5 September 2026 · June 2023 data review
Tantalus and Eagle Ridge provide two additional wind records for testing weather models along Howe Sound. Both retained June files contain all 720 expected hourly calendar labels and numeric wind fields. They broaden the regional context beyond the 20 × 20 km community view. Their clock, wind-bearing definitions and measurement history need confirmation before they can support an absolute-time wind comparison.
| Station | June calendar labels | Populated speed/direction records | Literal zero-speed records | Current inventory location |
|---|---|---|---|---|
| Tantalus 11092 | 720 | 720 | 0 | 49.84351, −123.14309; station elevation 320 m |
| Eagle Ridge 11191 | 720 | 720 | 139 | 49.3718, −123.26973; station elevation 140 m |
The original labels run from 2023/06/01 00:00 through 2023/06/30 23:00. No labels are missing, duplicated, malformed or outside that nominal month. This is a check of printed calendar labels; no timezone or elapsed-hour interpretation has been assigned. Inventory elevations are not sensor heights, and current metadata do not prove the historical installation.
All four wind columns—speed, direction, gust and standard deviation—are finite numeric in every row. The current archive describes speed and direction as preceding-hour vector averages, speed/gust in km/h, and direction in degrees N. The standard-deviation unit is not supplied. Numeric availability is not a measurement-quality certification.
Eagle Ridge's 139 literal 0.0 speed records retain numeric bearings, and all have the same 0.071 gust token. Those values are preserved. They do not establish an instrument detection threshold, a fault, full-hour calm or a meaningful wind direction at zero speed. The applicable logger and calm-wind definitions are needed to interpret them.
Each station has 660 RAW and 60 ST observation-type tokens. ST appears at the native 06:00 and 18:00 labels. These tokens are kept separate from quality flags. Snow fields are entirely blank; negative precipitation values are also retained without correction or physical interpretation.
The native-data audit froze its scope and original hashes before the full parse, then passed 2,936 integrity and accounting checks. It did not use earlier row counts as expected answers. A separate standard-library parser reconstructed both originals and confirmed all 27,360 native field tokens in the downloadable copy, across 1,440 rows and 19 native columns.
The copy includes original file, row, physical-line and byte provenance. CSV formatting changes, but named field strings remain exact. Original hashes and the checking scope are in the companion record. These are computational integrity checks, not independent calibration or meteorological validation.
Download the native-field copy · Checks and provenance
| Record to obtain | What it resolves |
|---|---|
| Upstream monthly-export specification and logger clock history | Timestamp zone, daylight-time policy, ambiguous/missing labels and the exact preceding-hour interval. The different archive's ECCC clock rule cannot be imported. |
| Wind channel and processing definitions effective in 2023 | Whether bearings mean wind-from or wind-to, north reference, alignment, scalar/vector averaging, sample recovery and zero/variable-wind handling. |
| Dated sensor location, height, exposure, maintenance and QA records | The actual quantity and installation being compared, including trees, terrain and road-cut exposure. RAW is not a substitute for these records. |
| WRF production, assimilation and evaluation records | Whether station observations were supplied, accepted and used by the model or upstream analyses. Network membership alone proves neither use nor independence. |
The current hourly model fields are states at their timestamps, while the road archive describes preceding-hour averages. A shared timestamp does not make those quantities identical. A future protocol must match the observation operator or state the remaining approximation; two hourly endpoints cannot recover unobserved subhourly wind reversals.
Station and episode selection, spatial sampling and exclusions should be fixed before inspecting errors. Choosing whichever time offset, bearing sign or neighbouring model cell fits best would not recover missing metadata. The retained 2025 road files could support later tests of the already selected HRRR episodes once qualified; 2025 observations do not overlap the 2019–2023 WRF release.
The archive's current pressure label and legacy documentation also disagree on units, and wind standard deviation lacks a verified algorithm. Neither field is used here to diagnose outflow or turbulence. No UTC assignment, wind rotation, forecast score or concentration calculation was performed.
The full audit code, field inventories and frozen original files remain in the research workspace. These downloads provide the checked data copy and provenance, rather than a standalone reproduction package. No request to a custodian was sent.