chapter 3

Future Possible Droughts

Elaine Wheaton, David Sauchyn, and Barrie Bonsal

Background and Rationale

Nothing is definite in the future, but drought is certain to play a role, as it is a part of the climate of the Canadian Prairies. Droughts can be casually defined as a worrisome lack of water or more formally defined as a prolonged period of abnormally dry weather that depletes water resources for human and environmental needs (Meteorological Service of Canada 1986). Droughts occur in many regions of North America and the world, but the agricultural region of the Canadian Prairie provinces of Alberta, Saskatchewan, and Manitoba is among the most susceptible to droughts and is the focus of this chapter.

Prairie people have considerable experience with climate extremes, such as drought and heat, but these extremes can still cause concern and damage. Drought is more costly than any other form of natural disaster (Wilhite 2000). This is especially true of the Canadian Prairies, where drought is very damaging to the economy, society, and the environment, even in recent years (e.g., Wheaton et al. 2008). Drought occurs in most years in some part of the Canadian Prairies, but it is the longer-duration and larger-area droughts that have the most severe impacts and provide the greatest challenges for adaptation. At least five major droughts have occurred in the Canadian Prairies during the past 120 years. These include multi-year droughts in the 1890s, 1910s, 1930s, and 1980s, and in 2000–2004 (Bonsal, Wheaton, Chipanshi, et al. 2011; Bonsal, Wheaton, Meinert, et al. 2011). During this last major drought, parts of the Prairies had some of the driest conditions in the historical record, and it was one of the first documented coast-to-coast droughts in Canada.

Each major drought on the Prairies appears to have several unique characteristics, including duration, area of coverage, intensity, and cause, but Bonsal, Wheaton, Meinert, et al. (2011) documented several similarities among the droughts, including their origin in the US northern plains and subsequent migration into the Canadian Prairies. The authors devised and used a six-stage drought classification system to compare the major droughts. A key difference of the 2000–2004 drought is that its peak in terms of area of severe drought was during winter, whereas the others peaked during the May-to-August growing season. Most of these major droughts lasted almost two years, but the 1928–32 drought lasted over 40 months.

More recently, a less severe and shorter drought occurred from 2008 to 2010 in the Canadian Prairies (Wittrock et al. 2010). A core of well-below-average rainfall appeared around Edmonton, Alberta, and northward in the summer of 2008. By that autumn, this core area had expanded westward to the British Columbia border and eastward into Saskatchewan. The drought intensified in the winter of 2008–9, and the most severe and largest dry area appeared in spring 2009. Rainfall eased the dryness by autumn 2009, but dryness continued in areas of Alberta. Spring rains in 2010 ended the meteorological drought for most areas, but effects lingered.

Reducing the negative impacts of drought requires considerable planning and preparation so that society can effectively adapt to future droughts. These activities require understanding the nature of future possible droughts, which is the rationale for work that projects important climate extremes, especially future drought. It is prudent and critical to advance knowledge of future possible droughts for effective adaptation, that is, to decrease these massive costs and to take advantage of any opportunities. Such opportunities can result from numerous benefits of drought, including increased quality of grain and hay, reduced levels of some insects and diseases, and fewer delays for construction of roads and buildings (Wheaton et al. 2011).

If the past were the only guide to the future, past information would suffice for estimating future droughts. However, the current risk of drought is changing, perhaps fairly rapidly, and research indicates that dry areas (such as the Prairies) are expected to become drier. The potential for future drought risk is increasing largely because of human-induced climate change (IPCC 2012).

Objectives and Methods

Information is needed regarding the nature of future droughts to facilitate adaptation and reduce vulnerability. Critical questions to address include: How will droughts change in terms of characteristics such as severity, duration, frequency, timing, cause, and area? The objective of this chapter is to address these questions by reviewing drought literature focused on the Canadian Prairies. Much work has emphasized the global scale, but considering that drought is an important hazard for the region, several papers have focused at the scale of the Prairie provinces. We attempt to emphasize the near future to about mid-century, but the relevant literature tends to use the following standard periods: the 2020s (2011–40), 2050s (2041–70), and the 2080s (2071–2100); some literature uses other scales, such as time-series results, for the period up to 2100.

For each study reviewed in this chapter, its methods are briefly described to help assess its results. More recent literature is used, where possible, with some reference to earlier literature for perspective. Generally, drought characteristics are measured using several indicators. The most common of these indicators are the moisture deficit (e.g., precipitation minus potential evapotranspiration), Palmer Drought Severity Index (PDSI), and Standardized Precipitation Index (SPI). Newer indices, such as the Standardized Precipitation Evapotranspiration Index (SPEI), are the subject of current research.

Future Possible Droughts of the Canadian Prairie Agricultural Region

A few studies have focused on the nature of future droughts in the Canadian Prairie provinces. Work at the global scale hints at future drought conditions on the Prairies, but focused assessments done at a finer resolution provide more detailed information. The reason for estimating future possible droughts is to improve adaptation as droughts have serious impacts and require considerable adaptation to reduce vulnerability (Kulshreshtha et al., this volume). In this section, we present methods and findings of recent work on future droughts in the Canadian Prairies and discuss this research within the context of earlier work.

One of the studies with findings for the Canadian Prairies is by Barrow (2010). She analyzed output from a set of regional climate models (RCMs) to determine characteristics of potential evapotranspiration (PET) and moisture deficit. PET was calculated using two methods for comparison: Thornthwaite and Penman-Monteith. RCMs were shown to simulate observed precipitation values better than global climate models (GCMs), as the RCMs have finer spatial resolution. All but one of the models used by Barrow (2010) showed increases in both intensity and area of moisture deficit (water-year October to September) for the 2041–70 period. Time-series analyses project increases in evaporative demand over time for all simulations driven by the projected temperature increases. For the Canadian Regional Climate Model (CRCM) results, annual moisture deficits for this future period range from about -400 mm in southwest Saskatchewan to -200 mm just north of about 50° North (Figure 1). These values are about double the annual moisture deficit for the 1971–2000 period. Also, the area with an annual moisture deficit of -200 to -400 mm expands considerably into the future, migrating from a narrow ribbon along the US border in southwestern Saskatchewan and southeastern to central Alberta to cover all of southern Saskatchewan past Regina.

Barrow’s (2010) findings of future expansion of arid areas confirm earlier work. For example, Sauchyn et al. (2005) used the aridity index (ratio of annual precipitation to PET) for the Canadian Prairies, and found that the area of aridity (ratio less than 0.65) increased by 50% and expanded northward.

Figure 1. Spatial pattern of the future annual moisture deficit for the water-year of October to September (2041–70, precipitation minus potential evapotranspiration, mm). The two different experiment identifiers of the Canadian Regional Climate Model (CRCM) are labeled as aev and aet
(Source: Barrow 2010: 21)

Another study focusing on the Canadian Prairies by Thorpe (2011) used a range of climate change scenarios from several GCMs to estimate future PET for the Prairie Ecozone. He also found that PET increases in the future. The average Prairie Ecozone PET for Saskatchewan and Alberta is about 550 mm for the baseline climate of 1961–90, increases to about 600 mm at about 2020, reaches almost 700 mm by 2040, and increases even more rapidly thereafter for the warm scenario (Figure 2). Changes in annual precipitation are projected to vary from only small increases in the warm scenario to small decreases in the cooler scenario. The changes in precipitation are projected with much lower confidence than for temperature. These potential increases in precipitation are insufficient to compensate for the increased atmospheric water demand, producing the greater moisture deficits as estimated by Barrow (2010), for example. Williams and Wheaton (1998) calculated that increased annual precipitation of about 7%–10% is needed to compensate for an increase in mean annual temperature of 3°C.

Figure 2. Average potential evapotranspiration for the Prairie Ecozone of Alberta (AB), Saskatchewan (SK), and Manitoba (MB) for the baseline climate (1961–90) and for two future scenarios
(Source: Thorpe 2011: 5)

Sushama et al. (2010) used the CRCM and the number of dry spells or dry days with precipitation less than 2 mm (and other thresholds) to explore future drought characteristics. Results indicate that the number of dry days will increase by up to about five days in the 2050s for southern Saskatchewan. The 10- and 30-year return levels of maximum dry spell length are projected to increase during the 2050s and 2080s in the Canadian Prairies, especially in the south.

Price et al. (2011) developed high-resolution climate scenarios for Canada from several GCMs. Besides increases in temperature and only modest increases in precipitation, they project solar radiation levels will increase slightly during summer in the Canadian Prairies’ semi-arid ecozone. These changes would contribute to increasing dryness. Vapour pressure levels are also projected to increase and would offset some of the effects of warming on evaporative demand, but overall evaporation rates are expected to increase. Generally, Price et al. (2011) estimate that temperature, precipitation, and solar radiation will increase, along with some increases in inter-annual variation, which indicate that multi-year droughts will become more common and more intense, especially with higher emission scenarios by 2100.

Figure 3. Projected changes to a) severity (%), b) frequency, and c) maximum duration (months) for 10-month drought events at the watershed scale, and d) classification of watersheds based on projected changes to the severity and frequency of 10-month events for the 47 watersheds in the Canadian Prairies for five pairs (i–v) of Canadian Regional Climate Model simulations
(Source: PaiMuzumber et al. 2012: Figure 12)

PaiMazumber et al. (2012) estimated future durations of drought severity; their results show that 6- and 10-month long droughts will become more severe over southern Saskatchewan and Manitoba in the 2050s compared with the 1971–2000 baseline. The 10-month droughts are expected to increase in frequency by as many as four events in the 2050s. Maximum durations of long-term droughts are projected to increase for a large part of the southern Prairies, and the largest increases are expected for droughts lasting 10 months or longer. The most vulnerable watersheds were found to have future possible increases in both severity and frequency of 10-month droughts for five pairs (GCM/RCM) of climate simulations for the 2050s (Figure 3). The CRCM and the high-emission scenario (A2) were used to develop climate scenarios, and monthly precipitation deficits were used to measure drought severity. Limitations of the research include the CRCM’s ability to simulate precipitation, the use of only one model, and the use of precipitation alone to describe drought.

Bonsal et al. (2013) produced one of the most comprehensive descriptions of future possible drought for the Canadian Prairies and were the first to use three time periods—pre-instrumental period, instrumental (or observational) period, and future—spanning the years 1365–2100. Their study area was Alberta and western Saskatchewan from the US border north to past Edmonton (i.e., 54° North). They used five climate scenarios downscaled from two versions of the CGCM and the UK Hadley climate model (HadCM3), as well as the baseline period of 1961–90. Summer (June, July, August) self-calibrated PDSI and SPI values were averaged over the study area, and time series were produced from 1900 to 2100. They examined the time series of the areal averaged PDSI and SPI for the 1901–2099 period for the five GCMs, their means, and the nine-year running means (Figure 4).

Bonsal et al.’s (2013) results indicate that the pattern of the future mean PDSI values shows drying from the present to 2020, followed by a slight improvement with much variability to 2040. After 2040, persistently negative values occur with a downward trend, reflecting drier to drought conditions. The authors suggest this trend indicates a permanent regime shift to a more arid climate. In contrast, the SPI time series for the future period reveal no strong change compared with the instrumental period to about 2040; however, a higher persistence of multi-year droughts is found in the central and southern portion of the study area. This result occurs because SPI is calculated using only precipitation and not temperature. Drought indicators that consider precipitation alone are insufficient to determine future drought characteristics (e.g. Bonsal et al. 2013).

Figure 4. Summer a) Palmer Drought Severity Index (PDSI) and b) Standardized Precipitation Index (SPI) area-averaged values for the instrumental period (1901–2005) and the future (2011–2099). The black lines are the future ensemble-mean values from the five climate model runs, and the red lines are the nine-year running means. The minimum and maximum climate projections for each summer are shown in grey. (Source: Bonsal et al. 2013: Figure 9)

Drought area was also estimated by Bonsal et al. (2013); they found a substantial increase in the area and frequency of severe drought and worse (i.e., PDSI of -3 or less) in the area of the Canadian Prairies they studied. Even SPI shows that most future summers have severe drought conditions in some portion of the study area. These patterns suggest that severe droughts will become a more permanent feature in some areas of the southwestern Canadian Prairies in terms of characteristics such as occurrence, duration, and/or severity.

Multi-year droughts were also investigated by Bonsal et al. (2013) using the PDSI and found to be more frequent in the future period compared with the instrumental period (105 years). The length of a drought was considered to be the average number of consecutive summers with a negative value. Summer droughts of five years and longer have a frequency of 1.9 occurrences per 100 years during the instrumental period. This frequency is expected to more than double to 4.2 per 100 years in the future. The frequency of droughts of 10 years or longer increases to 3.1 per 100 years in the future. This result is even worse than the paleo record frequency of 3.0 per 100 years (see Chapter 2 by Sauchyn and Kerr in this volume). A worst-case situation is for increased frequency of drought of 10 years with consecutive summer droughts (i.e., negative PDSI values).

Although the general climate is projected to become drier, substantial variability could occur. The IPCC (2012) has identified areas of expected changes for the return period of intense daily rainfall events globally. For central North America, including the agricultural prairies, a decrease of 5 to 10 years is projected for the return period of a maximum 20-year rainfall event. This means that an extreme daily rainfall event could occur as much as twice as often as during 1981 to 2000. This result is for the middle 50% of models for the medium (A1B) to extreme (A2) emission scenarios. Dai (2010) also reports that the type of rainfall is expected to change with continued warming to more intense rainfall events and fewer light rainfalls. This pattern would tend to exacerbate drought, because intense rainfalls do not recharge soil moisture as well as more gentle rainfalls. Drier soils also tend to increase the risk of drought, because heating is used to warm soils to higher temperatures instead of evaporating water. This effect is similar to the cooling effect of water on one’s skin as compared with dry skin that stays warmer.

Wheaton et al. (2013) reviewed projections of extreme precipitation globally and for the Canadian Prairies and found consistent estimates from several sources of increases in future extreme rainfall. Therefore, it seems that long periods of dry to drought conditions would be punctuated by periods of extreme rainfall. Some of the mechanisms behind this trend include higher temperatures, shorter snow-cover seasons, longer warm seasons, and changing atmospheric and oceanic circulation patterns, as shown conceptually in Figure 5.

Figure 5. Dry times become drier and wet times become wetter
(Source: Wheaton 2013)

Based on research estimating future drought on a regional or global basis, there is a clear evidence for increasing risk of more common and persistent severe future droughts, including on the Canadian Prairies. A summary of projections of probable future drought characteristics emphasizes the consistency in projections of dry times and places becoming drier (Table 1). The Canadian Prairies will not be the only area facing more severe future drought. The IPCC (2012) states that drought will intensify in the twenty-first century in some seasons and areas. These regions include central North America, southern and central Europe, the Mediterranean, Central America and Mexico, northeastern Brazil, and southern Africa.

Lessons for the Future from Paleoclimates

At the end of this chapter, we conclude that current and past droughts may seem mild compared with future possible droughts. By “past” droughts we mean those that have affected the Canadian Prairies since agriculture was introduced, that is, those occurring after the settlement of the Prairies by Euro-Canadians. Also, nearly all planning and resource management that involves weather and water is based on direct observations and information collected from water gauges and weather instruments. This direct observation of weather and water began soon after the railroad was built and settlers arrived, so these records appear to be long, but they actually are very short compared to the age of the Prairie landscape and stream network that formed with the retreat of the continental ice sheet between 12,000 and 18,000 years ago. Climate varies over a large range of temporal scales, spanning seasons to climatic cycles that may last for tens of thousands of years. A weather record that spans decades to at most about 100 years will reveal only the shorter cycles. These short weather records are embedded in longer cycles that can be detected only from indirect study or inference of climate from geological and biological indicators (proxies) of climate variability and change.

The past climate or paleoclimate of the Canadian Prairies has been reconstructed from various climate proxies, including trees growing at the margins of Prairie grasslands and in island forests like the Cypress Hills, and the types and relative abundance of certain minerals, plant remains (e.g., pollen, spores, seeds), and aquatic organisms (e.g., diatoms, ostracodes) found in buried soils and lake sediments. The sampling and analysis of these remnants of prior ecosystems has revealed shifts in climate, in some cases abruptly, over the past 10,000–12,000 years of relative landscape stability. For example, the paleoecology of the Peace–Athabasca Delta (Wolfe et al. 2012) and the paleolimnology of Humboldt Lake, Saskatchewan (Michels et al. 2007) show systematic shifts in moisture regime, including extended dry periods (megadroughts) during the Medieval Climate Anomaly in the ninth to eleventh centuries. These past periods of higher temperature and aridity have been used as temporal analogues of the warmer climate emerging as a result of anthropogenic effects.

In Chapter 2, Sauchyn and Kerr look in detail at the nature of these various climate proxies and how they are used to infer past climate and water conditions. Here we are interested in what the paleoclimate of the Prairies can reveal about the climate to expect in coming decades. In the future, our climate will be increasingly influenced by human modifications of the atmosphere and Earth’s surface. Anthropogenic emissions of greenhouse gases have been apparent only since the mid-nineteenth century and have become a major factor affecting climate only in recent decades. Knowledge of the regional climate regime is extremely important to detect an anthropogenic signal and to separate natural climate variation from what is human-induced. Future climate will be affected by both, although at some point the distinction between natural and anthropogenic will become irrelevant because the “natural” drives of climate (excluding volcanic eruptions), notably ocean-atmosphere circulation anomalies, are part of an increasingly artificial climate system. The paleoclimate record gives us a baseline; it shows the climate cycles as they exist in a mostly natural climate regime. Climate scientists expect that, for at least the next few decades, regional climate fluctuations will mostly consist of natural climate variability (Deser et al. 2012). This scenario applies, in particular, to regions like the Canadian Prairies that have a high degree of climate variability, and thus where the anthropogenic signal is more difficult to detect against the background of extreme inter-annual and decadal variability.

Based on the research above, we can expect that prolonged and severe droughts, similar to those that are evident in the paleoclimate record and discussed in the previous chapter, will reoccur in the coming decades. These droughts were of longer duration and, in some cases, greater severity than the worst droughts of the post-settlement period—those recorded by weather and water gauges. In the absence of global warming, we would expect unprecedented drought conditions. Global warming only amplifies the probability that future droughts will be more severe than those that have produced much of the adaptation of our communities and economy to a dry climate.

Summary and Conclusions

This chapter reviews recent literature regarding characteristics of future drought in the Canadian Prairies. Overall, research results, especially for the Prairies, indicate that dry times are expected to become much drier, and wet times wetter. Probable future droughts in the Canadian Prairies are likely to be drought types that, although perhaps not catastrophic, have the power to slowly erode adaptive capacity of both human and natural capital. Alternatively, the worst-case scenarios for future droughts may have low probability but could be catastrophic.

Current and past droughts may seem mild compared with future possible droughts, and the disruption of the climate by increasing greenhouse gases might result in some additional surprising effects on climate. The nature of future drought is particularly concerning because of insufficient water for increasing atmospheric demands and increasing (and even stable) societal demands. Much-improved adaptation to extremes, such as drought, is needed.

Estimating future droughts and extreme precipitation has several limitations, but projections using several different indicators, climate models, and emission scenarios provide compelling evidence of the risk of increased intensity, duration, frequency, and area of future droughts and extreme precipitation.

References

Barrow, E. 2009. Climate Scenarios for Saskatchewan. Prepared for the Prairie Adaptation Research Collaborative (PARC). Regina, SK: Climate Research Services.

Barrow, E. 2010. Hydroclimate Data for the Prairies: An Analysis of Possibilities. Regina, SK: Climate Research Services.

Bonsal, B., R. Aider, P. Gachon, and S. Lapp. 2013. “An Assessment of Canadian Prairie Drought: Past, Present, and Future.” Climate Dynamics 41, no. 2: 501–16. doi: 10.1007/s00382-012-1422-0.

Bonsal, B., E. Wheaton, A. Chipanshi, C. Lin, D. Sauchyn, and L. Wen. 2011. “Drought Research in Canada: A Review.” Atmosphere-Ocean 49, no. 9: 303–19.

Bonsal, B., E. Wheaton, A. Meinert, and E. Siemens. 2011. “Characterizing the Surface Features of the 1999–2005 Canadian Prairie Drought in Relation to Previous Severe Twentieth Century Events.” Atmosphere-Ocean 49, no. 9: 320–38.

Dai, A. 2010. “Drought under Global Warming: A Review.” WIREs Climate Change 2, no. 1: 45–65. doi: 10.1002/wcc.81.

Deser, C., R. Knutti, S. Solomon, and A.S. Phillips. 2012. “Communication of the Role of Natural Variability in Future North American Climate.” Nature Climate Change 2: 775–79.

IPCC (Intergovernmental Panel on Climate Change). 2012. “Summary for Policymakers. Pp. 1–19 in C.B. Field, V. Barros, T.F. Stocker, Q. Dahe, D.J. Dokken, K.L. Ebi, M.D. Mastrandrea, K.J. Mach, G.-K. Plattner, S.K. Allen, M. Tignor, and P.M. Midgley (eds.), Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation. Cambridge, UK: Cambridge University Press.

Meteorological Service of Canada (MSC) Drought Study Group. 1986. An Applied Climatology of Drought in the Canadian Prairie Provinces. Meteorological Service of Canada, Canadian Climate Centre, Report no. 86-4. Downsview, ON: MSC.

Michels, A., K.R. Laird, S.E. Wilson, D. Thomson, P.R. Leavitt, R.J. Oglesby, and B.F. Cumming. 2007. “Multi-decadal to Millennial-scale Shifts in Drought Conditions on the Canadian Prairies over the Past Six Millennia: Implications for Future Drought Assessment.” Global Change Biology 13: 1295–1307.

PaiMazumber, D., L. Sushama, R. Laprise, M. Khaliq, and D. Sauchyn. 2012. “Canadian RCM Projected Changes to Short- and Long-term Drought Characteristics over the Canadian Prairies.” International Journal of Climatology 33, no. 6: 1409–23. doi:10.1002/joc.3521.

Price, D., D. McKenney, L. Joyce, R. Siltanen, P. Papadopol, and K. Lawrence. 2011. High-resolution Interpolation of Climate Scenarios for Canada Derived from General Circulation Model Simulations. Information Report NOR-X-421. Edmonton, AB: Northern Forestry Centre, Canadian Forest Service.

Sauchyn, D., S. Kennedy, and J. Stroich. 2005. “Drought, Climate Change, and the Risk of Desertification on the Canadian Plains.” Prairie Forum 30, no. 1: 143–56.

Sushama, L., N. Khaliq, and R. Laprise. 2010. “Dry Spell Characteristics over Canada in a Changing Climate as Simulated by the Canadian RCM.” Global and Planetary Change 74: 1–14.

Thorpe, J. 2011. Vulnerability of Prairie Grasslands to Climate Change. Prepared for the Prairies Regional Adaptation Collaborative. Saskatoon, SK: Saskatchewan Research Council.

Wheaton, E. 2013. “Future Risks of Wet and Dry Climate Extremes in Saskatchewan, Canada.” Presentation to the Climate Extremes and Mining Advisory Group. Regina, SK, April 2013.

Wheaton, E., B. Bonsal, and V. Wittrock. 2013. Possible Future Dry and Wet Extremes in Saskatchewan, Canada. Prepared for the Water Security Agency. Saskatoon, SK: Saskatchewan Research Council.

Wheaton, E., V. Wittrock, G. Koshida, E. Siemens, and D. Smeh. 2011. What are the Benefits of Drought? Concepts and Examples to Promote Effecive Climate Adaptation Information Services. Prepared for Adaptation and Impacts Research Section, Environment Canada. Saskatoon, SK: Saskatchewan Research Council.

Wheaton, E., S. Kulshreshtha, V. Wittrock, and G. Koshida. 2008. “Dry Times: Lessons from the Canadian Drought of 2001 and 2002.” The Canadian Geographer 52, no. 2: 241–62.

Wilhite, D. 2000. “Drought as a Natural Hazard: Concepts and Definitions.” Pp. 3–18 in D. Wilhite (ed.), Drought: A Global Assessment. Vol. 1. New York, NY: Routledge Press.

Williams, D., and E. Wheaton. 1998. “Estimating Biomass and Wind Erosion Impacts for Several Climatic Scenarios: A Saskatchewan Case Study.” Prairie Forum 23, no. 1: 49–66.

Wittrock, V., E. Wheaton, and E. Siemens. 2010. More than a Close Call: A Preliminary Assessment of the Characteristics, Impacts of and Adaptations to the Drought of 2008–2010 in the Canadian Prairies. Prepared for Environment Canada. Saskatoon, SK: Saskatchewan Research Council.

Wolfe, B.B., R.I. Hall, T.W.D. Edwards, and J.W. Johnston. 2012. “Developing Temporal Hydroecological Perspectives to Inform Stewardship of a Northern Floodplain Landscape Subject to Multiple Stressors: Paleolimnological Investigations of the Peace-Athabasca Delta.” Environmental Reviews 20: 1–20.