Skip booklet index and go to page content

Technical Thematic Report No. 6. - Trends in large fires in Canada, 1959 to 2007

National Trends

Fire frequency is commonly represented by the percent of the total area that is burned annually. On average from 1959 to 2007, 18,471 km² burned in Canada each year. This is about 0.35% of the forested area in Canada (Table 1). This number is quite variable from one year to the next (Figure 1), ranging from a low of 1,524 km² in 1963 to a high of 75,377 km² in 1989.

Table 1. Average annual area burned by large fires, the percent of the forest land base that burns each year, and the contribution to the total area burned annually in Canada by ecozone+, 1959-2007.
Ecozone+Average annual area burned (km²)Annual forested area burned (%)Relative contribution of the ecozone+ to the total area burned in Canada (%)
CANADA18,4710.43100.0
Boreal Shield6,4680.4936.9
Newfoundland Boreal1240.130.8
Boreal Plains2,2140.4711.4
Taiga Plains2,8580.7113.8
Taiga Shield3,7890.7717.7
Hudson Plains5470.173.3
Taiga Cordillera8570.474.5
Boreal Cordillera1,2060.387.8
Montane Cordillera3160.102.6
Western Interior Basin540.110.4
Pacific Maritime200.020.3
Atlantic Maritime380.020.5

Top of Page

Figure 1. Annual area burned by large fires in Canada, 1959-2007.

Source: data from 1959-1994 are from the large fire database; data from 1995-2007 are derived from remote sensing.

Long Description for Figure 1

This bar graph shows the annual area burned by large fires in Canada from 1959 to 2007. The annual area burned is quite variable from one year to the next, ranging approximately as low as 1,500 km2 in 1963 to a high of 75,000 km2 in 1989.

Top of Page

Area burned also varies spatially across the country (Figure 2). The largest areas burned are ecozones+ with the least human influences and most severe fire weather (Stocks et al., 2003; Gillett et al., 2004; Parisien et al., 2006). This includes the Boreal Shield, which makes up 37% of the area burned in Canada annually, followed by the Taiga Shield and Taiga Plains which make up 32% of the area burned in Canada collectively (Table 1). One should be aware that the fire regime is not consistent across the Boreal Shield or the Taiga Shield. There is large variability from east to west, with the west side of each having significantly higher fire frequency. Therefore these ecozones+ are commonly treated as two separate regions within the fire literature (Amiro et al., 2001; Stocks et al., 2003; Parisien et al., 2006; Burton et al., 2008; Amiro et al., 2009). Fire is also frequent in the Boreal Plains and Boreal Cordillera as a result of severe fire weather. Fire is less common in the wetter coastal climates of the Pacific Maritime, Atlantic Maritime, and Newfoundland Boreal. Lastly, large forest fires seldom occur in the Arctic, Prairies, and Mixedwood Plains due to a lack of fire prone fuels and/or discontinuity of fuels.

Figure 2. Distribution of large fires across Canada, 1980s-2000s.
Each colour represents the total area burned in the corresponding decade (1980s, 1990s, and 2000s).

The 2000 decade includes only data from 2000-2007.
Since the analysis for this report was completed, trends for total area burned in Canada by decade were calculated including data up to 2010. Results can be found on page 96 of Canadian Biodiversity: Ecosystem Status and Trends 2010 (Federal, Provincial and Territorial Governments of Canada, 2010).

Long Description for Figure 2

This map shows the distribution of large fires across Canada from the 1980s to the 2000s. The largest areas burned include the Boreal Shield (37% of area burned in Canada), the Taiga Shield, and the Taiga Plains (32% of area burned in Canada, collectively). The western regions of the Boreal Shield and the Taiga Shield have significantly higher frequency than their eastern regions. Fire is also frequent in the Boreal Plains and the Boreal Cordillera. Fire is less common in the wetter coastal climates on each side of the country, as well as across the Arctic, Prairies, and Mixedwood Plains ecozones+

Top of Page

Over the last 50 years the contribution to annual area burned by ecozone+ is variable, with the exception of Pacific Maritime, Western Interior Basin, and Atlantic Maritime, which are consistently small contributors (Figure 3). The trends reflect the averages shown in Table 1. The Boreal Shield is regularly the dominant contributor to area burned, followed by the Taiga Shield for all decades except the 1960s (Figure 3). The low number for the 1960s may be due to poor monitoring in northern parts of the country (Stocks et al., 2003). Some trends not evident in Table 1 include an increasing contribution by the Hudson Plains ecozone+ over time. In the 1960s, this ecozone+ contributed 1.9% to the total area burned rising to 4.8% in the 2000s, but again this could be related to insufficient monitoring in the early period. Another trend is the decreased significance of burned area in the Newfoundland Boreal Ecozone+ since the 1960s. The high value in this decade is attributed to a number of large human caused fires that occurred in 1961, which have not occurred since (see Newfoundland Boreal section on page 15 for more details).

Figure 3. Canadian area burned statistics by ecozone+, 1960s-2000s.

The 2000 decade includes only data from 2000-2007.

Long Description for Figure 3

This stacked bar graph shows the percent contribution to the total area of Canada burned by ecozone+ by decade from the 1960s to 2000s. The percent area burned by ecozone+ varies each decade, with the exception of Pacific Maritime, Western Interior Basin, and Atlantic Maritime, which are consistently small contributors. The Boreal Shield is the dominant contributor to area burned, followed by the Taiga Shield for all decades except the 1960s.

Top of Page

Changes to long-term area burned over the last five decades are shown in Figure 4. There is an increasing trend in annual area burned from the 1960s to 1980s, follow by a levelling off in the 1990s, and a decline in the 2000s (Figure 4). Stocks et al. (2003) attribute the increase up to the 1980s to the expanded use of forests by humans combined with advances in fire detection methods and monitoring. Other studies have shown that the increase is not just an artefact of changes to fire detection methods, but is linked to increased temperatures over the last 40 years (Podur et al., 2002; Gillett et al., 2004; Skinner et al., 2006; Girardin, 2007). At first glance the recent decline does not appear to be in line with the preceding increase and with predictions that area burned is expected to continue to increase with warmer global temperatures (Weber and Flannigan, 1997; Gillett et al., 2004; Flannigan et al., 2005; Flannigan et al., 2009). Further investigation of these predictions shows increases in area burned with global warming are not expected to be linear, nor are they anticipated to be consistent across the country. One of the potential reasons for the decline in area burned in the 2000s may be other climatic influences that affect fire occurrence. These include large scale ocean circulation patterns such as the El Niño - Southern Oscillation (ENSO), the Pacific Decadal Oscillation (PDO), and the Atlantic Multidecadal Oscillation (Flannigan and Wotton, 2001; Skinner et al., 2002; Girardin et al., 2006b). The rise in area burned from the 1970s through to the 1980s corresponds to a warm (positive) phase of the PDO. Skinner et al. (2006) showed a positive relationship between the warm phases of ENSO and PDO and higher Seasonal Severity Rating (SSR – part of the Canadian Forest Fire Danger Rating System) in western, northwestern, and northeastern Canada. The SSRs are calculated to estimate the control difficulty of fires by predicting their potential intensity based on fire weather (Van Wagner, 1987; Flannigan et al., 2000). Starting in the mid-1990s there was a shift to a cool phase PDO that has subsequently been flipping between warm and cool phases until recently. Based on historic data, Skinner et al. (2006) found a cold phase PDO led to wetter summers and lower SSRs in western Canada. Changes to these large scale atmospheric oscillations may result in variability within the expected increase in area burned attributed to warming global temperatures. Further research is required to determine if this is the reason behind the recent decline in area burned (for more information on large scale climatic oscillations, see Bonsal and Shabbar, 2011).

Figure 4. Total area burned by large fires per decade for Canada, 1960s-2000s.

The value for the 2000s decade was pro-rated over 10 years based on the average from 2000-2007. Since the analysis for this report was completed, trends for total area burned in Canada by decade were calculated including data up to 2010. Results can be found on page 96 of Canadian Biodiversity: Ecosystem Status and Trends 2010 at www.biodivcanada.ca/ecosystems. The shape of the histogram remained the same (Federal, Provincial and Territorial Governments of Canada, 2010).

Long Description for Figure 4

This bar graph shows the total area burned by large fires per decade for Canada from the 1960s to the 2000s Areas are approximately 87,000 km2 in the 1960s, 138,000 km2 in the 1970s, 270,000 km2 in the 1980s and the 1990s, and 178,000 km2 in the 2000s.

Top of Page

At a smaller scale, regional variations in weather control the seasonality of fires. The fire season generally starts in April in the southern parts of the country and lasts through mid-October, as seen in the duration of active fires in the Boreal Plains, Boreal Shield, and Montane Cordillera ecozones+ (Table 2). The peak month of the fire season is July. The duration of active fires is much shorter in more northern parts of the country, such as the Taiga Plains and Taiga Shield, and shorter still in some ecozones+ predominately affected by human caused fires, such as the Atlantic Maritime, Pacific Maritime, and Newfoundland Boreal. Based on statistics presented in Weber and Flannigan (1997) that included all fires (that is, those smaller than 2 km² in addition to those larger than 2 km²), humans cause approximately 65% of the total number of fires in Canada, but are responsible for only 15% of the area burned (Stocks et al., 2003). Most of these fires are smaller than 2 km² and are therefore not included in the analysis for this report. Humans ignite fires either through recreational activities (for example, camping), arson, or industrial practices (for example, timber production and railways) (Wotton et al., 2003). These activities are generally close to human settlement resulting in early detection and quick response by fire crews limiting the size of most human caused fires. Human caused fires are also characterized by their seasonality as they occur predominately in the spring (Stocks et al., 2003; Kasischke and Turetsky, 2006; Burton et al., 2008). This is evident in the peak month of fire activity in the Atlantic Maritime and Newfoundland Boreal where the ratio of human to lightning caused fires is high (Table 2). Further details on trends and changes in fire seasonality and cause are presented in each ecozone+ section.

Table 2. Cause, duration, and seasonality of large fires by ecozone+, 1959-1999.
Ecozone+Ratio of number of human to lightning ignitionsDuration of the active fire season (days)Month of peak fire activity
CANADA0.36199July
Boreal Shield0.26143June
Newfoundland Boreal27.035May
Boreal Plains1.40158May
Taiga Plains0.1681July
Taiga Shield0.0875July
Hudson Plains0.1040July
Taiga Cordillera0.0544July
Boreal Cordillera0.2937June
Montane Cordillera1.00101August
Western Interior Basin3.3845July
Pacific Maritime1.8554July
Atlantic Maritime4.0632May

Fire cause sis depicted as the ratio of human to lightning ignitions.
Seasonality is represented by the duration of active fires (average difference between fire start dates and end dates of each year), and the peak month of fire activity.
Source: data are from the large fire database.

Top of Page

The other predominant cause of fires in Canada is lightning, which was responsible for the ignition of fires that resulted in 85% of the total area burned in Canada from 1959 to 1997, based on an analysis by Stocks et al. (2003). Lightning ignitions are most prevalent in northern parts of the country including the Taiga Plains, Taiga Shield, Boreal Shield, Boreal Cordillera, and Hudson Plains ecozones+ (Table 2). Fire suppression is limited to non-existent in some of these ecozones+ allowing lightning fires to burn naturally over large areas. Unlike human caused fires, lightning ignitions tend to occur later in the fire season in the summer months (Flannigan and Wotton, 2001; Stocks et al., 2003). These fires are often more severe than those that occur earlier in the spring as fuel moisture conditions are low enough to permit fires of greater severity and intensity (Amiro et al., 2001). Amiro et al. (2004) calculated the Head Fire Intensity (HFI) Index that is part of the Canadian Forest Fire Danger Rating System that estimates head fire intensity based on weather inputs. They found that the greatest HFIs were in the western boreal ecoregions. They also found a significant positive trend in HFI in the Taiga Shield only for data from 1959 to 1999.

The fire season severity as measured by the SSR varies greatlyacross Canada. Parisien et al. (2006) calculated an average SSR for ecozones described in the National Ecological Classification System (Ecological Stratification Working Group, 1995) (slightly different than the ecozone+ framework used in this report -- see Preface on page i) based on fire weather from 1959 to 1997. SSR values range from 0 to 5 for long-term averages, with higher numbers indicating higher intensity potential. From these data, the Montane Cordillera had the highest SSR value at 4.7, followed by the Boreal Plains at 3.7. The lowest SSR, 0.7, was found in the eastern side of the Taiga Shield, compared to an SSR of 2.8 for the western side. Despite having the greatest potential for intense fires, the Montane Cordillera had the least area burned of all ecozones in the analysis due to its mountainous terrain that limits fire spread. Burton et al. (2008) also used the SSR values calculated by Parisien et al. (2006) as part of an analysis to look at burn severity across the country, which is also a limited topic covered at larger scales within the fire literature. They found the most severe fires occurred in the western Taiga Shield and Taiga Plains based on a comparison of pre- and post-fire net primary productivity. The lowest burn severities occurred in the Boreal Plains. Based on their analysis of individual large fires across the country they concluded that burn severity is a primary driver of the ecological diversity of these forests based on the mosaic of burned and unburned sites created.

Top of Page

Summary

An analysis was done using forest fire data from the large fire database (Stocks et al., 2003) and remote sensing to look at the status and trends in large fires (> 2 km²) across the country. This analysis does not document all fires that occur in Canada each year, only large fires, which represent approximately 3% of the total number of fires, but account for approximately 97% of the area burned. Area burned varies greatly from one year to the next across the country. During low years as little as 1,500 km² has burned, compared to extremely high years where 75,000 km² of forest has burned. The majority of large fires occur in the boreal and taiga regions, in remote parts of the country, where suppression efforts are limited, and extreme fire weather conditions are common. The majority of fires in these areas are caused by lightning ignitions, compared to human ignitions. The predominance of lightning caused fires results in a large fire season that peaks in July nationally; while human caused ignitions are the predominate reason for the long duration of the fire season, which starts as early as April lasting into mid-October.

Despite limitations of the large fire database prior to the 1980s, the literature has shown that the increase in area burned from the 1960s to 1990s, has been linked to warmer temperatures across the country (Podur et al., 2002; Gillett et al., 2004; Skinner et al., 2006; Girardin, 2007). More recent data show a decline in area burned from 2000 to 2007. Although this may seem counterintuitive to the expectations of continued warmer temperatures due to climate change, further investigation of the literature demonstrates that the impacts of warmer temperatures are not expected to is be consistent across the country, nor are they anticipated to be linear (Weber and Flannigan, 1997; Gillett et al., 2004; Flannigan et al., 2005; Flannigan et al., 2009). Further research and a complete dataset for 2000s are required to conclude that area burned has significantly declinedFootnote6 and to determine the causes.

Due to the large variability in annual fire statistics a longer time period is required to fully elucidate trends in the fire regime in all areas of Canada. Continued advancements in the monitoring and documentation of large and small (< 2 km²) forest fires is therefore essential. Technologies like remote sensing that allow monitoring and detection in parts of the country that were previously limited, such as the North, are important assets to forest fire datasets. Ideally these data should continue to be supplemented and corroborated by qualitative and quantitative data collected by forest fire management agencies. Consistency in mapping methods and synthesis of both data types are required to provide a complete documentation of changes to the forest fire regime across Canada.

Footnotes

Footnote 6

 Analysis by decade including the full 2000s dataset was completed for Canadian Biodiversity: Ecosystem Status and Trends 2010 (Federal, Provincial and Territorial Governments of Canada, 2010). Results confirm the decline.

Return to footnote 6 referrer