Canada has the third largest renewable freshwater resources globally.  Analysis of historical time series of steam flow observations across Canada and groundwater recharge rates in Eastern Canada indicates that they have decreased by approximately 10% over the latter half of the 20th century.  This phenomenon might be explained by increases in evapotranspiration, in regions where moisture is not limiting, due to global warming trends.  Increasing evapotranspiration could mean a net decrease in renewable freshwater supply if precipitation does not also increase. A drop in renewable freshwater supply could have significant impacts on water availability in the relatively arid Prairie provinces of Canada.  Additionally, areas of the country that rely on groundwater for municipal and agricultural uses could face the need for additional infrastructure (deeper wells or new water supply lines) as evapotranspiration continues to increase.

Evapotranspiration (ET) is defined as the net exchange of water vapour between the Earth's surface and the atmosphere.  ET includes transpiration, dew deposition, evaporation and sublimation. ET is an important part of the water cycle.  In Canada, between 25% and 95% of the water received in the form of precipitation leaves the surface as ET.  In this sense, ET can and does often exceed annual stream flow.  However, unlike stream flow, there is no straightforward approach to measure ET over large areas.

Goals

This activity will produce, for the first time, Canada wide estimates of annual ET for the period 1960 to 2000, based on a numerical modelling approach, using observed climate and land cover information.

Methods

A detailed, validated numerical simulation model ‘EALCO' is being used to estimate historical and projected water budgets.  Projected trends in ET are also being produced using numerical simulation models driven with a range of Global Circulation Model forecasts for climate conditions under different emission scenarios.

Results

1. ET has consistently increased over most climate zones in Canada between 1960 and 2000 EXCEPT the Prairies.
   
   

   

   Figure 1:  Trends in ET over Canada between 1960 and 2000 for high-quality climate stations.
   

   larger image
   [GIF, 100.8 kb, 960 X 720, notice]

   

   Figure 2:  Climate stations where historical ET trends for 1960 to 2000 are significantly different from the ‘no change' case based on statistical analysis of model results.
   

   larger image
   [GIF, 103.1 kb, 960 X 720, notice]
   
   The increase in ET is related to temperature in southern Canada and increased total radiation in northern Canada.  In contrast, the Prairies do not show widespread increasing trends chiefly because precipitation, that has so far not shown significant trends over time, limits AET.  There is currently insufficient long term climate data over the Arctic climate zone to have high confidence in historical trends.
   
2. Projected changes in Prairie water supply based on trends in precipitation less ET do not indicate ‘high confidence' of either significant increases or decreases.
   
   

   

   Figure 3: Modeled trends in climate and renewable water supply over the 21st century for a 50km x 50km region in the vicinity of eight locations in the Prairie provinces of Canada.
   

   larger image
   [GIF, 20.9 kb, 720 X 960, notice]
   Renewable water supply can be estimated using long term averages of annual precipitation less ET (P- ET).  We ran our ET models using climate inputs from three GCMs for each of three climate scenarios (A1 – business as usual, A2 – increased emission rates, B1 – reduced emission rates) for emissions between 2000 and 2100.  Trends in P-ET vary between location, climate scenario, and even the GCM used to specify the projected climate.  Increasing trends in water supply are often more frequent than decreasing trends except for Medicine Hat where soil and land surface conditions result in consistently high ET under all trials.  It is noteworthy that most modeled trends are of the same order of magnitude of the uncertainty in out modelling results (~0.5mm/year).
   
3. Using simpler models to estimate AET can often result in substantial biases in estimates of water supply trends.
   
   

   

   Figure 4: Comparison of modeled trends in renewable water supply using potential evapotranspiration (x-axis) or actual evapotranspiration (y-axis) models.  Trends for the potential evapotranspiration approach are based on climate inputs from the Canadian GCM (CGCM) model for an intensive emission scenario (‘A2') for the 21st century.  Trends for actual evapotranspiration are based on three different models (Canadian CGCM, US National Centre for Atmospheric Research CCM and Hadley Centre HADCM3).
   

   larger image
   [GIF, 11.2 kb, 720 X 960, notice]
   
   
   Other studies have used alternate numerical estimates of ET that either ignore land cover (potential ET or PET) or assume a fixed land cover irrespective of location or climate(‘reference ET').  Here we compare estimates of renewable water supply, in terms of precipitation less evapotranspiration (P-ET), using our detailed numerical model versus estimates, published in Schindler and Donahue 2006, based on a simpler potential evapotranspiration model.  The estimates using the potential evapotranspiration model forecast large (~200mm over 100 years or 2mm/year) decreases in water supply in the Western Prairie Provinces corresponding to almost half of the available precipitation.  In contrast, our estimates form a cloud around the zero change level with the scatter due primarily to land cover and soil conditions and the GCM forecast model used.  Historical trends over the 20th century (see item 1 above) tend to support our estimates.
   
   
   
   
   
   
   
   
   
   
   
   
   
   
   
   
   
   
   
4. Projected changes in water supply over areas with historically high precipitation totals often indicate decreases in water availability although the extent of the decrease varies with the GCM model inputs and climate scenario used.
   
   

   

   Figure 5(a)
   

   larger image
   [GIF, 17.0 kb, 720 X 960, notice]

   

   Figure 5(b)
   

   larger image
   [GIF, 25.2 kb, 720 X 960, notice]
   
   

   Figure 5: Projections of ratio of evapotranspiration to precipitation (a) and base flow (b) (groundwater flow within streams and rivers) over the Annapolis Valley watershed based on EALCO model runs using different GCM forecasts for different climate scenarios (A1 – business as usual; A2 – increasing rate of emissions; B1 – decreasing rate of emissions).

   In contrast to the Prairies, our projections indicate decreasing trends in renewable water supply over areas in southern Canada with substantial precipitation.  For example, evaporative fraction (defined as the ratio of evapotranspiration to precipitation) is projected to increase over the 21st century for the Annapolis Valley watershed in Nova Scotia.  This increase implies a decrease in groundwater flow in streams (base flow); although the range of the modeled decrease depends varies with climate scenario adopted and GCM used to produce climate inputs to our water budget model under each scenario.

Top of page

• More information on historical estimates of evapotranspiration is available on the Sustaining the Environment and Resources for Canadians Web site
• More information on historical trends of evapotranspiration over Canada is contained in the following publication:  Trends in land evapotranspiration over Canada for the period 1960-2000 based on in-situ climate observations and a land surface model (Fernandes et al.) Abstract
• Please contact Dr. Richard Fernandes for more details historical and projected trends in water supply.
• This work was supported by the Natural Resources Canada RCVCCP and PERCC programmes, the RESEAU Government-OnLine initiative and the Canadian Space Agency.