Before the Sky Turns Orange
The C.L.E.A.R. System: Canadian Lead-Time Early Air Response
Providing up to 87 Hours (24.7 h average) of Advance Warning for Wildfire Smoke
Source: Bui & Zander (2026), “Before the Sky Turns Orange”
Project Summary
C.L.E.A.R. is a wildfire PM2.5 early warning system for Toronto, Edmonton, Montréal, and Vancouver that repurposes existing NAPS air quality stations located 100–650 km away to provide up to 87 hours (24.7 h average) of advance warning before dangerous smoke arrives.
Research Question
"Can hourly PM2.5 readings at distant monitoring stations predict a city's PM2.5 levels with enough lead time to issue meaningful public health warnings?"
Purpose
Develop an early warning system that uses existing NAPS and U.S. EPA monitoring infrastructure — stations located 100–650 km upstream — to detect approaching wildfire smoke plumes and issue colour-coded health alerts up to 87 hours in advance (24.7 h average).
The system targets four major Canadian cities: Toronto, Edmonton, Montréal, and Vancouver, covering diverse geographic and smoke exposure patterns.
Three-Rule Detection System
C.L.E.A.R. organizes its 56-station operational network into three detection rules (Bui & Zander, 2026):
Rule 1: Regional Alert
Trigger: Station >40 µg/m³, predicted city ≥20 µg/m³
Ontario & Quebec stations 100–650 km. Primary detection for nearby fires. Regression predicts city PM2.5; alert only if predicted ≥20 µg/m³.
Rule 2: NW Ontario Sequential
Trigger: Thunder Bay >35 µg/m³ + Intermediate >20 µg/m³
Two-stage detection for NW Ontario fires (1,000–1,800 km). Thunder Bay triggers; intermediate stations (200–600 km) confirm smoke transport toward Toronto within 96 hrs.
Rule 3: Quebec Upstream
Trigger: Same as Rule 1 (station >40 µg/m³)
Québec fires 500–1,400 km NE. Montréal, Ottawa, Quebec City, Cornwall, etc. Eastern monitoring network.
Data Sources & Analysis
- Station selection: R ≥ 0.30, p < 0.001, N ≥ 100 (inclusion criteria per Bui & Zander, 2026)
- Toronto reference: NAPS 60430 (primary), 60410 (secondary); highest-station approach
- NAPS Network: 348 Canadian NAPS + 131 U.S. EPA stations analyzed (56 operational after statistical screening); 36M+ hourly observations
- U.S. EPA AQS: Border stations in NY, PA, VT, WA, OR for cross-border smoke
- Study period: 2003–2023 (21 years of wildfire seasons)
- Regression model: PM2.5city = slope × PM2.5station + intercept
- R-weighted averaging: Stations with higher correlation (R²) have more influence on predictions
- Evaluation window: 120 hours (5 days) to confirm smoke arrival
- Event cooldown: 168 hours (7 days) between distinct events
Independent Validation — 2024 Out-of-Sample
The finalized 56-station network was tested on the 2024 wildfire season — events not used during model development (Bui & Zander, 2026):
8 independent 2024 wildfire smoke episodes (5 true positives, 2 true negatives, 1 false positive, 0 false negatives). The system detected every hazardous smoke event that reached Toronto.
Three-Rule Detection — 2003–2023 Development
The three rules provide 360° coverage, each detecting wildfire smoke approaching Toronto from a different direction and distance band:
| Detection Rule | Coverage | Direction detected | Role |
|---|---|---|---|
| Rule 1: Regional | 100–650 km | Central Ontario, Ottawa Valley & Québec | Primary detector — earliest, most reliable warnings |
| Rule 2: NW Sequential | 1,000–1,800 km | Northwestern Ontario (via Thunder Bay) | Two-stage trigger-and-confirm for distant NW fires |
| Rule 3: NE Corridor | 500–1,400 km | Northeastern Québec corridor | Upstream Québec detection — mirrors Rule 1 |
Combined system performance (2003–2023, 33 wildfire events)
Accuracy = (26 TP + 4 TN) ÷ 33 events. The three false positives were non-fire regional haze events; no smoke event that reached Toronto was missed (100% sensitivity).
Real-World Example — Toronto, June 2023
During the catastrophic Quebec wildfire event (peak 241 µg/m³ — Toronto's highest ever):
| Date | Rule Triggered | Station | PM2.5 | Alert |
|---|---|---|---|---|
| June 3 | Rule 3 (NE Corridor) | Ottawa (341 km NE) | 156 µg/m³ | EXTREME |
| June 6 | Rule 1 (Regional) | Cornwall (387 km ENE) | 160 µg/m³ | EXTREME |
| June 28 | — | Toronto (Peak) | 241 µg/m³ | EXTREME |
Seven major Quebec fires burned simultaneously, producing record-breaking smoke across Eastern Canada.
Burn Area Analysis
Analysis of the Canadian National Fire Database shows statistically significant increases in national burn area over the study period. The 2023 season burned over 18 million hectares, more than double any previous year.
Methodology Citation
All detection rules, thresholds, and validation metrics follow the C.L.E.A.R. research paper — Bui & Zander (2026), “Before the Sky Turns Orange: The C.L.E.A.R. System.” The paper is the authoritative source of truth for this system.
AI Platform Validation
Regression analyses were independently replicated across four AI platforms to ensure mathematical accuracy:
All four platforms produced mathematically equivalent results, which were then hand-verified against raw data.
C.L.E.A.R. Alert Levels
The system uses a five-tier alert scale based on predicted PM2.5 concentrations:
| Alert Level | PM2.5 Level | Public Health Action Plan |
|---|---|---|
| LOW | 0–20 µg/m³ | No significant risk. No action required. |
| MODERATE | 20–60 µg/m³ | Sensitive groups (children, elderly, respiratory conditions) should reduce outdoor activity. |
| HIGH | 60–80 µg/m³ | General population affected. Reduce prolonged outdoor exertion. Use N95/KN95 mask outdoors. |
| VERY HIGH | 80–120 µg/m³ | Significant risk for all. Avoid outdoor exertion. Keep doors and windows closed. |
| EXTREME | >120 µg/m³ | Emergency conditions. Stay indoors. Close windows. Run HEPA filter. No indoor pollution sources. |
How It Works
For each included station, the system uses a simple linear regression:
Alert thresholds at each station are computed by inverting the formula:
When a station's live PM2.5 reading exceeds its computed threshold, the corresponding alert is triggered — providing advance warning based on the station's distance and tier classification. Each station is identified by its NAPS ID (visible on the dashboard and live map) and can be queried individually via the API.
Conclusion
Novel Contribution
First system to repurpose existing NAPS infrastructure specifically for wildfire smoke early warning across multiple Canadian cities.
Proven Reliability
100% sensitivity with zero missed events in the 2024 out-of-sample validation (87.5% accuracy, 66.7% specificity). The three-rule system provides robust detection with minimal false alarms.
Actionable Results
Colour-coded alerts with specific health recommendations give the public clear guidance hours before smoke arrives.
Scalable Solution
The methodology can be extended to any city with nearby upstream monitoring stations.
Future Work
- Satellite integration for real-time smoke plume tracking
- More cities added to the network
- Live season validation during upcoming 2026 wildfire season
- Expanded NAPS coverage to address monitoring gaps (e.g., Edmonton NNW)
- Wind direction/speed incorporation for improved predictions
- Public API — already live at
/api/v1/with station-level queries
References
- Anthropic (2026). Claude AI Platform.
- Cruz, M. G. et al. (2019). Fire dynamics and behaviour.
- Environment and Climate Change Canada. National Air Pollution Surveillance (NAPS) Program.
- U.S. Environmental Protection Agency. Air Quality System (AQS) & AirNow.
- IQAir / World Health Organization. PM2.5 health guidelines.
- National Forestry Database. Canadian wildfire statistics.