5 September 2026 · Follow-up to the historical screening study
The three selected historical screening peaks depend almost entirely on AERMOD's treatment of plume material that penetrates above the mixed layer. At each selected peak, that component accounts for more than 99.99% of the calculated concentration at 1.5 m above flat ground. This identifies a controlling model assumption to investigate. It does not establish how an actual Woodfibre plume behaves or validate community exposure.
We reproduced the original peak concentration tokens exactly using the unchanged model and the same source parameters. The new runs select one original synthetic weather record per source and request the model's supported diagnostic output. No plume-rise option, source heat, diameter or receptor was adjusted to make the result appear plausible.
| Historical case | Mass assigned above the mixed layer, as printed | Share of calculated ground peak from that component | Synthetic mixed-layer height | Penetrated-component mean-height parameter at the selected distance |
|---|---|---|---|---|
| WWGF | 100.0% | 100% | 520 m | 858.4 m |
| LPCF | 96.7% | >99.99% | 437 m | 703.1 m |
| CDGF — diameter warning | 86.8% | >99.99% | 989 m | 1,507.9 m |
All heights are model quantities above the same hypothetical flat plane. The last column is a mathematical component parameter at the selected receptor distance, not a measured altitude, the top of a plume, or a trajectory. The three cases use different synthetic weather and cannot be combined as a single atmospheric event.
Mass fraction and concentration contribution are different. For example, the CDGF calculation assigns about 86.8% of emitted mass to the penetrated component, yet this component supplies almost all of its calculated ground peak. Dispersion determines how much each component contributes at that particular place and height. Adding the direct and indirect components' displayed source masses would also be wrong: they are terms representing the same below-layer material in the formulation.
The earlier AERSCREEN report's separate final-height diagnostics—approximately 9.14, 9.27 and 39.52 km—therefore do not explain the ground concentration as literal plume altitudes. The newly recovered component values clarify which part of the actual concentration calculation controls these peaks. They do not remove the need to evaluate highly buoyant plume behaviour. CDGF's equivalent-diameter warning remains unresolved. Earlier applicability review
EPA documents direct, indirect and penetrated convective plume components, with a separate blend of coherent and meandering contributions. It also describes an experimental highly buoyant plume option addressing premature downward mixing under some conditions. That option depends on evolving mixing heights; we have not enabled it on synthetic matrix records or assumed a known correction for Woodfibre. EPA formulation, §§5.2, 5.6 and Appendix B.2.12
The three selected cases come from the previously frozen highest-peak source cases, at 4,715 m for WWGF, 3,955 m for LPCF and 7,355 m for CDGF. Each replay retains all 302 original refined receptors, source geometry, the 1.5 m receptor height and the effective model-option list. The original surface/profile files are preserved, and each selected record is copied byte-for-byte with its original artificial time label.
The three final peak tokens match the original AERMOD plot output exactly: 0.135102, 0.397879 and 0.0884742 µg/m³, respectively, for 1 g/s of passive pollutant at fixed historical heat. These are the original calculation's more precise printed values; the earlier AERSCREEN summary rounds them further. They are not current chemical concentrations or finite-event exposure averages.
Across 906 receptor calculations, the parsed direct, indirect and penetrated concentrations sum to the corresponding coherent/meander components within printed precision. Their weighted blend matches the final diagnostic concentration and its plot-file value. The penetrated share at the peak is reconstructed from total minus the very small direct/indirect contributions, with rounding limits carried through. These rounding limits are numerical reporting limits, not atmospheric uncertainty intervals.
The parser uses the original Fortran field widths because large adjacent values run together in the diagnostic text. Asterisks in overflowing indirect-component height fields remain missing, with their original text retained. They are not replaced by zeros or invented heights. Intermediate rows contain a not-yet-updated AERVAL field; the check uses the subsequent final blend rather than interpreting those intermediate zeros as results. Original AERMOD source, calc1.f DEBOUT, CPLUME and coherent/meander blend
One failed selection attempt is preserved. Applying a calendar start/end filter to the complete synthetic matrix returned no target result despite no model warnings. The matrix's artificial date labels are not chronological: in the WWGF case, record 140 exceeds the requested end label, before the intended record 350. AERMOD stops reading there. Extracting the exact selected record resolved that selection problem and restored the original concentration. This reinforces why these files cannot represent real inversion evolution. EPA user guide, §§3.2.18 and 3.5.4
The immediate modelling question is now specific: does the selected model reliably predict penetration and subsequent ground-level contributions for the relevant flare heat, local vertical profiles and event duration? More map detail cannot answer that question.
Prioritize the current flare's linked heat/emission history and simultaneous wind and temperature profiles above the source and transport corridor. Those observations must resolve the mixed-layer/inversion structure and changes through an event. Seek relevant validation studies with comparable buoyancy and operating conditions; a successful source equation or a smaller coefficient is insufficient. Evaluate the chosen transport formulation against those conditions before proposing any correction or promoting screening curves to neighbourhood contours.
The research plan and model-method roadmap retain the full priorities. The community map remains illustrative, with its original solver and concentration colours unchanged. No B.C. objective, safe-distance or health-risk conclusion follows from this diagnostic.
The compact diagnostic summary, all-receptor table and frozen diagnostic evidence pack provide the numerical evidence. The pack preserves the successful replays, failed attempt, diagnostic files, source references, processing and checks; previous archives remain unchanged.