Temperature trends in Canada

Jeff Fritzsche, Environment Accounts and Statistics Division

The data in this article consist of annual and seasonal time series of temperature departures from normal over a 62 year period (1948 to 2009) for eleven climatic regions as well as for Canada as a whole (Map 1). The temperature departure from normal is the difference between the observed temperature values and a temperature 'normal' which is the average of observed temperatures over a specified time period. 1  The period used to calculate the normal employed in this analysis is 1961 to 1990, as reported by Environment Canada in the Climate Trends and Variations Bulletin (CTVB). 2 

A new data collaboration

This article is the second of an ongoing series in EnviroStats showcasing data related to Canada's climate and the impacts of climate change. The focus of these articles is short statistical analyses of climate-related data, such as sea ice extent and snow cover. The first in the series was released in September, 2010 (www.statcan.gc.ca/pub/16-002-x/2010003/part-partie2-eng.htm) and examined the cumulative mass balance of six Canadian glaciers.

The articles are the product of an ongoing collaboration among Statistics Canada, Environment Canada and Natural Resources Canada.

The data featured in the articles will be made available through the Statistics Canada website, both in free CANSIM data tables and through new articles re-examining trends in the data every few years.

Surface air temperature is considered by the World Meteorological Organization-Global Climate Observing System as an Essential Climate Variable, 3  part a group of variables related to the atmosphere. Air temperature is also one of several variables used by the United Nations Framework Convention on Climate Change (UNFCCC) and the Intergovernmental Panel on Climate Change (IPCC) to assess long-term changes to climate. 4 

Background and methodology

The departures from normal data used in this study were taken directly from the Climate Trends and Variations Bulletin for Canada (CTVB). 5  The data consists of annual mean, and seasonal mean, maximum and minimum temperature departures from normal for the period 1948 to 2009 for each of eleven climatic regions as well as for Canada as a whole (Map 1). To compile a set of data that reflects both the national and regional variations, daily temperature data from more than 330 weather stations were used to compute seasonal and annual departures from normal. These data are housed in the Adjusted and Homogenized Canadian Climate Data (AHCCD) 6  archives. For each station, monthly mean temperatures were computed from the record of daily minimum and maximum temperature readings.

Departures from normal are defined as departures from the 1961 to 1990 normal 7  in Celsius degrees (°C). Using departures rather than actual temperatures makes it possible to relate all regional data to the same reference point. The annual departure is the average of all monthly departures and the seasonal departure is the average of the monthly departures in the corresponding season. The seasons are defined as follows: winter (December of the previous year, January, February), spring (March, April, May), summer (June, July, August), and fall (September, October, November).

Since weather stations are not evenly distributed across the country, the temperature departures are first interpolated using Gandin's Optimal Interpolation 8  method using a grid covering the entire country. The gridded departures are averaged over each region and the nation. Next, mean temperature departures are calculated from the gridded maximum and minimum temperature departures. Finally, the gridded mean, maximum, and minimum temperature departures are averaged over Canada and each climatic region to produce national and regional time series used in the trend analysis.

A number of analytical techniques were applied to the data to determine if statistically significant trends exist in the annual and seasonal departure from normal data. These techniques included ordinary least squares analysis and non-parametric analysis using Sen's method. 9  A trend cycle (smoothed time series) was generated from the original time series using a 17-term Henderson filter. Linear regression was run on both the original time series and on the smoothed time series, while Sen's method was applied to the original time series only. All techniques showed that the annual and seasonal departures from normal showed statistically significant increases for all climatic regions over the study period. The Sen's method was used to derive the departure from normal trend results used in this study.

The annual data were tested for the presence of serial correlation and were found not to be correlated to a significant degree. Analysis of the seasonal departure from normal data was undertaken using the SAS/Autoreg procedure, which takes into account the possibility of serially correlated errors in the data. 10 



Analysis of the national annual mean temperature departure from normal time series (Chart 1) shows a warming trend 11  over the period 1948 to 2009. The linear trend for annual mean temperature departures between 1948 and 2009 moved above the 1961 to 1990 normal beginning in 1973. The linear trend indicates an increase in mean temperature of 1.4°C over the 62 years in the record.

Analysis of seasonal national mean temperature departures from normal shows that mean winter and spring temperatures got milder over the study period with these trends showing increases of 2.4°C and 1.8°C over the past 62 years. 12  Mean summer and fall temperatures showed a smaller increase in departure from normal. This indicates that increased winter and spring temperatures contributed to the warming trend to a greater degree than other seasons (Chart 3).


Annual departure from normal

Although all eleven climatic regions showed positive warming trends over the study period, there were regional differences (Charts 1 and 2). The climatic regions showing the strongest warming trends are found in the far north of Canada; namely, the Arctic Tundra, Arctic Mountains and Fiords, Mackenzie District, and Yukon and North British Columbia Mountains climatic regions. The trends for these regions show increases in temperatures of 1.6°C to 2.2°C over the study period.

The Mackenzie District climatic region recorded the strongest warming trend from 1948 to 2009 rising a total of 2.2°C, while the Atlantic Canada climatic region recorded the smallest trend increase in mean temperature over the period, 0.5°C in total.

Across southern Canada (the Great Lakes and St. Lawrence, the Prairies and the South British Columbia Mountains climatic regions) and the west coast (Pacific Coast climatic region), the mean temperature departure trend increased between 0.9°C and 1.7°C over the study period. The Northeastern Forest climatic region, which encompasses a section of Manitoba, most of Ontario and Quebec as well as Labrador, recorded a smaller warming trend (0.8°C) over the study period.

Seasonal departure from normal by region

On a regional basis, analysis of seasonal mean temperature departures from normal indicates that the Atlantic Canada climatic region experienced significantly cooler winters over the period while seven of the eleven climatic regions experienced warmer winters over the study period.

Analysis of national seasonal mean minimum temperature and maximum temperature departures from normal provides further support for these findings. Results show that the trends in the mean minimum winter and spring temperature departures increased faster than the mean maximum temperature departures. The trends in fall and summer mean minimum and mean maximum departures also increased, but at slower rates (Chart 4). 13 

Date modified: