How we measure climate and hazard risk
The Climate score summarizes physical-hazard exposure across seven hazards: flood, wildfire, earthquake, landslide, radon, severe weather (hail and tornado) and heat. Every layer is built from a named national dataset, scored per dissemination area on a 0–100 scale where higher means lower hazard, and each hazard's sub-score is published separately so you can drill into the risk that matters to you.
Last updated
What we measure
Flood exposure is measured from Natural Resources Canada's Flood Susceptibility Mapping raster. For each dissemination area we take the 90th-percentile pixel — the riskiest tenth of the area — rather than the mean, which dilutes worst-case exposure. Historical flood events from NRCan's Historical Flood Events database (records back to 1696) add a recency-weighted event-density signal.
Wildfire exposure combines three signals: the fraction of the area burned since 1986 (National Burned Area Composite), the density of fire ignitions within 10 km over 40 years (Canadian National Fire Database), and the mean flammability of the vegetation fuel present (Fire Behaviour Prediction fuel raster).
Earthquake hazard uses the national seismic model behind Canada's building code: CanadaSHM6 spectral acceleration Sa(0.2 s) at the 2%-in-50-year probability level (NBC 2020, Site Class C), weighted 70%, plus a magnitude-scaled sum of historical events within 100 km over 100 years from NRCan's earthquake bulletin, weighted 30%.
Landslide hazard is measured from the Pan-Canadian Landslide Database (severity-weighted incident density within 5 km) and, inside Québec, the provincial ZPEGT regulatory zones. Radon potential comes from Health Canada's 2012 Cross-Canada Survey of radon in homes, at health-region resolution. Severe weather covers hail and tornado from the Northern Hail Project, NOAA MRMS radar-derived hail swaths, and the National Tornado Database (1980–present); hail carries 70% of the severe-weather blend because it is the dominant insured-loss driver where the hazard is material — Calgary's 2020 and 2024 hailstorms were both top-5 Canadian insured-loss events. Heat is measured as Humidex threshold days (30, 35 and 40 °C) from the Ouranos PAVICS climate ensemble, blending the observed 1971–2000 baseline (60%) with the SSP2-4.5 mid-century projection (40%).
Data sources and update cadence
- Natural Resources Canada — Flood Susceptibility Mapping; Historical Flood Events
Flood hazard raster and event history
CadenceCurrent-condition raster; no open future-scenario raster is published yet
- Natural Resources Canada — National Burned Area Composite, Canadian National Fire Database, FBP fuel raster
Wildfire burn history, ignition density and fuel flammability
CadenceHistorical 1986–present, refreshed as NRCan publishes
- Natural Resources Canada — CanadaSHM6 seismic hazard model and National Earthquake Database
Design-level ground motion (NBC 2020) and 100-year event history
CadenceTied to the National Building Code cycle; next model update expected around 2030
- Pan-Canadian Landslide Database (PCLD v12) and Québec ZPEGT zones
Landslide incident inventory and regulatory exposure zones
CadenceHistorical inventory, updated as published
- Health Canada — 2012 Cross-Canada Radon Survey, with Statistics Canada health-region boundaries
Share of homes at or above the 200 Bq/m³ guideline, and geometric-mean radon levels
Cadence2012 survey; health-region resolution
- Northern Tornadoes Project / National Tornado Database, Northern Hail Project, NOAA MRMS
Tornado tracks and intensities (1980–present); radar-derived hail swaths and sizes
CadenceHistorical to present, refreshed as the projects publish
- Ouranos PAVICS — Humidex ensemble (downscaled from ERA5-Land)
Days per year above Humidex 30, 35 and 40, observed and projected
Cadence1971–2000 baseline and 2036–2065 SSP2-4.5 projection, blended 60/40
How the score is computed
Each hazard is scored with the same skeleton: compute the raw metric per dissemination area, clip the extreme 1% tails, then rank nationally to a 0–100 percentile, inverted so higher means lower hazard. Where a raw zero genuinely means "no recorded hazard" — burned area, tornado counts, flood-event density — zero-value areas receive the safest score and the ranking runs over the non-zero areas only. Radon, measured from a small-sample survey, is statistically shrunk toward the national mean so sparsely sampled regions cannot dominate the extremes.
The seven hazard scores combine into the Climate score with published weights (below). If a hazard has no coverage for an area — Québec's ZPEGT zones outside Québec, for example — its weight is dropped and the remaining weights are rescaled, rather than pretending absence of data is absence of hazard.
One difference from the other four lenses matters when reading the number: climate sub-scores are national percentiles, not within-city ranks. They pass into the scoring pipeline without re-ranking, so a flood score of 40 in Calgary and 40 in Toronto describe comparable national standing. The composite Climate score rolls up to neighbourhoods area-weighted, and reports show the same number in both framings: climate score (higher = safer) and climate risk (100 minus the score).
Share of the Climate composite, effective 2026-05-08. Weights sum to 100% and every hazard's sub-score is also published individually.
What this score does not capture
Stated plainly, because a score is only worth trusting if its limits are.
- Most hazards are current or historical only. Heat is the exception, blending a 2036–2065 projection at 40%; flood, wildfire, earthquake, landslide and radon carry no future-climate scenario because no open projection dataset of adequate quality exists yet for them.
- The landslide inventory is biased toward regions with LiDAR coverage (the BC coast, the Ottawa Valley, southern Québec). An area with no recorded landslides may lack hazard — or lack documentation. The score reads absence of records as low hazard and says so.
- Radon is measured at health-region resolution, so a score describes the typical tested home in a large region, not your house. Radon is also the one hazard a homeowner can mitigate directly; Health Canada recommends testing the specific home.
- Radar-derived hail detection is weak on the Pacific coast, so Vancouver's hail record is less complete than Calgary's, Edmonton's or Toronto's.
- Flood susceptibility is modelled at the dissemination-area level from a national raster. It is not a substitute for a property-level flood-plain determination, an insurance assessment or an engineering study.
- Scores describe hazard exposure, not consequence. Building construction, mitigation works and insurance terms determine what an event costs, and none of those are in the data.
Definitions
- Climate score vs climate risk
- The same number in two framings: the score runs 0–100 with higher meaning lower hazard; risk is 100 minus the score. Reports state which framing a figure uses.
- 90th-percentile (p90) flood exposure
- The flood-susceptibility value of the riskiest tenth of pixels in a dissemination area. Used instead of the mean so a dangerous riverbank corner is not averaged away by dry uplands.
- Sa(0.2 s), 2% in 50 years
- Spectral acceleration at a 0.2-second period with a 2% chance of exceedance in 50 years — the ground-motion measure Canada's building code designs to, from the CanadaSHM6 national model.
- Humidex threshold days
- Days per year on which the Humidex reaches at least 30, 35 or 40 °C. The heat score weights the three thresholds 40/40/20 and blends observed baseline with mid-century projection 60/40.
- Winsorised national percentile
- The raw hazard metric with its extreme 1% tails clipped, then ranked 0–100 across all dissemination areas nationally. Unlike other lenses, climate ranks are national, not within-city.
- Zero-is-no-hazard treatment
- For event-count metrics, a raw zero is treated as genuine evidence of no recorded hazard: those areas score safest, and the percentile ranking runs over non-zero areas only.
- 200 Bq/m³
- Health Canada's guideline level for indoor radon. The radon sub-score's primary input is the share of tested homes at or above this level per health region, from the 2012 Cross-Canada Survey.
Frequently asked questions
Which hazards does the climate score include?
Seven: flood, wildfire, earthquake, landslide, radon, severe weather (hail and tornado combined) and heat, with weights of 24, 20, 13, 10, 10, 10 and 13 percent respectively, effective 2026-05-08. Each hazard's sub-score is published separately alongside the composite.
Does the score account for climate change?
Partially, and we state where. The heat score blends the observed 1971–2000 baseline (60%) with a 2036–2065 projection under the SSP2-4.5 scenario (40%). The other six hazards are current or historical only, because no open future-scenario dataset of adequate quality exists yet for them — Natural Resources Canada, for example, has not published an open future-scenario flood raster.
Is a climate score comparable between cities?
Yes — unlike the other four lenses. Climate sub-scores are national percentiles, so a wildfire score of 35 means the same national standing whether the neighbourhood is in Calgary or Vancouver. This is the one lens where cross-city comparison is supported by the method.
Can I use the flood score instead of a flood-plain map?
No. The score summarizes a national susceptibility raster at dissemination-area resolution. For a purchase decision, consult the municipal or provincial flood-plain designation for the specific property and, where relevant, an engineering assessment. The score tells you which neighbourhoods warrant that closer look.
Why does hail dominate the severe-weather score?
Hail carries 70% of the severe-weather blend and tornado 30%, because hail is the materially dominant insured-loss driver in the one covered region where severe weather scores high: Alberta. Calgary's 2020 and 2024 hailstorms were both among the five costliest insured-loss events in Canadian history.
How should I read a radon score?
As a regional prior, not a measurement of your house. The data is Health Canada's 2012 survey at health-region resolution: it says what share of tested homes in the region exceeded the 200 Bq/m³ guideline. Radon varies house to house and is fixable; Health Canada recommends testing any specific home.