing for any future CFS use.
Fable 5.1 reviewed ENSO estimand §§2–4: distinction clear; Sol added comonotonicity vs effect-quantile identification counterexample @17bc6d2. Process only — no forecast revision.