4.4.1.2: 2016 Fort McMurray wildfire
In early May
2016, a large wildfire burned almost 600,000 ha (a land area covering 6000
square kilometers) in northern Alberta. This fire resulted in the evacuation of
all of the residents of Fort McMurray (over 80,000 people) and halted
production in the oil sands (Government of Alberta, 2016). Insured losses are
estimated at $3.5 billion (IBC, 2016). The total cost of the event is still
being determined, but it is expected to be considerably higher.
The fire
ignited near the Horse River amid very dry fuel conditions. High winds a few
days later resulted in rapid spread and fire growth. A study has used event
attribution to assess the influence of human-induced climate change on several
measures of wildfire risk (see Box 4.2), albeit not extreme fire itself, in
this region (Kirchmeier-Young et al., 2017a).
Like the
previous example, the study used large ensembles of model simulations, in this
case employing the Canadian Earth System Model (CanESM2). To assess human
influence, the model was run with only natural forcings (solar and volcanic
effects) and also with a combination of natural and anthropogenic forcings. The
anthropogenic component includes GHG emissions, aerosols, atmospheric ozone
changes, and land-use change.
Fire weather
(see Box 4.2), fire behavior, and fire season measures were calculated to
characterize fire risk from climate model output. To quantify the anthropogenic
contribution, a risk ratio (NASEM, 2016) was calculated as the ratio of two
probabilities: one for the event’s occurrence when the human component is
included, and one for the occurrence of the same event with only natural
factors. The risk ratio can be interpreted as how many times as likely the event
is as a result of anthropogenic factors. For example, a risk ratio of 1 implies
no change in the probability of occurrence, and a risk ratio of 2 implies the
event is twice as likely, or that there has been a 100% increase in the
probability of the event compared with the unperturbed climate.
Results of
the analysis show that three of the fire risk indices — extreme values of the
Fire Weather Index, high number of spread days, and long fire seasons — all
show risk ratio values greater than 1, indicating extreme values of each
measure of wildfire risk are more likely when anthropogenic warming is
included. Risk ratios vary among the different fire risk indices analyzed.
However, extreme values of all measures describing wildfire risk are more
likely with anthropogenic forcing.
Increasing
temperatures, like those observed across Canada, will lead to drier fuels, and
thus increased fire potential, as well as longer fire seasons. It would require
increases in precipitation well beyond what is expected with climate change to
offset increasing temperatures in terms of the FWI indices (Flannigan et al.,
2016). The study demonstrated that the extreme Alberta wildfire of 2016
occurred in a world where anthropogenic warming has increased fire risk, fire
spread potential, and the length of fire seasons across parts of Alberta and
Saskatchewan.
5.2: Snow cover
The portion
of the year with snow cover decreased across most of Canada (very high
confidence) as did the seasonal snow accumulation (medium confidence). Snow cover
fraction decreased between 5% and 10% per decade since 1981 due to later snow
onset and earlier spring melt. Seasonal snow accumulation decreased by 5% to
10% per decade since 1981 with the exception of southern Saskatchewan, and
parts of Alberta and British Columbia (increases of 2% to 5% per decade).
It is very
likely that snow cover duration will decline to mid-century across Canada due
to increases in surface air temperature under all emissions scenarios.
Scenario-based differences in projected spring snow cover emerge by the end of
the century, with stabilized snow loss for a medium emission scenario but
continued snow loss under a high emission scenario (high confidence). A
reduction of 5% to 10% per decade in seasonal snow accumulation is projected
through to mid-century for much of southern Canada; only small changes in snow
accumulation are projected for northern regions of Canada (medium confidence).
Snow cover
is a defining characteristic of the Canadian landscape for a few months each
winter along the southern margins of the country and for up to nine or 10
months each year in the high Arctic. Snow is responsible for a cascade of
interactions and feedbacks that affect the climate system, freshwater
availability, vegetation, biogeochemical activity including exchanges of carbon
dioxide and trace gases, and ecosystem services (Brown et al., 2017). To
understand changes in snow, it is necessary to consider multiple variables,
including snow cover fraction (SCF), which is affected by the timing of snow
onset and snow melt, and the maximum seasonal snow water equivalent (SWEmax),
the amount of water stored by snow and available for melt in spring. These
variables affect the exchange of energy between the surface and the atmosphere
(with important feedbacks to the global climate system) and freshwater
availability, as nearly all Canadian watersheds are snow-dominated in the
winter. Snow is critical to winter travel and tourism in many regions of the
country and is a key requirement for the construction of winter roads that
connect remote communities and mines, particularly in the Northwest
Territories, northern Manitoba, and northern Ontario.
Surface
observations of snow depth from climate monitoring stations (such observations
are referred to as “in situ data”) are not well suited for detecting trends and
variability in snow cover because they measure snow only at individual points
(Brown and Braaten, 1998). Snow depth can vary significantly at the local scale
because of interactions with vegetation and topography (typically driven by
winds), which means single point measurements may not capture the mean snow
depth on the landscape (Neumann et al., 2003). In addition, climate stations
are exceptionally sparse above 55º north latitude in Canada and are biased to
lower elevations in mountainous areas and in coastal areas in the sub-Arctic
and Arctic. It is, therefore, challenging to use the conventional Canadian
climate observing network for a national-scale assessment of snow. Satellite
observations and land surface models are available that provide daily,
spatially continuous data across all of Canada, extending back for decades.
These products have a coarse spatial resolution (25–50 km), which presents
problems for alpine areas and regions with mixed land cover. Researchers have
made significant efforts to determine the agreement among datasets to ensure
robust analysis of trends (Mudryk et al., 2018).
5.2.1: Observed changes in snow cover
Based on an
analysis of multiple datasets covering 1981–2015, SCF (characterized as the
proportion of days within each month that snow was present on the ground)
decreased by 5% to 10% across most of Canada during most seasons (Mudryk et
al., 2018), notably, for eastern Canada in spring (April/May/ June) and most of
the Canadian land area in the fall (October/November/December). This loss of
snow cover is consistent with previous studies using in situ datasets covering
a longer time period (Brown and Braaten, 1998; Vincent et al., 2015), but the
1981–2015 period is characterized by stronger reductions in snow cover during
the snow onset period for eastern Canada in response to enhanced fall warming
(consistent with Brown et al., 2018). Decreasing SCF trends over high latitudes
of Canada are consistent with documented reductions in annual snow cover
duration (SCD; the number of days with snow cover) across circumpolar Arctic
land areas of two to four days per decade (approximately 1% to 2% per decade,
assuming 250 days mean snow cover) (Brown et al., 2017). Some studies (Derksen
and Brown, 2012; Derksen et al., 2016; Brutel-Vuilmet et al., 2013;
Hernández-Henríquez et al., 2015; Mudryk et al., 2017) identified spring snow
cover losses slightly stronger, because different datasets and time periods
were considered. Despite these differences, all studies consistently show
reductions in spring SCF.
Analysis of
surface temperature from a blend of six atmospheric reanalysis datasets showed
that warming trends over the 1981–2015 period are found in all Canadian land
areas with SCF reductions (Mudryk et al., 2018). Cooling trends in winter and
spring are associated with the regions of increasing SCF. Observations from
climate stations in the regions where SCF trends increased over 1981–2015 also
show decreased maximum snow depth and SCD over the longer 1950–2012 period
(Vincent et al., 2015), so the positive trends over 1981–2015 reflect nature
variability in regional surface temperatures and precipitation.
While SCF
information is important for identifying changes in where snow covers the
ground, from a water-resources perspective, it is important to understand how
much water is stored in the form of snow. This is determined from the pre-melt
SWEmax. SWEmax declined by 5% to 10% across much of Canada during the period
1981–2015, according to the multi-dataset analysis shown in Figure 5.3 (Mudryk
et al., 2018). This is consistent with snow depth trends from surface
measurements (Brown and Braaten, 1998; Vincent at al., 2015) and other
observational studies (for example, Mudryk et al., 2015). Increases in SWEmax
are evident across parts of British Columbia, Alberta, and southern
Saskatchewan. The influences of temperature and precipitation changes need to
be separated to understand the driving mechanisms behind trends in SWEmax
(Raisanen, 2008; Brown and Mote, 2009; Mankin and Diffenbaugh, 2014;
Sospedra-Alfonso and Merryfield, 2017).
No comments:
Post a Comment