Warning This Web page has been archived on the Web.

Archived Content

Information identified as archived on the Web is for reference, research or recordkeeping purposes. It has not been altered or updated after the date of archiving. Web pages that are archived on the Web are not subject to the Government of Canada Web Standards. As per the Communications Policy of the Government of Canada, you can request alternate formats on the Contact Us page.

Skip booklet index and go to page content

Threats to Water Availability in Canada

11. Climate Variability and Change - Rivers and Streams

Paul H. Whitfield,1 Paul J. Pilon,2 Donald H. Burn,3 Vivek Arora,4 Harry F. Lins,5 Taha Ouarda,6 C. David Sellars7 and Christopher Spence8

1 Environment Canada, Meteorological Service of Canada, Vancouver, BC
2 Environment Canada, Meteorological Service of Canada, Burlington, ON
3 University of Waterloo, Department of Civil Engineering, Waterloo, ON
4 Environment Canada, Meteorological Service of Canada, Canadian Centre for Climate Modelling and Analysis, Victoria, BC
5 United States Geological Service, Water Resources Division, Reston, VI
6 Institut national de la recherche scientifique-Eau, Ste Foy, QC
7 Water Management Consultants, Vancouver, BC
8 Environment Canada, Meteorological Service of Canada, Yellowknife, NT


Current Status

Canadians have a strong cultural link with water, a major facet of our heritage, our spirituality, and our economy. We are viewed globally, perhaps with envy, as having vast water resources in a pristine setting. While this is true in general, there are areas within Canada that are semi-arid and sensitive to even small departures from average precipitation. Other areas with visibly abundant resources, such as the Great Lakes basin, can suffer from local over-exploitation of the resource. Concern is growing that these precious water resources are further threatened by what is popularly referred to as climate change. Several recent studies (Adamowski and Bocci, 2001; Burn and Hag Elnur, 2002; Cunderlik and Burn, 2002; Pilon and Yue, 2002; Whitfield and Cannon, 2000; Yue et al., 2002a,b; Zhang et al., 2001) have investigated trends in the flow of rivers and streams occurring in natural basins across Canada. Different aspects of the hydrological regime have been analyzed, such as seasonal patterns, average flow conditions and their extremes.

These studies have found that different areas display markedly different trends and tendencies in streamflow, and there is no simple description possible for natural rivers and streams across Canada. Individual station results indicate that annual minimum and mean daily flows are increasing significantly in northern British Columbia, Yukon Territory and southern Ontario, with evidence of decreasing tendencies concentrated in southern British Columbia (see Fig. 1). Studies show that maximum flows are tending to decrease significantly across most of Canada, similar to those reported for the U.S. by Lins and Slack (1999). In essence, changes are occurring, but not in simple ways. At this time, it is not possible to attribute these results fully either to a changing climate, resulting from increased greenhouse gas concentrations, or to natural climate variability. These analyses have focussed on observations over the last several decades, and at times contradict hydrological simulations using Global Circulation Models (GCMs). The gap between retrospective analysis and model results is, from a hydrologic perspective, an area of concern.

Image: trends in annual minimum daily flow data for the RHBN (1957-1996)

Fig. 1 Trends in annual minimum daily flow data (1957-1996) for the Canadian RHBN (Yue et al., 2003).

All the above cited studies have used streamflow data from the Reference Hydrometric Basin Network (RHBN) (Harvey et al., 1999; Pilon and Kuylenstierna, 2000). This network has decreased in size from its original design and currently contains 200 plus stations representing basins with at least twenty years of record under stable or pristine conditions. Hydrometric stations within the RHBN, such as the subset shown in Fig. 1, are of particular importance for studies oriented to climate variability and change; however, analysis of the characteristics of stations indicates certain limitations. The network tends to be composed of large basins in the north and smaller basins in the south, with certain provinces having large gaps in spatial coverage. This underscores the importance of long-term continued funding to targeted hydrological data collection for monitoring ambient conditions, and the need to supplement the current network to allow smaller basins to be brought on-line in the North and to fill major geographic gaps. The current Canadian approach of a data collection process driven by immediate needs (e.g., hydropower production, flood control reservoir operation, apportionment) is unlikely to result in an adequate network for providing basic data to evaluate impacts of climate variability and change on water resources. Monetary commitment to such long-term data collection is needed.

Since the last hydrological atlas for Canada (1977), there has not been an update that reflects changes to historic patterns. Today, more dynamic techniques are needed to produce relevant estimates of current water availability. Canada should be in a position to know current water availability and changes that might be affecting regional and local availability for both regulated and natural basins.

Much effort has been expended in advancing knowledge of small-scale process hydrology (Buttle et al., 2000). Canadian researchers have also developed modelling capabilities for water resources management. These larger scale models, available to practitioners, such as WATFLOOD (Kouwen and Mousavi, 2002) and SLURP (Kite, 1995), have not built upon the understanding of physical processes established through small-scale hydrological studies. An added complexity to modelling streamflow conditions is accounting for human activities (e.g., land-use changes, consumptive use practices, diversions) as these can affect local water availability. Such modelling efforts require knowledge of local intervention (e.g., withdrawals for irrigation, changes in land use, etc.). Understanding climate change and variability in the broad Canadian context requires an enhanced understanding and a modelling capability that builds upon hydrological process knowledge and correctly represents human interventions. The separation of impacts of interventions from those of climate change and variability is the basis for estimation and prediction of water availability under Canada’s diverse and changing circumstances.

The process of modelling water availability requires adequate hydrological data. In Canada, the density of existing networks is low, and sparsely populated areas have even fewer stations. Currently, our analysis and modelling capabilities suffer from a bias against observing streamflow in small watersheds and a lack of linkages among hydrological observing networks (e.g., streamflow, precipitation, humidity, evaporation). The lack of accurate and high-resolution input data limits our ability to model the response of catchments. In essence, detailed knowledge about the spatial and temporal distribution of precipitation is lacking, mainly due to the very low density of our climate network in several regions of Canada and to our inability to translate remotely sensed information into meaningful hydrological fields. This is also valid for a number of other essential parameters (e.g., permafrost conditions, soil moisture, radiation).

Overall, we seem to be addressing some important issues and making progress toward improving our knowledge base on water availability. However, Canadian researchers are not explicitly pursuing work aimed at increasing our understanding of water availability. There are areas in which Canada has very good knowledge of important processes, and there are other areas where there are rather obvious knowledge, information, or product gaps. Canada needs a national program that coordinates our scientific expertise, data collection, modelling, and research to position us to be able to estimate water availability and to deal with impacts of climate variability and change in streams and rivers.

On the banks of the Yukon River.

On the banks of the Yukon River. Studies show that annual flows are increasing in northern British Columbia, Yukon Territory and southern Ontario, while decreasing in southern British Columbia.


The general public has become more aware of climate change and climate variability and their potential impacts on water availability. This has led to greater acceptance of the need for further study and may, in turn, lead to actions to address perceived impacts.

Concern has increased that climate change and variability might have impacts on hydrologic extremes (i.e., floods and droughts) as well as affecting the overall level of water availability. This realization has led to greater appreciation of the importance of understanding extremes of water availability and the impact change might have on society, the economy and the environment.

We do not yet know, with certainty, the impacts of climate change and variability on water bodies of importance to Canadians. Efforts are being made to define potential impacts based on generated future scenarios, in many cases based on results of one or relatively few GCMs, with different GCMs typically providing different potential future conditions. Disturbingly, the modelled impacts (i.e., prognostic analyses) tend not to be consistent with retrospective analyses of hydrological data (i.e., diagnostic analyses) for the same periods. For example, the Mackenzie Basin Impact Study (Cohen, 1998) noted that streamflow in tributaries to the Mackenzie will decrease with warming, but Whitfield and Cannon (2000) suggest that streamflow may initially be increasing in these same streams. There are also some signs of consistency between prognostic and diagnostic analyses, such as earlier freshets for river basins characterized by snowmelt. As long as a gap exists between the results of prognostic and diagnostic analyses, caution should be exercised when developing coping strategies and policies based on seemingly conflicting evidence.

Diagnostic studies require statistical testing methodologies. There has been a tendency to develop increasingly sophisticated approaches that more fully address limitations of earlier approaches and allow spatial inferences to be made. These approaches have led to an increased understanding of the patterns of trend in streamflow (Burn and Hag Elnur, 2002; Cunderlik and Burn, 2002; Whitfield and Cannon, 2000; Yue et al., 2002b).

However, the number of sites in various hydrological networks available for diagnostic studies is decreasing. Statistically based diagnostic studies require networks with long-term stations with adequate geographical coverage to provide the capacity to discern whether a trend is or is not occurring and where a trend might be occurring (Yue et al., 2002a). Long-term data are needed to depict and separate accurately the characteristics of natural climate variability from climate change.

Increasingly, we need a more global perspective on the driving forces generating streamflow. There has been an increased understanding of the teleconnections between large-scale oceanic and atmospheric processes and meteorological and hydrological processes. Examples of large-scale oceanic and atmospheric processes include the El Niño - Southern Oscillation (ENSO) and the Pacific Decadal Oscillation (PDO). These processes and the impacts they can have on hydrology and, hence, water availability, are becoming better known as new research explores such relationships (Barlow et al., 2001; Neal et al., 2002).

Knowledge and Program Needs

There are significant difficulties in distinguishing between natural climate variability and long-term change. To address this issue more fully, attention must focus on three broad areas: data collection, technique development, and coordinated and funded programs.

Data Collection

As a nation, our surface water and precipitation monitoring systems are poorly resourced, designed and coordinated. For example, the World Meteorological Organization (1994) recommends minimum densities for various networks and has a minimum for precipitation observations of between 1000 and 4000 locations based upon land mass, while Canada has less than 500 reported sites (WMO, 1995). These deficiencies should be addressed for a number of reasons, including providing a solid foundation for studies directed to separating climate change effects from those of normal variability. Some key steps include:

  1. increasing linkages among observing networks to improve understanding of causes and effects of impacts within the hydrological cycle. For example, glacier mass balance represents an opportunity to increase the hydrologic understanding of transient changes in runoff regime (Moore and Demuth, 2001)
  2. conducting a thorough review and design of our observing networks to ensure they are effective and can provide adequate information upon which to base analyses. With additional investment to long-term observing systems, Canada will have the data and knowledge necessary for sound decision making
  3. collecting flow and related data at additional sites in unregulated and relatively pristine watersheds to fill geographic gaps and more adequately represent smaller areas. The enhanced observation of small basins increases knowledge of the effect of scale on hydrologic processes in Canada
  4. enhancing data collection in northern Canada where the changes in local climates are predicted to be larger than in the south. With this focus, we are more likely to be able to verify that change is occurring as predicted.

Technique Development

The dilemma Canada faces with respect to climate change and variability and their associated impacts indicates the need for development of new approaches and methodologies. Current approaches are inadequate for dealing with the unprecedented ramifications of potential changes. There has been a heavy reliance on simple statistical testing approaches and GCMs. Currently, very little research effort is placed on hydrological aspects of climate change and development of statistical trend detection approaches relevant for hydrological analyses. There is a need to:

  1. develop statistical testing approaches to compare spatial and temporal model results with observed spatial and temporal trends
  2. develop more powerful retrospective (diagnostic) procedures.

There is also a need to develop analysis and modelling techniques to increase understanding of the impacts of climate change, natural variability and human-induced interventions. Beyond development of new modelling approaches, this necessitates obtaining data on consumptive uses, diversions, regulation and land-use changes. Remotely sensed data have tremendous potential for aiding assessment of general water availability in Canada; investments are needed to develop methodologies and capitalize on this potential. Such capabilities are fundamental to establishing current and future water availability throughout Canada. These concepts are important elements of integrated watershed management and can lead to adaptive practices that will help Canadians face changes in water availability.

To establish local trends and conditions, national, regional (e.g., hemispheric, North American), and global analyses must be performed. Patterns emerge and are uncovered when this issue is approached from different scales and perspectives. Countries should establish specialized networks with characteristics similar to those used to establish the RHBN. This provides a common basis for analysis, and for evaluating and understanding variability and change within the global, regional and Canadian contexts.

At present, hydrologists are concerned that the GCMs do not accurately reflect hydrological conditions. GCMs do not provide outputs at the temporal and spatial scales deemed important for water resources applications. There is a need to:

  1. improve operational modelling of Canadian hydrological systems with a particular focus on the feedbacks inherent in this system
  2. improve operational watershed models by inclusion of small-scale hydrological process knowledge to replicate nature’s processes more closely
  3. move beyond using GCM outputs as inputs to hydrological models as this is an over-simplification of the hydrological cycle, given the complex feedbacks within water and energy cycles
  4. establish the performance of the GCM and GCM-hydrological models over known conditions, particularly with respect to replicating trends in streamflow conditions over the past 30 to 50 years to increase model credibility by closing the gap between prognostic and diagnostic analyses
  5. increase GCM resolution so that important land features known to influence local climatology, such as the Great Lakes and Rocky Mountains, are reasonably represented in the modelling system.

Previous approaches for assessing risk in hydrological systems have explicitly assumed that the hydrological regime is stationary and will continue to be stationary in the future. This assumption is not necessarily valid, due to natural variability and climate change. However, there is insufficient evidence, particularly because of a lack of data upon which to assess variability and change and from a cause-effect perspective, to warrant revisiting current design approaches. As a precaution, it would be wise to consider within risk management that future risks may be more or less than anticipated, based on a historical analysis of data. At present, it is difficult to be certain if recently observed patterns and trends will persist into the future.

Coordinated and Funded Programs

Funding currently provided for research and development on available water within Canada’s streams and rivers and the influence of climate variability and change on these waters needs strengthening. Environment Canada should focus and coordinate its water programs, developing a clear direction on understanding Canada’s water availability.

Research and development associated with water quantity are important for developing policy and programs to assist Canadians and their economy with this issue. Potential future impacts of variability and climate change on water availability could have tremendous impacts on society, the economy, and the environment. To ensure that Canadians are not inadvertently placed under increased risk, greater investment in ascertaining water availability is needed. Leadership is required to help focus efforts in research and development to gain a better understanding of hydrological conditions associated with natural variability, and those emanating from climate change. Emphasis should be placed on development of prediction (diagnostic and prognostic) systems for overall infrastructure design, and this would also assist in preventing loss of life and reducing damages to property and livelihood from floods and droughts.

Increased emphasis should be placed on retrospective diagnostic analysis, in addition to the current focus on generation of future scenarios and their impacts. Powerful spatial and temporal statistical testing approaches need to be developed. GCM performance requires validation with historical data, using sound statistical tools. There is a need to develop a much better understanding of the cause-effect relationships that have led to pronounced hydrological trends and patterns over recent periods. These needs accentuate the requirement for increased program coordination and funding for research and development programs. There is a need for increased data collection targeted at hydrometric stations on natural basins and for increased observation of meteorological variables to develop the knowledge and hydrological modelling capability vital to quantifying the impacts of climate on water availability.


Adamowski, K. and C. Bocci. 2001. Geostatistical regional trend detection in river flow data. Hydrolog. Processes 15: 3331-3341.

Barlow, M., S. Nigam and E.H. Berbery. 2001. ENSO, Pacific decadal variability, and U.S. summertime precipitation, drought, and stream flow. J. Climate 14: 2105-2127.

Burn, D.H. and M.A. Hag Elnur. 2002. Detection of hydrologic trend and variability. J. Hydrol. 255: 107-122.

Buttle, J.M., G. Jones, P. Marsh and M.K. Woo. 2000. Recent advances in Canadian hydrology. Hydrolog. Processes 14: 1537-1538.

Cohen, S. 1998. Mackenzie basin impact study (MBIS). Environment Canada, No. En 50-118/1997-1E, 1998.

Cunderlik, J.M. and D.H. Burn. 2002. Local and regional trends in monthly maximum flows in southern British Columbia. Can. Water Resour. J. 27: 191-212.

Harvey, K.D., P.J. Pilon and T.R. Yuzyk. 1999. Canada’s reference hydrometric basin network (RHBN). In Proceedings of the CWRA 51st Annual Conference, Nova Scotia.

Kite, G.W. 1995. The SLURP model, computer models of watershed hydrology, chapter 15. In V.P. Singh (ed.), Water Resources Publications, Littleton, Colo.

Kouwen, N. and S.-F. Mousavi. 2002. WATFLOOD/SPL9 hydrological model and flood forecasting system, chapter 15, p. 649-686. In V.P. Singh and D.K. Frevert (ed.), Mathematical models of large watershed hydrology. Water Resources Publications, Littleton, Colo.

Lins, H.F. and J.R. Slack. 1999. Streamflow trends in the United States. Geophys. Res. Lett. 26(2): 227-230.

Moore, R.D. and M.N. Demuth. 2001. Mass balance and streamflow variability at Place Glacier, Canada, in relation to recent climatic fluctuations. Hydrolog. Processes 15: 3473-3486.

Neal, E.G., M.T. Walter and C. Coffeen. 2002. Linking the pacific decadal oscillation to seasonal discharge patterns in Southeast Alaska. J. Hydrol. 263: 188-197.

Pilon, P.J. and J.K. Kuylenstierna. 2000. Pristine river basins and relevant hydrological indices: essential ingredients for climate-change studies. WMO Bulletin 49(3): 248-255.

Pilon, P.J. and S. Yue. 2002. Detecting climate-related trends in streamflow data. Water Sci. Technol. 8(45): 89-104.

Whitfield, P.H. and A.J. Cannon. 2000. Recent variation in climate and hydrology in Canada. Can. Water Resour. J. 25: 19-65.

World Meteorological Organization (WMO). 1994. Guide to hydrological practices. Fifth Edition. WMO-No. 168. Table 20.1, p. 270.

World Meteorological Organization (WMO). 1995. INFOHYDRO manual. WMO-No. 683. Table 4.4.07.

Yue, S., P.J. Pilon, G. Cavadias and B. Phinney. 2002a. Power of the Mann-Kendal and Spearmans rho tests for detecting monotonic trends in hydrological series. J. Hydrol. 259: 254-271.

Yue, S., P.J. Pilon and B. Phinney. 2003. Canadian streamflow trend detection: impacts of serial and cross correlation. J. Hydrol. Sci. 48(1): 51-64.

Yue, S., P.J. Pilon, B. Phinney and G. Cavadias. 2002b. The influence of autocorrelation on the ability to detect trend in hydrological series. Hydrolog. Process. 16(9): 1807-1829.

Zhang, X., K.D. Harvey, W.D. Hogg and T.R. Yuzyk. 2001. Trends in Canadian streamflow. Water Resour. Res. 37(4): 987-998.