Frost-Free Season Shift in Canada
Frost-free season shift in Canada, city by city: how the growing season changed between Environment and Climate Change Canada's 1981–2010 and 1991–2020 climate normals. Every city is measured at the same long-record station in both periods — no estimates, no substitutions, every number traceable to ECCC.
At a glance: Across 14 major Canadian cities, each measured at the same ECCC station in both normal periods, the average frost-free growing season changed by +1.2 days between 1981–2010 and 1991–2020 (median +1.5). 9 cities lengthened, 3 shortened, and 2 stayed the same. The largest gain was Toronto at +6 days; the largest loss was Edmonton at -5 days. The shift is real but modest — smaller than the year-to-year swing at any one station — so read it as a slow drift, not a green light to plant weeks earlier. Source: Environment and Climate Change Canada climate normals, 1981–2010 and 1991–2020.
📥 Download the dataset
The full 14-city table — station, Climate ID, frost-free days in each period, the shift, frost dates and ECCC codes — as a free CSV.
Download CSV →📝 How to cite this study
Frost-free season shift, Canada (1981–2010 vs 1991–2020): ECCC Canadian Climate Normals, compiled by GrowersGuide.ca
The Shift, City by City
Each row is one city measured at the same station in both normal periods (the Climate ID is identical in each). "Frost-free days" is ECCC's Average Length of Frost-Free Period; "shift" is the 1991–2020 value minus the 1981–2010 value. Sorted from the biggest gain to the biggest loss.
| City | Station (ECCC) | 1981–2010 | 1991–2020 | Shift |
|---|---|---|---|---|
| Toronto, ON | TORONTO LESTER B. PEARSON INT'L A Climate ID 6158733 |
168 | 174 | +6 |
| Fredericton, NB | FREDERICTON A Climate ID 8101500 |
130 | 135 | +5 |
| Quebec City, QC | QUEBEC/JEAN LESAGE INTL A Climate ID 7016294 |
145 | 148 | +3 |
| St. John's, NL | ST JOHN'S A Climate ID 8403506 |
139 | 142 | +3 |
| Calgary, AB | CALGARY INT'L A Climate ID 3031093 |
117 | 119 | +2 |
| Montreal, QC | MONTREAL/PIERRE ELLIOTT TRUDEAU INTL A Climate ID 7025250 |
165 | 167 | +2 |
| Victoria, BC | VICTORIA INT'L A Climate ID 1018620 |
211 | 213 | +2 |
| Ottawa, ON | OTTAWA MACDONALD-CARTIER INT'L A Climate ID 6106000 |
159 | 160 | +1 |
| Winnipeg, MB | WINNIPEG RICHARDSON INT'L A Climate ID 5023222 |
121 | 122 | +1 |
| Charlottetown, PE | CHARLOTTETOWN A Climate ID 8300300 |
153 | 153 | 0 |
| Saskatoon, SK | SASKATOON DIEFENBAKER INT'L A Climate ID 4057120 |
117 | 117 | 0 |
| Regina, SK | REGINA INT'L A Climate ID 4016560 |
115 | 114 | -1 |
| Vancouver, BC | VANCOUVER INT'L A Climate ID 1108447 |
237 | 235 | -2 |
| Edmonton, AB | EDMONTON INT'L A Climate ID 3012205 |
110 | 105 | -5 |
| Mean shift across 14 cities | +1.2 | |||
Frost-free days are whole-day averages; a shift of 0 means the rounded average did not change. The figure is ECCC's Average Length of Frost-Free Period — the average of each year's frost-free run (last spring frost to first fall frost, daily minimum ≤ 0 °C), not the gap between the two mean dates.
Why this differs from our frost-dates dataset. These numbers use ECCC's published Average Length of Frost-Free Period statistic. Our frost-dates dataset and gardening-data hub instead count the days between the average last and first frost dates — a different, also-valid calculation, so a city can read a few days longer there. Both are ECCC-sourced; this study deliberately uses ECCC's own frost-free-period stat because it's the value ECCC re-computes for each normal period, making the 1981–2010 vs 1991–2020 comparison apples-to-apples.
How to Read This — Honestly
The headline is a +1.2-day average lengthening, with most cities (9 of 14) gaining frost-free days. That's consistent with the direction of long-term warming — but the honest read is that the change is small and uneven:
- It's smaller than year-to-year variability. A single station's frost-free period routinely swings 20–40 days from one year to the next. A few days of change in the 30-year average is a genuine signal, but it is dwarfed by the noise of any individual spring or fall.
- Not every city moved the same way. 3 of the 14 stations actually shortened. Station relocations, changes in instrument siting, and local microclimate effects can push one station against the broad regional grain — which is exactly why comparing the same station in both periods matters.
- The two periods overlap. 1981–2010 and 1991–2020 share 20 years, so this compares the decade added (2011–2020) against the decade dropped (1981–1990) — a decadal nudge to a 30-year mean, not a before-and-after snapshot.
- For planting, trust the forecast, not the trend. A longer average season doesn't cancel a hard late frost. Use these normals to understand the direction of travel; use your own year's forecast and local last-frost date to decide when to plant.
Why some cities are missing
A city only appears here if the exact same long-record station published a mean frost-free-period value in both the 1981–2010 and 1991–2020 normals. Where it didn't, the city was left out rather than patched with a different station — swapping stations between periods would turn a real climate shift into an artefact of which weather site you picked.
Halifax is the clearest example: Halifax Stanfield Int'l A reports a frost-free average in the 1981–2010 normals but only frost-probability figures (no published mean) in 1991–2020, and no alternate Halifax station carries the mean in both periods — so Halifax is excluded. This is the integrity rule doing its job, not an oversight.
The Frost Dates That Moved With It — & Data Quality
A longer season generally tracks an earlier last spring frost, a later first fall frost, or both. This table shows how each city's average frost dates shifted alongside the frost-free stat (they're computed independently by ECCC, so they won't subtract exactly to the day counts above), plus ECCC's own data-completeness code (A = strictest coverage … D = at least 15 years) so you can weigh the record quality behind every figure.
| City | 1981–2010 | 1991–2020 | ||||
|---|---|---|---|---|---|---|
| Last frost | First frost | Code | Last frost | First frost | Code | |
| Toronto | Apr 30 | Oct 16 | A | Apr 27 | Oct 20 | C |
| Fredericton | May 17 | Sep 25 | C | May 14 | Sep 27 | D |
| Quebec City | May 11 | Oct 4 | D | May 10 | Oct 5 | D |
| St. John's | May 30 | Oct 17 | A | May 28 | Oct 18 | C |
| Calgary | May 21 | Sep 16 | A | May 21 | Sep 18 | D |
| Montreal | Apr 29 | Oct 12 | A | Apr 29 | Oct 14 | C |
| Victoria | Apr 7 | Nov 5 | A | Apr 4 | Nov 4 | D |
| Ottawa | Apr 30 | Oct 7 | A | Apr 29 | Oct 7 | C |
| Winnipeg | May 23 | Sep 22 | A | May 23 | Sep 22 | C |
| Charlottetown | May 16 | Oct 17 | A | May 17 | Oct 18 | D |
| Saskatoon | May 21 | Sep 15 | A | May 19 | Sep 15 | C |
| Regina | May 20 | Sep 12 | C | May 23 | Sep 14 | C |
| Vancouver | Mar 18 | Nov 10 | A | Mar 19 | Nov 10 | D |
| Edmonton | May 24 | Sep 12 | A | May 27 | Sep 10 | A |
Methodology & Sources
- Source. Every figure is Environment and Climate Change Canada's published Average Length of Frost-Free Period (and its average last spring / first fall frost dates), taken from the Canadian Climate Normals for the 1981–2010 and 1991–2020 reference periods (climate.weather.gc.ca/climate_normals).
- Same-station rule. For each city the same long-record station — identified by its ECCC Climate ID — is used in both periods. If the same station did not publish a mean frost-free value in both, the city was excluded, never silently swapped for a nearby station.
- The shift is simply the 1991–2020 mean frost-free days minus the 1981–2010 mean frost-free days, per station.
- Data-quality codes. ECCC assigns each normal a completeness code (A = WMO "3 and 5" rule, B = ≥25 years, C = ≥20 years, D = ≥15 years). Both periods' codes are shown so weaker records are visible rather than hidden.
- Aggregates are computed live from the downloadable CSV on every page load, so the mean, counts and extremes on this page always match the source rows — nothing is hand-typed.
- Caveat. A single station's normal reflects that station's exact location and history. Station moves and siting changes can shift a record independently of climate; treat each city as one station's story, and the 14-city mean as the more robust signal.
For researchers, journalists, garden clubs & educators
This study is free to quote and republish — in articles, newsletters, class handouts or your own site — provided you credit GrowersGuide.ca with a link to this page and cite Environment and Climate Change Canada as the primary data source. Want a wider city list, a specific province, or the raw per-station pulls before you publish? Get in touch at zusashicanada@gmail.com.
Every number here is traceable to a named ECCC station and Climate ID — we'd rather you cite it correctly than approximate it.
Use of This Data
The compiled study and aggregates on this page are free to use with attribution to GrowersGuide.ca (a link to this page is sufficient). The underlying climate normals are the work of Environment and Climate Change Canada; please credit that primary source when the context calls for it. Provided as-is, without warranty — verify against ECCC directly for any critical or commercial use.