Freshwater Quantity and Quality in Canada:

Ecosystem interaction with a Changing Atmosphere

John W. Pomeroy Hydrological and Aquatic Sciences Division

National Hydrology Research Institute Environment Canada

Saskatoon, Saskatchewan

INTRODUCTION

The topic of this presentation is the impact of climate change on freshwater resources

in Canada, focussing on the interaction of ecosystems with a changing atmosphere

through the medium of surface water. The purpose of the discussion is to:

1) outline the role of freshwater in ecosystem response to and interaction with

atmospheric change, and

2) suggest the data required by Canada to anticipate and react sensibly to the impacts

of climate change.

The discussion will use examples that highlight Environment Canada scientific

advances in the present understanding of ecosystem-water-atmosphere interactions.

The data requirements will be those appropriate for a federal agency mandated with

identifying and ameliorating a national threat of this broad nature.

Freshwater has three irr~portant roles in ecosystem response and interaction with

atmospheric change. Freshwater is a

1) transmitter of atmospheric change to Canada's environment,

2) mediator of this change., and

3) host for aquatic ecosystems that are affected by this change.

As a transmitter, water is a flow of mass, energy and biochemical constituents through

and between ecosystems and between the surface and the atmosphere, as water,

water vapour, snow and ice; hence it transmits climate change impacts across the

country and across ecological and jurisdictional boundaries. It transmits the effect of

drought to soils as soil water and the effects of heavy precipitation downstream as

floods. Freshwater can mediate climate change to some degree because it is stored on

the landscape as lakes, snowcovers, glaciers, wetlands, rivers, and is a store of latent

energy. The latent energy of freshwater is extremely large; the energy used to melt a

gram of ice would raise the temperature of a grani of water 80°C and the energy used

to evaporate a gram of water would raise its temperature approximately 600°C if

evaporation did not occur. Water stored as snow reflects 80% of incoming solar energy

and therefore has a strong cooling effect. Despite the obvious importance of such a

surface energy mediator, the role of surface water in controlling climate change has not

had sufficient attention. As a host for organisms water is unsurpassed in Canada and

the subject of intense study. As water quantity and quality are affected by the changing

climate there is a direct impact on aquatic organisms with potential feedbacks to the

atmosphere through their transpiration and trace gas uptake or release.

FRESHWATER-ATMOSPHERE EXCHANGE

Freshwater atmospheric exchange can be viewed from various perspectives that are

driven by traditional scientific disciplines. From the point-of-view of physics, water is a

substance with a unique behaviour that covers the land as a liquid, solid and gas; it

travels easily, has a highly variable reaction with radiant energy and can store large

quantities of latent energy. Distinctive to water are its incompressability as a liquid and

that the solid form is less dense than the liquid. Life on our planet would be

unrecognizable without these water properties. From a chemical perspective, water is

an extremely strong solvent and transport medium for various acids, geochemicals and

organic molecules. It transports nutrients, pollutants and geochemicals to, from and

around the Earth's surface, and is critical to the geochemical cycling of the planet.

Biologists have long documented the strict requirements that life forms have for water,

either as habitat or for consumption. The organisms have adapted to certain quantity

and quality conditions but also exert a controlling feedback on water quality. With

respect to society - and Canadian society is outstandingly dependent upon large

aniounts of water - Canada is known to have a disproportionately-large amount of the

Earth's fresh water resource. This resource is not efficiently used however, as we have

adapted to the present generous supplies. Our neighbours are well aware of this

situation. Economic demand for water is confounding to its management as this

demand is highly inelastic. Plentiful, clean water has little or no value but when in short

supply it becomes almost priceless as it is necessary for life on a daily basis.

Therefore, if the balance between supply and demand of North American water

changes because of changes in precipitation, quality or evaporation, we will have to

respond to dramatic internal and external pressures. The outcome of such pressures

on our industry, agriculture and recreation may not be positive or pleasant.

Hydrology brings together the various perspectives of water-atmosphere-ecosystem

exchange by consideration of the hydrological cycle. Figure 1 shows the hydrological

cycle over Canada, in which we have atmospheric water vapour transport (clouds,

vapour) evaporation feeding the atmospheric vapour from oceans and various

mountain, forest, agricultural and arctic ecosystems, and then water returning to the

surface as either rain or snow. Much precipitation infiltrates the soil where it can be

used by plants for evapotranspiration or can drain deeply to replenish groundwater.

Precipitation in excess of infiltration, evapotranspiration and drainage requirements is of

very great interest because it moves along the land surface as runoff or streamflow.

Because of runoff, continental water returns to the oceans and can cross normally dry

land surfaces. Hence even if there is a local drought, a river flowing through to a lake

or a wetland can produce evapotranspiration and provide a habitat. When irrigated this

water can replenish soil water supplies. An irr~portant store of water is in groundwater

aquifers, so that even if we temporarily run out of surface supplies, we can withdraw

from this "bank account" for awhile.

WATER IN GLOBAL CLIMATE SIMULATIONS

It is important to consider how global circulation models (GCMs) handle surface water

in .their predictions of climate change and their (in)ability to predict the impacts of this

change. Of first note is that the present simulations are unable to obtain a correct

surface water balance. These models inacc~~rately represent surface hydrology, by

causirig simulated rivers to appear to flow uphill, or causing simulated plants to

completely shut down during droughts because the models are atmospherically derived

and do not sufficiently consider hydrological systems yet. We hope that they will soon,

with assistance from hydrologists.

Another point that GCMs need to consider is that many natural ecosystems manage

their climate by managing their water because it is the most practical and efficacious

way to store energy on the surface. Canadians try to do this in agricultural ecosystems

by spacing plants, irrigating, snow management, crop combinations, etc. Because of

the important control and response of regional ecosystems to water fluxes and storage,

links between ecosystem and atmospheric dynamics must focus on fresh water as the

mediator and transmitter of that link.

An example of how GCMs handle the water balance is given for the Mackenzie River

Basin of northwestern Canada. Hydrologists are interested in this basin for a number of

reasons. It has a lot of very Canadian distinctions to it - glaciers, wetlands, permafrost,

snow melt, etc. It is also of interest because it flows into the Arctic Ocean. The

Mackenzie River provides the largest input of fresh water to the Arctic Ocean from the

North American continent, and is very important for ocean and icepack dynamics. It is

one the world's largest river basins, in an area expected to experience a lot of climatic

change; hence, it is the Canadian focus of the Global Energy And Water Cycling

Experiment of the World Climate Research Programme. As shown in Fig. 2, the

Mackenzie basin is covered by 30 grid points of the Canadian Climate Centre (CCC)

GCM. The resolution of any climate simulation of the basin is therefore very coarse.

Each of the grid cells is one GCM grid point for which there is a climate and water

sinlulation. Each cell has a water balance with the excess simply "flushed" directly

through the basin without hydrologic controls. Figure 3 shows a ten-year "present

climate" average of the CCC GCM outputs of water flow routed through the Mackenzie

River, the results of the GCM climate parameters fed into a process-based hydrological

model developed at by Dr. Geoff Kite of NHRl called "SLURP", and measured average

discharge of the Mackenzie. It is evident that the GCM has serious deficiencies in

predicting hydrology. It is predicting the peak runoff event several months early and

almost an order of magnitude larger than it actually is, and late summer flows are badly

underestimated. These errors can be largely corrected by inclusion of a hydrological

component such as that developed by Dr. Kite.

A hydrological model with the GCM data can be used to simulate a large basin

relatively easily because storage and routing of water dorr~inate the basin response. It

is not as simple to simulate the runoff of a smaller basin with a hydrological model

because other hydrological processes become important at small scales. These scales

are much smaller than the present GCM grid cell which calls into question the intrinsic

ability of GCMs to help predict surface hydrology irnpacts. lniprovements to this

situation are being led by land-surface schemes that are being attached to GCMs.

Canada is fortunate to have developed a model called "CLASS" - the Canadian Land

Surface Scheme, to represent the interaction of GCMs and the Earth's surface. In a

recent intercomparison, CLASS was found to be the most hydrologically accurate of

any land surface process model; however, when we examine it in detail, we find that

the hydrological processes are not always realistically represented in this model and

hence it can produce some unrealistic results.

An important aspect of climate change impacts in Canada is the depletion of a snow

covered area during snowmelt. This is very important for the albedo feedback to the

energetics calculation of GCMs, for calc~~lating snowmelt runoff in the spring, and also

for various ecological factors: birds need snowfree land to nest, the rate of melt

determines runoff versus infiltration and hence the filling of wetlands. Comparing

CLASS with a snow hydrology model developed by Dr. Kevin Shook of Saskatoon in

Fig. 4 shows that CLASS predicts complete snow cover disappearance on the Prairies

when the snow covered area is actually about 90%, and about three weeks before the

actual disappearance. This error has a substantial impact through the whole

hydrological and ecological system. The simulation was run for Bad Lake, where we

have good measured data dating from the International Hydrological Decade in 1972.

One of the research programmes at NHRl right now is an attempt to improve models

such as CLASS so that they would adequately represent this type of phenomena.

A phenomenon often neglected by hydrological models and atmospheric models is

snow accumulation. There are a lot of very important processes here that are

extremely important to Canada because most of the country is snow covered for about

half of the year and northern areas for even longer. Hence the stored seasonal snow

cover leads to the largest annual runoff events in most of Canada. Snow cover

provides about 80% of runoff in the Prairies, in the Boreal forest about 50%, and up in

the Arctic it creeps up again to 60, up to 80-90% in the high Arctic, and somewhat less

in some of the more temperate regions. This is critical because snowmelt recharges

the lakes, various wetlands, bogs and fens around the country. The amount and timing

of snowmelt runoff are very strongly influenced by ,the amount of snow on the ground.

The amount of snow on the ground in spring is not that which falls, but rather that which

falls and is not redistributed by either the forest cover or by blowing snow, and that

which is not melted over the winter. What we have found is that it is very rare in fact

that you find a situation where the amount of snow that falls equals that which is on the

ground, and no atmospheric models have incorporated this effect yet. Figure 5

presents all exarr~ple from an Arctic basin where there are dramatic losses in the

amount of snow, or gains in certain areas, that are governed by blowing snow

processes and their interaction with the vegetation cover. This sort of variation needs

to be incorporated into hydrology models and GCM and NHRl has a programme to do

just that.

In Fig. 5, data are presented from the GEWEX experimental watershed in the Western

Arctic north of Inuvik. Here the ecosystem exerts a tremendous hydrological control on

the precipitation, amount of snow melt and runoff. If there is a climate change

response that would result in a change in the spatial distribution of this vegetation, we

would see a dramatic change of runoff, even if there is no change in the snowfall.

CLIMATE WARMING - THREATENED REGIONS.

Is all of Canada equally threatened? The U.K. Meteorological Office Unified (Hadley

Centre) GCM includes a coupling to oceans and the cooling effect of sulphate aerosol

and shows that areas that are strong sources of sulphate are not going to experience

the strongest warming. This is matching historical trends very well. The result is that

only moderate warming is anticipated for southeastern Canada, in the Great Lakes, St.

Lawrence area: + I to +2OC. Western and northern Canada apparer~tly will experience

the strongest warming in the Northern Hemisphere (+2 to +5 C), and we might want to

target these very strongly threatened ecosystems. Figure 6 is a map of predictions of

mean temperature change In 45 years by the Hadley Centre GCM. It is the most recent

run of the model for Canada, and you can see ,that eastern Canada is looking at just lo

C warming under that prediction, whereas western and northern Canada will still

experience a fairly strong warming. The impact of climate change on the water

resources of western and northern Canada should therefore be targeted by the

Department.

CLIMATE VARIABILITY - ECOSYSTEM RESPONSE

All of Canada may be expecting increased variability of weather and frequency of

destructive weather, often manifested by destructive hydrological phenomena. The

ecosystem manifestations of this weather will be moderated or exacerbated by

freshwater impacts. The frequency of frost is affected by soil moisture and atmospheric

moisture; frequency of blizzards is strongly affected by air temperature and snowfall;

the flooding recharge of delta lakes is affected by upstream snow cover and the rate of

snowmelt; forest fire incidence will depend upon soil moisture. In Saskatchewan 10%

of the commercial forest burned in the spring of 1995 due to abnormally hot dry

weather. In spring of 1996, there occurred substantial snowmelt floods in the same

province. Depletion of oxygen in lakes is linked to inflow timing and quality. An

increasing frequency of droughts can reduce summer streamflow and increase the

solute concentration in certain surface water supplies. That may lead to salinization in

irrigation areas, increased dissolved solid load in some situations and toxicity in others.

These manifestations are typical of increased variability but it cannot easily be

statistically verified yet.

FRESHWATER QUANTITY RESPONSE, CONTROL AND FEEDBACK

When we consider climate change and freshwater quantity, it can be examined with

respect to the response of the freshwater system, the control exerted by the freshwater

system, and the feedback to the climate system from freshwater. In Canada in general

there could be a shorter snowcovered season, and an earlier melt. In areas where the

snowcover forms and melts periodically over the winter, there will be greater winter

flooding; as in eastern Canada and the western mountain regions. Greater frequency

of summer droughts is expected, particularly in the Prairies and Southern Ontario, and

in response to the quantity change, a biomigration of species from areas limited by

water. One would be a northward movement of the rrrixed wood forest into the present

Boreal forest, northward movement to the Prairies into the mixed wood forest belt,

rangeland perhaps into Prairie zones, and the retreat of trees from B.C. interior valleys.

We are also going to have movement of the temperate forests in the east into some of

the Boreal forest areas. As a result, there will be a reduction of wetlands and lakes in

general across Canada. In the Prairies there will be a drying of many Prairies sloughs

dependent upon the snowmelt runoff; a reduction in Boreal lakes which are also

dependent upon this local runoff, and the drying of river delta lakes that depend on the

spring flood. Dr. Terry Prowse of NHRl finds that there is some indication that the

drying of the Peace-Athabasca Delta lakes is due to a decline in Alberta snow cover.

The floodplain lakes in the Mackenzie delta are an extremely important aquatic habitat

for muskrat and waterfowl and important for many other species. A simulation done by

Dr. Phil Marsh at NHRl on the Mackenzie delta lakes showed that with a 2xCO,climate,

in about ten years the lakes begin a substantial retreat in size because of reduced

flooding. This is one example of increased and protracted variability in water supply

across the country. Another example is Fig. 7 showing discharge of the Kootenay River

in British Columbia. Dr. Kite's SLURP Model is run using the outputs of the Canadian

Climate Centre GCM for the present climate (IxCO,) over a series of years, and you

see in the top graph, the simulated and recorded flows are matched quite well over this

period of record. In the bottom graph the IxCO, simulation indicates a single spring

flooding event, usually in MayIJune. For comparison is a 2xC0, simulation which

indicates floods at any time in the winter. In certain cases the warmer climate produces

serious winter floods and higher flow in tlie fall. Essentially what is going on here is that

precipitation, instead of falling as snow, is falling as rain, and flushing through the

system much more rapidly.

For streams draining small basins on the Prairies, spring is the time of runoff and often

the only time when runoff is produced. One of the predictions with a warmed climate in

winter and spring, is ,that we may not have fully frozen soils as they are now. Because

there are normally summer shortages of water on the Prairies, this limits the annual

runoff considerably. Dr. Raoul Granger of NHRl has produced a frozen soil infiltration

routine and incorporated this routine in a standard hydrological model. If we take that

frozen soil routine out and replace it with an unfrozen soil calculation, a good simulation

of hydrological response of prairie basins to a warmer climate is developed. The

change is shown in Fig. 8 with the two matching hydrographs showing measured and

presently predicted runoff and the very low hydrograph showing what is predicted under

a warmer climate. The warmer climate prediction suggests that runoff will drop to

almost nothing, so if you are a duck, it is going to be a hard time. This is also very bad

if one is a farmer dependant upon surface water supplies. The runoff from mountains

and the Prairies, though small, has a very important effect on small towns, on the

agricultural area, and upon the wildlife.

There are some interesting links through evaporation between ecosystem productivity

and climate. Dr. Raoul Granger has measured the temperature and vegetation density

of a boreal forest landscape using satellite remote sensing, the results are shown in

Fig. 9. It is evident that there is an inverse relationship between mid-summer

temperature and vegetation density. NHRl research in the Model Forest Programme

has shown that this is due to evapotranspiratioli .from the more heavily vegetated

surfaces. Evapotranspiration cools the surface, resulting in the temperature

differences. Hence, by clearcutting the boreal forest we are causing local climate

changes that are quite severe. To illustrate the severity of the climate change, Fig. 10

shows the surface temperature measured over one particular day in June, right around

this time when 10% of the commercial forest in Saskatchewan was burning. The

surface temperature goes up to about 30" C everywhere except the clearcut where it

goes up to 45" C, 15" higher than is "natural". 'This is a sort of local climate change

that can be created at the surface if we are not too careful. The disturbing thing for

forestry is that the wilting point for spruce seedlings is about 36" C and hence by

clearcutting they have created an envirol-~ment unfavourable for growing trees.

FRESHWATER QUALITY RESPONSE TO ATMOSPHERIC CHANGE

It is anticipated that as well as climatic change, there is a potential chemical change

that would results in increased availability of nitrogen to terrestrial and aquatic

ecosystems. In Canada, this could be due, in part, to warmer snow packs that will

sustain a greater deposition velocity of nitric acid vapour and hence a greater

fertilization of snow by nitrogen species, and as well as the simple increase of

anthropogenic nitrogen in the air, which has not been reduced like sulphate. As a

result, there may be productivity increases in certain ecosystems, perhaps

eutrophication in a few situations.

As snow melts earlier under a warmer climate, infiltration will increase because of the

development of unfrozen soils, reducing runoff to streams and transferring much of the

acid snow impact to land. Many ecosystem impact models suggest increased soil and

plant productivity because of increased retention of nitrogen on land, but also the

decline of certain forest species because of nitrogen saturation in certain ecosystems.

Nitrogen saturation is not an issue in Canada, but it might become one in the future.

Impacts to lakes and streams will din-~inish simply because many acids will not reach

the lakes and streams with a reduced spring snowmelt. Modelling results suggest a

reduction in the loadirlg of acids to lakes and the reduced episodic releases of

contaminants. For eastern Canada a 10-50% reduction in episodic acid shock is

predicted because of a less flashy spring snow melt. This prediction considers only

climate change without any change in atmospheric sulphur.

Other points to consider are an increased solution of basin geochemicals with a

decreased stream transport capability, leading to an increase in the solute load and

salinity of certain aquatic systems . Warming of and a higher evaporation from lakes

will increase eutrophication and solute levels in the Prairies, southern Ontario and

possibly the Arctic, while snow melt reduction to the Boreal forest lakes will reduce the

solute load and cause possible UV-B penetration increases. There will be a change in

wetland distribution as oligotrophic bogs may degrade into fens in certain regions.

WATER-ECOSYS'TEM FEEDBACKS TO ATMOSPHERIC CHANGE

Positive feedbacks to the system are decreased snow cover resulting in decreased

albedo, increased net radiation at the surface and hence surface warming; and

decreased forest/ wetland1 lake cover causing decreased evaporation and hence,

enhanced surface warming. That's bad. What's good? Increased productivity will

increase carbon fixation in the boreal forest and the Arctic in particular. This would

lower greenhouse gas levels beyond what they might be. Increased evaporation rrlight

increase cloud cover and increase precipitation which would result in a cooling of wetter

surfaces and higher productivity again. Figure 11 shows ,the relationst-lip between

primary productivity and actual evaporation or evapotranspiration. The higher the

productivity, the greater the evapotranspiration that can be generated. Figure 12

illustrates a positive feedback: the relationship between North American snow cover in

nill lions of square kilometers and temperature anomalies in winter. The wide line is the

North American maximum temperature in the snow regions showing a lo increase. That

results in about a million square kilometer reduction in North American snow cover: a

dramatic relationship we can already measl-ire, and we would expect that to continue.

RESEARCHANDDATANEEDS

What do we need from modelling right now? We require Canadian atmospheric models

with realistic sulphate aerosol, cloud, surface hydrology, snow and ice, ocean and

biophysical land cover interactions. That's quite a shopping list, but all are required if

we are to determine hydrological impacts in Canada. For water quantity, models must

be physically based and in agreement with field measurements. We need good field

measurements of temperature, precipitation, and radiation for the regions of Canada on

a seasonal basis. We must link continental scale hydrological models to atmospheric

models, and within these continental scale models, nest mesoscale and small scale

hydrological models so we can determine impacts at a local scale: then link these down

to hydrological process models that will tie into terrestrial and aquatic ecosystem

models.

For water quality, in addition to the these requirements, we need to know the

atmospheric deposition of contarrlinants and major geochemicals, and we need river,

lake and wetland hydroecology models, and terrestrial hydroecology models linked to

aquatic and a'tmospheric models.

Regarding data, for water quantity we need to focus on threatened areas: the West

and the North, the ecotones, the transitions between the present biomes, critical

habitats, sensitive environments, and small catchments that will show these changes

quickly. Where appropriate, remote sensing could be used as a cost-effective means to

gather vegetation, snow cover, soil moisture, and surface temperature data. For water

quality, with flow measurements we could also obtain nitrogen and phosphorous,

dissolved solids and salinity measurements: nitrogen export indicates the health of

the hydroecological system. For lakes, rivers and wetlands we require dissolved

oxygen, nutrients, water temperature and species composition (not only for commercial

species) data. We must monitor change in the distribution of aquatic and terrestrial

habitats and the biophysical characteristics of these habitats.

AMELIORATION OF FRESHWATER IMPACTS

How might we ameliorate the freshwater impacts? We can preserve forests in

threatened areas: for exarr~ple, the mixed wood of Western Canada. We need

integrated land use and stream flow regulation management in threatened basins. An

example here is the drying of the Peace Athabasca delta, which is regulated by a dam

and undergoing climatic change. The dam provides a chance to restore the natural

flow. We can also promote carbon-fixing land uses; water conservation measures in

areas that will experience shortages, perhaps urban areas; irrigation where supply and

salinization warrant according to our best predictions; snow management on the

Prairies, which might be one way to decrease the albedo and to augment supplies

without irrigation or major dams; and this niight be controversial but we may have to

introduce plant species or genetically engineer plant species that manage water more

effectively to promote negative feedback.

CONCLUSIONS

To conclude, the impacts of what are goiog to be unprecedented, at least in recent

time, atmospheric change impacts on water resources of Canada, their prediction and

mitigation, are going to present quite a challenge to our government agencies and we

must coordinate internally, internationally and interprovincially. The issues - local,

national and international - cannot be separated. The impacts are multi-scale, and

there are severe and occasionally catastrophic local implications for the ecological

health of Canada and the livelihood of Canadians. Strategies for prediction and

mitigation, therefore, must be national with regional emphasis on vulnerable areas. In

closing, I would like to extend thanks to all the people I was able to collaborate with to

gather this material.

MACKENZIE RIVER BASIN CCC - GCM GRID POINTS

I GCM water excess - Recorded SLURP wfth GCM data

0 1 I I I I I I I I I I I

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Snow Accumulation (mm)

A Anticipated Mean Temperature Change due to MADLEV CENTRE FOR CLIMATE PREDICTI[N W RESEARCH Greenhouse Gases and Sulphate Loading

with Ocean Coupling, 2040

-6 -4 -2 0 2 4 6

Temperature Change OC

Environment Environnement PSI Canada Canada

Discharge (m3. s-') Discharge (m3. s-')

Aspen

Pine

Pine lantation *

Clearcut

Fresh clearcut

Normalized Index V

Surface Temperature, "C

NEGATIVE FEEDBACK: Primary Productivity - Actual Evaporation

2 2.5 3 Actual Evaporation {Evapotranspiration)

( w l )

POSITIVE FEEDBACK: Snow Cover - Air Temperature

- North American Mean Maximum Temperature - snow regions

- Northern Hemisphere Mean

~emperatur6 Anomaly (C)