1. Introduction
Mountains play a key role in the global hydrological cycle and are a main source of water for many of the world’s river systems (Beniston et al. 1997). It is expected that climatic change may have a significant impact on mountain snowpack and, subsequently, the snow-derived water supply (Barnett et al. 2005). Water supply on the western prairies of Canada is highly dependent on snowmelt from the east slopes of the Rocky Mountains (Schindler and Donahue 2006). The potential impacts of climate change present a significant concern for the semiarid prairies, as the annual water supply is often exceeded by domestic, industrial, agricultural, and ecosystem water demand.
Snowpack provides temporary storage of water, released at important times of the year (Hamlet et al. 2005). Mountain snow accumulation is expected to decline with continued atmospheric warming (Hamlet and Lettenmaier 1999), resulting in a reduction of available water from snowpack in mountainous regions (Barnett et al. 2005; Lapp et al. 2005). Numerous studies have already shown hydrological changes in snow-dominated regions, with the earlier onset of melt (Cayan et al. 2001; Mote et al. 2005; Stewart et al. 2004, 2005; Stewart 2009) and decreases in mean annual streamflow (Zhang et al. 2001; Rood et al. 2005).
This study investigates the potential effects of climate change on mountain snowpack in the St. Mary River watershed, Montana. The International Joint Commission, a cooperative body overseeing the management of international waterways between Canada and the United States, received 115 submissions during public consultation between 2003 and 2005 from parties on both sides of the border regarding current water allocations from the St. Mary River system (IJC 2009). Many of these submissions expressed concern about the need for changes to water allocation, while other submissions warned about climate change and declining water supplies. The water supply in the St. Mary River is already fully allocated for human water use. The St. Mary system will likely come under substantial stress if the water supply is decreased in the future. There is a critical need to establish the potential impacts of climate change on this watershed as water demand stress grows and parties in each country discuss current and future allocation policies.
To assess the potential changes in mountain snowpack in the St. Mary River watershed, the Generate Earth Systems Science (GENESYS) input finescale hydrometeorological model is applied. To enable predictions of future snowpack conditions, temperature and precipitation outputs from general circulation models (GCMs) are used. This combination of GCMs and finescale hydrometeorological modeling provides a structure for investigating the relationships between climate and water resources (Leavesley 1994; Xu 1999).
GCMs provide the most sophisticated, physically based approach to simulating large-scale responses of the climatic system to projected changes in greenhouse gas (GHG) emissions (Laprise et al. 2003). GCMs operate at an hourly to subhourly time step over several centuries, and have a spatial scale on the order of thousands of square kilometers. GCMs simulate annual and seasonal climate patterns over large regions; however, due to their coarse spatial resolution, they lack the ability to model local climatic conditions, particularly in regions of diverse terrain (Xu 1999; Hay et al. 2000). Appropriate representation of local climatic conditions is needed to make suitable assessments of the impacts of climate change on freshwater ecosystems (Hauer et al. 1997).
The GENESYS model was chosen to represent finescale hydrometeorological processes in this analysis due to the fact that it has been shown to provide realistic estimates of hydrometeorological conditions in complex mountain terrain. GENESYS has been successfully used to simulate the daily snow water equivalent (SWE) for the St. Mary (MacDonald et al. 2009) and Oldman River watersheds (Lapp et al. 2002, 2005).
A number of GCM scenarios are used to perturb the GENESYS model and test the sensitivity of winter snow hydrology to a range of possible future climates. The objective of this work is to assess potential changes in snowpack timing, volume, and spatial coverage in the St. Mary River watershed for 30-yr periods centered around the years 2025, 2055, and 2085 relative to a historical (1961–90) period. To determine the available water from snow, the estimated total annual snow accumulation (mm of SWE) is used, as this is a good predictor of potential spring runoff (Barnett et al. 2005). Changes in rain-to-snow ratios are also assessed as the rain-to-snow ratio is an important measure of the impacts from climate warming on snow accumulation (Stewart 2009; Knowles et al. 2006). The temporal change in the future snowpack is assessed using the day of year where the maximum snowpack occurs over the watershed, after this date snow ablation begins; therefore, it is assumed that this is a surrogate for the onset of spring melt. To evaluate the impacts from climate change on the spatial coverage of snowpack, changes in the date of complete snow removal and the long-term mean annual snow cover are analyzed. This study also provides insights into the spatial and temporal responses of mountain hydrometeorological conditions to changes in climate.
2. Study area
The headwaters of the St. Mary River watershed lie on the eastern slopes of the Rocky Mountains with the majority of the upper watershed residing within Glacier National Park, Montana (Fig. 1). The Many Glacier snow telemetry (SNOTEL) site (NRCS 2007), the Preston snow course (D. Fagre 2006, unpublished manuscript), and the St. Mary climate station (NOAA/NCDC 2006) are shown. The St Mary River flows from the continental divide, through the upper and lower St. Mary lakes, and ends in southern Alberta, Canada, where it meets the Oldman River. The watershed is 1100 km2, topographically diverse, and ranges in elevation from 1249 to 3031 m. The region is snowfall dominated, with over 70% of the annual precipitation received as snow at higher elevations (Selkowitz et al. 2002). The St. Mary River watershed supplies water for approximately 200 000 ha of irrigation for southern Alberta. The St. Mary River is an important international watershed, part of the St. Mary and Milk River water supply management arrangement between Canada and the United States.
3. Methods
a. Hydrometeorological model
Daily temperature and precipitation data from the St. Mary climate station are used to drive the GENESYS model. To estimate spatial precipitation values, GENESYS applies a monthly spatial correction determined using Parameter Elevation Regression on Independent Slopes Model (PRISM) monthly precipitation values (Daly et al. 2008); this method is described in MacDonald et al. (2009). The form of precipitation is determined using the rain and snow partitioning algorithm derived by Kienzle (2008). Snow interception is calculated using a formula derived by Hedstrom and Pomeroy (1998) that uses leaf area index (LAI) from the MODIS LAI dataset (LP DAAC 2008) and air temperature derived by applying lapse rates described by Pigeon and Jiskoot (2008). Rain interception is calculated empirically using the Von Hoyningen-Huene (1983) formula. Sublimation calculations are made using equations developed by Déry et al. (1998). Runoff is determined as snowmelt and rainfall, occurring after complete saturation of the soil. Snowmelt is determined using a temperature index melt routine developed by Quick and Pipes (1977). The melt factor used in the snowmelt routine was calibrated to 1.0 cm °C−1; this factor was assumed to stay constant over the simulation period. Infiltration to the soil is determined as a proportion of the daily snowmelt and rainfall, until saturation is reached. Although runoff is calculated, GENESYS does not have runoff routing capabilities, largely due to the complexities involved with defining the subsurface characteristics of mountain watersheds, as demonstrated in a similar region by Magruder et al. (2009). Therefore, runoff output from GENESYS was not used in this analysis.
To simulate hydrometeorological variables spatially over the watershed, the GENESYS model uses terrain categories (TCs), which are regions of similar hydrological response derived using a GIS. TCs for the St. Mary River watershed were derived using a combination of 100-m-elevation bands and seven land cover types (closed coniferous forest, open coniferous forest, deciduous forest, dry herbaceous, mesic herbaceous, water, and barren rock or soil), resulting in 82 individual categories ranging in area from 100 m2 to 88 km2. For each TC the daily winter hydrological balance was calculated. For a detailed description of the GENESYS model and verification for the St. Mary River watershed, refer to MacDonald et al. (2009).
MacDonald et al. (2009) applied the GENESYS model to simulate daily SWEs for the 1961–2008 time period over the St. Mary River watershed. SWE simulations were compared with daily SWE values from the Many Glacier SNOTEL site (NRCS 2007). The model simulated daily SWEs from 1 October 1976 to 26 April 2001 at the Many Glacier SNOTEL site with an r2 = 0.73 (RMSE = 73 mm).
Spatial SWE simulations were assessed by comparing simulated snow-covered areas with the Moderate Resolution Imaging Spectroradiometer (MODIS)/Aqua snow cover 8-day global 500-m grid, version 5 (Hall et al. 2007). Eight dates representing the entire 2000/01 snow year were compared. MODIS snow cover extent was reproduced with an average Hanssen–Kuipers skill score (KSS) of 0.66 for all dates simulated (MacDonald et al. 2009). Although limited in the number of years of verification due to data availability, these results suggest the GENESYS can be applied to represent SWE over the St. Mary River watershed and, therefore, can be used to make predictions of the potential impacts of climate change on SWE.
b. GCM data
The monthly GCM data used were made available by the Pacific Climate Impacts Consortium (PCIC 2008). Following the recommendations of Solomon et al. (2007), a number of scenarios were used to capture a range of possible future climates. To select the range of scenarios, a method adapted from Barrow and Yu (2005) was used. This method enables a guided analysis of the sensitivity of SWE to future changes in temperature and precipitation. Scenarios were selected using annual mean temperature change and percent precipitation change for the 30-yr period centered around the year 2025. This period was selected because temperature and precipitation estimates from GCMs for this period are likely more certain than for projections farther into the future. Due to the fact that the variability in temperature change projected by all GCMs was low relative to the variability in precipitation (Fig. 2), the selection of each scenario was more reliant on the predicted precipitation change.
Following von Storch et al. (1993), estimates of monthly temperature and precipitation were derived as the average of the four grid cells closest to the study site. Five future climate scenarios, representing a range of plausible future temperature and precipitation conditions, were chosen (Table 1). The future climate scenarios in Table 1 represent a GHG emissions scenario and GCM combination.
c. Downscaling
The “delta” method of downscaling was used to couple the GENESYS model to GCM output. This method applies monthly changes derived from GCM output to the observed climate data (Hay et al. 2000; Wood et al. 1997; Xu 1999). The delta method has been used to downscale GCM output in numerous hydrological impacts studies (Hamlet and Lettenmaier 1999; Morrison et al. 2002; Andreasson et al. 2004; Loukas et al. 2004; Cohen et al. 2006; Merritt et al. 2006). The limitation to this method is that changes in the variability of climatic conditions are not accounted for (Leavesley 1994). However, the local variability of the driving climate station is preserved (Hamlet and Lettenmaier 1999). Given the uncertainty in future variability and range of plausible future climates, the delta method provides a conservative estimate of the impacts of climate change on water resources (Merritt et al. 2006).
Changes in GCM-derived temperature and precipitation for 30-yr periods centered around 2025, 2055, and 2085 relative to the 1961–90 time series were made. To determine daily minimum and maximum temperature changes, a Fourier transform was applied, similar to methods used by Epstein and Ramirez (1994) and Morrison et al. (2002). The Fourier transform was applied to the maximum and minimum temperatures for each month and each GCM scenario, creating 365 continuous predicted maximum and minimum temperature changes. These values were then added to observed temperature values for every day. Percent changes in precipitation amounts were determined relative to the 1961–90 period, where the mean monthly percent change was multiplied by the observed values on days when precipitation occurred. New 30-yr datasets representing changes in temperature and precipitation predicted by the GCMs were used to perturb the hydrometeorological model for 15 simulations of future climate impacts on mountain snowpack.
d. Data analysis
To estimate trends in maximum snowpack (mm of SWE) and the date of spring snowmelt (day of year), the Mann–Kendall nonparametric test was used. This test is frequently used for trend detection of hydrological variables (Burn and Hag Elnur 2002). Sen’s nonparametric slope test was used to determine the slope of the trend line (change in the variable per year). To assess spatial change in mean annual snowpack (mm of SWE), mean annual values of SWE averaged over 30-yr periods centered around the years 2025, 2055, and 2085 were calculated for 500 m × 500 m grid cells. Thirty-year surfaces were compared visually and the percent change in mean annual snowpack is calculated relative to the 1961–90 base period. The spatial change in snowmelt timing (day of year) was also assessed using the 30-yr average day of complete snowmelt.
4. Results and discussion
a. Climate change scenarios
All five scenarios show increases in mean annual temperature; however, predictions of changes in mean annual precipitation are more variable. The magnitude of the predicted temperature change for the period centered around 2025 ranged from 1.0° to 1.7°C relative to the 1961–90 period. The predicted relative precipitation change for the period centered around 2025 ranged from a 4.0% decrease to a 5.0% increase (Table 2). The variability in temperature and precipitation predictions is greatest for the period centered around 2085, demonstrating a greater influence of climate warming and greater uncertainty as GCM predictions extend farther into the future.
b. Rain-to-snow ratio
Predicted changes in the timing and magnitude of annual maximum snowpack resulting from climate change are affected by the projected increases in temperature and the resulting increases in the proportion of precipitation that falls as rain rather than snow. The GENESYS model is sensitive to temperature increases due to the fact that the snow–rain partitioning algorithm developed by Kienzle (2008) uses the mean daily mean temperature estimates for each TC to determine whether precipitation falls as rain or snow. For the simulated historical period, 70% of the total annual precipitation was snow. The GENESYS simulations show that as temperatures increase, the proportion of snow to total annual precipitation decreases from the historical period in every scenario (Table 3). Merritt et al. (2006) found similar results in the Okanagan River basin, British Columbia, where they predict that a greater proportion of rain during transitional months will lead to an increase in the rate of snowpack recession under the climate change projections. There is very little difference between all scenarios for the period centered around 2025; however, variability increases between all scenarios by the period centered around 2085 (Table 3).
c. Spatial change in mean annual SWE
Three scenarios were selected to represent the spatial change in 30-yr mean annual SWE (mm of SWE). The Canadian Centre for Climate Modelling and Analysis (CCCma) Coupled General Circulation Model’s (CGCM) A2(3) future climate scenario represents the greatest change, the National Center for Atmospheric Research’s (NCAR) B2(1) future climate scenario represents the least change, and the Commonwealth Scientific and Industrial Research Organisation’s (CSIRO) A1(1) future climate scenario represents a median change in 30-yr mean annual SWE (Fig. 3).
Figure 3 illustrates spatial changes in SWE over the watershed for three time periods. Figure 3 uses a shading palette where greater snow depths are a lighter shade of gray. Hence, a decline in spatial SWE is represented by increased darkening of the watershed. The CSIRO-A1(1) and CGCM-A2(3) future climate scenarios both result in an overall decrease in mean annual SWE over the watershed with an increase in the area with a mean annual snowpack of 0–100 mm and a decrease in the area covered by greater than 300 mm. Under these scenarios, the GENESYS model predicts that increases in temperature will reduce the extent and depth of SWE over the St. Mary River watershed. The NCAR-B2(1) future climate scenario results in very little spatial change in the mean annual snowpack even by the period centered around 2085. This is due to the fact that this scenario has the lowest increase in temperature and greatest increase in precipitation relative to the other scenarios (Table 2). Based on these results, the St. Mary River watershed under the NCAR-B2(1) future climate scenario would likely experience little decline in mean annual SWE in the future.
d. Spatial change in snowmelt date
Figure 4 demonstrates the spatial change in the 30-yr average snowmelt date for three representative scenarios over the entire watershed. Figure 4 uses a shading palette where an earlier melt date is represented by a darker gray shade. Hence, the later dates of melt are represented by increased whitening of the watershed. Simulations for all three scenarios result in earlier dates of complete snowmelt, with a clear elevation gradient. The CGCM-A2(3) and CSIRO-A1(1) future climate scenarios show substantial advancement in the date of complete melt, with substantial increases in the area melting between 27 March and 4 May by the period centered around 2085. The NCAR-B2(1) future climate scenario shows less change in the date of complete melt with the most significant change at midelevations where historically complete melt occurred between 19 July and 30 September; future predictions are for these regions to completely melt between 5 and 29 May.
The GENESYS model results show the decrease in mean annual SWE at lower elevations is in large part a function of temperature change, similar to results found by Mote (2003). As the climate warms, low-elevation portions of the watershed with historical winter average daily air temperatures near the freezing point will warm above freezing, changing both the phase of the precipitation and the rate of snowmelt (Regonda et al. 2005; Nolin and Daly 2006).
e. Depth of maximum SWE
Figure 5 presents the maximum SWE time series for the watershed for all five scenarios. The Mann–Kendall test detects decreasing trends over time in maximum annual SWE for the Center for Climate Systems Research’s (CCSR) A1T and the CGCM-A2(3) future climate scenarios at the 99% confidence level (Sen’s slope estimates = −4.10 and −5.81, respectively), and at the 95% confidence level in the CSIRO-A1(1) future climate scenario (Sen’s slope = −2.91). No significant trends are detected in the NCAR-B2(1) and ECHAM-A2(1) future climate scenarios (Sen’s slope estimates = −0.10 and −1.45, respectively). The difference between the scenarios shows the sensitivity of the system to the range of possible future conditions and demonstrates the importance of selecting a range of plausible future climate scenarios.
The lack of trend in the NCAR-B2(1) future climate scenario is likely due to the relatively large increases in precipitation and small increases in temperature (Table 2). Future annual maximum SWE actually increases under the NCAR-B2(1) future climate scenario. The lack of trend in the ECHAM-A2(1) future climate scenario again is likely from the relatively small increase in temperature when compared with the CCSR-A1T, CGCM-A2(3), and CSIRO-A1(1) scenarios.
In the CCSR-A1T, CGCM-A2(3), and CSIRO-A1(1) future climate scenarios, modest changes in precipitation and increases in temperature result in significant reductions overall in maximum SWE. These results are consistent with observed historical trends in SWE (Hamlet et al. 2005; Mote et al. 2005; Mote 2006). Leung and Wigmosta (1999) also found under a 2 × CO2 future climate scenario that a similar watershed on the west slopes of the Rockies could experience an 18% reduction in SWE; they too show that snowpack is highly susceptible to changes in temperature. This study suggests that higher temperatures will likely result in a decrease in maximum annual SWE unless higher temperatures are accompanied by large increases in precipitation.
Changes in the variability of maximum SWE are also observed, where although only changes in mean monthly climate were applied, the variability in maximum SWE increased over time in both scenarios. The simulated annual maximum SWE had a 30-yr mean of 708 mm with a standard deviation of 254 mm for the 1961–90 historical period. For the period centered around 2085, the means of the maximum annual SWE decreased to 465, 444, 648, and 692 mm for the CCSR-A1T, CGCM-A2(3), CSIRO-A1(1), and ECHAM-A2(1) future climate scenarios, respectively, and increased to 789 mm for the NCAR-B2(1) scenario. Under all scenarios, standard deviations increased. This increase in variability of maximum annual SWE in the future suggests that even if historical climatic variability is maintained in future climates, increases in temperature could enhance the extreme SWE years.
f. Timing of maximum SWE
An earlier onset of snowmelt has already been recorded in numerous studies over western North America (Burn 1994; Cayan et al. 2001; Mote et al. 2005; Regonda et al. 2005; Stewart et al. 2005). All five scenarios are used to show changes in the date of maximum SWE over the St. Mary River watershed (Fig. 6).
The Mann–Kendall test identifies significant decreasing trends in the simulated date of maximum SWE in the CCSR-A1T, CGCM-A2(3), CSIRO-A1(1), and ECHAM-A2(1) future climate scenarios at the 99% confidence level (Sen’s slope estimates = −0.61, −0.66, −0.41, and −0.23, respectively). A significant decreasing trend is also shown in the NCAR-B2(1) future climate scenario at the 90% confidence level (Sen’s slope estimate = −0.09). Under the CCSR-A1T, CGCM-A2(3), CSIRO-A1(1), and ECHAM-A2(1) future climate scenarios, the dates of maximum snowpack could approach early January by the period centered around 2085, while under the NCAR-B2(1) future climate scenario the dates of maximum snowpack could occur in February.
The mean date of maximum SWE over the watershed is 8 April for the historical 1961–90 period. All scenarios are consistent in predicting an earlier mean date of maximum SWE over the period centered around 2085. This shift in means along with significant trends toward earlier dates of maximum SWE infers an overall earlier onset of spring in the St. Mary River watershed.
An earlier date of maximum SWE in all scenarios is, again, likely a function of increased temperature. Studies of trends in the onset of spring support this, showing the important role of increased temperatures on spring snowpack (Hamlet et al. 2005; Mote 2006). The influence of temperature can be seen by looking at the NCAR-B2(1) future climate scenario, where there is no significant trend in maximum SWE depth; however, there is a decreasing trend for the date of maximum SWE. This result is consistent with Clair et al. (1998) and shows the sensitivity of snowmelt to temperature and that an annual increase of 14% in precipitation by the period centered around 2085 can be offset by a 2°C increase in temperature.
g. Uncertainty in future snowpack predictions
There are a number of sources of uncertainty in projecting future snow conditions. The input data present a significant source of uncertainty as temperature and precipitation observations from the St. Mary climate station are likely not representative of the temperature and precipitation over the entire study area. However, this is the only long-term climate record in the region and, therefore, is the best available dataset. The distribution of temperature and precipitation to TCs also presents a level of uncertainty. For example, using Precipitation-elevation Regressions on Independent Slopes Model (PRISM) monthly precipitation estimates to distribute daily values of temperature and precipitation to each TC does not account for the temporal variability in these climate variables at the daily time step. MacDonald et al. (2009), through detailed daily SWE simulations did, however, show that this method was able to capture the seasonal differences in precipitation between the mountainous portions of the watershed and the low-elevation St. Mary station.
Other variables used as input into the GENESYS model, including LAI, temperature lapse rates, and spatial precipitation adjustments, are assumed to be constant at the monthly time step. The melt factor was assumed to remain constant for all simulations. It is acknowledged that the assumption that these variables remain constant under future conditions presents a source of uncertainty. However, at the spatial scale of this study and without adequate hydrometeorological data in mountain environments, these assumptions seem reasonable. It is also important to recognize the need for adaptive management, and without information of potential future changes in SWE, management strategies would be difficult to adopt.
Perhaps the greatest source of uncertainty is in the scenario projections of future temperature and precipitation. An objective of this study was to capture the range of variability in GCM and emission scenario projections available from the PCIC (2008). This study has demonstrated that, depending on the scenario, SWE could respond very differently in the future. All climate change scenarios used in this study are consistent with projections of an earlier onset of spring melt. However, there are differences between scenarios, especially in mean annual and maximum SWE (Figs. 3 and 5), demonstrating the effects of scenario selection on future predictions.
5. Summary
Given the 1°–4°C increase in mean temperature already observed over the last century in North America (Schindler and Donahue 2006), it is likely that warming may be more than 2°C by the period centered around 2085. It is, therefore, reasonable to assume that the lack of change in snowpack resulting from the NCAR-B2(1) model using the B2(1) emissions scenario is highly unlikely even with substantial emission controls. Similar caution is important to consider in interpreting all scenarios.
Emissions scenarios provide insights into the types of adaptation that might be required to mitigate the effects of climate change. The A1 and A1T emissions scenarios used in the CSIRO- Mark version 2 (CSIRO-Mk2) and CCSR/National Institute for Environmental Studies (NIES) climate models, respectively, represent a future with very rapid economic growth, a growing global population that peaks in midcentury and subsequently declines, and the rapid introduction of new and more efficient technologies. The A2 emissions scenario used in the CGCM2 and ECHAM4 climate models represents a world with a continually increasing global population and regionally oriented economic growth that is fragmented and slower relative to other emissions scenarios (Nakicenovic et al. 2000). GENESYS simulations using the CSIRO-A1(1), CCSR-A1T, CGCM-A2(3), and ECHAM-A2(1) future climate scenarios show that even under substantial adaptation, the mean and maximum annual SWE over the St. Mary River watershed may decline (Figs. 3 and 4).
The B2 emissions scenario used in NCAR’s Parallel Climate Model (PCM) represents a world where the emphasis is on local solutions to economic, social, and environmental sustainability, with intermediate levels of economic development. It also assumes increases in population, however, at a lower population growth rate than in the A2 scenario (Nakicenovic et al. 2000). Model results from the GENESYS simulations using the B2(1) emissions scenario in the PCM show that significant changes to current social and environmental policies may mitigate the effects of climate change on snowpack in the St. Mary River watershed.
The St. Mary River watershed in northern Montana is a critical international watershed where decreasing snowpack poses a significant threat under a range of future climate scenarios.
The GENESYS hydrometeorological model was applied to predict the impacts of climate change on snowpack in the St. Mary River watershed. Five scenarios of future climate were derived, representing a range of plausible future conditions. Results indicate that snowpack in the St. Mary River watershed is highly vulnerable to changes in temperature and, to a lesser extent, changes in precipitation. This can be demonstrated by comparing the CSIRO-A1(1) and NCAR-B2(1) future climate scenarios. The 10% and 14% increases in precipitation and 4.8° and 2.0°C temperature changes represented by the two future climate scenarios result in, respectively, very different outcomes for SWE. GENESYS simulations using the CSIRO-A1(1) future climate scenario represent a change in the 30-yr average maximum SWE from 708 mm historically to 648 mm by the period centered around 2085, while simulations using the NCAR-B2(1) future climate scenario result in a change to 789 mm. These differences in SWE demonstrate that even with high precipitation increases, temperature plays a key role in determining future SWE conditions. Key outcomes of this study include the following:
Future predictions are highly dependent on GCM and emissions scenario selection. This is due to both variability between GCM representations of climate and the global GHG emission scenarios used in the projections. Given these two sources of uncertainty, it is difficult to quantify which future scenario is most probable. However, the selected range of scenarios covers the range of probable outcomes as projected by GCMs and GHG scenarios recommended by Solomon et al. (2007). Hence, the results of this study reflect the likely range of climate change impacts on snowpack in the St. Mary River watershed.
For each future climate scenario, the proportion of rain to snow increases relative to the 1961–90 period. This is important, as with increased rain-to-snow ratios, it is likely that the storage of water in the form of snow will decrease.
All future climate scenarios in this study project a reduction in the spatial extent of the mean annual SWE by the period centered around 2085. The GENESYS model using the CGCM-A2(3) future climate scenario predicts a 63% decline, the CSIRO-A1(1) future climate scenario estimate is 17%, and the NCAR-B2(1) future climate scenario indicates a 3% decline. The greatest reductions in mean annual SWE are predicted to occur at lower elevations, where critical temperature thresholds are likely to be exceeded in the future.
The CGCM-A2(3) and CSIRO-A1(1) future climate scenarios show a substantial spatial change in the date of complete snowmelt while the NCAR-B2(1) future climate scenario does not. This change to an earlier date of complete melt could have important implications for water resources as increased evaporation from open soils and drier early season conditions could occur.
Significant decreasing trends in maximum SWE over the watershed are predicted by the GENESYS model using the CCSR-A1T, CGCM-A2(3), and CSIRO-A1(1) future climate scenarios, while there are no significant trends in maximum SWE predicted for the NCAR-B2(1) and ECHAM-A2(1) future climate scenarios. The decreasing trend in maximum SWE over the watershed demonstrates the sensitivity of the snow accumulation to increased air temperature. The nonsignificant trend in maximum SWE is likely a function of the relatively modest increases in temperature projected by the NCAR-B2(1) and ECHAM-A2(1) future climate scenarios.
This study supports observed trends toward an earlier onset of spring. Significant decreasing trends in the date of maximum SWE are projected for all scenarios.
Results from this study suggest that limiting global GHG emissions, as presented in the NCAR-B2(1) future climate scenario, may reduce the impacts of climate change on maximum SWE in the St. Mary River watershed. However, changes in the timing of peak SWEs are predicted to occur, independent of the future climate scenario.
This study suggests that SWE is much more sensitive to changes in temperature than to changes in precipitation. This is demonstrated by the ECHAM-A2(1) future climate scenario, as this scenario has the greatest reduction in precipitation and a relatively low increase in temperature but not the greatest reduction in SWE.
The high sensitivity of snow processes to changes in climate poses important questions about water resources in the future. This modeling effort provides insights into future snow conditions in the St. Mary River watershed but leaves much to be resolved. Current monitoring efforts of changes in mountain ecosystems exist at spatial and temporal scales that are inadequate. To properly quantify and adapt to future changes in mountain hydrometeorology, increased monitoring of meteorological variables over time and space is required.
Acknowledgments
Funding support from the Alberta Ingenuity Centre for Water Research and the Natural Sciences and Engineering Research Council of Canada is much appreciated. The advice and direction of the reviewers is also appreciated, as their suggestions greatly improved this manuscript. One reviewer in particular provided detailed comments that played a key role in this manuscript’s final version.
REFERENCES
Andreasson, J., Lindstrom G. , Grahn G. , and Johansson B. , 2004: Runoff in Sweden—Mapping of climate change impacts on hydrology. Nord. Hydrol. Prog., 48 , 625–632.
Barnett, T. P., Adam J. C. , and Lettenmaier D. P. , 2005: Potential impacts of a warming climate on water availability in snow-dominated regions. Nature, 438 , 1–7.
Barrow, E., and Yu G. , 2005: Climate change scenarios for Alberta. Prairie Adaptation Research Collaborative Rep. prepared in cooperation with Alberta Environment, Regina, SK, Canada, 73 pp.
Beniston, M., Diaz H. F. , and Bradley R. S. , 1997: Climatic change at high elevation sites: An overview. Climatic Change, 36 , 233–251.
Burn, D. H., 1994: Hydrologic effects of climate change in west-central Canada. J. Hydrol., 160 , 53–70.
Burn, D. H., and Hag Elnur M. A. , 2002: Detection of hydrologic trends and variability. J. Hydrol., 255 , 107–122.
Cayan, D. R., Kammerdiener S. A. , Dettinger M. D. , Caprio J. M. , and Peterson D. H. , 2001: Changes in the onset of spring in the western United States. Bull. Amer. Meteor. Soc., 82 , 399–415.
Clair, T. A., Ehrman J. , and Higuchi K. , 1998: Changes to the runoff of Canadian ecozones under a doubled CO2 atmosphere. Can. J. Fish. Aquat. Sci., 55 , 2464–2477.
Cohen, S. J., and Coauthors, 2006: Learning with local help: Expanding the dialogue on climate change and water management in the Okanagan region, British Columbia, Canada. Climatic Change, 75 , 331–358.
Daly, C., Halbleib M. , Smith J. I. , Gibson W. P. , Doggett M. K. , Taylor G. H. , Curtis J. , and Pasteris P. A. , 2008: Physiographically-sensitive mapping of temperature and precipitation across the conterminous United States. Int. J. Climatol., 28 , 2031–2064.
Déry, S. J., Taylor P. , and Xiao J. , 1998: The thermodynamic effects of sublimating, blowing snow in the atmospheric boundary layer. Bound.-Layer Meteor., 89 , 251–283.
Epstein, D., and Ramirez J. A. , 1994: Spatial disaggregation for studies of climatic hydrologic sensitivity. J. Hydraul. Eng., 120 , 1449–1467.
Hall, D. K., Riggs G. A. , and Salomonson V. V. , cited. 2007: MODIS/Aqua snow cover 8-day L3 global 500 m grid, version 5: 16 Oct, 2000; 17 Nov, 2000; 17 Jan, 2001; 14 Mar, 2001; 15 Apr, 2001; 17 May, 2001, 12 Jul, 2001; 14 Sept, 2001. National Snow and Ice Data Center, Boulder, CO, digital media. [Available online at http://nsidc.org/data/myd10a2v5.html].
Hamlet, A. F., and Lettenmaier D. P. , 1999: Effects of climate change on hydrology and water resources in the Columbia basin. J. Amer. Water Resour. Assoc., 35 , 1597–1623.
Hamlet, A. F., Mote P. W. , Clark M. P. , and Lettenmaier D. P. , 2005: Effects of temperature and precipitation variability on snowpack trends in the western United States. J. Climate, 18 , 4545–4561.
Hauer, R. F., and Coauthors, 1997: Assessment of climate change and freshwater ecosystems of the Rocky Mountains, USA and Canada. Hydrol. Processes, 11 , 903–924.
Hay, L. E., Wilby R. L. , and Leavesley G. H. , 2000: A comparison of delta change and downscaled GCM scenarios for three mountainous basins in the United States. J. Amer. Water Resour. Assoc., 36 , 387–397.
Hedstrom, N. R., and Pomeroy J. W. , 1998: Measurements and modelling of snow interception in the boreal forest. Hydrol. Processes, 12 , 1611–1625.
IJC, cited. 2009: Public consultation documents. International Joint Commission. [Available online at http://www.ijc.org/rel/boards/smmr/public_consult-e.htm].
Kienzle, S. W., 2008: A new temperature based method to separate rain and snow. Hydrol. Processes, 22 , 5067–5085.
Knowles, N., Dettinger M. D. , and Cayan D. R. , 2006: Trends in snowfall versus rainfall in the western United States. J. Climate, 19 , 4545–4559.
Lapp, S., Byrne J. M. , Kienzle S. W. , and Townshend I. , 2002: Linking global circulation model synoptics and precipitation for western North America. Int. J. Climatol., 22 , 1807–1817.
Lapp, S., Byrne J. M. , Kienzle S. W. , and Townshend I. , 2005: Climate warming impacts on snowpack accumulation in an alpine watershed: A GIS based modeling approach. Int. J. Climatol., 25 , 521–526.
Laprise, R., Caya D. , Frigon A. , and Paquin D. , 2003: Current and perturbed climate as simulated by the second-generation Canadian Regional Climate Model (CRCM-II) over northwestern North America. Climate Dyn., 21 , 405–421.
Leavesley, G. H., 1994: Modelling the effects of climate change on water resources—A review. Climatic Change, 28 , 159–179.
Leung, L. R., and Wigmosta M. S. , 1999: Potential climate change impacts on mountain watersheds in the Pacific Northwest. J. Amer. Water Resour. Assoc., 35 , 1463–1471.
Loukas, A., Vasiliades L. , and Dalezios N. R. , 2004: Climate change implications on flood response of a mountainous watershed. J. Geophys. Res., 112 , D11124. doi:10.1029/2006JD007561.
LP DAAC, cited. 2008: Leaf area index—Fraction of photosynthetically active radiation 8-day L4 global 1 km. Land Processes Distributed Active Archive Center. [Available online at https://lpdaac.usgs.gov].
MacDonald, R. J., Byrne J. M. , and Kienzle S. W. , 2009: A physically based daily hydrometeorological model for complex mountain terrain. J. Hydrometeor., 10 , 1430–1446.
Magruder, I. A., Woessner W. W. , and Running S. W. , 2009: Ecohydrologic process modeling of mountain block groundwater recharge. Ground Water, 47 , 774–785.
Merritt, W. S., Alila Y. , Barton M. , Taylor B. , Cohen S. , and Neilsen D. , 2006: Hydrologic response to scenarios of climate change in sub watersheds of the Okanagan basin, British Columbia. J. Hydrol., 326 , 79–108.
Morrison, J., Quick M. C. , and Foreman M. G. G. , 2002: Climate change in the Fraser River watershed: Flow and temperature projections. J. Hydrol., 263 , 230–244.
Mote, P. W., 2003: Trends in snow water equivalent in the Pacific Northwest and their climatic causes. Geophys. Res. Lett., 30 , 1601. doi:10.1029/2003GL017258.
Mote, P. W., 2006: Climate-driven variability and trends in mountain snowpack in western North America. J. Climate, 19 , 6209–6220.
Mote, P. W., Hamlet A. F. , Clark M. P. , and Lettenmaier D. P. , 2005: Declining mountain snowpack in western North America. Bull. Amer. Meteor. Soc., 86 , 39–49.
Nakicenovic, N., and Coauthors, 2000: Special Report on Emissions Scenarios. Cambridge University Press, 599 pp.
NOAA/NCDC, cited. 2006: Web climate service. [Available online at www.ncdc.noaa.gov/oa/climate/stationlocator.html].
Nolin, A., and Daly C. , 2006: Mapping “at risk” snow in the Pacific Northwest. J. Hydrometeor., 7 , 1164–1171.
NRCS, cited. 2007: SNOTEL data and products. Natural Resources Conservation Service. [Available online at http://www.wcc.nrcs.usda.gov/snow/].
PCIC, cited. 2008: PCIC regional analysis tool. Pacific Climate Impacts Consortium. [Available online at http://pacificclimate.org/].
Pigeon, K., and Jiskoot H. , 2008: Meteorological controls on snowpack formation and dynamics in the southern Canadian Rocky Mountains. Arct. Antarct. Alp. Res., 40 , 716–730.
Quick, M. C., and Pipes A. , 1977: UBC watershed model. Hydrol. Sci. Bull., 306 , 215–233.
Regonda, S. K., Rajagopalan B. , Clark M. , and Pitlick J. , 2005: Seasonal cycle shifts in hydroclimatology over the western United States. J. Climate, 18 , 372–384.
Rood, S. B., Samuelson G. M. , Weber J. K. , and Wywrot K. A. , 2005: Twentieth-century decline in streamflows from the hydrographic apex of North America. J. Hydrol., 81 , 1281–1299.
Schindler, D. W., and Donahue W. F. , 2006: An impending water crisis in Canada’s western prairie provinces. Proc. Natl. Acad. Sci. USA, 103 , 7210–7216.
Selkowitz, D. J., Fagre D. B. , and Reardon B. A. , 2002: Interannual variations in snowpack in the Crown of the Continent ecosystem. Hydrol. Processes, 54 , 3651–3665.
Solomon, S., Qin D. , Manning M. , Marquis M. , Averyt K. , Tignor M. M. B. , Miller H. L. Jr., and Chen Z. , Eds. 2007: Climate Change 2007: The Physical Sciences Basis. Cambridge University Press, 996 pp.
Stewart, I. T., 2009: Changes in snowpack and snowmelt runoff for key mountain regions. Hydrol. Processes, 23 , 78–94.
Stewart, I. T., Cayan D. R. , and Dettinger M. D. , 2004: Changes in snowmelt runoff timing in western North America under a ‘business as usual’ climate change scenario. Climatic Change, 62 , 217–232.
Stewart, I. T., Cayan D. R. , and Dettinger M. D. , 2005: Changes toward earlier streamflow timing across western North America. J. Climate, 18 , 1136–1155.
von Hoyningen-Huene, J., 1983: Die Interzeption des Niederschlages in landwirtschaftlichen Pflanzenbeständen. Dtsch. Verb. Wasserwirtsch. Kulturbau, 57 , 1–66.
von Storch, H., Zorita E. , and Cubasch U. , 1993: Downscaling of global climate change estimates to regions scales: An application to Iberian rainfall in wintertime. J. Climate, 6 , 1161–1171.
Wood, A. W., Lettenmaier D. P. , and Palmer R. N. , 1997: Assessing climate change implications for water resources planning. Climatic Change, 37 , 203–228.
Xu, C., 1999: From GCMs to river flow: A review of downscaling methods and hydrologic modelling approaches. Prog. Phys. Geogr., 23 , 229–249.
Zhang, X., Harvey K. D. , Hogg W. D. , and Yuzyk T. R. , 2001: Trends in Canadian streamflow. Water Resour. Res., 37 , 987–998.
Changes in mean annual temperature and precipitation for each of the 5 scenarios relative to the 1961–90 base period.
Changes in snow as a percentage of the total precipitation for all five scenarios.