RUN-A3 provenance + seed failure
seed 42003 · schema earth-state-0.1 · emulator emu-set-0.1 · tolerances tol-0.1 · constitution v1.0
prepared sample · not engine output
Intentional failure: one derived quantity has unresolved provenance, and a repeat of the seed produced a divergent trajectory. Affected quantities are withheld.
Withheld from display (2)
- global_mean_surface_tempSeed reproducibility failed: trajectory not deterministic under its declared seed. Withheld pending engine fix.
- sea_levelProvenance incomplete: an assumption in the sea-level chain is unresolved. Withheld rather than presented as auditable.
Scientific honesty over completeness: withheld, never dimmed.
Earth state by system layer
Constitutional gates — criteria evaluated for this run
This table lists only the criteria active for this run's tolerance set. It is not the full Constitutional Gate catalog: 7 catalogued gates are not evaluated for engine-0.2 and are listed below with their v1.0 disposition. They are neither passed nor failed and are excluded from every passed-gate count.
| Gate | Criterion | Measured | Result |
|---|
Catalogued gates not evaluated for this tolerance set
Carbon mass conservation
UPSTREAM · NOT RE-VERIFIEDUpstream property — not independently re-verified by SHFT v1.0
Carbon conservation is a property of the pinned upstream SURFER v3.0 formulation and is evidenced only by the reproduced upstream carbon-budget benchmark (carbon_budget_2000_2009_and_cumulative). The earth-state-0.2 schema stores no carbon reservoir masses. Of the two catalogued flux-like quantities, net_primary_productivity is defined but not represented; weathering_rate is represented. SHFT therefore holds no complete reservoir/flux inventory and cannot construct an independent closure test. Completeness is not inferred.
Energy balance closure
UPSTREAM · NOT RE-VERIFIEDUpstream property — not independently re-verified by SHFT v1.0
Energy closure is a property of the pinned upstream formulation. The adapter/payload stores only global mean surface temperature and ocean heat content; no top-of-atmosphere or component energy fluxes are stored, so no auditable closure residual can be computed by SHFT v1.0.
Hindcast: last deglaciation
DEFERRED · NOT VALIDATEDDeferred — not independently validated in v1.0
Deferred in v1.0. No reconstruction-based ice benchmark has been sourced or reproduced for the pinned backbone, and no engine-0.2 criterion evaluates ice volume against any target. No pass is inherited from tol-0.1.
Hindcast: sea level
DEFERRED · NOT VALIDATEDDeferred — not independently validated in v1.0
Deferred in v1.0. The upstream publication compares thermosteric sea level with IPCC evidence, but no sourced, reproducible adapter-path comparison exists in SHFT and no engine-0.2 criterion evaluates it. The seeded tol-0.1 demonstration result is hand-authored and is not evidence about the SURFER-backed engine.
Hindcast: surface temperature
UPSTREAM · NOT RE-VERIFIEDUpstream property — not independently re-verified by SHFT v1.0
Upstream reproduction evidence only. The reproduced pinned-model benchmarks atmospheric_co2_curve_1850_2015_rmse_ppm and gmst_anomaly_2014_k compare upstream output with published historical evaluation, and are recorded under upstream_reproduction. No engine-0.2 criterion compares served SHFT adapter-path output against a sourced historical temperature target, so there is no independent SHFT hindcast gate.
Long-run numerical stability
DEFERRED · NOT VALIDATEDDeferred — not independently validated in v1.0
Deferred in v1.0. No long-run SURFER-backed stability experiment has been executed and stored: the longest reproduced evidence is the LTMIP long-tail CO2 comparison, and the interface offers horizons far short of the 1 Myr integration the catalogued gate describes. The seeded tol-0.1 demonstration does not count.
Water mass conservation
NOT APPLICABLENot applicable to the active production backbone
SURFER v3.0, the sole active production backbone, models no hydrological cycle and SHFT Layer 1 stores no water inventory. The catalogued river_discharge quantity is defined but not represented. There is no quantity against which water mass closure could be tested, so the gate is outside the scope of the active backbone and must never be displayed as passed.