Woodfibre Explorer community comparison — 5 November 2026

PURPOSE
Compare the existing historical and proposed 2026 source scenarios under five
identical, selected weather settings. These are source-only conditional results,
not observed exposure, calibrated forecasts, illness probabilities or permit tests.
The tables use ten community reference points, not every home or community maxima.

QUICK FILE CHECK (Python 3, standard library only)
  python3 verify.py
This checks every bundled hash, 3,500 scaled values, CSV consistency, units and
averaging labels. It does not rerun atmospheric transport.

EXACT SOFTWARE REPLAY (Node.js 22 or newer, plus Python 3)
  node rerun.cjs
  python3 analyze-rerun.py
  python3 verify.py --rerun
The runner reads only the frozen inputs/ files and writes results to rerun/.
It uses no network or third-party package. Recorded results remain untouched.
27 unit runs may take a few minutes. Runtime fields naturally vary, so the final
check compares scientific output arrays/tables rather than output-file hashes.

INPUTS AND METHOD
inputs/explorer.html freezes the exact public Explorer source, including its
embedded terrain, reference points, historical inventory, engine and worker.
inputs/data/proposed-flare-sources-2026.json freezes the proposed source catalogue.
The engine SHA256 is ecec70ed2c1c308dd201360fd8058417eadadb4265d4cd301e10bd0e0031719f.
The current source notice PDF is included; the 120.6 MB full application and old
inventory PDF are linked in sources/findings.md rather than silently changed.
The recorded/comparison.json records hashes, all inputs/settings, individual
results and domain/overlap summaries. sources/applicant-results.json is a separate
bounded reading of the applicant's CALPUFF report; its numbers do not enter ours.

Five fixed winds: inversion inflow from 200 degrees at 3 m/s; inversion outflow
from 20 degrees at 3 m/s; mixed inflow/outflow at those directions and speeds;
and weak inflow from 200 degrees at 0.5 m/s. Inversion altitude is 500 m above
sea level when present. Upper wind follows the same direction. PM removal is off.
Each calculation covers 24 hours with constant selected rates and weather.
Current continuous cases begin from clean air for a 24-hour analysis window.
The vent case shows only the first 24 hours of its 48-hour default release.
Historical source injection: 200 m above its coarse source terrain, 205.1552 m ASL.
Current source injection: the derived physical tip altitude, 142.5 m ASL.
Heat/plume rise, exit momentum, chemistry and background are not calculated.
Source geometry, duration and rate all change between the old and new defaults.

The 27 unit runs include five weather settings by five geometry/duration variants,
plus HP/LP same-cell and first-day 24/48-hour equivalence checks. These are numeric
equivalences at the coarse grid, not a claim of identical physical sources.
Each source is scaled separately by g/s. No concurrent HP+LP or whole-facility
emissions total is inferred. Values are means of the lowest air cell above terrain;
these layers span about 9–47 m at the reference points, not a 2 m breathing height.

CSV DATA DICTIONARY
wind: one of the five identifiers above (stagnant means the selected weak inflow).
profile: current source identifier or historical WWGF/CDGF/LPCF.
community: the selected named reference point, not an area maximum.
species: PM25, NOx, CO, VOC, SO2, H2S or TRS.
metric: PM25 uses the full first-day mean; others use the largest of 24 hourly means.
valueUgM3: source contribution in micrograms per cubic metre. Blank means unavailable,
not zero. A tiny positive value does not demonstrate actual exposure.
benchmarkUgM3/benchmarkRatio: numeric screen where configured. PM25=25 (24h),
SO2=170 (1h), CO=14300 (1h legacy reference). These do not implement multiyear
attainment statistics. NOx has no default NO2 conversion or benchmark here.
lowestLayerDepthM: depth of the sampled coarse model air layer.
VOC is total unspeciated mass, not benzene or mixture ppm. H2S and TRS can overlap
and must not be added. A rate or concentration mass ranking is not a toxicity ranking.

INTERPRETATION LIMITS
No probability, population weighting or comprehensive worst-case search is supplied.
Some near-source cells exceed numeric screens, while current cases do not cross
PM25/SO2/CO screens at these ten points in the five weather tests. Residential polygon
coverage is incomplete. The transport engine has known grid-spreading sensitivity;
a low point value or an unshaded map is not evidence of safety or regulatory compliance.
The separate historical heat-aware calculator and saved CALMET wind viewer are not
used by this experiment. Better source data does not constitute field validation.

All source documents retain their own authorship. This is AI-assisted software and
source verification, not independent human scientific review or institutional approval.