With its headwaters in the water towers of the western Cordillera of North America, the Fraser River is one of the continent’s mightiest rivers by annual flows, supplies vital freshwater resources to populous downstream locations and sustains the world’s largest stocks of sockeye salmon along with four other salmon species. Here we show the Variable Infiltration Capacity (VIC) model’s ability to reproduce accurately observed trends in daily streamflow for the Fraser River’s main stem and six of its major tributaries over 1949-2006 when air temperatures rose by 1.4 °C while annual precipitation amounts remained stable. Rapidly declining mountain snowpacks and earlier melt onsets result in a 10-day advance of the Fraser River’s spring freshet with subsequent reductions in summer flows when up-river salmon migrations occur. Identification of the sub-basins driving the Fraser River’s most significant changes provides a measure of seasonal predictability of future floods or droughts in a changing climate.
The Fraser River Basin (FRB) forms one of the largest and most important watersheds draining the western Cordillera of North America1. The FRB spans 240,000 km2 of diverse landscapes including 11 different biogeoclimatic zones, from dry interior plateaus to wet, snowy mountains in its headwaters of the northern Rockies, the Columbia and Coast Mountains (Fig. 1). Owing to its rich diversity and abundance of natural resources, the FRB offers a plethora of economic opportunities (e.g., in agriculture, fishing, forestry, mining and hydropower production), provides cultural and societal values and attracts many tourists and recreationists in its largely undisturbed, scenic landscapes1. It also provides extensive terrestrial and aquatic habitat for iconic large mammals and fish species such as grizzly and black bears, mountain caribou, wolves, wolverines, cougars, moose, white sturgeon and five species of salmon. Indeed, the Fraser River and its tributaries sustain the largest migrations of Pacific Ocean sockeye salmon2.
There is mounting evidence that climate change has influenced the FRB’s hydrologic regime during the 20th and early 21st centuries, affecting its economic vitality and pristine landscapes3,4,5. Mean annual air temperatures warmed by 1.4 °C between 1949 and 2006 across the FRB while total annual precipitation remained stable despite a significant change in its type from snowfall to rainfall6. These changes in the FRB’s climate have impacted the evolution and duration of its seasonal snowpack, resulting in a 19% decline in the contribution of snow to its hydrological regime6,7,8. This suggests the FRB is currently transitioning from a nival to a pluvio-nival hybrid system6,9,10, raising concerns about the impacts of its changing hydrology on terrestrial and aquatic species and their habitats11,12 as well as on flood and drought occurrences.
While trends in streamflow timing are commonly used as indicators of climate change13, identifying the driving factors for observed changes remain quite challenging14. Rising air temperatures coupled with changing precipitation amounts and types affect mountain snowpack accumulation and ablation processes that, in turn, lead to modifications in streamflow timing. Large systems, such as the FRB, integrate spatially and temporally complex physical processes that can confound climate change signals and their hydrological impacts. To resolve some of these issues, the application of a semi-distributed, macroscale hydrological model can provide insights on the various agents of change from the grid- to basin-scale. Thus this paper addresses three key research goals: 1) to evaluate the performance of a semi-distributed, macroscale hydrological model in simulating FRB observed streamflow trends at a daily timescale; 2) to quantify the relative contribution and its change over time of the FRB’s main sub-basins to total streamflow on the main-stem lower Fraser River; and 3) to assess the impacts of rising air temperatures, of changes in precipitation types and amounts and of rapidly declining mountain snowpacks on changes in streamflow timing and amounts across the FRB. To achieve these objectives, both observational and modelling datasets are applied to investigate changes to the hydrologic regime of the FRB from 1949 to 2006.
The six sub-basins examined in the present study contribute 75.0% (67.9%) of the annual observed (simulated) Fraser River discharge at Hope, British Columbia (BC) with the largest contributions from the Thompson-Nicola (TN), Upper Fraser (UF) and Quesnel (QU) sub-basins (Table 1). Based on observations, the relative contributions of the FRB’s western sub-basins (Stuart (SU), Nautley (NA) and Chilko (CH)) to the annual discharge of the Lower Fraser River at Hope (LF) declined whereas those of the eastern sub-basins have generally increased over the study period. The Variable Infiltration Capacity (VIC) model accurately simulates trends in daily streamflow observed across the FRB, with Nash-Sutcliffe Efficiency (NSE) scores between 0.47 and 0.97, statistically significant correlations and relatively low measures of error (Table 2). Note however the VIC model’s reduced performance over the Stuart, Nautley and Chilko river sub-basins, owing perhaps to deficiencies in the forcing data and the lack of the representation of large, deep lakes, glaciers and land cover changes in the current model’s formulation.
The observed (simulated) daily streamflow for the Fraser River at Hope during April and May exhibits increasing trends with a maximum of 1481 m3 (s 58 yrs)−1 (2003 m3 (s 58 yrs)−1) in response to an earlier onset of snowmelt (Fig. 2a). In contrast, the results also show observed (simulated) decreasing trends in streamflow with a minimum of −1780 m3 (s 58 yrs)−1 (−1495 m3 (s 58 yrs)−1) but over a prolonged period (June, July and August). Trends in daily streamflow throughout the remainder of the water year are relatively modest in magnitude and are not statistically significant (p > 0.05; see Supplementary Fig. 1). Integrating over time the observed and simulated daily streamflow trends reveals the insignificant changes in total annual discharge values, confirming the reported changes are driven by phase shifts as the FRB is transitioning from a nival to a hybrid/pluvial regime.
Grid-scale trends in simulated daily streamflow for 1 May (Fig. 2b) and 1 June (Fig. 2c) provide insights on the spatial distribution of these changes. On 1 May, most areas show positive trends with the more amplified values > 200 m3 (s 58 yrs)−1 found in the lower Coast Mountains, parts of the Upper Fraser and the windward side of the Canadian Rockies. While the 1 May positive streamflow trends are, in general, spatially coherent, the 1 June negative streamflow trends are isolated to the Coast Mountains and the Canadian Rockies, in the mountainous terrain of the Fraser River.
Given the FRB’s vast area, it is imperative to investigate the contributions of each of its major sub-basins to the observed and simulated trends in daily streamflow. The strong positive trends in daily streamflow in April and May followed by their declining trends in summer are driven mainly by changes in the UF and TN sub-basins (Fig. 3a,b). These two sub-basins have the largest mean annual flows (Table 1). The summations of the time series of daily trends in observed and simulated streamflow for the six major sub-basins match closely the values for the Fraser River at Hope, BC with r = 0.94 and 0.96 (both with p < 0.05), respectively (Fig. 3c,d). This clearly illustrates the upscaling of streamflow timing changes that occurs from individual sub-basins to the overall larger FRB and is a feature that the VIC model represents faithfully.
Using the slopes determined by the Mann-Kendall Test (MKT) method and median values of the annual daily streamflow as a y-intercept for the study period, the streamflow in a single water year is reconstructed for 1949 and 2006 using observed and simulated streamflow, respectively at Hope, BC (Fig. 4). In 2006, streamflow in April and May reconstructed using the MKT increased by at least 1500 m3 s−1 (based on both the observed and simulated data) owing to the earlier onset of the spring freshet. Lagging the 2006 reconstructed hydrographs by 10 days relative to the 1949 ones confirm the recent 10-day advances of the onset of the spring freshets for the Fraser River at Hope (see also Supplementary Table 1). In contrast, the duration of diminishing streamflow extends from June to August, suggesting declining summer flows across the FRB. The declines persist during the recession to lower flows in autumn just when the salmon are migrating up the Fraser River. Despite the phase shift of daily streamflow values, the total annual streamflow does not change significantly.
Climatic trends in air temperature, precipitation (including snowfall), snow water equivalent (SWE), runoff and evapotranspiration explain the streamflow trends for water years 1949–2006 (Supplementary Fig. 2). Air temperatures warmed mainly during winter (up to 5 °C in January) as snowfall declined by up to 35 mm month−1 with a seasonal deficit of 103 mm. In turn, SWE declined by 105 mm by the time of its peak accumulation. The combined effects of the warmer air temperatures, reduced snowfall and hence peak SWE induce earlier melt of the snow and a rapid advance of the spring freshet detected in observations and the VIC model simulations. Increased spring and summer rainfall, however, compensates part of the precipitation decline during winter, leading to little change in annually-integrated streamflow amounts. Evapotranspiration also increases in spring and summer, further offsetting concurrent increases in precipitation and thus streamflow. Advance of the spring freshet induced by rising air temperatures is confirmed by a series of VIC model simulations forced by detrended air temperatures (Dtd-T) and precipitation (Dtd-P) over the 1949–2006 time period (see Supplementary Fig. 2). Comparing the VIC simulations driven by the forcings with trends removed from both precipitation and air temperature (Dtd-PT), from precipitation only (Dtd-P) and from air temperature only (Dtd-T) reveal that the shifts in the timing of the spring freshet and the recession to lower flows are eliminated only when daily air temperatures are detrended in the VIC model simulations. Similar features are not seen when only precipitation alone is detrended, illustrating the control of air temperature trends on streamflow timing in the Fraser River Basin.
This study provides new insights on the causes of streamflow timing changes observed in the FRB during the late 20th and early 21st centuries. Although total annual precipitation in the FRB has remained nearly stable over water years 1949 to 2006, changes in its type and timing coupled with warmer air temperatures have led to a significant decline in peak seasonal snowpack accumulation and an advance in the onset of melt during spring. Further to this, evapotranspiration rates during spring and summer have increased significantly, reducing runoff during the warm season despite concurrent increases in rainfall (Supplementary Fig. 2). Partitioning these results across six major sub-basins of the FRB reveals the dominance of the UF and TN sub-basins leading to these changes. This suggests that the timing of the spring freshet and the return to low summer flows of the lower Fraser River in any given year can be forecasted reasonably well by tracking conditions solely within these two upstream sub-basins. This can then provide a measure of predictability for possible spring floods generated by high snowpack levels or summer droughts when low snowpack levels are recorded15.
A key finding arising from this work is the VIC model’s ability to reproduce observed trends in daily streamflow for the Fraser River and six of its major sub-basins. Previous studies typically have evaluated a hydrological model’s performance in simulating observed climatological hydrographs or annual to monthly long-term trends in streamflow16. This effort further demonstrates the VIC model, when driven by a robust long-term meteorological forcing dataset, achieves a high degree of accuracy (e.g., NSE scores ranging from 0.47–0.97 along with other metrics; see Table 2) in simulating FRB daily streamflow trends. This is important as assessing trends in monthly streamflow could conceal changes occurring on shorter time scales. Small alpine and Arctic watersheds where snowmelt occurs rapidly causing “flashy” hydrographs could be especially susceptible to such issues17. This work confirms the central role land surface and hydrological models play in assessing the direction of hydrological changes (both past and future) and their driving factors, which is especially critical for the many ungauged basins of remote regions such as northern Canada18.
The Fraser River and its many tributaries continue to host one of the largest Pacific Ocean salmon populations19. Due to the FRB’s transition to a hybrid regime, its potential impacts such as lower summer flows and warmer streamflow temperatures will lead to a degradation of salmon habitat along with additional physiological stress during up-river salmon migrations. This retrospective study using the VIC hydrological model (with a focus on streamflow phase shifts in response to rapidly diminishing snowpack levels) establishes an important element for both the historical evaluations and the future predictions of the regime changes in hydroclimate variability across the FRB with potential impacts to keystone species such as salmon. Although this study suggests there are ongoing changes to the hydrological regime of the FRB, future climate change may lead to further alterations. As such, additional efforts are needed to investigate future climate scenarios and the response of the FRB hydrological system to enhanced atmospheric forcing. This will lead to the development of climate adaptation strategies20 for communities situated in the basin, improved water resources management and better decision-making at all levels of governments.
VIC model application
To assess the FRB’s changing hydrological regime, the macroscale VIC model is employed to conduct retrospective hydrologic simulations from (water years) 1949 to 2006 6. The VIC model21,22 (along with more recent modifications) has been broadly applied to evaluate hydrological responses to a changing climate over a wide range of global river basins6,23,24,25. Driven by a forcing dataset of daily minimum and maximum air temperature, precipitation and wind speed26,27, the VIC model is integrated on a daily time step at 0.25° spatial resolution over the entire domain of the FRB using ten 100-m elevation bands in each grid cell. VIC model outputs used in the present study include time series of daily grid-scale and the routed streamflow (m3 s−1) for the FRB’s major sub-basins following Wu et al.28 as well as evapotranspiration. Details of the VIC model implementation and application to the FRB including its forcing dataset, parameters, model outputs, calibration (performed over water years 1949–1968) and validation (performed over water years 1969–2006) are described in Kang et al.6 (see also Supplementary Figs 3 and 4). Since the model calibration is conducted by minimizing streamflow differences against the observations, the simulated streamflow for these 20 years is not independent of the observed streamflow.
Processes not considered in these VIC model simulations are changes in land cover or land use (e.g., forest harvesting or deforestation by the recent mountain pine beetle outbreak in the FRB29), lakes as natural reservoirs and glacier melt contributions to streamflow. Deforestation leads to less interception of precipitation by a vegetation canopy while glacier melt contributions are high in warm, dry years, offsetting low precipitation amounts5. As such, both these processes augment runoff productivity for the same amount of precipitation. Lakes also store water, thereby delaying streamflow generation in downstream rivers and enhancing evaporation. There is evidence of higher observed versus simulated flows in the last two decades of the study period (1986–2006) in some sub-basins (SU, NA and CH) compared to previous decades that could indicate a growing influence of deforestation, lakes and glaciers to streamflow generation in the FRB (Supplementary Fig. 4). Such processes may be incorporated in future versions of the VIC model and applications to the FRB.
Daily streamflow trends simulated by the VIC model are assessed with corresponding observations compiled by the Water Survey of Canada. Six unregulated major sub-basins of the FRB, namely the UF, SU, NA, QU, CH and TN, are employed in this study (Table 1). Hydrometric data for the main-stem Fraser River (LF) at Hope, BC are also used in this study. This location is selected since it has the longest record of observed streamflow for the Fraser River’s main stem closest to its outlet to the Salish Sea in the Pacific Ocean, covering 94% of the basin’s drainage area.
Statistical and trend analyses
Mean annual streamflow for the FRB’s six major sub-basins and their relative contributions to total mean annual streamflow for the Fraser River at Hope spanning water years 1949 to 2006 are first quantified. Monotonic trends in observed and simulated daily streamflow covering the period of study at each site of interest are then evaluated using the Mann-Kendall Test (MKT)30,31,32,33. The non-parametric MKT is commonly used for hydrological trend analyses as it is robust to outliers and can be applied to non-normal data31. Trends are considered to be statistically significant when p < 0.05 with a two-tailed test while slope magnitudes are extracted from the associated Kendall-Theil Robust Lines31. Comparisons between observed and simulated trends in daily streamflow are assessed using correlation coefficients (statistically significant when p < 0.05), NSE coefficients32, root mean square errors (RMSEs) and relative biases. Trends in observed and simulated daily streamflow for the six major FRB sub-basins are summed for comparison with those for the Fraser River at Hope. Changes in the relative contributions of the six main sub-basins to total Fraser River discharge are evaluated over 1949–2006. Grid-scale trends in daily streamflow on 1 May and 1 June are computed to establish its spatial variations across the FRB during the spring freshet. The slopes and end-points of the Kendall-Theil Robust Lines for daily streamflow are then used to re-construct the water year, daily hydrographs for the Fraser River at Hope, 1949 and 2006. This allows an exploration of the observed and simulated hydrographs for those years if the changes were entirely monotonic over the period of study as based on the MKT. A 5-day moving average is applied to the reconstructed hydrographs to remove short-term fluctuations in streamflow. The 2006 reconstructed hydrographs are also shown with a 10-day lag to better illustrate the recent advances in the onset of the spring freshets. Shifts in the timing of the spring freshet are assessed from the average number of days between the 1949 and 2006 rising limbs of the hydrograph when streamflow surpasses 2000 m3 s−1, 3000 m3 s−1 and 4000 m3 s−1, respectively34,35.
How to cite this article: Kang, D. et al. Impacts of a Rapidly Declining Mountain Snowpack on Streamflow Timing in Canada’s Fraser River Basin. Sci. Rep. 6, 19299; doi: 10.1038/srep19299 (2016).
Reynoldson, T. B., Culp, J., Lowell, R. & Richardson, J. S. [Fraser River Basin] Rivers of North America. [ Benke, A. C. & Cushing, C. E. (Eds)] [607–732] (Elsevier Press, New York, 2005).
Eliason, E. J. et al. Differences in thermal tolerance among sockeye salmon populations. Science 332, 109–112 (2011).
Danard, M. & Murty, T. S. On recent climate trends in selected salmon-hatching areas of British Columbia. J. Climate 7, 1803–1808 (1994).
Foreman, M. G. G. et al. Simulations and retrospective analyses of Fraser Watershed flows and temperature. Atmos. Ocean 39, 89–105 (2001).
Déry, S. J., Hernández-Henríquez, M. A., Owens, P. N., Parkes, M. W. & Petticrew, E. L. A century of hydrological variability and trends in the Fraser River Basin. Environ. Res. Lett. 7, 024019 (2012).
Kang, D., Shi, X., Gao, H. & Déry, S. J. On the changing contribution of snow to the hydrology of the Fraser River Basin, Canada. J. Hydrometeorol. 15, 1344–1365 (2014).
Déry, S. J. & Brown R. D. Recent northern hemisphere snow cover extent trends and implications for the snow-albedo feedback. Geophys. Res. Lett. 34, L22504 (2007).
Choi, G., Robinson, D. A. & Kang, S. Changing northern hemisphere snow seasons. J. Climate 23, 5305–5310 (2010).
Morrison, J., Quick, M. C. & Foreman, M. G. G. Climate change in the Fraser River watershed: Flow and temperature projections. J. Hydrol. 263, 230–244 (2002).
Shrestha, R. R., Schnorbus, M. A., Werner, A. T. & Berland, A. J. Modelling spatial and temporal variability of hydrologic impacts of climate change in the Fraser River basin, British Columbia, Canada. Hydrol. Processes 26, 1840–1860 (2012).
Waylen, P. & Woo, M. Annual floods in southwestern British Columbia, Canada. J. Hydrol. 62, 95–105 (1983).
Leith, R. M. M. & Whitfield, P. H. Evidence of climate change effects on the hydrology of streams in south-central BC. Can. Water Resour. J. 23, 219–230 (1998).
Stewart, I. T., Cayan, D. R. & Dettinger, M. D. Changes toward earlier streamflow timing across western North America. J. Climate 18, 1136–1155 (2005).
Déry, S. J. et al. Detection of runoff timing changes in pluvial, nival and glacial rivers of western Canada. Water Resour. Res. 45, W04426 (2009).
Barnett, T. P., Adam, J. C. & Lettenmaier, D. P. Potential impacts of a warming climate on water availability in snow-dominated regions. Nature 438, 303–309 (2005).
Stahl, K., Tallaken, L. M., Hannaford, J. & van Lanen, H. A. J. Filling the white space on maps of European runoff trends: Estimates from a multi-model ensemble. Hydrol. Earth Syst. Sci. 16, 2035–2047 (2012).
Déry, S. J., Stieglitz, M., Rennermalm, A. K. & Wood, E. F. The water budget of the Kuparuk River Basin, Alaska. J. Hydrometeorol. 6, 633–655 (2005).
Mlynowski, T. J., Hernández-Henríquez, M. A. & Déry, S. J. An evaluation of hydrometric monitoring across the Canadian pan-Arctic region, 1950–2008. Hydrol. Res. 42, 479–490 (2011).
Jacob, C., McDaniels, T. & Hinch, S. Indigenous culture and adaptation to climate change: sockeye salmon and the St’át’imc People. Mitig. Adapt. Strat. Glob. 15, 859–876 (2010).
Picketts, I. M., Curry, J., Déry, S. J. & Cohen, S. J. Learning with practitioners: determining community climate change adaptation priorities. Climatic Change 118, 321–337 (2013).
Liang, X., Lettenmaier, D. P., Wood, E. F. & Burges, S. J. A simple hydrologically based model of land surface water and energy fluxes for general circulation models. J. Geophys. Res. 99, 14415–14428 (1994).
Liang, X., Lettenmaier, D. P. & Wood, E. F. Surface soil moisture parameterization of the VIC-2L Model: evaluation and modification. Global Planet. Change 13, 195–206 (1996).
Nijssen, B., O’Donnell, G. M., Lettenmaier, D. P., Lohmann, D. & Wood, E. F. Predicting the discharge of global rivers. J. Climate 14, 3307–3323 (2001).
Maurer, E. P., Wood, A. W., Adam, J. C., Lettenmaier, D. P. & Nijssen, B. A long-term hydrologically based dataset of land surface fluxes and states for the conterminous United States. J. Climate 15, 3237–3251 (2002).
Haddeland, I., Skaugen, T. & Lettenmaier, D. P. Hydrologic effects of land and water management in North America and Asia: 1700 ~ 1992. Hydrol. Earth Syst. Sci. 11, 1035–1045 (2007).
Shi, X., Groisman, P. Y., Déry, S. J. & Lettenmaier, D. P. The role of surface energy fluxes in pan-Arctic snow cover changes. Environ. Res. Lett. 6, 035204 (2011).
Shi, X., Déry, S. J., Groisman, P. Y. & Lettenmaier, D. P. Relationship between recent pan-arctic snow cover and hydroclimate trends. J. Climate 26, 2048–2064 (2013).
Wu, H., Kimball, J. S., Mantua, N. & Stanford, J. Automated upscaling of river networks for macroscale hydrological modeling. Water Resour. Res. 47, W03517 (2011).
Kurz, W. A. et al. Mountain pine beetle and forest carbon feedback to climate change. Nature 452, 987–990 (2008).
Mann, H. B. Nonparametric tests against trend. Econometrica 13, 245–259 (1945).
Kendall, M. G. Rank Correlation Methods (Oxford University Press, New York, 1975).
Déry, S. J., Stieglitz, M., McKenna, E. C. & Wood, E. F. Characteristics and trends of river discharge into Hudson, James and Ungava Bays, 1964–2000. J. Climate 18, 2540–2557 (2005).
Hatcher, K. L. & Jones, J. A. Climate and streamflow trends in the Columbia River basin: Evidence for ecological and engineering resilience to climate change. Atmos. Ocean 51(4), 436–455 (2013).
Moore, J. T., Harper, J. T. & Greenwood, M. C. Significance of trends toward earlier snowmelt runoff, Columbia and Missouri basin headwaters, Western United States. Geophys. Res. Lett. 34, L16402 (2007).
Burn, D. H. Climatic influences on streamflow timing in the headwaters of the Mackenzie River Basin. J. Hydrol. 352, 225–238 (2008).
Thanks to Dennis P. Lettenmaier (UCLA), Justin Sheffield (Princeton University), Francis Zwiers and colleagues (Pacific Climate Impacts Consortium) for their assistance and contributions. Thanks to Michael Allchin (UNBC) for drafting Figure 1 as well as Marco Hernández-Henríquez (UNBC) for reviewing the manuscript. This work was supported by the Canada Research Chair (CRC) program, NSERC Discovery and Accelerator Grants to SJD and the NSERC-funded Canadian Sea Ice and Snow Evolution (CanSISE) Network.
The authors declare no competing financial interests.
Electronic supplementary material
About this article
Cite this article
Kang, D., Gao, H., Shi, X. et al. Impacts of a Rapidly Declining Mountain Snowpack on Streamflow Timing in Canada’s Fraser River Basin. Sci Rep 6, 19299 (2016). https://doi.org/10.1038/srep19299
This article is cited by
Determining contemporary and historical sediment sources in a large drainage basin impacted by cumulative effects: the regulated Nechako River, British Columbia, Canada
Journal of Soils and Sediments (2019)
Theoretical and Applied Climatology (2019)
Theoretical and Applied Climatology (2018)
Climatic Change (2017)