Alberta climate record
Alberta has been measuring its own weather since 1880. Here is what it recorded.
The short version: Environment and Climate Change Canada. The federal department that runs Canada's weather service. Nearly every dataset on this page is its published measurement: the station records, the river gauges, the record book, and the climate projections.'s own analysis puts warming over 1948–2016 at +1.8 °C for Edmonton and +1.5 °C for Calgary — and most of that lives in winter, which has warmed about three times as much as summer.
Every chart on this page is built from published government measurement: weather-station records back to the 1880s, a map of every wildfire the province has recorded, river gauges that have run for over a century, and the federal government's own climate projections. No modelling of our own, no imagery — the record itself. How each number was made is documented at the bottom.
The record at a glance
One stripe per year, five founding stations
Each Warming stripes. A chart form with one coloured stripe per year — blue for cooler than the baseline, red for warmer — introduced by the climate scientist Ed Hawkins in 2018. It trades exact numbers for a shape a reader can take in at a glance. is one year's average temperature at that station, coloured against the station's own 1961–1990 average (the baseline). The reference period every comparison on this page is measured against: three full decades, long enough to average out swings, covered by nearly every long-running station, and the World Meteorological Organization's standard baseline for tracking climate change. — blue cooler, red warmer. These five stations have been reporting since 1890 or earlier. (Measurements begin in 1880; the yearly series starts in 1884, the first year with all twelve months reported.)
Plain words · What's an anomaly, and why not just temperatures?
Every chart of past temperature on this page shows differences, not temperatures. Each year at each station is compared with that station's own 1961–1990 average, and the chart plots how far above or below it landed. Climate scientists call that difference an anomaly.
The reason is that absolute temperatures mostly measure geography. Banff sits high in the mountains and Medicine Hat low on the prairie, so their thermometers will never agree — but when a warm year comes, both depart from their own normals together. Comparing each station with itself cancels out elevation and siting, which is what lets ten very different places be read, and averaged, as one record.
Table view — each year compared with the 1961–1990 average, across all ten stations
| Year | Difference from 1961–1990 average (°C) |
|---|---|
| 1902 | -1.08 |
| 1903 | -0.8 |
| 1904 | -0.9 |
| 1907 | -1.64 |
| 1909 | -1.95 |
| 1910 | +0.26 |
| 1911 | -1.7 |
| 1912 | +0.05 |
| 1913 | -0.64 |
| 1914 | -0.45 |
| 1915 | +0.46 |
| 1916 | -2.15 |
| 1917 | -1.67 |
| 1918 | -0.22 |
| 1919 | -1.08 |
| 1920 | -0.64 |
| 1921 | -0.12 |
| 1922 | -1.1 |
| 1923 | +0.26 |
| 1924 | -0.61 |
| 1925 | -0.37 |
| 1926 | +0.67 |
| 1927 | -2.09 |
| 1928 | +0.8 |
| 1929 | -1.01 |
| 1930 | +0.63 |
| 1931 | +1.38 |
| 1932 | -0.99 |
| 1933 | -1.82 |
| 1934 | +0.66 |
| 1935 | -1.25 |
| 1936 | -1.39 |
| 1937 | -1.02 |
| 1938 | +0.4 |
| 1939 | +0.18 |
| 1940 | -0.27 |
| 1941 | +0.74 |
| 1942 | -0.34 |
| 1943 | +0.11 |
| 1944 | +1.01 |
| 1945 | -0.97 |
| 1946 | -0.25 |
| 1947 | -0.49 |
| 1948 | -0.83 |
| 1949 | -0.75 |
| 1950 | -2.45 |
| 1951 | -2.68 |
| 1952 | +0.13 |
| 1953 | +0.72 |
| 1954 | -0.71 |
| 1955 | -2.12 |
| 1956 | -0.7 |
| 1957 | -0.51 |
| 1958 | +0.76 |
| 1959 | -0.57 |
| 1960 | +0.13 |
| 1961 | +0.36 |
| 1962 | +0.31 |
| 1963 | +0.8 |
| 1964 | -0.45 |
| 1965 | -1.16 |
| 1966 | -1.28 |
| 1967 | -0.18 |
| 1968 | -0.43 |
| 1969 | -0.96 |
| 1970 | -0.52 |
| 1971 | -0.51 |
| 1972 | -1.72 |
| 1973 | -0.06 |
| 1974 | 0 |
| 1975 | -0.89 |
| 1976 | +1.12 |
| 1977 | +0.53 |
| 1978 | -0.41 |
| 1979 | -0.38 |
| 1980 | +0.5 |
| 1981 | +2.03 |
| 1982 | -1.68 |
| 1983 | +0.25 |
| 1984 | +0.3 |
| 1985 | -0.38 |
| 1986 | +1.21 |
| 1987 | +2.67 |
| 1988 | +1.53 |
| 1989 | +0.12 |
| 1990 | +0.31 |
| 1991 | +1.14 |
| 1992 | +1.12 |
| 1993 | +0.7 |
| 1994 | +0.36 |
| 1995 | -0.35 |
| 1996 | -1.89 |
| 1997 | +0.9 |
| 1998 | +1.49 |
| 1999 | +1.47 |
| 2000 | +0.15 |
| 2001 | +1.49 |
| 2002 | -0.07 |
| 2003 | +0.56 |
| 2004 | +0.72 |
| 2005 | +1.44 |
| 2006 | +1.7 |
| 2007 | +0.98 |
| 2008 | +0.42 |
| 2009 | -0.04 |
| 2010 | +1.04 |
| 2011 | +0.28 |
| 2012 | +0.9 |
| 2013 | +0.89 |
| 2014 | +0.16 |
| 2015 | +2.03 |
| 2016 | +2.2 |
| 2017 | +1.04 |
| 2018 | +0.06 |
| 2019 | -0.22 |
| 2020 | +0.45 |
| 2021 | +1.02 |
| 2022 | +0.71 |
| 2023 | +2.61 |
| 2024 | +1.38 |
| 2025 | +1.42 |
Measured here, not modelled elsewhere
The trend is in Alberta's own station logs
The warmest year Composite. An average across several stations, with each station first compared against its own baseline. Working in differences before averaging keeps any one station's altitude or siting from tilting the result. was 1987 (+2.67 °C), with 2023 second (+2.61 °C); three of the ten warmest years have come since 2015. Environment Canada's own analysis of the Gridded record (CanGRD). Environment and Climate Change Canada's national analysis that interpolates homogenized station records onto an even grid, so a trend can be read for any location. This page's headline warming figures for Edmonton and Calgary come from it. puts warming over 1948–2016 at +1.8 °C for Edmonton and +1.5 °C for Calgary — measured, not projected. The chart below is Edmonton, the province's longest continuous record: any single year bounces around, and the 10-year average (running mean). The average of the most recent ten years, recomputed at each step along the record. Any single year swings for reasons that have nothing to do with climate; averaging ten at a time smooths the swings so the underlying direction shows. is what shows the direction.
Plain words · Where these numbers come from
A weather station that reports for a century does not stand still. Thermometers get replaced, the site moves across town, the hour of observation shifts — and each change can move readings by tenths of a degree without the climate moving at all. Homogenization finds those artificial steps, by comparing each station against its neighbours, and corrects them. That is why trend work uses the adjusted series rather than the raw logs.
Canada's homogenized dataset ends in December 2020 and was not continued. To bring the charts to the present, this page joins on each city's current station, shifted by the average difference between the two records over the years both were reporting — the splice. Those recent years are raw rather than homogenized; the method section below says exactly how the joining was done and why the last few points deserve proportionate care.
Edmonton yearly temperature, compared with the 1961–1990 average
Adjusted station record 1884–2020, recent years joined from the successor station (see method)
The asymmetry
Winters have warmed about three times as much as summers
In degrees, that is about 3.6 °C for winters against about 1.1 °C for summers, fitted to the 10-year averages shown across ten long-running stations — the five above plus five more that came online by the early 1900s. Alberta's warming is concentrated in the cold half of the year — milder Januarys, later freeze-up — which is why a single annual number understates what residents actually notice.
Winter and summer temperatures, averaged across ten stations
10-year averages, compared with each station's own 1961–1990 average
What isn't trending
Precipitation isn't drifting — it's swinging
Calgary's driest year on record and its wettest are 10 years apart: 1892 and 1902. Unlike temperature, precipitation shows Variability vs trend. Two different things a record can show. Variability is the year-to-year swing; a trend is the slow drift underneath it. A record can have plenty of one and none of the other, and telling them apart is the central problem of reading climate data. — wet and dry years swing around a long-run average of 424 mm that never moved. The Dust-Bowl 1930s and the drenched 1951 sit inside the same flat record.
Plain words · Trend, or just a noisy record?
Ask two questions of any climate series. How big are the year-to-year swings? And underneath the swings, is the record drifting anywhere? The swings are variability; the drift is a trend. The two are independent: Alberta's temperature record has modest swings and a clear drift, while its precipitation record has enormous swings and no detectable drift.
Telling them apart is a statistics problem. Fit a straight line through the record, then ask whether its slope is bigger than what pure chance would produce in a record that swings this much. Where the slope clears that bar, this page draws the trend; where it doesn't — Calgary's precipitation here, and the river-timing chart further down — the swings are wider than any drift the record can resolve.
Calgary yearly precipitation, compared with the long-run average
Daily totals summed per year, 1885–2025 · years with fewer than 300 reported days excluded
Fire
2023 burned about 17 times the area of an average year
That is 2.6 million Hectare. The metric unit of land area: a square 100 metres on a side, about two and a half acres. A million hectares is a square roughly 100 kilometres on a side., roughly double the previous record (1981). An average year across 1961–2022 burned about 150 thousand hectares. The purple line is each decade's average year — the level a normal season drifted around. Burned area reflects both fire weather and fire management, and both have changed over these decades — but 2023 is off every historical scale.
Alberta area burned by wildfire
Every fire the province has mapped, summed by year · provincial fire boundaries, prescribed burns and out-of-province portions excluded · 1961–2025
Water
The Athabasca's peak flow scatters too widely to call a trend
Across 73 recorded years since 1922, the date of the river's Peak flow. The single highest flow a river gauge records in a year. On Alberta rivers it usually arrives in late spring, when the mountain snowpack melts — hydrologists call that seasonal surge the freshet. lands anywhere from late April to early September. The Athabasca is Alberta's longest major river Regulated river. A river with dams that store water and release it on a schedule, so its gauge reads operations as much as weather. The Athabasca has no dam on its main stem, which is why this page reads it rather than the longer Bow record., so what its gauge records is climate, not dam releases. A Trend line. The straight line that passes closest to all of a chart's points at once; its slope is the average change per year. This page draws one only where that slope is distinguishable from chance, which is why some charts deliberately have none. through the dots suggests the peak may be drifting about 7 days earlier per century — but the year-to-year scatter is so wide that the record cannot yet Statistical significance. Whether a measured drift is larger than what chance alone would produce in a record this variable. A slope that fails the test is not shown to be zero — the record simply cannot yet tell the difference..
Athabasca River at Athabasca — date of the year's highest flow
1922–2025 · Water Survey of Canada station 07BE001
Extremes
Edmonton's record book leans recent. Calgary's doesn't.
A daily heat record can be set in any of the 146 years a station has been reporting. If the climate weren't changing, the 26 years since 2000 would hold about 18% of them. Edmonton's hold 25%, which is a genuine tilt.
Calgary's hold 20%, which is close enough to 18% that this Record book (LTCE). Environment and Climate Change Canada's Long Term Climate Extremes database — the official registry of daily temperature records. It joins each city's successive stations into one continuous history, so a record can stand across the whole period of measurement. alone shows nothing — Calgary's biggest decades for standing heat records are still the 1910s through the 1930s.
Daily heat records cluster — one long heatwave sets a run of them at once — so Edmonton's counts carry less independent weight than their size suggests. And a record can only be broken, never un-broken, so the older decades hold only what has survived 146 years of challenges. The cleaner signal is the asymmetry underneath: the all-time coldest days in both cities were set in the 1880s and 1890s and have never been touched since.
Edmonton
Daily temperature records — Environment Canada's official record book
Hottest day ever recorded
37.2°C1937
Coldest day ever recorded
-49.4°C1886
Daily heat records still standing that were set since 2000
91
25% of all 366 — chance alone would give those years about 18%
Calgary
Daily temperature records — Environment Canada's official record book
Hottest day ever recorded
36.7°C2018
Coldest day ever recorded
-45.0°C1893
Daily heat records still standing that were set since 2000
73
20% of all 366 — chance alone would give those years about 18%
What the models project
The future is a range that depends on emissions
Each band spans the Percentile band (p10–p90). Line up the answers from dozens of models and the band spans from the 10th percentile to the 90th — the middle eight answers in every ten — with the dot marking the median (p50). of results from dozens of Climate model. A physics simulation of the atmosphere, ocean, ice, and land, run forward in time under an assumed emissions future. Modelling centres around the world build them independently, which is why projections are read across many models rather than from any single one. run for one SSP scenarios. Shared Socioeconomic Pathways: standard what-if storylines of future emissions that climate models are run under. SSP1-2.6 is a low-emissions future, SSP2-4.5 a middle one, SSP5-8.5 a high one. They are conditions to explore, not predictions. over a 30-year period, and the dot marks the middle projection. The models are the federal government's own, and the method section names them. One switch to note: the charts above show differences from a 1961–1990 average, while these bands show the yearly average temperature itself. Find the two reference lines first — the gap between them is the warming already observed — then read each band as how much further.
Plain words · How a climate projection works
A climate model is a physics simulation: the atmosphere, ocean, ice, and land divided into a grid, with the equations for energy and moisture stepped forward through time. Dozens of research centres around the world build them independently, and CMIP — the Coupled Model Intercomparison Project — has them all run the same coordinated experiments so their answers can be compared. The projections on this page come from the sixth round, CMIP6.
No model can know future emissions, so each is run under standard what-if storylines called SSPs: a low-emissions path, a middle path, a high path. Within each storyline the models disagree with one another, and that disagreement is information. The bands on these charts span the middle range of the models' answers, with the dot at the median. A single line would claim a precision nobody has.
2031–2060
2051–2080
2071–2100
2031–2060
2051–2080
2071–2100
Show the work
Method and sources
Temperature. Annual and seasonal series use Environment and Climate Change Canada's Homogenized data (AHCCD). Station records corrected for changes that alter readings without the climate changing — new instruments, station moves, different observing hours. AHCCD is Environment and Climate Change Canada's homogenized dataset, built specifically for trend analysis. It ends in December 2020. — station records corrected for instrument changes, station moves, and observing-time changes, which is the dataset built for trend analysis. AHCCD ends in December 2020. Later years are raw monthly averages from each city's active successor station, The splice. How this page bridges the homogenized record's end in 2020: each city's current station is joined on, shifted by the average difference between the two records over the years both were reporting, so the joined years line up with the adjusted series. between the two series over their overlap years (offsets range -0.2 to 0.8 °C; Banff's is 0.0 over a 25-year overlap because the station never changed). Spliced years are raw, not homogenized — treat the last five points with proportionate care. Comparisons are against each station's own 1961–1990 average. AHCCD encodes missing values as −9999.9; those are excluded, and annual averages require all twelve months.
Precipitation. Sum of raw daily totals at Calgary Int'l A (station succession 3031093 → 3031092), years with fewer than 300 reported days dropped. Raw, not homogenized — precipitation has no AHCCD-grade homogenized series with province-wide coverage.
Wildfire. Alberta Wildfire historical fire perimeters (Open Government Licence – Alberta), summed per year using the provincial burn code only — prescribed burns and portions burned in BC, NWT, Saskatchewan, and National Parks are excluded. Mapping before 1961 is incomplete, so the chart starts there.
Rivers. Water Survey of Canada (HYDAT). The federal river-gauge network, run with the provinces, and the database its measurements live in. The river chart's peak-flow dates come from it. annual instantaneous peak discharge — the single highest flow recorded each year. The Athabasca is shown because it is unregulated; the Bow's record is longer but reads dam operations as much as climate. The fitted timing slope (−7 days/century) has a standard error near ±12 days/century, so the page treats it as not yet distinguishable from zero.
Projections. CanDCS-U6 — ECCC's statistically downscaled CMIP6. The sixth round of the Coupled Model Intercomparison Project: the current coordinated exercise in which climate-modelling centres worldwide run the same experiments so their projections can be compared and combined. ensemble — accessed via ClimateData.ca. 30-year window averages, 10th/50th/90th percentiles across models, under SSP1-2.6, SSP2-4.5, and SSP5-8.5. These are conditional pathways, not forecasts.
What this page doesn't do. It documents what changed, not why. Attribution — how much of the change is greenhouse-gas driven — is its own body of work; the federal government's Canada's Changing Climate Report covers it for these same datasets.
Data: Environment and Climate Change Canada (AHCCD, climate observations, HYDAT, LTCE, CanGRD, CanDCS-U6 via ClimateData.ca); Alberta Wildfire (OGL-Alberta).
Plain-words glossary — every term on this page, in one place
- 10-year average (running mean).
- The average of the most recent ten years, recomputed at each step along the record. Any single year swings for reasons that have nothing to do with climate; averaging ten at a time smooths the swings so the underlying direction shows.
- 1961–1990 average (the baseline).
- The reference period every comparison on this page is measured against: three full decades, long enough to average out swings, covered by nearly every long-running station, and the World Meteorological Organization's standard baseline for tracking climate change.
- Anomaly.
- The difference between a measurement and a reference average. Every past-temperature chart on this page shows anomalies: each year compared with the same station's own 1961–1990 average, rather than the temperature itself.
- Climate model.
- A physics simulation of the atmosphere, ocean, ice, and land, run forward in time under an assumed emissions future. Modelling centres around the world build them independently, which is why projections are read across many models rather than from any single one.
- CMIP6.
- The sixth round of the Coupled Model Intercomparison Project: the current coordinated exercise in which climate-modelling centres worldwide run the same experiments so their projections can be compared and combined.
- Composite.
- An average across several stations, with each station first compared against its own baseline. Working in differences before averaging keeps any one station's altitude or siting from tilting the result.
- Environment and Climate Change Canada.
- The federal department that runs Canada's weather service. Nearly every dataset on this page is its published measurement: the station records, the river gauges, the record book, and the climate projections.
- Gridded record (CanGRD).
- Environment and Climate Change Canada's national analysis that interpolates homogenized station records onto an even grid, so a trend can be read for any location. This page's headline warming figures for Edmonton and Calgary come from it.
- Hectare.
- The metric unit of land area: a square 100 metres on a side, about two and a half acres. A million hectares is a square roughly 100 kilometres on a side.
- Homogenized data (AHCCD).
- Station records corrected for changes that alter readings without the climate changing — new instruments, station moves, different observing hours. AHCCD is Environment and Climate Change Canada's homogenized dataset, built specifically for trend analysis. It ends in December 2020.
- Peak flow.
- The single highest flow a river gauge records in a year. On Alberta rivers it usually arrives in late spring, when the mountain snowpack melts — hydrologists call that seasonal surge the freshet.
- Percentile band (p10–p90).
- Line up the answers from dozens of models and the band spans from the 10th percentile to the 90th — the middle eight answers in every ten — with the dot marking the median (p50).
- Prescribed burn.
- A fire set deliberately, under planned conditions, to manage vegetation or reduce the fuel available to future wildfires. The province maps prescribed burns alongside wildfires, so this page's burned-area totals exclude them.
- Record book (LTCE).
- Environment and Climate Change Canada's Long Term Climate Extremes database — the official registry of daily temperature records. It joins each city's successive stations into one continuous history, so a record can stand across the whole period of measurement.
- Regulated river.
- A river with dams that store water and release it on a schedule, so its gauge reads operations as much as weather. The Athabasca has no dam on its main stem, which is why this page reads it rather than the longer Bow record.
- SSP scenarios.
- Shared Socioeconomic Pathways: standard what-if storylines of future emissions that climate models are run under. SSP1-2.6 is a low-emissions future, SSP2-4.5 a middle one, SSP5-8.5 a high one. They are conditions to explore, not predictions.
- Statistical significance.
- Whether a measured drift is larger than what chance alone would produce in a record this variable. A slope that fails the test is not shown to be zero — the record simply cannot yet tell the difference.
- The splice.
- How this page bridges the homogenized record's end in 2020: each city's current station is joined on, shifted by the average difference between the two records over the years both were reporting, so the joined years line up with the adjusted series.
- Trend line.
- The straight line that passes closest to all of a chart's points at once; its slope is the average change per year. This page draws one only where that slope is distinguishable from chance, which is why some charts deliberately have none.
- Variability vs trend.
- Two different things a record can show. Variability is the year-to-year swing; a trend is the slow drift underneath it. A record can have plenty of one and none of the other, and telling them apart is the central problem of reading climate data.
- Warming stripes.
- A chart form with one coloured stripe per year — blue for cooler than the baseline, red for warmer — introduced by the climate scientist Ed Hawkins in 2018. It trades exact numbers for a shape a reader can take in at a glance.
- Water Survey of Canada (HYDAT).
- The federal river-gauge network, run with the provinces, and the database its measurements live in. The river chart's peak-flow dates come from it.
Where to go next. Look up your own town: ClimateData.ca serves these same projections for every community in Canada, and the Climate Atlas of Canada maps them across the Prairies. This page is updated as each new year of records is published. If you want the same period read through the price of oil instead of the temperature record, that page is here.
Built by Tamrack — the stories Alberta's data tells. See all of Tamrack's reports.