# Current flare elevations and regional-weather support

5 November 2026. A source-to-weather compatibility check; no dispersion or weather-validation claim.

The current application’s source coordinates still select the same native regional-weather cell as the earlier site proxy. The new physical and equivalent source elevations expose which levels that cell can and cannot supply. Raw nearest-cell winds should not be substituted into the Explorer as source-height winds.

## Evidence and method

The October application gives HP and LP physical stack heights of 134 m above an 8.5 m ASL grade, and HP/LP upset equivalent POINT heights of 208.3 and 154.3 m above grade. These equivalent parameters initialize a model representation of a flare; they are not final plume-rise heights. Source data and coordinate transformation are retained in `work/flare-document-2026-11-05/geometry-audit/current-application-geometry.json` and `work/current-source-screening-2026-11-05/source-records.json`.

Source: https://woodfibrelng.ca/wp-content/uploads/2026/10/Woodfibre-LNG-Project_WDA-Air-Emissions-Permit-Application-Package.pdf (PDF194–195, DMP Table6.5).

All 1,800 retained regional cell centres were ranked by great-circle distance. The nearest cell is I28,J74 for both sources, about 319 m from HP and 321 m from LP. Its rounded terrain is 159 m MSL. Nearby cells have very different terrain: 54 m about 1.02 km away and 488 m about 1.30 km away. No cell was selected to achieve a preferred height or wind.

The analysis reuses all 24 UTC hours of the previously selected 2 June 2023 native-profile sample. Its original 40 levels per hour, field meanings and independent extraction checks are retained. No winter, inversion, outflow or climatological representativeness is claimed for this June day. Full source lineage is in `outputs/woodfibre-wrf-profile-sample-2023-06-02-sources.json` and the associated methods/checks files. Provider: https://wrf.nrs.gov.bc.ca/.

## Nominal elevation support

| Source reference | Application height + grade, m ASL | Nearest native column, all 24 sample hours |
|---|---:|---|
| HP/LP physical tip, including HP vent | 142.5 | Below the column’s 159 m terrain; no wind lookup produced |
| HP upset equivalent POINT centre | 216.8 | Bracketed by native profile levels; diagnostic u/v only |
| LP upset equivalent POINT centre | 162.8 | Above model terrain but below its lowest 171 m profile level; no extrapolation produced |

These compare as-listed vertical references. The application’s ASL grade has no recovered vertical-datum tie, while native profile heights are rounded model geopotential heights. No precise engineering vertical transformation is claimed. Even nominal height bracketing does not establish that winds at the displaced cell represent the source or rising plume.

Using 134 m above **model** terrain would sample nominally at 293 m MSL, not at the application’s 142.5 m tip elevation. The two terrain origins differ by 150.5 m as listed. The same origin mismatch applies to the equivalent heights. AGL is not interchangeable across grids with different terrain.

The code retains missing/out-of-support values. For the bracketed HP equivalent target only, it linearly interpolates earth-oriented horizontal velocity components, rather than averaging direction angles. These values are native-column diagnostics, not qualified source winds. A 10 m surface field, extrapolation below terrain, or choosing a farther lower-terrain cell does not by itself repair the physical representation.

## Consequence for the prediction work

The archive may still be useful as input to a suitable terrain/weather modelling chain. This check rules out a simple nearest-column source-wind substitution; it does not rule out regional-model downscaling. Our earlier CALMET and station comparisons show that downscaling arithmetic and larger terrain coverage alone have not established local wind skill.

The immediate priorities remain:

1. Recover the applicant’s receptor-keyed concentration outputs and source/weather control package to reproduce its community results. This is the prepared, unsent BCER records request, pending user approval.
2. Recover the actual site grading surface and vertical-datum tie, then use an explicit terrain-to-weather height mapping.
3. Obtain coincident source/corridor observations and elevated wind/temperature profiles to test shoreline flow and inversion/outflow structure. The previously identified ECCC Squamish profile archive needs custodian access; the downloaded public ZIP supplied surface observations only.

Until those inputs are available, do not relabel the Explorer’s assumed winds or air-layer mean concentrations as ground-level community predictions. The existing applicant-results report and conditional scenario display remain available separately.

## Verification and retained files

`protocol.json` freezes input hashes, targets, selection and missing-value rules before classification. `nearest-current-source-cells.json` retains the four nearest cells per source. `vertical-support.csv` retains all 72 target-hour records, with blank winds where unsupported. `verify.py` separately checks every support classification using decimal heights and all eight retained distances using a spherical-vector calculation. All checks pass; this is data/coordinate arithmetic verification, not validation of the atmosphere.

No source rates, model winds or public Explorer calculations changed in this turn. This research result is now linked from the current public roadmap. The source geometry and weather methods remain available in the evidence library.

Public data: woodfibre-current-source-weather-support-2026-11-05.csv
Arithmetic checks: woodfibre-current-source-weather-support-2026-11-05-checks.json