|
Fire Research » Fire & Climate Change » National & International Activities
Spatial and Temporal Characteristics of Historical Fire Weather and
Fire Danger in Canada
Introduction
Forest fires are recognized as a key factor driving the boreal
forest carbon balance across the northern hemisphere. During their
flaming and smoldering phases, fires release carbon in the form
of CO2, CO, CH4, and trace gases and particulates.
Carbon is also released from burned areas over the longer term through
the decomposition of organic matter. Because fire burns such a large
portion of the boreal forest every year (2 million to 3 million
ha annually) it is important to know if the Canadian forests are
a sink or a source of carbon. Examination of the fire record for
Canada shows an increasing trend in area burned from the mid-1970s.
This trend is not statistically significant at this stage, however,
because of the short period of reliable fire records in Canada and
the large variability in annual area burned. Studies of long-term
temperature records by Environment Canada have shown a statistically
significant increase in most regions of Canada. This increase has
been stronger in the spring than the summer. Precipitation records
show similar trends.
|
|
Figure 1: Significance graph for spring noon
temperature. Red represents a positive trend at the 90% confidence
level. Blue represents a negative trend at the 90% confidence level.
Actual time series are shown for three selected stations.
|
|
Figure 2: Significance graph for spring precipitation.
Red represents a positive trend at the 90% confidence level. Blue
represents a negative trend at the 90% confidence level. Actual
time series are shown for three selected stations.
|
The Canadian Forest Fire Weather Index (FWI)
System is used across Canada and in many other countries to project forest
fire potential. Outputs from the system are good indicators of potential
wildland fire occurrence and spread. It is a weather-based system that
tracks daily changes in moisture in several fuel layers and then integrates
these fuel moisture values to create indices of fire spread, fuel consumption,
and potential fire intensity. We decided to look for historical trends
in values of these outputs, for although it is a good general indicator
of fire potential, the system's output generally has considerablly less
variation than the record of area burned.
|
Methods
Weather
Daily noon weather observations of air temperature, 24-hour accumulated
precipitation, relative humidity, and wind speed were obtained from
Environment Canadas Atmospheric Environment Service (AES)
for the period 1958 to 1995. Over this 38-year period, 65 weather
stations operated at the same observation location for the full
period.
Fire Danger
For each year, a season start date and fuel moisture model start-up
values were determined at each station from the AES snow, temperature,
and precipitation records. The FWI System was then used to calculate
daily values of fuel moisture and a set of fire behavior codes.
These moisture values and behavior codes were then used in conjunction
with a fuel type classification within the Canadian Forest Fire
Behavior Prediction (FBP)
System to calculate actual fire behavior quantities such as Rate
of Spread, Head Fire Intensity and Crown Fraction Burned.
Trend Analysis
For the 38-year period of record, we looked for trends in fire danger
indices, as well as several fire behavior quantities. These summaries
led to a series of annual yearly values for each variable at each
weather station. In this initial study we performed linear regressions
on these annual time series data and looked for increasing or decreasing
trends at each station by testing the significance of a linear regression
coefficient at the 90% confidence level. A series of maps of Canada
are presented for a number of the variables studied (Figures 1–5).
These significance maps show station locations as points and are
shaded with colors to indicate whether trends were significantly
positive (red) or significantly negative (blue). Positive values
not significant at the 90% confidence level but above the 68% level
are shown in orange, and negative trends between the 68% and 90%
levels are shown in green. Areas where trend slopes were not significantly
different from 0 with less than 68% confidence are shown in gray.
|
Figure 3: Significance graph for fire season
start date. Red represents a positive trend at the 90% confidence
level. Blue represents a negative trend at the 90% confidence level.
Actual time series are shown for three selected stations.
|
|
Figure 4: Significance graph for the 90th percentile
level of Daily Severity Rating over the full fire season. Red represents
a positive trend at the 90% confidence level. Blue represents a
negative trend at the 90% confidence level. Actual time series are
shown for three selected stations.
|
|
Figure 5: Significance graph for frequency of
exceeding a Fine Fuel Moisture Code of 90 over the full fire season.
Red represents a positive trend at the 90% confidence level. Blue
represents a negative trend at the 90% confidence level. Actual
time series are shown for three selected stations.
|
Discussion and Summary
Most of the variables analyzed in this study failed to show strong or
consistent trends across the country for the 38-year period studied.
In general, the grassland area of the prairies showed negative trends
in fire danger, while most of the boreal forest showed either increasing
fire danger or little change. Fire season start date (Figure 3) did seem
to have weak but consistent trend toward earlier values (leading to a
longer season), because of perhaps the increase in spring temperature
(Figure 1) and precipitation (Figure 2). General fire index means and
90th percentile levels failed to show consistent trends (Figure 4). The
frequency of exceeding critical fire spread thresholds such as the Fine
Fuel Moisture Code (FFMC) seemed to be increasing consistently across
the entire boreal region of Canada (Figure 5). Of the fuel moisture indices
in the FWI System, FFMC is perhaps the most crucial, as it strongly affects
not only rate of spread but also the probability of fire occurrence. Further
analysis of fire and weather interactions are necessary to determine if
these trends in weather are driving the observed trends in the fire record
for Canada.
For more information, contact:
B.M. (Mike) Wotton |
|
Canadian Forest Service
Great Lakes Forestry Centre
1219 Queen Street East
Sault Ste. Marie, ON P6A 2E5 |
Phone: (705) 541-5700
Fax: (705) 541-5701
e-mail: Mike.Wotton@nrcan.gc.ca |
M.D. (Mike) Flannigan |
|
Canadian Forest Service
Great Lakes Forestry Centre
1219 Queen Street East
Sault Ste. Marie, ON P6A 2E5 |
Phone: (705) 541-5541
Fax: (705) 541-5701
e-mail: Mike.Flannigan@nrcan.gc.ca |
K.A. (Kim) Logon |
|
Canadian Forest Service
Great Lakes Forestry Centre
1219 Queen Street East
Sault Ste. Marie, ON P6A 2E5 |
Phone: (705) 541-5760
Fax: (705) 541-5701
e-mail: Kim.Logan@nrcan.gc.ca |
D.L. (Dave) Martell |
|
Faculty of Forestry
University of Toronto
Earth Sciences Center
33 Willcocks Street
Toronto, ON M5S 3B3
|
Phone: (416) 978-6960
Fax: (416) 978-3834
e-mail: martell@smokey.forestry.utoronto.ca |
|