Reviewed 5 September 2026. Scope: selection and preparation of the acquired 2023–2025 Squamish Elementary record as a candidate dispersion background. No station statistics or model results were calculated for this review, and no monitor was accepted as representative of Woodfibre or the whole Howe Sound domain.
The controlling general source is the saved July 2026 B.C. Air Quality Dispersion Modelling Guideline, particularly §8.1, printed pp.70–73 / PDF pp.77–80. Its §8.2 expressly refers NO₂ methods to the separate Guidance for NO₂ Dispersion Modelling in British Columbia. The official separate PDF retrieved for this review is July 2022, not July 2026; its §2 explicitly replaces several older ARM/OLM/background instructions still reproduced in the general guideline. This distinction matters: use its revised chemistry and background sequence rather than treating the general guideline's older conversion paragraphs as the final method. General guideline, NO₂ guidance, §§1.3 and 2, printed pp.1, 5–6 / PDF pp.5, 9–10.
Background includes natural and human sources other than the sources explicitly modelled. Cumulative concentration is background plus the modelled increment. In preference order, §8.1 selects a nearby long-term monitoring network, representative long-term monitoring elsewhere, then a modelled background. A single station value applied across the entire domain is an assumption, not a demonstrated spatial field. If some monitored contributors are also explicitly modelled, their contribution must be handled consistently to avoid counting them twice; §8.1 gives the definition but no universal monitor-subtraction algorithm. General guideline, §8.1, p.70 / PDF p.77.
For off-site observations, examine similarity of meteorology, topography and surface features; emission types and changes; monitor distance from influential sources; and instrumentation and collection protocols. Distance alone does not establish suitability. A town station may describe an urban receptor better than the air encountered by an elevated plume over water. That last statement is a representativeness inference for investigation, not a finding about the selected station. NO₂ guidance adds local chemistry, ozone, marine activity, industrial sources, traffic and population/land-cover similarity; mixed coastal/urban/industrial categories warrant discussion with the Ministry. General guideline, §§8.1.1–8.1.2, pp.70–71 / PDF pp.77–78, NO₂ guidance, §3.3.2, pp.16–17 / PDF pp.20–21; §4.2, pp.19–20 / PDF pp.23–24.
The general guideline recommends the most recent year of representative hourly monitoring, requires at least one year, and recommends 75% completeness in each quarter. For this acquired record, 2025 is the latest complete calendar year available in the project; 2023 and 2024 can be identified separately as interannual sensitivities. A three-year objective-achievement value is not the default dispersion background. Averaging periods of background, model increment and comparison criterion must match. General guideline, §8.1.4, p.71 / PDF p.78.
| Requested cumulative period | Fixed background specified by §8.1.4 |
|---|---|
| Level 1 short-period assessment | Maximum observed concentration for the matching averaging period, using the screened or unscreened record. Thus an hourly maximum is not a daily background. |
| Level 2/3, 1-hour NO₂ | Annual 98th percentile of daily maximum 1-hour observations. |
| Level 2/3, 1-hour SO₂ | Annual 99th percentile of daily maximum 1-hour observations. |
| Level 2/3, 1-hour other pollutants | 98th percentile of hourly observations. |
| Level 2/3, 24-hour | 98th percentile of daily 24-hour values. |
| Annual | Mean of hourly observations over the year; this bullet is not restricted to a stated assessment level. |
These rules are on printed pp.71–72 / PDF pp.78–79. For multiple representative stations, calculate each station's relevant statistic first, then take the arithmetic average of the station statistics; do not pool their hourly samples or average 1-hour, 24-hour and annual statistics together. Present both cumulative and incremental modelling results. The annual hourly mean is distinct from the quarter-weighted PM₂.₅ objective-achievement calculation previously documented. General guideline, §8.1.4.
Calculation details not settled by §8.1.4: It does not prescribe a percentile estimator for incomplete records, daily-maximum validity, daily-average minimum hours, rounding, RAW_VALUE versus ROUNDED_VALUE, time-varying 3-/8-hour backgrounds, or the completeness denominator after source-screening. Do not silently substitute objective-achievement exceptions that retain incomplete high-concentration days or years.
There is useful primary support for a transparent candidate convention:
K = N - floor(p*N), retains tied observations separately and uses p=0.98. This can be adopted and labelled as the candidate percentile convention; the dispersion guideline does not expressly import it for every pollutant, hour-of-day subset or background statistic. Do not use an interpolated software default without stating it.Field Sampling Manual, NO₂ objective implementation guide.
Pre-screening is allowed for intermittent sources with a significant monitor influence. Wildfire periods may be removed only with supporting analysis of both air quality and meteorological records showing smoke was the major contributor. There is no automatic exclusion of a wildfire year, concentration cutoff, or instruction to remove every advisory day. Nor is a smoke-excluded value a replacement for the unscreened ambient record. General guideline, §8.1.4, p.71 / PDF p.78.
The NO₂ guidance's Appendix C, p.35 / PDF p.39, specifically found that removing 2017–2018 Northeast Air Zone ozone observations for wildfire influence was unwarranted in its example. That regional finding cannot be transferred to Squamish; it demonstrates the need to establish a pollutant- and location-specific effect. A PM₂.₅ smoke flag alone does not prove that co-occurring NO₂ or O₃ should be removed. NO₂ guidance, Appendix C.
For a later screened candidate, preserve the original record and a dated exclusion mask with evidence, report both screened and unscreened statistics, and report coverage before and after exclusions. Retaining the full calendar denominator after exclusion is a conservative, reproducible proposed convention; the cited section does not explicitly resolve it. The planned unscreened candidates require no smoke exclusions.
For Level 2/3 1-hour NO₂, §3.3.2 directs sequential consideration of these options, not selection of whichever gives the lowest answer:
The annual background remains the mean of hourly observations over one year. A time-varying background is added before identifying cumulative hourly peaks and percentile results, not added only to the already-selected modelled peak. NO₂ guidance, §3.3.2, pp.15–17 / PDF pp.19–21, especially p.16 and footnote 23.
Incomplete-bin ambiguity: The document gives complete-year ranks but does not specify minimum valid samples in each month/hour bin, rank adjustment for missing samples, treatment of a missing year-cell, or weighting of incompletely sampled year-cells. Overall quarterly coverage does not guarantee evenly sampled clock-hour bins. The proposed candidate tables should retain each cell's expected/valid count, missing fraction, selected observation timestamp(s), years contributing and precision. Declare whether annual-hour cells use the complete-year eighth-highest rule or an N-adjusted rank; label any sensitivity. Monthly maxima remain maxima, but poorly sampled cells can understate them. Never silently pool years before taking a maximum: that differs from averaging yearly selected maxima. No NO₂ gap-filling analogue of O₃ Appendix D is authorized by this section.
The seasonal method is outside the proposed calculation. If later used for winter 2023, it needs December 2022; an incomplete winter must not be presented as a complete season from the 2023–2025 archive.
Monte Carlo is also outside this calculation. It requires prior Ministry approval, applying simpler methods first, three years of monitored daily maxima, month-conditioned random combinations with model daily maxima, 1,000 repetitions, and the median of the resulting P98 values. The script must be supplied. It is not a reason to replace daily maxima with randomly paired hourly NO₂/O₃ records. NO₂ guidance, §3.3.2.1, pp.17–18 / PDF pp.21–22; §4.2, p.20 / PDF p.24.
Background NO₂ is the additive existing concentration. O₃ is a reactant controlling conversion of the modelled NO plume. For ARM2, OLM and PVMRM, add background NO₂ after applying chemistry to the modelled plume. Do not add baseline NOx first and then apply the old ARM curve, or count baseline NO₂ a second time. Apply chemistry at each hour and receptor before aggregation, including annual aggregation. The updated OLM relation is NO₂ = ISR*NOx + min(O₃, (ER-ISR)*NOx) + background NO₂, with compatible volume-ratio units, source-specific in-stack ratio and default equilibrium ratio 0.9. This equation does not justify deriving an ISR from ambient station ratios. NO₂ guidance, §§2, 3.2.1.2–3.2.1.3 and 3.3.1–3.3.2, pp.5–6, 11–15 / PDF pp.9–10, 15–19.
Appendix C supplies two O₃ options:
Appendix C requires attention to urban proximity, major NOx emitters and traffic: local NO can remove ozone, so a near-road/town monitor may not represent ozone available to a plume elsewhere. Its O₃-specific urban criterion is within 10 km of a settlement of at least 5,000 people, with case-by-case consideration 10–20 km from a larger city. That is an O₃ selection criterion, not the AERMOD urban-dispersion setting. If representative hourly observations are unavailable, discuss use or development of a profile with the Ministry. NO₂ guidance, Appendix C, pp.32, 36–37 / PDF pp.36, 40–41.
The guidance does not prescribe constructing a joint observed NO₂–O₃ percentile, selecting simultaneous maxima of their sum, or requiring a common valid NO₂/O₃ mask for the separate fixed-background statistics. Maintain the native paired chronology for chemistry/representativeness diagnostics, but do not invent joint-pair filtering. For chronological chemistry, O₃ must align with model meteorology; the chosen NO₂ background may instead be a justified fixed or clock-hour statistic. NO₂ and NOx mass conversions must use consistent molecular bases; the guidance specifies µg/m³ = ppb × 1.880 for NO₂/NOx-as-NO₂ at its standard conditions. That factor does not apply to O₃. NO₂ guidance, §3.2.1.2.1, p.11 / PDF p.15; §3.3.2 and Appendix C.
Appendix D requires a complete input with no missing hours for Tier 3 OLM/PVMRM. Its sequence is:
NO₂ guidance, Appendix D, p.42 / PDF p.46.
The appendix leaves endpoint implementation unspecified. For the proposed copy, interpolate only when both bracketing original hours are observed; do not use a substituted neighbor recursively or wrap January 1 to December 31. Edge gaps without both neighbors should use the remaining-gap rule and retain a distinct provenance flag. Do not count substituted values as observations for the quarterly gate. Select fallback maxima from observed values, and preserve the original missing-hour mask. If an entire month has no observations, the monthly fallback is undefined; do not silently invent one. These are explicit implementation safeguards, not additional quoted regulatory requirements.
Field Sampling Manual Part B1 §7.1 prescribes Pacific Standard Time, ending-hour reporting. Freeze the mapping from native timestamps to days and hour-of-day bins, and document the subsequent mapping to model hours. Do not introduce daylight-saving shifts. In particular, a 00:00 ending hour represents the previous elapsed hour; the file adapter must document whether the corresponding model date/hour is the previous day/hour 24. The guidance's profile table uses a midnight label but does not resolve every provider's timestamp encoding. Field Sampling Manual, p.26 / PDF p.29.
For a regulatory project, the general guideline's modelling plan is required (§1.6, step 8, p.6 / PDF p.13); baseline methodology is a listed review item (§11.1, p.92 / PDF p.99). A new ambient-monitoring program needs Ministry review and acceptance (§8.1.3). The NO₂ guidance explicitly requires approval of the background approach and justified dataset (§3.3.2, p.17), chemistry approach/curve/ISR/O₃ inputs (§3.2.1.2–3.2.1.3), and prior approval for Monte Carlo. A proposed site-specific ARM2 curve needs consultation and adequate data; its example is ten representative complete calendar years, and province-wide curves are generally preferred. No new fit follows merely from acquiring three years. General guideline, NO₂ guidance.
The general §8.1 wildfire paragraph does not expressly require separate approval of each excluded day; it does require supporting evidence. Avoid turning an inferred documentation need into an invented blanket permission requirement. Conversely, computing research candidates does not mean regulatory acceptance has been obtained.
Before any background is added to Woodfibre predictions, resolve station representativeness for the plume and receptor zones; source overlap; future emissions changes (§8.1.6); pollutant and averaging-period match; completeness and rank conventions; screened-period evidence if any; and chronological compatibility where O₃ is used. Report all chosen assumptions in the plan. The current bounded work can produce unscreened candidate statistics and a flagged 2025 O₃ preparation copy while those fitness-for-use questions remain open. It should not add them to a town map or label them a regulator-approved background.
The original general guideline and other reused primary PDFs remain under work/ambient-baseline/objective-methods/sources/; their existing acquisition manifest records URLs and hashes. The newly retrieved NO₂ PDF, its extracted text and a source manifest are under work/background-selection/sources/. Fresh general-guideline text and rendered page checks are under work/background-selection/; printed pp.71–72 of the general guideline and pp.16, 37 and 42 of the NO₂ guidance were visually checked against the extracted instructions. Original station files and prior scientific outputs were not modified.