Prepared 5 September 2026. Use 36 separate cases: three historical heat sources × six surface examples × two temperature sets. This is a normalized, sustained-source sensitivity using AERSCREEN 21112 and MAKEMET 16216. This review ran no models and changed no existing visualization or source defaults.
Each case releases a 1 g/s inert pollutant tracer while retaining the full historical flare heat. It does not represent 1 g/s of fuel. Physical stack height is 100 m and radiant heat loss is 0.55. Total heat is 829,441,536 cal/s for WWGF, 214,146,240 for LPCF and 2,235,346,960 for CDGF, as reconstructed in the historical source audit. Later linear scaling by a pollutant rate is valid only while heat, geometry and other model settings remain fixed. This does not resolve the historical hourly-emissions ambiguity or establish current emissions.
Use the following explicit common settings:
| Setting | Selected value | Meaning |
|---|---|---|
| Minimum wind speed | 0.5 m/s | EPA routine screening default; true zero wind is excluded |
| Wind reference height | 10 m | Nominal MAKEMET height; not an asserted onsite sensor height |
| Adjust surface friction velocity | N | Unadjusted option; consistent with BC’s low-wind guidance |
| Rural/urban classification | Rural | Explicit generic screening assumption |
| Receptor height | 1.5 m | Enable flagpole receptors explicitly |
| Minimum receptor distance | 30 m | Analyst-selected computational start; public-access boundary unknown |
| Maximum probe distance | 20,000 m | Distance-curve scope; check for a peak at the outer limit |
| Terrain, building downwash | Off | Flat hypothetical setting |
| Inversion-breakup and shoreline fumigation | Off | These processes remain outside this sensitivity |
EPA documents the wind/reference-height defaults and flagpole option. BC identifies 1.5 m as breathing height and recommends leaving the adjustable low-wind options disabled; that latter recommendation occurs in its AERMET/AERMOD calms discussion. AERSCREEN’s default receptor is 0 m if flagpole height is not enabled. Verify the emitted input and final output show 1.5 m before describing these as breathing-height curves. EPA guide, printed pp5 and 31–32/PDF pp16 and 42–43, BC July 2026 guideline, printed pp44 and 65/PDF pp51 and 72.
Use both temperature sets for every source/surface pair:
| Set | Entered range | MAKEMET candidate temperatures |
|---|---|---|
| EPA default | 250–310 K | 250, 280, 310 K |
| Expanded sensitivity | 250–320 K | 250, 285, 320 K |
The expanded range, −23.15 to 46.85°C, retains EPA’s lower default and rounds the upper endpoint outward to the next 5 K increment above 316.15 K. The inspected Squamish Airport series reports −17.6°C and 38.2°C as its 1982–2020 long-term extrema, and 43.0°C on 28 June 2021. These are regional context, not temperature measurements at the flare or guaranteed future bounds. The June 2021 peak has a null native flag; that alone does not establish a final validation level. ECCC normals and long-term extremes, ECCC June 2021 daily records.
EPA asks for regional record extremes when overriding its defaults. MAKEMET samples the minimum, midpoint and maximum, with additional internal filtering. The expanded interval therefore does not contain every old meteorological combination. Retaining both sets preserves their temperature candidates; even their union is not a continuous temperature search. EPA guide, printed pp21–24/PDF pp32–35, MAKEMET source, lines 403–407. The temperature evidence and June daily sample preserve provenance.
Use these six separate homogeneous surface analogues, entered as manual values:
| Surface and condition | Albedo | Bowen ratio | Roughness, m |
|---|---|---|---|
| Water, summer analogue | 0.10 | 0.10 | 0.001 |
| Water, snow-free autumn analogue | 0.14 | 0.10 | 0.001 |
| Grassland, summer analogue | 0.18 | 0.80 | 0.10 |
| Grassland, snow-free autumn analogue | 0.20 | 1.00 | 0.01 |
| Coniferous forest, summer analogue | 0.12 | 0.30 | 1.30 |
| Coniferous forest, snow-free autumn analogue | 0.12 | 0.80 | 1.30 |
These use BC’s seasonal tables with the average-moisture category assumed. They are a small sensitivity sample, not mapped current conditions or a proven concentration bound. Grassland is not established as the correct analogue for an industrial clearing. Summer/autumn choices also omit wet/dry and snow-cover sensitivities. BC Tables 4.2–4.4b, printed pp24–26/PDF pp31–33.
One discrepancy must remain visible: BC’s water roughness table gives 0.0001 m, while AERSCREEN imposes a 0.001 m floor and uses that value in its built-in water table. Enter 0.001 m explicitly. For these manual cases use Surf=0 and Clim=0; the climate field is ignored. Built-in land-use codes are 1 water, 2 deciduous forest, 3 coniferous forest, 4 swamp, 5 cultivated land, 6 grassland, 7 urban and 8 desert shrubland; moisture codes are 1 average, 2 wet and 3 dry. The standard input has no individual season selector. Built-in mode cycles through seasons, including winter states that should not silently be assigned to an unfrozen fjord. EPA guide, printed pp22–24/PDF pp33–35, AERSCREEN source, lines 1168–1245 and 2385–2390.
The synthetic matrix uses discrete stable wind candidates up to 18 m/s and convective candidates up to 10 m/s, with internal filtering; it supplies no local frequency or real event dates. A “water” setting changes surface parameters but does not simulate a shoreline crossing. AERSCREEN has separate shoreline fumigation equations for eligible point sources, including flares, with specific flat-ground assumptions. Leaving them off means coastal fumigation has not been assessed. BC recommends CALPUFF Hybrid where coastal effects matter. EPA guide, printed pp57–60/PDF pp68–71, BC guideline, printed p88/PDF p95.
The useful result is a reproducible comparison of concentration per unit emission, peak distance and sensitivity to source heat and assumed surfaces. Each distance can reach its maximum under different synthetic conditions; the curve is not one simultaneous plume. It cannot locate neighbourhood exposure, validate the experimental valley map, or establish a safe distance. A 1.5 m point receptor also differs from that map’s near-surface cell mean. Preserve both temperature-set results and all warnings, check the actual switches and receptor height, and report any outer-distance truncation. The accompanying machine-readable input basis contains exact settings, code references, source hashes and acceptance checks.