1. Introduction
Although there is a long history of speculation that forests affect rainfall patterns, the idea that terrestrial vegetation may significantly influence climate at regional to global scales received little attention during much of the twentieth century (Thompson 1980; Hayden 1998). Over the past two decades, however, advances in technology have enabled studies of vegetation feedbacks on atmospheric and climatic processes over a wide range of scales (Pielke et al. 1998) through simulation modeling and interdisciplinary field experiments such as the boreal ecosystem–atmosphere study (BOREAS), which focused on a large study area in the western Canadian boreal forest (Sellers et al. 1997). Such studies have resulted in major changes in perspective concerning the nature of vegetation–climate interactions. For example, it had been suggested by Bryson (1966) that the distribution of the Canadian boreal forest is determined by the average positions of the arctic front in winter and in summer. More recently, however, Pielke and Vidale (1995) proposed that the boreal forest itself has a significant influence on the position of the arctic front. This interpretation was based on BOREAS measurements showing that low albedo of the boreal forest causes significantly greater heating of the overlying air mass relative to that over arctic tundra, leading to a significant warming of regional climate (Otterman et al. 1984; Foley et al. 1994; Bonan et al. 1995). Another potentially important feedback of vegetation is the recycling of precipitation and soil moisture through evapotranspiration. Indeed, recent studies indicate that regional rainfall may decline significantly following deforestation and other losses of vegetation cover, especially in continental interiors such as the Amazon basin (Shukla et al. 1990), the Sahel region of west Africa (Savenije 1995), and in northern China and Mongolia (Xue 1996).
In the continental interior of western Canada, the boreal forest occupies an area of more than one million square kilometers and is composed of a variety of forest types, including evergreen conifers [e.g., black spruce, Picea mariana (Mill.) BSP, and jack pine, Pinus banksiana Lamb.] and deciduous hardwoods, of which trembling aspen (Populus tremuloides Michx.) is by far the most abundant (Peterson and Peterson 1992; Hogg 1994). One of the major observations of the BOREAS experiment (Sellers et al. 1997) was that the midsummer evaporative fraction (the ratio of latent heat flux to available energy) was much smaller over the coniferous forest (ca. 0.3–0.45) than over trembling aspen forest (ca. 0.6–0.9). Despite the slightly lower albedo of coniferous forest (Betts and Ball 1997), midsummer transpiration rates were nearly twice as great from the trembling aspen forest as from the coniferous forest ecosystems studied in BOREAS (Black et al. 1996). For deciduous vegetation such as aspen, however, there was strong seasonal variation in energy partitioning associated with leaf emergence and senescence (Blanken et al. 1997), as transpiration rates decreased to near zero during the seasons when the aspen canopy was leafless (typically early October to mid or late May).
Aspen-dominated deciduous and mixed-wood forests occupy a significant proportion of the southern boreal forest of western Canada (Peterson and Peterson 1992), especially in the province of Alberta (Fig. 1), where these forest types occupy about half of the total forested area based on maps from the Canadian Forest Inventory (Lowe et al. 1994). Trembling aspen is also the predominant tree species in the aspen parkland, a transitional vegetation zone located between the boreal forest and the prairies to the south (Bird 1961; Hogg and Hurdle 1995).
There has been increasing recognition of the importance of adequately characterizing seasonal changes in land surface characteristics, such as the leaf area index of deciduous vegetation, for realistic regional representation of climate in general circulation model (GCM) simulations (e.g., Xue et al. 1996). For example, Schwartz (1992, 1996) showed that the onset of spring leafing in eastern North America leads to discontinuities in the time series of meteorological conditions near the surface, including humidity, wind, and diurnal temperature range. Thus, the seasonal cycle of vegetation leafing and senescence may exert a strong control over land–atmosphere interactions (Moulin et al. 1997). Despite the abundance of aspen forests and other deciduous vegetation in the western Canadian interior, the potential role of leaf phenology as a cause of feedback on the climate of this region has received little attention. Based on model simulations showing that terrestrial evapotranspiration promotes rainfall and causes summer cooling (e.g., Shukla and Mintz 1982; Shukla et al. 1990; Dirmeyer 1994), we postulated that leaf phenology of deciduous vegetation may exert a significant influence on seasonal patterns of temperature and precipitation in the west-central Canadian interior east of the Rocky Mountains (Fig. 1).
Even if such feedbacks are important, they may be difficult to detect over one or two growing seasons because of high temporal and spatial variability in meteorological conditions, particularly precipitation. An analysis of the long-term climate record, however, might provide evidence of vegetation feedbacks in this region. The objective of this study was to examine the geographic variation in long-term averages of seasonal changes in temperature and precipitation patterns as a preliminary means of determining the likely importance of deciduous vegetation phenology on the climate of the western Canadian interior. We also compared the results of this analysis with observed seasonal changes in energy partitioning and transpiration at the BOREAS Old Aspen tower site, located centrally within the region of interest near the southern edge of the Canadian boreal forest.
2. Methods
a. Climate data
Climate data used in the analyses included the 1961–90 monthly mean temperature and precipitation normals for Canadian stations from Environment Canada (1994). Additional climate data for 1961–90 were used to generate maps of spatial variability across North America (see below). A more detailed analysis was also conducted using daily temperature and precipitation data for nine climate stations located in or near the major commercial zone of trembling aspen in the southern boreal forest of the Canadian prairie provinces of Alberta, Saskatchewan, and Manitoba (Table 1; Fig. 1). These stations were selected based on having a relatively long period of record (30–107 years) and proximity to grid points of GCM outputs from the Canadian Climate Centre (see below).
b. Pattern of seasonal temperature variation
The seasonal cycle of long-term, mean temperature closely resembles a simple sinusoidal function at most midlatitude locations, primarily owing to the sinusoidal nature of the annual variation in solar forcing resulting from the earth’s orbit around the sun and its tilted axis of rotation (List 1958; Jones 1992). However, in a preliminary analysis, we noticed a distinctive pattern of seasonal variation in mean monthly temperature at several climate stations in the western Canadian interior. This pattern is characterized by long-term mean temperatures that are anomalously warmer during April and October, relative to what would be expected from a sinusoidal function, with correspondingly cooler-than-expected mean temperatures in midsummer (example shown for Prince Albert, Saskatchewan, in Fig. 2).
The best-fitting sinusoidal function was determined for each climate station through an iterative, nonlinear regression procedure in which the values of the three coefficients were sequentially varied to obtain the minimum sum of squares of the deviations between observed and modeled mean temperatures for each month. Coefficients a and b were initially set to the corresponding values of observed mean temperature and its seasonal amplitude, while the lag coefficient (c) was varied in steps of 0.1 day from 0 to 35 day to obtain the smallest sum of squares of the deviations. Using the optimum value of c, the amplitude (b) was then varied in steps of 0.1°C across all values within ±5°C from its initial value. The optimum value of b was then used and the procedure was repeated until the sum of squares of the deviations reached a constant minimum value. Coefficients b and c were then set to their optimum values while varying the value of a in 0.05°C steps, but the optimum value of a was always within ±0.1°C of its initial (observed) value. The deviations (ΔT) of observed minus modeled values of mean monthly temperature were then determined for each station.
c. Seasonal variation in solar radiation
Seasonal changes in observed values of monthly mean global solar radiation (Environment Canada 1982) were also examined from available data for the region of interest. The effect of latitude on the seasonal pattern of potential (extraterrestrial) solar radiation incident on a horizontal surface was also modeled using a simple sinusoidal function for solar declination (annual amplitude of 23.44°) with an assumed seasonally constant solar flux density of 1370 W m−2. Values of either observed or potential solar radiation were fitted to a sinusoidal function using the same method as that used for temperature [Eq. (1)].
d. Seasonal pattern of precipitation
e. Mapping of spatial variation in climatic characteristics
Monthly normals for the period 1961–90 were obtained (Environment Canada 1994) for 2831 climate stations for temperature and for 3154 climate stations for precipitation, supplemented by data for 1726 British Columbia stations obtained from C. Daly at Oregon State University (http://www.ocs.orst.edu/pub/daly/bc_p2.1st). In addition, mean monthly values for the 1961–90 period for the United States and Mexico were calculated using adjusted data for 2009 temperature and 2675 precipitation stations extracted from the Global Historical Climate Network database (http://www.ncdc.noaa.gov/cgi-bin/res40.pl). The combined dataset provided good spatial coverage for most of North America, especially over most southern portions of Canada (Fig. 3). Indices of anomalous warmth in April and October [IT; Eq. (3)] and of summer precipitation patterns [IP; Eq. (4)] were calculated for each station location. Elevation data were obtained from the hydrologically corrected USGS GTOPO30 digital elevation model (http://edc.usgs.gov/landdaac/gtopo30/hydro/na_dem.html) and resampled using the ARC/INFO GRID Program (Environmental Systems Research Institute, Inc., Redlands, CA) to a 10-km square grid on the Lambert Conformal Conic projection. Values of IT and IP were then interpolated to this 10-km grid by applying the gradient plus inverse distance squared (GIDS) weighting interpolation method (Nalder and Wein 1998; Price et al. 1998; Price et al. 2000), treating latitude, longitude, and elevation as independent variables.
f. Comparison with GCM outputs
The average observed value of IT was estimated for the commercial zone of trembling aspen in the southern boreal forest of western Canada, based on the 1961–90 monthly climate normals from the nine representative stations (Table 1) used in the more detailed analyses indicated above. This was compared with the corresponding average value of IT calculated at the nine adjacent grid points from the 1 × CO2 outputs of the Canadian Climate Centre GCM2 (McFarlane et al. 1992;http://www.cccma.bc.ec.gc.ca).
g. Field measurements at a boreal aspen site
A large international field experiment (BOREAS) was conducted in the region, primarily during 1994 and 1996. BOREAS has provided a wealth of information on mass and energy exchange between the boreal forest and the atmosphere (Sellers et al. 1997). Measurements included tower-based monitoring of sensible and latent heat (water vapor) fluxes at sites in several forest types, including the Old Aspen site (53°38′N, 106°12′W), located about 70 km northeast of Prince Albert, Saskatchewan (Fig. 1). This site is almost exclusively dominated by a pure stand of trembling aspen about 70 years old and 18–22 m tall, with an understory shrub layer dominated by 2-m-tall beaked hazelnut (Corylus cornuta Marsh.). Eddy correlation measurements of sensible and latent heat fluxes were made at the site in 1993–94 (Black et al. 1996; Blanken et al. 1997), while water fluxes were also estimated from sap flow measurements within individual aspen stems at the same site (Hogg et al. 1997). However, these measurements did not fully cover the autumn leaf-fall period; thus in this paper, we report similar results for 1996, when both sap flow and eddy correlation measurements were made continuously from mid-April to late October. For each day, 24-h averages of sensible and latent heat flux were calculated from half-hourly measurements made at the top of a 39-m tower above the aspen canopy using the methods described by Blanken et al. (1997). Sap flow was monitored hourly at the 1.3-m height on two aspen trees at each of two locations using the heat pulse method (Hogg and Hurdle 1997; Hogg et al. 1997). Results from the four trees were averaged and reported as daily averages of sap flow per unit area of conducting sapwood (Qs; mm3 h−1 m−2) (Edwards et al. 1996; Hogg et al. 2000).
3. Results and discussion
a. Seasonal patterns in monthly mean temperature
Long-term monthly mean temperatures showed a distinctive seasonal pattern over the southern boreal forest of the west-central Canadian interior, where trembling aspen form a major component of the vegetation (Fig. 1). The general nature of this pattern is shown in the climate of Prince Albert, Saskatchewan, where spring and autumn temperatures are warmer and summer and winter temperatures are cooler than expected from the best-fitting sinusoidal equation (Fig. 2). The seasonal pattern and magnitude of monthly deviations (ΔT) were similar for representative climate stations located in or near major areas of aspen forest cover over a geographic area extending at least 1500 km, from northwestern Alberta to southern Manitoba (Fig. 4a). The months with the greatest positive values of ΔT were in April (+1.7° to +2.6°C) and October (+2.4° to +3.5°C), giving average April and October temperature anomalies (IT) of 2.1° to 2.9°C (Table 1). Values of ΔT were consistently negative during the summer months of June, July, and August at each of these stations. Seasonal patterns in ΔT were also similar at other stations located elsewhere in the Canadian boreal forest across a distance of >3000 km (Fig. 4b), although the trend was less pronounced at the easternmost location in western Quebec (IT = 1.5°C). At the selected Canadian climate stations outside the range of the boreal forest, however, the seasonal patterns in ΔT were inconsistent, with little or no tendency for positive temperature anomalies in April and October (IT from −0.4° to +1.0°C) (Fig. 4c).
Mapping of spatial variation in IT (Fig. 5) shows a zone of elevated values (+1.4° to +3.0°C) that extends >4000 km from the Yukon–Alaska boundary to northern Quebec, encompassing most of the Canadian boreal forest, aspen parkland, and northern prairies. The areas with largest IT (>2.2°C) include most of the major areas with significant aspen-dominated forest cover in the portion of the Canadian boreal forest west of 95°W longitude (Fig. 1), although high values also occurred in the Yukon and the northern third of Saskatchewan, where aspen is common but not a regionally dominant component of the vegetation. In east-central North America, values of IT appeared to be reduced by about 1°C in the Great Lakes region, including the major aspen areas in Ontario and Quebec. Slightly positive values of IT (0° to +1.4°C) occurred over most of the other forested areas in Canada and also across much of the central and eastern United States. Values of IT were slightly negative (−1° to 0°C) along the Atlantic and Pacific coasts and across most of the western United States and strongly negative over western Alaska and the high arctic islands, especially Ellesmere Island (IT < −4°C).
b. Solar radiation patterns
As might be expected, seasonal change in observed global solar radiation closely approximates a sinusoidal relationship based on the long-term monthly average of four climate stations with available data in the boreal forest and aspen parkland of western Canada (latitudes ranging from about 50° to 55°N) (Fig. 6a) Thus, observed solar radiation patterns cannot explain the distinctive seasonal pattern for mean temperature in this region, where conditions in spring and autumn are warmer and summer temperatures are correspondingly cooler than would be expected from a sinusoidal relationship.
If the effects of cloud and other atmospheric absorption influences are neglected, the theoretical seasonal pattern of potential solar radiation incident to a horizontal surface is nearly sinusoidal at a latitude of 45°N, but for more northerly latitudes, potential solar radiation exhibits negative anomalies from a best-fitting sinusoidal equation during spring and autumn (as shown for 60°N in Fig. 6b). Thus, it appears that the expected latitudinal influences on solar radiation patterns have a relatively minor influence on the observed spatial variation in IT across Canada and the northern United States (Fig. 5) except in areas north of the Arctic Circle, where the observed strongly negative values of IT may be explained, at least partially, by the prolonged winter period of polar night.
c. Seasonal patterns in monthly mean precipitation
Climate stations in the southern boreal forest and aspen parkland of western Canada show strong seasonal variation in precipitation (Fig. 7). Mean monthly precipitation generally peaks at 60–100 mm during the summer months (June–August) but is typically low (15–30 mm) from October to April. The spatial distribution of IP (Fig. 8) shows that the area of the Canadian boreal forest tends to exhibit greater summer enhancement of precipitation than the areas located either to the north (subarctic) or south (prairie or temperate forest). Over most of the boreal forest and aspen parkland, there is a significant summer enhancement of precipitation (IP > 40 mm) except in the more northerly areas (e.g., Northwest Territories) or in areas adjacent to large bodies of water (e.g., the Great Lakes, Lake Winnipeg, and along the Atlantic coast). Values of IP are particularly high (>60 mm) over north-central Alberta (Fig. 1), where areal coverage of trembling aspen is highest based on the Canadian Forest Inventory (e.g., Lowe et al. 1994). It should be noted, however, that other areas of North America showed values of IP that were nearly as great (e.g., northern Quebec) or even greater (e.g., southern Alaska and along the Atlantic coast of the southeastern United States) than those recorded in Alberta.
d. Observed seasonal changes in a boreal aspen forest
Tower-based, eddy correlation measurements made at the BOREAS Old Aspen site (location shown in Fig. 1) during 1996 showed dramatic seasonal changes in energy partitioning and transpiration associated with spring leafing and autumn leaf fall of the aspen canopy (Fig. 9). The flux of sensible heat greatly exceeded that of latent heat (water vapor) during the spring (April and May) when the aspen was leafless. However, sensible heat flux decreased significantly during the period of increasing leaf area from late May to early June, when there was a corresponding increase in latent heat (water vapor) flux that continued to dominate over sensible heat until the beginning of leaf senescence in early September. Measurements of upward movement of water (sap flow) within individual aspen stems showed a similar seasonal pattern to that of latent heat flux from the ecosystem. Both sets of measurements support the hypothesis that leaf phenology is a major determinant of seasonal changes in evapotranspiration from the aspen-dominated landscape. Average daily evapotranspiration rates from the forest were more than five times greater (2.6 mm day−1) during the period with significant sap flow when the aspen canopy was in leaf (29 May–28 September) than during the leafless periods in spring and autumn (average of 0.46 mm day−1 for 19 April–28 May and 29 September–30 October). Similar results were reported at this site for 1993–94 (Blanken et al. 1997) except that the increase in latent heat flux started earlier, as warmer temperatures in May evidently resulted in earlier spring leafing of the aspen canopy (Chen et al. 1999).
Five years of measurements at this aspen site, including studies of seasonal change in leaf area index (A. G. Barr et al. unpublished), have shown that the spring increase in water vapor flux was closely linked to the emergence and expansion of leaves, which started as early as the end of April during the unusually warm spring of 1998. The range of dates observed for the first emergence of aspen leaves was comparable to that recorded for this species by Ahlgren (1957) over a 5-yr period in northeastern Minnesota (27 April–21 May). It should be noted, however, that there is often a 2–3-week variation in leaf phenology among aspen clones (Witter and Waisenen 1978) such that leaf emergence can be delayed until early June in late-leafing clones during cool years.
e. Seasonal change in average daily climate at the regional scale
Mean seasonal changes in temperature and precipitation for each day of the year, shown in Fig. 10, based on averages of the long-term climate records from the nine stations located in or near major areas of aspen cover in the Canadian prairie provinces of Alberta, Saskatchewan, and Manitoba (Fig. 1). These seasonal patterns have a higher temporal resolution than those shown in Figs. 4 and 7, but are otherwise similar. The positive anomalies in spring and autumn mean temperature (ΔT) are clearly evident, reaching their maximum values in mid-April and mid- to late October (Fig. 10). These events coincide approximately with the start and end of the growing season in the region, when snow is generally absent from the landscape. One aspect of these positive temperature anomalies in spring and fall is that the length of the growing season was significantly greater than that obtained from a best-fitting sinusoidal function with the same mean annual temperature. For example, the seasonal period when long-term, mean daily temperature exceeds 5°C is 18 days longer (23 April–13 October) than that obtained from the sinusoidal model (1 May–3 October). Despite the longer growing season, however, observed temperatures during the summer are cooler than those estimated by the same sinusoidal model (mean ΔT = −1.4°C during June and July).
The summer period with negative values of ΔT (late May to mid-September) coincides with the typical period when green leaves are present in the forest canopies of aspen and other deciduous species in the region (Ahlgren 1957). This period also coincides with the season of maximum precipitation, with daily values of 1.5–3.0 mm day−1. In contrast, mean daily precipitation is remarkably constant at a much lower value, near 0.7 mm day−1, from early October to mid-April, which includes the typical 1-month period in autumn between leaf fall and the development of continuous snow cover (Fig. 10).
f. Postulated feedbacks of vegetation phenology on regional climate
Our results show some intriguing coincidences, both spatially and temporally, between seasonal climate patterns and the distribution and leaf phenology of trembling aspen forests in the west-central Canadian interior. The area occupied by aspen is difficult to quantify precisely, but an analysis of the Canadian Forest Inventory (Lowe et al. 1994; Penner et al. 1997) indicates that in the region west of Ontario, aspen forests occupy an area of at least 200 000 km2. If aspen-dominated mixed woods are also included, the total area may exceed 500 000 km2. Such areas are at least comparable to the total area occupied by the Great Lakes (240 000 km2), which have been recognized as having a significant effect on the climate in that region (e.g., Changnon and Jones 1972).
Aspen-dominated areas in the west-central Canadian interior (Fig. 1) generally exhibited larger April and October temperature anomalies (Fig. 5) than those recorded anywhere else in North America, including Alaska, the continental United States, and Mexico. It should be noted, however, that the anomalies were nearly as great in adjacent areas of the Canadian prairies to the south and in conifer-dominated portions of the boreal forest to the north.
Although less striking at the continental scale (Fig. 8), there was also a tendency for a large summer enhancement of mean precipitation in aspen-dominated areas, especially in central Alberta. The analysis of mean daily climate for the nine stations representing aspen-dominated areas in western Canada revealed more than a doubling of mean precipitation (from <1 to 2 mm day−1) (Fig. 10) during the typical period, when spring leafing of aspen occurs (early May to early June), and a halving of mean precipitation (from 2 to <1 mm day−1) during the autumn period, when aspen leaves turn color and abscise (early September to early October). Spring leaf phenology of aspen and other deciduous tree species is determined primarily by cumulative thermal sums (Lechowicz 1984); thus, the onset of spring rainfall may be expected to contribute little to leaf phenology, as forest soils in the region are normally wet following snow melt (typically during April).
Although aspen forests are a major component of the landscape in this region, it is important to recognize that there are other deciduous vegetation types that may be expected to exert similar feedbacks on the climate of this region. These would include other deciduous tree species with similar leaf phenology (e.g., see Ahlgren 1957) as well as boreal fens with a significant component of deciduous vegetation, where seasonal changes in evaporative fraction (Lafleur et al. 1997) appear to be similar to those over aspen forests. Another important component of the landscape of western Canada is represented by the aspen parkland and prairie grassland zones, where a large proportion of the area is under cultivation for summer wheat and other agricultural crops. During the period of winter snow cover, typically from November to March or April, high albedos of these open landscapes (e.g., Betts and Ball 1997) may lead to seasonally cooler temperatures than would be expected if forest cover was present (e.g., Bonan et al. 1992). At other times of year, micrometeorological measurements of evapotranspiration and energy partitioning over natural grassland in southern Saskatchewan (Ripley and Saugier 1978), and even as far south as Kansas (Verma et al. 1992), show seasonal patterns that are somewhat similar to those observed for aspen (Fig. 9);however, strong decreases in grassland transpiration were noted in both areas during summer periods with low soil moisture. Even when irrigated, water use by agricultural crops varies strongly with the stage of development, and transpiration rates typically reach their seasonal maximum for only a few weeks after full cover is achieved (see Jensen et al. 1990). Thus, although leaf phenology of grassland and agricultural crops may contribute to the observed seasonal climate patterns in the west-central Canadian interior, the resultant vegetation feedbacks are presumably smaller, especially over the climatically drier regions of the Canadian prairies (Hogg 1994).
Evergreen, coniferous forests predominate over large areas of the western Canadian interior, especially in cordilleran regions and in the mid- and northern boreal forests of Saskatchewan and Manitoba (Fig. 1). These forest types have previously been shown to exert a significant feedback through the warming of the regional climate due to low albedo and high sensible heat fluxes (e.g., Bonan et al. 1995; Pielke and Vidale 1995). Their potential contribution to the observed seasonal climate anomalies is less clear, however, because leaf area and albedo (Betts and Ball 1997) of these forest types show relatively little seasonal variation compared with aspen-dominated forests. As previously noted, midsummer water vapor fluxes from landscapes dominated by boreal conifers are apparently much lower than those dominated by deciduous vegetation such as aspen. Nevertheless, frozen soils during the spring can delay the onset of transpiration in the boreal coniferous forest, leading to large sensible heat fluxes (Betts et al. 1998, 1999) and thus potentially contributing to the high values of IT that extend into the more northerly, conifer-dominated areas of this region.
If deciduous forest phenology is a major contributor to the April and October temperature anomalies in the southern boreal forest of western Canada, then similar effects might also be anticipated in other continental areas dominated by deciduous trees, notably the temperate forests of eastern North America. Differences in the seasonal timing of such anomalies might be expected, however, because of earlier spring leafing and later autumn leaf fall under the milder climatic conditions. In a preliminary analysis (not shown), we have noted that positive temperature anomalies occur in March and November across much of the northeastern United States and adjacent southern Canada, where these temperate deciduous forests are abundant. Further examination of relationships between leaf phenology and seasonal temperature would therefore be useful in revealing evidence of vegetation feedbacks in eastern North America and elsewhere. The index (It) used in the present study would need to be modified, however, before it would be appropriate for investigating postulated temperature feedbacks of vegetation phenology in temperate regions.
In the absence of model simulations specifically tailored to address questions of how seasonal changes in transpiration may affect atmospheric processes in this region (e.g., Xue et al. 1996), it is difficult to assess the degree to which vegetation feedbacks, including seasonal leafing of deciduous forests, may contribute to the large summer enhancement of rainfall in the western Canadian interior. Based on a global analysis by Trenberth (1998), the proportion of precipitation recycled by vegetation may be less than 10% when measured over distances of 500 km. Estimates by Brubaker et al. (1993) indicated a recycling rate of 34% for July rainfall over a 10°latitude by 20°longitude region of the central United States, an area comparable to that of the Canadian prairie provinces. Part of the difference in these estimates, however, results from the spatial scale being considered. When considered collectively over large areas within continental interiors, terrestrial vegetation may be found to exert a dominant influence over rainfall patterns; for example, in the upper Amazon basin (Lettau et al. 1979) and in the Sahel region of west Africa (Savenije 1995), where it was estimated that evapotranspiration by the regional vegetation contributes about 90% of local rainfall. Although comparable estimates are not available for the western Canadian interior, global analyses by Charles et al. (1994) indicate that moisture recycling at the continental scale might account for more than 50% of the annual precipitation in this region.
Earlier generations of GCMs, which lack sophisticated algorithms for vegetation–atmosphere interactions, may be expected to show limitations in their ability to characterize seasonal changes in regional climate, even under present conditions. For example, an examination of the 1 × CO2 output from the Canadian Climate Centre’s GCM2 (Boer et al. 1992; McFarlane et al. 1992) shows no evidence of the distinctive positive temperature anomalies that we detected in the observed climate record for April and October in the western Canadian interior. In fact, the value of IT was found to be slightly negative (−0.6°C) for the average modeled climate of the GCM2 for the nine grid nodes (see Methods) in aspen-dominated areas of the Canadian prairie provinces, whereas the long-term climate data indicated a strongly positive value of +2.5°C (based on data in Fig. 10 for the nine climate stations located near these same grid nodes). A comparable analysis for precipitation patterns indicates that observed seasonal variation in monthly precipitation is larger than that modeled by the GCM primarily because the modeled winter snowfall (43–50 mm water equivalent per month) is much greater than observed (16–21 mm month−1). This difference is significant even when allowing for a possible underestimation of about 30% in the historical measurements of snowfall amounts (A. G. Barr, Environment Canada, 1999, personal communication). The errors in winter precipitation are smaller in the more recent Canadian Regional Climate Model (CRCM) (Laprise et al. 1998) presumably because the higher spatial resolution of the CRCM more accurately characterizes the topographic influence of the mountain ranges that block the passage of moisture from the Pacific into the western Canadian interior. Based on the results of the present analysis, significant further improvements in the simulation of seasonal variation in both temperature and precipitation may be expected through the implementation of more sophisticated schemes for representing vegetation–atmosphere interactions (e.g., Verseghy 1996; Xue et al. 1996; Bonan 1997), especially the inclusion of deciduous forest phenology (e.g., Foley et al. 1996).
4. Conclusions
The analysis of long-term climate records shows distinctive seasonal patterns in both temperature and precipitation that broadly coincide with the major areas of aspen-dominated forest in the west-central Canadian interior. Observed seasonal patterns of energy partitioning and forest transpiration at the BOREAS Old Aspen site have led us to the hypothesis that these observed patterns of regional climate may be significantly influenced by leaf phenology of aspen and associated deciduous vegetation. Although the magnitude and significance of the postulated feedbacks cannot be established conclusively from the results of this study, the analysis of seasonal climate patterns appears to be consistent with our hypothesis. Nevertheless, further investigation is warranted in view of the potentially major influences of other factors on the observed seasonal climate patterns, notably the seasonal movements of air masses and associated fronts as well as regional feedbacks of other vegetation types, including boreal coniferous forest and cropland.
To our knowledge, the distinctive seasonal pattern of mean temperature in this region, characterized by anomalously warm temperatures in April and October, has not been previously reported. In the present study, this seasonal pattern was quantified, and its spatial and temporal distributions were examined through analyses of deviations from a best-fitting sinusoidal equation. The indices developed from these deviations should provide a useful means of testing the ability of models to simulate the seasonal features of climate in this region and elsewhere. A preliminary analysis indicated that the distinctive April and October temperature anomalies observed in western Canada are not represented by at least one of the earlier-generation GCMs. Similar tests on more recent models would be useful in determining their ability to simulate these seasonal temperature anomalies; however, it appears unlikely that any GCM will reproduce these anomalies unless its internal representation of surface characteristics explicitly considers the seasonal changes in the leaf area of deciduous forests in this region.
One of the implications of the observed seasonal temperature anomalies is that the length of the growing season across much of the western Canadian boreal forest is about 10% greater than what would be expected from a best-fitting sinusoidal equation with the same mean annual temperature. During the summer, however, temperatures are cooler than what would be expected from the same sinusoidal equation, while mean precipitation is much greater than at other times of year. Both of these characteristics would be expected to favor forest growth because the high vapor pressure deficits that typically occur during hot weather lead to reductions in stomatal conductance and CO2 uptake (e.g., Dang et al. 1997) while higher midsummer temperatures would tend to increase plant respiration rates. Thus, if the postulated feedbacks on regional climate are significant, then the presence of trembling aspen and other deciduous vegetation could play a role in maintaining the current distribution of the boreal forest in western Canada (cf. Hogg 1994, 1997). Determining the importance of these postulated effects, however, will likely await the application of regional land surface models that include phenological changes of deciduous vegetation within global models of the earth’s climate system.
Acknowledgments
This work was supported by funding from the Climate Change Network of the Canadian Forest Service and the Natural Sciences and Engineering Research Council of Canada (NSERC), with logistical support for the BOREAS field measurements from Atmospheric Environment Service, Parks Canada, and the National Aeronautics and Space Administration. Sap flow measurements were implemented by P. A. Hurdle, and analyses of tower-based flux measurements were conducted with the support of A. G. Barr and Z. Chen. M. Siltanen provided assistance with the spatial mapping, and helpful comments on the manuscript were provided by M. D. Flannigan, B. D. Amiro, I. D. Campbell, A. G. Barr, and two anonymous reviewers.
REFERENCES
Ahlgren, C. E., 1957: Phenological observations of nineteen native tree species in northeastern Minnesota. Ecology,38, 622–628.
Betts, A. K., and J. H. Ball, 1997: Albedo over the boreal forest. J. Geophys. Res.,102 (D24), 28 901–28 909.
——, P. Viterbo, A. Beljaars, H.-L. Pan, S.-Y. Hong, M. Goulden, and S. Wofsy, 1998: Evaluation of land-surface interaction in ECMWF and NCEP/NCAR reanalysis models over grassland (FIFE) and boreal forest (BOREAS). J. Geophys. Res.,103 (D18), 23 079–23 085.
——, M. Goulden, and S. Wofsy, 1999: Controls on evaporation in a boreal spruce forest. J. Climate,12, 1601–1618.
Bird, R. D., 1961: Ecology of the aspen parkland of western Canada in relation to land use. Contribution 27, Research Branch, Canada Department of Agriculture, Ottawa, ON, Canada, 155 pp.
Black, T. A., and Coauthors, 1996: Annual cycles of water vapour and carbon dioxide fluxes in and above a boreal aspen forest. Global Change Biol.,2, 219–229.
Blanken, P. D., and Coauthors, 1997: Energy balance and canopy conductance of a boreal aspen forest: Partitioning overstorey and understorey components. J. Geophys. Res.,102 (D24), 28 915–28 927.
Boer, G. J., N. A. McFarlane, and M. Lazare, 1992: Greenhouse gas-induced climate change simulated with the CCC second-generation general circulation model. J. Climate,5, 1045–1077.
Bonan, G. B., 1997: Effects of land use on the climate of the United States. Climatic Change,37, 449–486.
——, D. Pollard, and S. L. Thompson, 1992: Effects of boreal forest vegetation on climate. Nature,359, 716–718.
——, F. S. Chapin III, and S. L. Thompson, 1995: Boreal forest and tundra ecosystems as components of the climate system. Climatic Change,29, 145–167.
Brubaker, K. L., D. Entekhabi, and P. S. Eagleson, 1993: Estimation of continental precipitation recycling. J. Climate,6, 1077–1089.
Bryson, R. A., 1966: Air masses, streamlines, and the boreal forest. Geogr. Bull.,8, 228–269.
Changnon, S. A., Jr., and D. M. A. Jones, 1972: Review of the influence of the Great Lakes on weather. Water Resour. Res.,8, 360–371.
Charles, C. D., D. Rind, J. Jouzel, R. D. Koster, and R. G. Fairbanks, 1994: Glacial-interglacial changes in moisture sources for Greenland: Influences on the ice core record of climate. Science,263, 508–511.
Chen, W. J., and Coauthors, 1999: Effects of climate variability on the annual carbon sequestration by a boreal aspen forest. Global Change Biol.,5, 41–53.
Dang, Q.-L., H. A. Margolis, M. R. Coyea, M. Sy, and G. J. Collatz, 1997: Regulation of branch-level gas exchange of boreal trees:Roles of shoot water potential and vapor pressure difference. Tree Physiol.17, 521–535.
Dirmeyer, P. A., 1994: Vegetation stress as a feedback mechanism in midlatitude drought. J. Climate,7, 1463–1483.
Ecoregions Working Group, 1989: Ecoclimatic regions of Canada, first approximation. Ecological Land Classification Series 23, Ecoregions Working Group of the Canada Committee on Ecological Land Classification, 118 pp. and 1 map.
Edwards, W. R. N., P. Becker, and J. Cermak, 1996: A unified nomenclature for sap flow measurements. Tree Physiol.17, 65–67.
Environment Canada, 1982: Canadian climate normals 1951–80. Vol. 1, Solar radiation, Canadian Climate Program, Environment Canada, 57 pp.
——, 1994: Canadian Monthly Climate Data and 1961–90 Normals. Canadian Meteorological Centre, Environment Canada, CD-ROM.
Fisheries and Environment Canada, 1978: Hydrological Atlas of Canada. Surveys and Mapping Branch, Department of Energy, Mines and Resources, 34 maps.
Foley, J. A., J. E. Kutzbach, M. T. Coe, and S. Levis, 1994: Feedbacks between climate and boreal forests during the Holocene epoch. Nature,371, 52–54.
——, I. C. Prentice, N. Ramankutty, S. Levis, D. Pollard, S. Sitch, and A. Haxeltine, 1996: An integrated biosphere model of land surface processes, terrestrial carbon balance, and vegetation dynamics. Global Biogeochem. Cycles,10, 603–628.
Hayden, B. P., 1998: Ecosystem feedbacks on climate at the landscape scale. Philos. Trans. Roy. Soc. London,353B, 5–18.
Hogg, E. H, 1994: Climate and the southern limit of the western Canadian boreal forest. Can. J. For. Res.,24, 1835–1845.
——, 1997: Temporal scaling of moisture and the forest-grassland boundary in western Canada. Agric. For. Meteor.,84, 115–122.
——, and P. A. Hurdle, 1995: The aspen parkland in western Canada:A dry-climate analogue for the future boreal forest? Water Air Soil Pollut.,82, 391–400.
——, and ——, 1997: Sap flow in trembling aspen: Implications for stomatal responses to vapor pressure deficit. Tree Physiol.,17, 501–509.
——, and Coauthors, 1997: A comparison of sap flow and eddy fluxes of water vapor from a boreal deciduous forest. J. Geophys. Res.,102 (D24), 28 929–28 937.
——, B. Saugier, J.-Y. Pontailler, T. A. Black, W. Chen, P. A. Hurdle, and A. Wu, 2000: Responses of trembling aspen and hazelnut to vapor pressure deficit in a boreal deciduous forest. Tree Physiol., in press.
Jensen, M. E., R. D. Burman, and R. G. Allen, Eds., 1990: Evapotranspiration and irrigation requirements. Manuals and Reports on Engineering Practice 70, 360 pp. [Available from ASCE Publications, 1801 Alexander Bell Dr., Reston, VA 20191.].
Jones, H. G., 1992: Plants and Microclimate, a Quantitative Approach to Environmental Plant Physiology. 2d ed. Cambridge University Press, 428 pp.
Lafleur, P. M., J. H. McCaughey, D. W. Joiner, P. A. Bartlett, and D. E. Jelinski, 1997: Seasonal trends in energy, water, and carbon dioxide fluxes at a northern boreal wetland. J. Geophys. Res.,102 (D24), 29 009–29 020.
Laprise, R., D. Caya, M. Giguere, G. Bergeron, H. Cote, J.-P. Blanchet, G. J. Boer, and N. A. McFarlane, 1998: Climate and climate change in western Canada as simulated by the Canadian Regional Climate Model. Atmos.–Ocean,36, 119–167.
Lechowicz, M. J., 1984: Why do temperate deciduous trees leaf out at different times? Adaptation and ecology of forest communities. Amer. Nat.,124, 821–842.
Lettau, H., K. Lettau, and L. C. B. Molion, 1979: Amazonia’s hydrologic cycle and the role of atmospheric recycling in assessing deforestation effects. Mon. Wea. Rev.,107, 227–238.
List, R. J., 1958: Smithsonian Meteorological Tables. Vol. 114. Smithsonian Miscellaneous Collections, Smithsonian Institution, 527 pp.
Lowe, J. J., K. Power, and S. L. Gray, 1994: Canada’s forest inventory 1991. Canadian Forest Service Information Rep. PI-X-115, 67 pp. plus 9 maps. [Available from Pacific Forestry Centre, 506 Burnside Road, Victoria, BC V8Z 1M5, Canada.].
McFarlane, N. A., G. J. Boer, J.-P. Blanchet, and M. Lazare, 1992: The Canadian Climate Centre second-generation general circulation model and its equilibrium climate. J. Climate,5, 1013–1044.
Moulin, S., L. Kergoat, N. Viovy, and G. Dedieu, 1997: Global-scale assessment of vegetation phenology using NOAA/AVHRR satellite measurements. J. Climate,10, 1154–1170.
Nalder, I. A., and R. W. Wein, 1998: Spatial interpolation of climatic Normals: Test of a new method in the Canadian boreal forest. Agric. For. Meteor.,9, 211–225.
Otterman, J., M.-D. Chou, and A. Arking, 1984: Effects of nontropical forest cover on climate. J. Climate Appl. Meteor.,23, 762–767.
Penner, M., K. Power, C. Muhairwe, R. Tellier, and Y. Wang, 1997:Canada’s forest biomass resources: Deriving estimates from Canada’s forest inventory. Canadian Forest Service Information Rep. BC-X-370, 33 pp. [Available from Pacific Forestry Centre, 506 Burnside Road, Victoria, BC V8Z 1M5, Canada.].
Peterson, E. B., and N. M. Peterson, 1992: Ecology, management, and use of aspen and balsam poplar in the Prairie Provinces, Canada. Special Rep. 1, Forestry Canada, Northern Forestry Centre, 252 pp. [Available from Northern Forestry Centre, 5320-122 Street, Edmonton, AB T6H 3S5, Canada.].
Pielke, R. A., and P. L. Vidale, 1995: The boreal forest and the polar front. J. Geophys. Res.,100 (D12), 25 755–25 758.
——, R. Avissar, M. Raupach, A. J. Dolman, X. Zeng, and A. S. Denning, 1998: Interactions between the atmosphere and terrestrial ecosystems: Influence on weather and climate. Global Change Biol.,4, 461–475.
Price, D. T., I. A. Nalder, and R. M. Siltanen, 1998: A 10-km national climate surface for Canadian global change studies. Scaling and Modelling in Forestry: Applications in Remote Sensing and GIS, Proceedings of International Workshop, D. Marceau, Ed., Université de Montréal, 161–168.
——, D. W. McKenney, I. A. Nalder, M. F. Hutchinson, and J. L. Kestevan, 2000: A comparison of two statistical methods for spatial interpolation of Canadian monthly mean climate data. Agric. For. Meteor.,101, 81–94.
Ripley, E. A., and B. Saugier, 1978: Biophysics of a natural grassland:Evaporation. J. Appl. Ecol.,15, 459–479.
Savenije, H. H. G., 1995: New definitions for moisture recycling and the relationship with land-use changes in the Sahel. J. Hydrol.,167, 57–78.
Schwartz, M. D., 1992: Phenology and springtime surface-layer change. Mon. Wea. Rev.,120, 2570–2578.
——, 1996: Examining the spring discontinuity in daily temperature ranges. J. Climate,9, 803–808.
Sellers, P. J., and Coauthors, 1997: BOREAS in 1997: Experiment overview, scientific results, and future directions. J. Geophys. Res.,102 (D24), 28 731–28 769.
Shukla, J., and Y. Mintz, 1982: Influence of land-surface evapotranspiration on the earth’s climate. Science,215, 1498–1501.
——, C. Nobre, and P. Sellers, 1990: Amazon deforestation and climate change. Science,247, 1322–1325.
Thompson, K., 1980: Forests and climate change in America: Some early views. Climatic Change,3, 47–64.
Trenberth, K. E., 1998: Atmospheric moisture residence times and cycling: Implications for rainfall rates and climate change. Climatic Change,39, 667–694.
Witter, J. A., and L. A. Waisenen, 1978: The effect of flushing times among aspen clones on tortricid caterpillar populations. Environ. Entomol.,7, 139–143.
Verma, S. B., J. Kim, and R. J. Clement, 1992: Momentum, water vapor, and carbon dioxide exchange at a centrally located prairie site during FIFE. J. Geophys. Res.,97 (D17), 18 629–18 639.
Verseghy, D. L., 1996: Local climates simulated by two generations of Canadian GCM land surface schemes. Atmos.–Ocean,34, 435–456.
Xue, Y., 1996: The impact of desertification in the Mongolian and the inner Mongolian grassland on the regional climate. J. Climate,9, 2173–2189.
——, M. J. Fennessy, and P. J. Sellers, 1996: Impact of vegetation properties on U.S. weather prediction. J. Geophys. Res.,101 (D3), 7419–7430.
Map showing distribution of major vegetation zones (adapted from Ecoregions Working Group 1989) and selected climate stations in western and central Canada (codes defined in Table 1). Major aspen areas are generalized based on Canada’s forest inventory (e.g., Lowe et al. 1994)
Citation: Journal of Climate 13, 24; 10.1175/1520-0442(2000)013<4229:PFODFP>2.0.CO;2
Mean monthly temperatures at Prince Albert, Saskatchewan, for the period 1961–90 (indicated as symbols; data from Environment Canada). Observed mean temperatures in Apr and Oct are 2.6° and 3.1°C, respectively, warmer than those obtained from the best-fitting sinusoidal equation (indicated as solid line). Equation (1) parameters are given in Table 1
Citation: Journal of Climate 13, 24; 10.1175/1520-0442(2000)013<4229:PFODFP>2.0.CO;2
Distribution of climate stations used in the spatial analysis of mean monthly temperature and precipitation (1961–90). Additional climate stations reporting precipitation only, and also used in the spatial analysis, are not shown
Citation: Journal of Climate 13, 24; 10.1175/1520-0442(2000)013<4229:PFODFP>2.0.CO;2
Deviations (ΔT) of monthly mean temperature from best-fitting sinusoidal equations for selected Canadian climate stations located (a) near major areas of aspen forest cover in the Canadian prairie provinces showing a common seasonal pattern of monthly temperature anomalies (IT = 2.1° to 2.9°C); (b) other areas showing a similar seasonal pattern (IT = 1.5° to 2.8°C); and (c) areas lacking this seasonal pattern (IT = −0.5° to 1.0°C). Equation (1) parameters for each station are given in Table 1
Citation: Journal of Climate 13, 24; 10.1175/1520-0442(2000)013<4229:PFODFP>2.0.CO;2
Map showing spatial variation in the average Apr and Oct temperature anomaly (IT) [see Eq. (3) in text] in Canada and adjacent areas of the United States based on the interpolation of values from climate stations for the period 1961–90
Citation: Journal of Climate 13, 24; 10.1175/1520-0442(2000)013<4229:PFODFP>2.0.CO;2
(a) Seasonal change in observed monthly solar radiation for aspen-dominated areas of the western Canadian interior based on average values of four climate stations (symbols) from Environment Canada (1982). The climate stations used correspond to the codes TP, WI, ED, and BV in Fig. 1. Best-fitting sinusoidal equation is also shown (solid line). (b) Theoretical values of seasonal change in potential (extraterrestrial) global solar radiation, as would be measured incident to a plane parallel to the earth’s surface at three latitudes (thick solid lines). Best-fitting sinusoidal functions are also shown (thin solid lines; note that at 45°N, the two lines coincide)
Citation: Journal of Climate 13, 24; 10.1175/1520-0442(2000)013<4229:PFODFP>2.0.CO;2
Seasonal change in mean monthly precipitation for selected climate stations representing major areas of aspen cover in the Canadian prairie provinces (1961–90 normals from Environment Canada)
Citation: Journal of Climate 13, 24; 10.1175/1520-0442(2000)013<4229:PFODFP>2.0.CO;2
Map showing spatial variation in the index of summer-enhanced precipitation (IP) [see Eq. (4) in text] in Canada and adjacent areas of the United States based on the interpolation of climate station data for the period 1961–90
Citation: Journal of Climate 13, 24; 10.1175/1520-0442(2000)013<4229:PFODFP>2.0.CO;2
Seasonal change in daily averages of (a) sensible and latent heat fluxes at the 39-m height above an aspen forest and (b) upward water movement (sap flow) in four aspen trees. Both sets of measurements were made at the BOREAS Old Aspen site in 1996
Citation: Journal of Climate 13, 24; 10.1175/1520-0442(2000)013<4229:PFODFP>2.0.CO;2
Average seasonal variation in characteristics of climate for nine climate stations (see Table 1) representing major areas of aspen cover in the Canadian prairie provinces: (top) mean observed daily temperature (thick line) in relation to best-fitting sinusoidal equation; (middle) mean daily difference (ΔT) between observed temperature and the sinusoidal equation; and (bottom) mean observed daily precipitation. The typical seasonal periods with snow cover (Fisheries and Environment Canada 1978) and presence of green aspen leaves (Hogg personal observations) in the region are also shown
Citation: Journal of Climate 13, 24; 10.1175/1520-0442(2000)013<4229:PFODFP>2.0.CO;2
List of climate stations shown in Fig. 1, with coefficients (a, b, and c) for best-fitting sinusoidal model [Eq. (1)] and magnitude of mean April/October temperature anomaly (IT) for the period 1961–90
