Prepared 5 September 2026. The accompanying woodfibre-terrain-grid.json contains real NRCan elevation data, resampled to a compact 20 × 20 km model grid. It is not an invented landscape. Terrain alone does not specify the wind field.
The source is the Canadian Digital Elevation Model (CDEM), tile 092G, downloaded directly from Natural Resources Canada. The archive contains the native GeoTIFF and its XML/PDF metadata. The file covers the full study area. The archive's HTTP modification date is 23 July 2019; the included metadata gives publication 5 January 2017 and creation November 2012. These dates describe the mosaic/archive, not a 2017–2019 survey of every location. The official collection is a legacy historical product with underlying temporal coverage 1945–2011. Its metadata warns that elevations may describe either ground or a reflective surface. This is therefore not a current construction or engineering survey. The source licence is Open Government Licence – Canada. Official CDEM catalogue, downloaded tile archive.
The native GeoTIFF is cdem_dem_092G.tif, 9,601 × 4,801 cells with 0.75 arc-second spacing in each geographic direction, approximately 15 m east–west and 23 m north–south at the study latitude. Its horizontal reference is NAD83(CSRS), EPSG:4617. Its vertical reference is CGVD28, in metres. The elevations retain that datum; no conversion to ellipsoid height, EGM96, CGVD2013 or an assumed sea-level surface was applied. NRCan CDEM specifications.
NRCan now provides MRDEM as a supported successor, with a 30 m grid and both terrain/surface products. A future version can replace this legacy terrain and compare the result. That replacement would also require managing its different vertical reference. Official MRDEM catalogue.
| Property | Value |
|---|---|
| Horizontal domain | Exactly 20,000 × 20,000 m in a local azimuthal equidistant projection |
| Projection centre | Longitude −123.22°, latitude 49.703°, WGS84 |
| Boundary extents | x = −10,000 to +10,000 m; y = −10,000 to +10,000 m |
| Cell count | nx = 100, ny = 100 |
| Cell width/height | dxM = 200, dyM = 200 |
| Array | elevationsM[row][col] |
| Row direction | Row 0 is north; rows increase southward |
| Column direction | Column 0 is west; columns increase eastward |
| Cell-centre location | x = -9900 + 200 × col; y = 9900 - 200 × row |
| Elevation unit/datum | Metres, CGVD28 |
| Valid / missing values | 10,000 valid; 0 missing; nodata = null |
| Stored elevation range | 0.0–2,166.1 m |
| Mean stored cell elevation | 726.78 m |
The JSON contains corner longitude/latitude coordinates, the projection definition, original source specifications, a file SHA-256 hash, processing details and reference-point checks. Horizontal distances and simulation cells should use the local metre coordinates, rather than treating degrees of longitude and latitude as equal distances.
For each 200 m cell, 25 samples form a 5 × 5 lattice at 40 m spacing, with offsets −80, −40, 0, +40, +80 m from the centre. Each sample is transformed from the local WGS84 projection to the source NAD83(CSRS) coordinates and bilinearly interpolated from the native GeoTIFF. The arithmetic mean becomes the representative model-cell elevation, rounded to 0.1 m. The extra decimal is storage precision, not evidence of decimetre survey accuracy.
No synthetic mountains, arbitrary terrain deformation, coastline-height overwrite or zero clamping was added. Sea-level zeros come from the source DEM. A cell spanning a steep shoreline may average land and water heights. The 200 m grid smooths smaller cliffs, gullies and structures; its representative elevation is not a maximum obstacle height or a description of subcell air volume.
Four point values were independently checked against the public NRCan CDEM point-elevation endpoint. Bilinear source samples and the API's returned terrain-node heights agree within 0.9 m at these locations. This verifies decoding and placement, not the accuracy of the original terrain survey. NRCan elevation API documentation.
| Location | Exact-point bilinear CDEM (m) | NRCan API (m) | Representative 200 m model-cell mean (m) |
|---|---|---|---|
| Woodfibre site proxy | 8.83 | 9 | 11.6 |
| Squamish city marker | 5.06 | 5 | 4.9 |
| Squamish Airport station coordinate | 59.34 | 59 | 61.9 |
| Mountain sample at −123.32°, 49.755° | 1097.90 | 1097 | 1143.5 |
The airport terrain value is not forced to match the separate station catalogue elevation of 53.7 m. Those records have different source/measurement contexts. The difference is retained transparently.
The Woodfibre site proxy has local coordinates x = −2,467.88 m, y = −4,197.82 m. It falls in row 70, column 37. These are the same company-published facility proxy coordinates used by the first map; they are not a surveyed flare-stack location. A physical release height must be added to an explicitly chosen ground elevation, with the datum and cell representation made consistent.
Jackson and Steyn's field network, vertical soundings and three-dimensional simulations of Howe Sound gap winds found considerable variation along the fjord, across it and with height. In their studied conditions, winds tended to strengthen down-channel, and the strongest winds occurred below roughly 1,000 m depth. The research shows why a constant horizontal wind cannot stand in for this system. Those results do not establish a universal 1,000 m mixing height. Jackson and Steyn, 1994, Gap Winds in a Fjord. Part I.
A later Howe Sound study used satellite radar and the Squamish Airport wind profiler to examine horizontal and vertical structure. In one February outflow case, a layer about 0.7–1.5 km above the surface marked the top of the outflow, with a cold near-surface layer. This is evidence for time-varying vertical structure, not a fixed plume ceiling for all weather. Along-Channel Winds in Howe Sound: Climatological Analysis and Case Studies, 2016.
Research on the December 1992 severe gap-wind event also found different hydraulic behavior under weaker and stronger forcing, including controls and jumps in the wind layer. An obstacle-aware velocity calculation can illustrate terrain effects, but cannot reproduce those thermodynamic and pressure-driven processes without meteorological forcing and validation. Hydraulic Physical Modeling and Observations of a Severe Gap Wind, 1994.
The actual DEM therefore supports real terrain geometry, slopes, model-cell air/ground masks and map relief. It does not independently provide wind speed, direction, temperature stratification, turbulence, mixing depth, sea-breeze reversal, drainage flows, or chemical concentration validation. Any such settings in the exploratory engine remain model assumptions unless separately backed by observations or a meteorological model. The atmosphere should extend above the domain's approximately 2.17 km highest represented terrain.