Researcher review: Woodfibre plume explorer

29 September 2026 · Review edition 1

Purpose: help reviewers assess whether the public Explorer is useful for scenario education, identify misleading interpretations, and define the work needed for quantitative community exposure estimates. This package is an invitation to review, not evidence of endorsement by any researcher or university.

Download research package (ZIP, 13 MB) · Package contents and checksums · Open the Explorer

September review snapshot

The following tables and checks describe Review edition 1 as archived on 29 September. For the current source inputs, use the November update above.

What is—and is not—connected

The AERSCREEN comparison holds each 30-minute case at its active rate for a full hour, doubling its historical event mass for that screen. The compressor case retains the model’s equivalent-diameter warning (W320) and unresolved applicability. The 14-hour tank case represents one active hour. These choices are not an actual flare timeline.

Reproducing an EPA reference calculation verifies that executable and configuration. The separate Baldwin coal-stack comparison investigates a plume treatment under different conditions. Neither validates the browser model or establishes transferability to an LNG flare in Howe Sound.

Start with these unresolved findings

Concentration convergence failed. An archived 48-to-64-column refinement changed the maximum lowest-cell concentration in the snapshot at one hour from 0.543 to 0.838 µg/m³ per 1 g/s source, and the Squamish marker from approximately 0.00123 to 0.000671 µg/m³. The changes are roughly +54% and −45%. This was one specified test, not a universal error bar or a bound on every preset. It changes terrain representation, cell geometry and numerical spreading together. The test is retained in the package as historical evidence; the new small verification suite does not rerun or supersede it. Detailed method

The map's wind and release height remain assumptions. The browser imposes inflow/outflow and vertical-exchange restrictions; it does not generate drainage winds or a temperature inversion. It holds those conditions fixed even for long releases. Flare heat does not determine the browser injection height. The separate meteorological investigations have identified substantial local discrepancies and have not qualified a replacement wind field for this map. Meteorological research

Near-surface is a layer mean. The lowest air-cell thickness varies with terrain and can be large. It is not a concentration evaluated at a person's breathing height. A finer source DEM does not make the transport grid equally fine.

The sources are historical reconstructions. WWGF feed-gas depressurization, LPCF tank shutdown and CDGF compressor maintenance use the October 2014 inventory and 2015 clarifications. Rates divide maximum daily mass by listed active duration, assuming a uniform release. The original hourly model schedule, current flare design and joint heat/emission time histories are not supplied by this arithmetic. These separate maxima are not automatically one simultaneous event. Source audit and primary documents

A benchmark colour is not a health-risk classification. The map shows source contributions without ambient background. NOx expressed on an NO₂ mass basis is not calculated ambient NO₂; the optional full-conversion screen is explicit. Total VOC cannot identify benzene or support a single toxicological threshold. PM₂.₅ comparisons require their full averaging period. No probabilities of illness, safety boundaries or regulatory attainment claims follow from these scenarios.

Authorship and review status

This community project was developed with AI assistance in research, code, calculations and writing. Earlier references to “independent checks,” “independent reconstruction” or “independent review” describe separate automated implementations or agent reviews within this project unless a named external source explicitly establishes otherwise. They do not mean independent human scientific peer review. Shared inputs and reasoning can carry shared errors.

We have not received a completed external scientific evaluation or institutional endorsement. Reported interest in evaluation is not an endorsement. Source documents retain their original authorship; a citation does not imply the source author reviewed this tool.

Reproduce the small checks

Unzip the package and start with its README. It contains the extracted browser engine, worker, embedded input data, historical source table and a Node-only verification script. No account, network request or package installation is needed to run the supplied checks with a compatible Node runtime.

The script checks source-rate arithmetic and selected analytically tractable transport properties. These establish implementation consistency only. The package also preserves the failed-refinement discussion and identifies checks not rerun. Checksums identify exact files; they do not certify scientific correctness. The full browser HTML is included as the reviewed snapshot; opening it requires its documented browser dependencies. The offline numerical checks do not.

The archive is intentionally a small entry point. Larger original-weather and established-model archives remain available through the evidence library. It does not claim to reproduce every historical research study with one command.

Questions for a first scientific review

1. Is the reconstruction of each historical inventory row defensible, including duration, contingency, PM categories and NOx mass basis? What original files would resolve the remaining ambiguity?
2. Should the custom browser transport be retained solely for teaching, or replaced for quantitative mapping? Which source, terrain, shoreline, buoyancy and weak-wind treatments would the proposed alternative need?
3. Which source-site, shoreline, town and vertical observations could test meteorology without reusing assimilated observations as purportedly independent validation?
4. What receptor heights, community coverage, background treatment, chemistry and averaging rules are needed for the intended public use?
5. Which public statements or map behaviours should be corrected now, and what evidence would justify stronger claims later?

For actionable feedback, identify the file or page, scenario settings, observed issue, proposed correction, supporting source and whether it changes a number, a method or an interpretation. Record disagreement and unresolved findings as well as accepted changes. No request or message is sent by this package.

Next research work, in order

These steps improve the scientific basis. This review edition improves traceability and makes the current limits testable; it does not upgrade the existing map to a validated exposure prediction.

Current-record check and event correction

The bounded 29 September public-record check is included in the download, with its search limits. It is not proof that an unlocated engineering document does not exist or has not been submitted.

The newly located July 2026 onsite monitoring report, dated 11 September, adds baseline context. It reports Level 0 validation, instrument downtime and further TSP quality review. Its passive VOC measurement is hexane-equivalent over a multiweek period; it does not supply benzene speciation or short-term peaks. No new current flare source schedule was recovered, and no new emissions have been inserted into the Explorer.

Woodfibre's official notice states that the open house originally scheduled for 17 September is now 1 October 2026, 4:30–7:30 p.m., Executive Suites, Squamish. The public question guide has been corrected. Organizer's notice, dated 15 September