ipcc far wg ii chapter 04

8/13/2019 Ipcc Far Wg II Chapter 04

1/46

Chapter 4

Hydrology and water resources

Co-chairmen: I. Shiklomanov, H. Lins, E. Stakhiv and K Mostefa-Kara

Lead authors: H. Lins, I. Shiklomanov and E. Stakhiv

Contributors: M. Ayers; M. Beran; F. Bultot; L. da Cunha; G. Demar6e; B.

Foxworthy; G. Griffiths; K. Hanaki; J. Kelman; R. Lawford; H.Liebscher; C. Liu; P. Mosley; B.Stewart; R. Street; T. Yamada


2/46


3/46

Contents

1 Introduction 4-12 Cli mate scenarios 4-1

3 Some physical aspects of hydrology and water resources 4-13.1 Methodological approaches 4-13.2 Changes in annual and seasonal conditions 4-23.3 Water demand 4-33.4 Water balance and lake levels 4-43.5 Other hydrologic characteristics 4-4

4 Hydrologic and water resource changes in large regions and countries 4-54.1 Nor th Ameri ca 4-5

4.1.1 Uni te d States 4-54.1.2 Canada 4-6

4.2 Br azi l 4-74.3 Western Europe and Scandinavia 4-84.4 Un ion of Soviet Socialist Republ ics 4-114.5 Northe rn China 4-114.6 Japan 4-124.7 New Zealand 4-13

5 Case studies of effects in cri tical or sensitive environments 4-155.1 Large water bodies 4-15

5.1.1 The Great Lakes Bas in (US/ Canada ) 4-155.1.2 The Caspian Sea 4-16

5.2 The ari d and semi-arid zones of Nort h Afri ca , including the Sahel 4-175.3 Water conditions in crit ical agricul tural regions 4-18

5.3.1 The South Platt e River 4-185.3.2 The Murray- Darl ing Basi n (Australi a) 4-18

5.4 Wate r conditions in intensively urbanised areas - the Delaware RiverBasin (U SA ) 4-21

5.5 Regions of snowmelt-generated runoff: The Sacramento-San JoaquinRiver Basin (U SA ) 4-22

6 Conclus ions 4-24

TablesTable 4.1. Est imat ed hydrologic impacts of varying climatic change scenarios for

Canada (Marta, 1989) 4-7Table 4.2. Estimated percentage changes in mean annual runoff at three sites in the

UK in response to changes in temperature and precipitation, based on theTur c formula (Beran and Arne ll , 1989) 4-10

Table 4.3. Potential changes in seasonal runoff in response to a temperaturewarming of 1 C for selected rivers in the US SR 4-12

Table 4.4. Est imat ed percentage changes in Great Lakes net basin supplies betweencurrent climatic conditions (1 x C 0 2 or the BA S E case) and doubled C 0 2conditions as simulated by the GISS, G F D L and O S U GC Ms (Croley, in

press) 4-16Table 4.5. Changes in components of the water balance of the Caspian Sea in

response to global warming 4-17Table 4.6. Simulated percent of time in which the Delaware River Bas in is in a

drought warning or drought emergency condition for prescribed temperature and precipitat ion scenarios (Ayers et a l , 1990) 4-22

4-ii


4/46

Figures

Figure 4.1. Dependence of consumptive water losses, over large continental areas, onthe aridity index. 1 - Africa, 2 - Europe and Australia, 3 - North andSouth Amer ica, and Asia 4-26

Figure 4.2. Me an annual water supply (in bill ions of gallons per day) under currentand hypothetical climatic conditions for the major water resource regionsof the US 4-27

Figure 4.3. Percentage change in mean annual runoff by varying precipitation (upper)and potential vapot ranspira tion (lower) (Schaake, 1990) 4-28

Figure 4.4. Change in annual runoff (in centimetres) for the USSR as estimated fora warming of 1 C 4-29

Figure 4.5. Estimated percentage change in annual runoff under climate changescenario SI (2030-2050) in New Zealand (Griff iths, 1988) 4-30

Figure 4.6. The Great Lakes drainage area (Smith and Tirpak, 1989) 4-31Figure 4.7. GCM- es timated average seasonal and annual changes in temperature

(upper) and precipitation (lower) for the grid cells encompassing theGreat Lakes (2 x C 0 2 minus 1 x C 0 2 ) (Smith and Tirpak, 1989) 4-32

Figure 4.8. Comparison of actual levels of Lake Er ie for the period 1900-80 withestimated levels for the same period generated by superimposing onto theactual conditions a warmer climate and a warmer climate with futurelevels of consumptive use (Sanderson, 1989) 4-33

Figure 4.9. Estimated levels of the Caspian Sea, 1989-2020 under varying climatechange scenarios: 1 - stationary climate with man's impact, 2 -model-based anthropogenic change in climate including man's impact, 3- map-based anthropogenic change in climate including man's impact, 46- observed lake level variations before 1988 4-34

Figure 4.10. Plo t of normalised annual average rainfall (sol id line) and 3-year movingaverage (dashed Une) for the town of Saint Louis, Senegal (upper,1932-1982) and Oran es Senia, Alge ri a (lower, 1927-1987) 4-35

Figure 4.11. Drainage area of the South Platte Rive r basin (US Ar my Corps ofEngineers, 1977) 4-36

Figure 4.12. GCM- estimated average seasonal and annual changes in temperature(upper) and precipitation (lower) for the grid ceUs encompassing theGreat Plains (2 x C 0 2 minus 1 x C 0 2 ) (Smith and Tirpak, 1989) 4-37

Figure 4.13. The drainage area of the Murray-Dar ling basin in southeastern Austra lia(Stewart, 1989) 4-38

Figure 4.14. The Central Valley (shaded) and Central Valley Basin of Cali forn ia.Symbols refer to locations of the GISS , G F D L and OS U G C M gridpoints(Smith and Tirpak, 1989) 4-39

References 4-40

4-iii


5/46

Hydrology and water resources

1 Introduction

Changes in climatic conditions due to increasing

atmospheric concentrations of radiatively active tracegases wi l l probably alter land and water resources,their distribution in space and time, the hydrologiccycle of water bodies, water quality, and watersupply systems and requirements for water resourcesin different regions. Quantitative estimates of thehydrologic effects of climate change are essential forunderstanding and solving potential water resource

problems associated with domestic water use, industry, power generation, agriculture, transportation,future water resources systems planning and management, and protection of the natural environment.

Climate change can be expected to lead to changesin soil moisture and water resources. The mostimportant climate variable that may change isregional precipitation, which cannot be predictedwell. Water supply and use in semi-arid lands arevery sensitive to small changes in precipitation andevapotranspiration by vegetation, because the fraction of precipitation that runs off or percolates togroundwater is small. Increased heat wi l l lead tomore evapotranspiration, but the increase is expected to be partly offset by reduced plant water use ina COj-enriched atmosphere.

Higher temperatures may also have an impact in thetransitional winter snow zones. Mo re winter precipitation would be in the form of rain instead of snow,thereby increasing winter season runoff and decreasing spring and summer snowmelt flows. Where theadditional winter runoff cannot be stored because offlood control considerations or lack of adequatestorage, a loss in usable supply would be the result.

This chapter presents estimates of the influence ofclimatic change on hydrologic and water resourceconditions in various countries and regions.

2 Climate scenarios

Forecasts of changes in climatic conditions fordifferent regions and periods of time are required toestimate the hydrologic effects of increasing tracegas concentrations in the atmosphere. A i r temperature, precipitation, cloud cover or insolation, windspeed and humidity are the most important conditions. However, rel iable forecasts of regionalclimate change are unavailable. In their absence,

various approaches to the development of scenariosof future climatic conditions are used, including (1)

hypothetical (or prescribed) scenarios, (2) scenariosobtained from atmospheric general circulationmodels (G CM) , and (3) scenarios based on

historical and palaeoclimatic reconstructions.

The first approach is to prescribe climatic changesfor various regions or river basins in a simplifiedmanner. As a rule, such scenarios specify air tem

perature increases from 0.5 C to 4.0 C and precipitation changes (increase or decrease) in the range of10% to 25%. Some authors also prescribe hypothetical changes in evaporation.

The second approach is to obtain scenarios directlyfrom GC Ms in which the atmospheric concentration

of C 0 2 is doubled (2 x C0 2 ) . A problem with usingG C Ms is that the simulations for the same regions by different climate models may yie ld different andsometimes opposite results, especially for precipitation changes (Gleick, 1988) .The third approach is todevelop future climate analogs based on climaticreconstructions of past warm epochs, when atmos

pheric C 0 2 was above the present value. Forinstance, Soviet climatologists show in their studies(Budyko and Izrael, 1987) that the so-calledHolocene opt imum (about 5-6 K A ) can be considered as a 1C global warming analog (about inthe year 2000-2005); the last interglacial epoch (theMik ul ino, 125 KA ) can serve as an analog of theclimatic conditions with a 2C warming in the year2020-2025; and the Pliocene climatic optimum thattook place a few million years ago, with mean airtemperatures 3C-4C above present, can be ananalog of climatic conditions for the more distantfuture years 2040-2050.

There are difficulties and limitations in the application of palaeoclimatic reconstructions owing to theuncertainty of the climatic conditions of the remote

past and to the lack of reliable palaeoclimatic datafor many regions and countries. There is also acontinuing discussion of the relative merits of thethree approaches. In the absence of reliable predictions, the question of which scenario is best remainsunanswered.

3 Some physical aspects of hydrology and water resources

3.1 Methodological approaches

During the past 10 years hydrologists from many

countries have extensively studied hydrologic consequences of future anthropogenic climate change.

4-1


6/46

These studies are based on various methods that can be united into the following groups:

(i) Analysis of long-term variations in runoff andmeteorological elements over past periods. Thismethod can be approached in two ways. The first

consists of statistical analyses of the relations between runoff, air temperature, and precipitation.This approach has been used by Stockton andBoggess (1979) and Revelle and Waggoner (1983)for western regions of the US and the ColoradoRiver basin, and by Polyak and Speranskaya (1987)for annual river runoff in the USSR. The secondway is to study the hydrologic consequences of past

periods of very warm or cold, wet or dry condit ions.Such analyses have been carried out by Schwarz(1977) and Glantz (1988) for the US, by Liu (1989)for northern China, and by Demaree and Nicolis(1990) for the Sahelian region.

In the US, the first major assessment ofclimate-runoff relations was conducted by Langbeinet al. (1949), who produced a set of rainfall-runoffcurves widely used in crude impact studies. Theserelations were tested and updated by Karl andRiebsame (1989), based on an analysis of actualclimate fluctuations over 90 undisturbed drainages.They found that runoff was less sensitive to temperature changes than suggested by Langbein and hiscolleagues.

(ii) Use of water balance methods over a long period of time. In this approach the main task is toestimate future total evaporation. These methodshave been used by Glantz and Wigley (1987) for theUS, Babkin (Shiklomanov, 1988), Vinnikov et al.(1989) for the USSR, Griffiths (1989) for NewZealand.

(iii) Use of atmospheric G C Ms . In this approach,G C M s with prescribed increases in the concentrations of trace gases in the atmosphere (usually2 x C0 2 ) are used to obtain direct estimates ofchanges in the climatic and hydrologic characteristicsfor large regions. For example, possible changes inrunoff, soil moisture, and evaporation have beenestimated for the US and Canada using G C Ms( USEPA, 1984; Sanderson and Wong, 1987; andSingh, 1987).

(iv) Use of deterministic hydrologic models. In thisapproach, rainfall-runoff models for river basins areemployed with climatological data sets, includingG C M outputs, to determine changes in hydrologiccondit ions. Thi s approach was used by many

authors for the basins located in various hydroclima-tic environments (Nemec and Schaake, 1982; Gleick,1986, 1987; Mather and Feddema, 1986; Cohen,

1986; Flas hka et a l , 1987; Bultot et a l , 1988;Kuchment et al , 1989; Shiklomanov, 1989a; andCroley, in press).

The first and second methods have been widelyapplied to estimate changes in water resources over

large areas because a relatively small amount ofinitial data is required, usually annual runoff, precipitation, and air temperature. Cau tio n should beexercised in extrapolating regression relationshipsover past years to future periods. One cannotassume that a past interannual pattern of meteorological factors will be repeated in the future. It isalso true that for the same annual precipitation andtemperature, annual runoff can widely vary, depending on the distribution of the meteorological variables within months and seasons.

The results obtained in hydrologic simulations basedon different G C M s are inconsistent for certainimportant hydrologic conditions and regions. Thiscan be attributed to the low resolution of the currentgeneration of GCMs , and to their simplified description of hydrologic processes. Nevertheless, theapproach is very promising and studies of this typeshould be continued.

Deterministic hydrologic models have some desirable properties. They allow explicit study of causalrelations in the climate-water resources system forestimating the sensitivity of river basins to changingclimatic condit ions. In addi tion , when regionalclimatic forecasts are available, possible runoffchanges in different hydroclimatic environments may

be simulated for water planning and management.

Perhaps the most comprehensive assessment of theeffect of climate change on water resources was arecent report that focused on the US by theAmerican Association for the Advancement ofScience Panel on Climatic Variability, ClimateChange and the Planning and Management of USWater Resources (Waggoner, 1990). Thi s documentincludes 18 papers by more than two dozen authorsand encompasses methods and issues ranging fromclimate forecasting, the translation of climaticchange information into hydrologic consequences,vulnerability of water systems, impacts and responses, to future water use and decision making underclimate uncertainty. Its contents cut across most ofthe above stated methods.

3.2 Changes in annual and seasonal conditions

Since the late 1970s, changes in annual and seasonalrunoff have been extensively investigated and described in many publications. This is because annual

4-2


7/46

and seasonal river runoff are very important for providing adequate water supplies to meet thedemands of most regions. Quantitative estimateshave been obtained for many regions of the US, forthe USSR and New Zealand, for some regions ofCanada, river basins in England, Wales, Belgium

and other specific regions (see Sections 4 and 5).

Investigations highlight the great sensitivity of riverwatersheds to even small changes in climatic conditions. Watersheds located in arid and semi-aridregions are especially sensitive because annual runoffis highly variable. For watersheds where snowmeltis an important source of runoff, annual runoff andits seasonal distribution is vulnerable both to changing air temperature and to changing precipitation.In middle latitudes of the Northern Hemisphere witha 1C or 2C warming, the winter runoff is expectedto increase drastically and spring high water to belower owing to earlier snowmelt (see Section 4).

Many estimates of runoff change due to globalclimate warming, including those given in Sections 4and 5 do not consider the possible direct influenceof increased C 0 2 on vapotranspiration (with increased C 0 2 concentration vapotranspirationusually decreases). Thi s phenomenon is consideredin Idso and Brazel (1984) for five US river basins aswell as in Aston (1987) for river basins in Australia.In these studies, the results were opposite to thoseobtained by other researchers, namely, a doubling ofC0 2 concentration brought about a 40% to 60%increase in annual runoff for US rivers in question,and that for Australian rivers by 60% to 80%.Wigley and Jones (1985) and Palutikof (1987)

pointed out to the great role of taking into accountthe direct C 0 2 effects on vapotranspiration andtotal evaporation, which is ignored by many researchers. These findings are controversial and needfurther investigation as total evaporation from landis thought to be determined primarily by energyfactors.

An important hydrologie consequence of globalwarming is potential changes in runoff extremes,

both high and low. Changing high flow extremesraise the problems of runoff control , development offlood control works, and design of hydrologie structures. Changing low flow extremes may require areassessment of water storage requirements and ofwater allocation schedules, in particular in thelimiting periods of the year.

Estimation of extreme river runoff conditions inconjunction with anthropogenic climate change is a

very important problem in hydrologie engineeringcalculations, since the design of hydrologie structuresis based on the concept of stationarity of hydro-

meteorological conditions. It is assumed that theobservational data over past decades reflect hydro-meteorological conditions during the design life ofthe project. Changes in extremes wi ll modify statistical calculations to estimate sizes of constructionsand their conveying capacity and to maintain stabili ty

during disastrous natural events.

Since extreme meteorological events cannot beexplicitly forecast by G C M simulations, nor by using

palaeoclimatic analogs, in practice the quantitativeregional estimates of possible changes in runoffextremes are as yet unavailable. Nevertheless, thereare plenty of qualitative estimates, based usually onan assumption about the proportionality of changesin runoff extremes to those in annual, seasonal ormonthly runoff (see Section 5). More reliableconclusions about the runoff extremes in variousregions of the world could be drawn if detailedquantitative estimates of changes in meteorologicalcharacteristics over shorter periods of time wereavailable.

3.3 Water demand

Under global climate wanning, changes in demandfor water resources should be expected in manyregions of the wor ld. Considerable transformationin the structure and the character of water consumption by different branches of industry and agriculture, and deepening conflicts and contradictions

between individual water users, are possible. Withthe same level of economic activities in a region, thewater requirements and even actual consumption aredetermined by the extent of total moisture deliveryto the region.

The relation between water consumption and available water is illustrated in Figure 4.1 which shows

per capita water consumption in the year 2000 anda dryness index R D / L P (where R D is the radiation

balance of a wet surface, P is the precipitat ion, andL is the latent heat of condensation). The greater

the dryness index, the higher the specific waterconsumption. The relations depicted in Figure 4.1are derived for large natural-economic regions of theworld by estimating future water consumption invarious countries up to the end of the century understable climatic conditions (Shiklomanov andMarkova, 1986; Shiklomanov, 1988, 1989b). Usingthese graphs and knowing possible changes inclimatic parameters with global warming, it is easyto estimate approximate changes in total waterdemands in different world regions. G C M simulations with 2 x C 0 2 in the atmosphere allow direct

estimations to be made for regional changes in thedryness index; such estimates have been made forthe US (Stakhiv and Lins, 1989).

4-3


8/46

The actual values of water consumption in variousregions will depend on climatic factors, the extent ofdevelopment of water-consuming sectors of nationaleconomies (primar ily on irrigated lands), and on thelimitations in water resources. Wi th significantfuture climate change there wi l l probably be problems in planning and locating future irrigated areas,industries with high water consumption, reservoirs,and problems of water supply for current waterusers. These changes may exacerbate problems inarid and semi-arid regions, where at present thereare difficulties with water supply and where conflictstake place between various water consumers andmanagers.

Much is known about how water resource systemsoperate under climatic stress, and extrapolations ofthis knowledge will help us anticipate impacts onmanagement and demand for water, and on protec

tion afforded from floods and droughts in a changingclimate (Fiering and Rogers, 1989; Stakhiv andHanchey, 1989).

A useful literature has developed on water systemsensitivity and adaptability to climate fluctuations.Fi ering (1982) and Hashimoto et al . (1982) la id outcriteria for assessing system sensitivity, includingreliabili ty, resiliency and vulnerability. Matalas andFieri ng (1977) described an optimising approach todesigning water systems with climate fluctuations inmind, and Novaky et al. (1985) offered a conceptual

model of water system impact and response andcreated a matrix for assessing system sensitivity toclimate change. Klemes (1985) synthesised thewater system sensitivity literature, suggesting that thekey concept is reliability - the long-term ability of asystem to meet demand despite fluctuations inclimate.

Peterson and Keller (1990) have evaluated theinfluence of temperature and precipitation changeson the irrigation requirements and the possibilitiesof developing future irrigated areas in arid regions

of the western US . They came to the conclus ionthat a warming could exert an enormous effect onirrigat ion development in the region. By increasingtemperature by 3C and decreasing precipitation by10%, cultivated areas in the western US coulddecrease by 30% and efforts would be required toimprove efficiency of water use and to develop newfreshwater supplies (Gleick, 1989).

Studies on the effects of effects of anthropogenicclimatic changes on irrigation water consumptionhave been conducted by the Food and AgricultureOrganization ( F AO) along with the UK' s Institute ofHydrology for the 3240 km 2 Malibamatsama Basinin Lesotho (South Africa) (Nemec, 1989; Institute of

Hydrology, 1988). Future climate change simulations for this region have been accomplished usinga G C M with doubling C 0 2 . The G C M data indicated a 6C increase in mean monthly temperature, a4-23% decrease in monthly precipitation fromDecember to May, and a 10-15% increase in monthly precipitation from June to November . Estimatesof changing evaporation and river runoff were basedon a water-balance model with a 10-day time increment. The research showed that with a doubling ofC 0 2 , changes in meteorological conditions in theBasin led to a 65% increase in water demands forirrigation; this could bring about the shrinkage ofirrigated areas from 37,500 ha at present to 20,000ha.

Regional estimates of potential changes in waterdemand with global warming are presented below inSections 4 and 5. The cases indicate that estimating

future water requirements and arrangements should be made by taking into account the peculiari ties ofeach region. Then the rel iab ili ty of such estimateswil l be primarily dependent on the accuracy andcomprehensiveness of predicting the changes inclimatic characteristics and hydrology with globalwarming.

3 .4 Water balance and lake levels

The hydrologic cycle of large lakes, in particular ofenclosed lakes, integrates climatic variability over

vast areas, including the lake basins and adjacentregions. Future global warming due to increasingC 0 2 in the atmosphere would lead to changes in thelake water balance components (precipitation,evaporation, inflow and outflow), their levels andheat budget. These changes are different for thedrainage basins and the enclosed lakes. Given

below in Section 5, as an example, are some ap proximate estimates of possible changes in the water balance and hydrologic cycle of the Nort h AmericanGreat Lakes and of the Caspian Sea, the largestenclosed lake in the world.

3.5 Other hydrologic characteristics

The hydrologic consequences of anthropogenicglobal warming are not limited to changes in riverrunoff and water balance values. Other consequences include changes in total water amount andlevels, erosion in river basins and riverbeds, andmodificat ions of turbidity and sediment load. Waterquality in many water bodies could deteriorate.Decreasing river runoff and lake level declines coulddecrease the possibility of dissolving pollutants andflushing processes.

4-4


9/46

Sea-level rise and runoff changes could lead toincreased flooding of low-lying coastal regions,increased shoreline erosion, changes in deltaic

processes, salinity changes in estuaries and rivers,and contamination of aquifers through saltwaterintrusion (see Chapter 6). Answers to questionsassociated with the hydrologic consequences of

global warming can be found only through innovative and geographically diverse studies that take intoaccount forecasts of both changing regional climaticconditions and water use.

4 Hydrologic and water resource changes in large regions and countries

4.1 North America

4.1.1 United States

In the US, in the 1970s and 1980s, on the basis ofhypothetical climate scenarios and river runoffdependence on meteorological factors, future annualrunoff changes were determined for the northeast(Schwarz, 1977) and western regions of the US(Stockton and Boggess, 1979), and the ColoradoRiver basin (Revelle and Waggoner, 1983). Asimilar study, using an annual water-balance modelwas conducted by Flaschka (1984) for the GreatBasin rivers. A deterministic hydrological modelwas used by Nemec and Schaake (1982) to study twoidealised basins in arid and humid regions. Theresearch demonstrated a strong sensitivity of river

basins (in particular, in arid and semi-arid regions)even to small changes in climatic conditions.

For example, increasing the annual air temperature by 1C or 2 C and decreasing precipitation by 10%decreases the annual river runoff in regions withrelatively low precipitat ion by 40% to 70%. Subsequently, these results were supported by calculationsmade by Stockton et al. (1989) who used moreadvanced models simulating runoff formation inriver basins. These authors predict changes in waterresources for diverse water supply regions of thecountry using two hypothetical scenarios: warm anddry (temperature change of +2C, precipitationchange of -10%) and cold and wet (-2C and+ 10%). The results are given in Figure 4.2. Theyshow that for many regions of the US, water resources are expected to decrease by 1.5 to 2 timesunder the warm and dry scenario.

Schaake (1990) investigated how the sensitivity ofrunoff to given scenarios of climatic change wouldvary spatially with different climate conditions overthe southeastern US . A simple, monthly, non-linearwater-balance model was developed using a basin in

China and another in Oklahoma, US . Constantvalues were assumed for the five parameters of themodel over the entire region. The model and

parameter values were tested using more than 2000station-years of data from 52 basins in the region.

None of the test data were used to develop orcalibrate the model. Simulated mean annual runoff

for 49 of the 52 basins fell within error bounds of100%. The climate scenarios were for changes of10% in potential vapotranspiration and 10% in

precipitat ion. The study concluded that the hydro-logic processes amplify the effects of such changeson runoff.

A measure of this amplification is elasticity; the ratioof the relative change in a runoff variable to therelative change in a climate variable. To illustratethis point, maps showing the percentage change inmean annual runoff to a 10% change in precipitation

and potential vapotranspiration are presented inFigure 4.3. Elast icit ies as high as 5 were found inthe southeastern US. The study concluded that: dryclimates are more sensitive to change than humidclimates; elasticity to precipitation change is greaterthan to vapot ranspirati on change; low flows will bemore affected than high flows; reservoir yields will

be affected, but the elast icity of reservoir yield is lessthan the elasticity of the mean flow; and, becausewater quality problems tend to be coupled with lowflow conditions, water quality effects may prove to

be among the most significant, especially in aridareas.In 1984, the US Environmental ProtectionAgency (USEPA, 1984) used output from GCMsdirectly for analysing changing precipitation, soilmoisture, and runoff under doubled atmosphericC 0 2 conditions. The results indicated a significantincrease (20%-60%) in river runoff in the northwestern US, and a decrease (26%) in the centralregion. Manabe and Wetherald (1986) obtainedsimilar results using a G C M to estimate soil moisture changes in the mid-continent region of the US.In contrast, however, simulations with other G C Ms

produced opposite results for the same region(Schlesinger and Mitchell, 1987; Mitchell and

Warri low, 1987).

Developing reliable scenarios of potential climaticchanges for regional impact assessment is a major

problem. The use of output from the currentgeneration of GC Ms to estimate regional hydrologicimpacts of climatic change is highly suspect becauseof the coarse resolution of the models and the grosssimplifications used in the parameterisations of theterrestrial hydrology (Gleick, 1989; W M O , 1987).Since soil moisture and river runoff processes arenot well specified in GCMs , a more fruitful ap

proach for estimating hydrologic impacts is thecoupling of G C Ms and deterministic hydrologic

4-5


10/46

models. In this case, the G C M output (typicallytemperature and precipitation) becomes the inputdata for more comprehensive regional hydrologicmodels.

Gleick (1986,1987) was one of the first to apply thisapproach for estimating global warming effects on

the seasonal runoff of the Sacramento River(Cal ifornia). As the basis, Gle ick used climatechange scenarios (monthly air temperature and

precipitation) developed from three G CMs with2 x C 0 2 in the atmosphere. Simulation of hydro-logic impacts was carried out with a water-balance

basin model with a monthly time increment. Thesesimulations showed increased winter runoff from16% to 81% and decreased summer runoff from30% to 68%. These results can be explained by adrastic change in snowfall and snowmelt conditionsassociated with a rising air temperature. A similarapproach has been used by other authors. Withdoubling atmospheric C0 2 , Great Lakes basin runoffdecreased by 12-13% (Sanderson and Wong, 1987),Great Lakes net basin supplies decreased by 2-113%(Croley, in press), and the flow of large rivers in theProvince of Quebec, Canada, increased by 7-20%(Singh, 1987).

Of the four GC Ms most widely used in the Unit edStates, none is capable of simulating observedannual or seasonal precipitation with accuracy on aregional basis. The models also differ in the valuesof forecasted warming for US regions, particularlyfor the summer months. Nevertheless, in analysingthe conclusions drawn by different researchers, thefollowing regional tendencies are presented as

potential hydrologic effects of global warming:

Pacific Northwest: some increase in annual runoffand floods;

California: a considerable increase in winter anddecrease in summer runoff with insignificant rise inannual runoff;

Colorado and Rio Grande River Basins, the Great Basin: decreasing runoff;

Great Lakes Basin: decreasing runoff and increasingevaporation;

Great Plains, northern and southeastern states:uncertain changes in water resources.

The above inferences were also obtained by Sovietclimatologists (Budyko and Izrael, 1987) using

palaeoclimatic analogs. These results show a considerable decrease in moisture south of 55N and a

noticeable increase in precipitation in northernregions of Nor th Amer ica .

4.1.2 Canada

In recent years a number of studies have investigated the potential hydrologic and water resources

effects of a climatic warming for Canada over thecoming decades. Most of these studies have basedtheir impacts on climatic simulations produced bythe GISS, G F D L , and U K M O GC M s (Marta, 1989).A summary of the results of these studies is presented in Table 4.1. Most of the investigations focusedon one of three distinct areas in the south-central

part of the nation: the Great Lakes, James Bay, andthe Saskatchewan River sub-basin. A nationwidestudy of broad Canadian regions has also beencompleted. In the Great Lakes region, G C M-simulated temperatures range from +3.1C -+ 4.8C. The simulations for precipi tation are muchmore variable, ranging from -3% to +8%. Generally speaking, the mix of this range of conditions could portend a decline in the hydrologic and waterresource base or the region. Unde r these temperature and precipitation conditions, runoff to the lakesis estimated to decrease between 8% and 11%, lakelevels could drop between 21 cm and 59 cm, andwater supplies could fall from 18% to 21%.

Farther north, in the James Bay region, the waterresource picture appears more favourable.Model-simulated temperature increases for this areafall in the range of 3.5C to 4.7 C, with precipitationestimated more narrowly to increase from between15% to 17.5%. This range of climat ic conditionsfavours increases in runoff ranging from 8% to 16%with concomitant increases in water supply of 9% to22%. Moving west onto the Canadian Prairie,impacts on the hydrology of the Saskatchewan Riversub-basin appear to be strongly model-dependent.Both the GISS and G F D L simulations indicateimpacts of fairly large magnitude, especially in termsof changes in the timing, location, duration andextent of runoff as wel l as soil moisture. However,

the GISS simulations indicate increasing streamflowsand water supplies for the region while the GFDLsimulations indicate decreases. In terms of net basinsupply, incorporating projected consumptive use, theGISS simulation produced a range of increasesvarying from 29% to 40%. On the other hand, theG F D L simulation produced declines in net basinsupply ranging from 27% to 70% (Cohen et al,1989).

Water in the Saskatchewan River sub-basin isutilised for agriculture, hydroelectric power produc

tion, cooling at thermal power plants, recreation andindustrial processing. Agr iculture is a major con

4-6


11/46

Table 4 .1 Estimated hydrologic impacts of varying climatic change scenarios for Canada (Marta, 1989).

Author

Southam andDumont(1985)

Cohen(1986; 1987)

Singh (1988)

Cohen et al.(1989)

Haas andMarta (1988)

Ripley(1987)

Region

Great Lakes

Great Lakes

James Bay

SaskatchewanRiver

S. SaskatchewanRiver (to LakeDiefcnbaker)

ArcticAtlantic

GulfHudsonPacific

Canada

Drainagearea

(km J x 104)

76.0(32 is lake

surface)

76.0

17.5(18 is lakeand reservoir

surface)

36.4

15.7

331.0151.03.0

356.094.0

935.0

Est .

(TC)

+4 .5

+3.1 to3.7

+4.3 to4.8

+3 .5+4 .7

+2 .5+4 .4

+2 .5+4.4

+2 to 3.5+2 to 3.5+2.5 to 3+2 to 4.5+2 to 3

+2 to 3

climatic changes

(P ) (GCM )

-3 to + 8

+0 .8+6 .5

+ 17.5+ 15

+ 18+ 19

+ 18+ 19

+54+ 11+23+ 17+23

+23

GISS/84

GFDL/80GISS/84

GFDL/80GISS/84

GFDL/80GISS/84

GFDL/80GISS/84

UKMO/83UKMO/83UKMO/83UKMO/83UKMO/83

UKMO/83

Est. hydrologicalchanges

(Runoff- >

-8.2-10.9

+ 10.9+ 15.7

-58+ 3 7

-20+ 2 8

+95+ 10

+235+20+20

+32

Other

Levels (cm)-21 to -59Flows ( )+2 to -13

Supply ( )-18.4

Supply ( )-20.8

Supply ( )+ 13.9

Supply ( )+ 16.1

Supply ( )-65Supply ( )

+38

sumer of water, in which return flows arc considerably less than withdraw als from surface supplies.

Certain others, such as hydroelectric production andrecreation, arc in-strcam users that do not 'consume'water, that is, their use of water docs not result in a'loss' from surface s upp lies. If the status quo is tobe maintained, results from the study of this watershed indicate that for the GISS scenarios, an increase in agricultural demand for water would notsignificantly affect the availability of water for otherusers, since the net basin supply is estimated to berelatively high. For the G FD L scenarios, however,the low estimated net basin supply could seriouslycom prom ise all users. Such a situation wouldnecessitate a major policy response, such as theestablishment of a regional water allocation schemefor the sub-basin, involving government agencies,user groups and the public. Should there be increases in population, reservoir storage, industrial oragricultural development, such a scheme might notbe required to cope with the conditions estimatedunder the GISS scenario, but any increases in water

demand would only make the situation more criticalgiven the conditions estimated by the GFDL GCM.

Looking very broadly at the impact of climaticwarming on the water resources of Canada in theaggregate, the general outlook may be somewhatfavourable. With temp era ture rises generally in the2C to 4C range and precipitation generally increasing across the country from 1 1 to 54 ,runoff could be expected to increase in all majorregions by 10 to 23 5 (Ripley, 1987).

4.2 Brazil

The Brazilian government and scientific communityonly recently became concerned over the prospect ofa greenhouse-gas-induced climate warming and itsassociated consequences on the water resources ofSouth Am erica. Attention has been given to thepossible effects of deforestation of Amazonia onregional hydrology and atmospheric carbon dioxide.Very recently, the Brazilian government began toexert some control over the deforestation process.

4 7


12/46

Although this is a positive step toward decreasingthe atmospheric build-up of C0 2 , it is also necessaryto point out that fossil fuel combustion is about 3 to5 times more important than deforestation as a netcontributor to global warming (Moore and Bohn,1987).

As yet, there are no indications of a large scale

anthropogenically-induced change on precipitation inthe Brazilian Amazon (Sternberg, 1987). Thisstability is probably due to the relatively small

percentage of the total rainforest area that has beendeforested (7%). Nobre et al. (1989) simulated theclimatic effects of total deforestation using a modeldeveloped by Sellers et al. (1986). This modelincorporates the interactions between vegetation andthe general circulat ion of the atmosphere. Themodel results indicated that the ground temperaturein the Amazon region would increase between1-3C, vapot ranspira tion would decrease 20-40%,

and precipitation would decrease 20-30%.

Such a temperature increase, whether from deforestation, fossil fuel emissions, or both, will probablyintensify and concentrate tropical convective rainfalland, as a result, promote increased flooding anderosion, and decreased soil moisture. Thi s suggeststhat the firm yield from existing reservoirs coulddecrease and that the probability of dam overtoppingowing to extreme floods may increase.

4.3 Western Europe and Scandinavia

Effects of climate change on water resources have been estimated for Western Europe (primarily forthe northern regions) by using various scenarios andapproaches. Western Europe was similar to Nor thAmerica in that river basins were very sensitive tosmall variations in climatic conditions, especially

precipitation (Palutikof, 1987; Novaky, 1985; Beran,1986; and Verhoog, 1987).

Studies by Schnell (1984) and Beran (1986) aresignificant for forecasting future water resources.Schnell estimated the consequences of C 0 2 doubling

for river runoff in the countries of the EuropeanEconomic Community (E EC) using temperature and precipitation derived from the UK MeteorologicalOffice ( U KM O ) G C M . The runoff values obtained

by Schnell using waterbalance methods show possible significant increases in the north of Europe(the U K , The Netherlands, and Belgium) anddecreases of similar significance in central andsouthern Europe.

A very comprehensive assessment of the potentialwater resources response to climate change for over

12 countries within the E E C has been conducted by

da Cunha (1989). Detail ed informat ion about waterresources and water availability for each country,coupled with the scenarios of climatic change simulated by GC M s for the E E C region , indicate thatman-induced climatic change is likely to influencewater resources differentially in different parts of theregion. If the simulated G C M results turn out to becorrect, then average precipitation values in the

southern part of Spain, Portugal, and Greece can beexpected to decrease, while increases can be expected throughout the rest of the E E C region. Themodel results also indicate a tendency for reduced

prec ipitation variabil ity in some of the northern parts of the region during the entire year, and forincreased variability in Greece and southern Italyduring the summer . Such changes in variabilitycould mean that higher flood peaks are possible in

parts of the U K , Germany, Denmark, and The Netherlands, while more severe droughts are possible in Greece and southern Italy. Thus, it is

impl ied that a large part of the E E C region would benefit from what appear to be potent ially favourable changes in climate, while a smaller part in thesouth would suffer unfavourable impacts.

The impacts of climate change on water resourceshave been classified by da Cunha into three groups:economic, social, and environmental . Economicimpacts are often the most relevant. The changes inthe global amount of water resources available andin the space and time distribution of these resourcesmay lead to water shortages and cause water priceincreases. Changes in water quality are also likely asa result of climatic change. Temperature increaseswould also induce an increase in the demand forwater, thus contributing to a worsening situationresult ing from a decrease in water availabil ity. Thenegative impacts are expected to be particularlyserious on cultivated land areas. Examples of thesenegative impacts include crop losses or shortages,replacement of present agricultural species by otherspecies more resistant to water shortage but lessuseful, increased activity of insects and predators,and the abandonment of agricultural land. Suchimpacts may also affect livestock by reducing herd

sizes and increasing disease and problems of vectorcont rol. Other economic impacts relate to losses intimber production, fisheries, and recreational businesses. Important indirect impacts include sedimentation processes, ie erosion, transport and depositionof sediments, as well as agriculture, hydraulic workoperations, and estuarine and coastal areas.

The social impacts of climatic change on waterresources are also important, since it is essential thatdecision makers provide guidelines on the strategiesto be adopted by water users to enable them to copewith climate change. In fact, the various types of

4-8


13/46

water users (farmers, industrialists or domesticconsumers) are not, by themselves, able to assumeindividually the most positive attitudes in order toreduce the negative impacts of climate change onwater resources. The possible climatic impacts onwater resources will force society to modify itsrelation with the environment by introducing a

number of adjustments in its behaviour. Emigra tionfrom the southern parts of the E E C unfavourablyaffected by climatic change would be one immediateform of social adjustment. Other possible socialconsequences include unemployment, deteriorationof conditions of hygiene and public health in theaffected regions, as well as public safety from forestfires.

Environmental impacts may also be relevant, rangingfrom serious ecological disruption to less serious andmore localised effects. Examples of serious disrup

tions are desertification, which is a possibility in thesouthern region, and sea-level rise with possibleeffects on flooding in coastal areas, destruction ofengineering works and structures, and modificationof coastal and estuarine environments. Examples ofmore localised effects include damage to wildlifehabitat owing to water shortages, degradation ofvisual landscape quality, increased salinity of soil andgroundwater, soil, erosion, forest fires, plant diseases, and insect and predator plagues.

Bultot et al. (1988, 1989) made the most detailed

estimations of river runoff changes for Belgium witha doubling of the C 0 2 concentration. The authorsused the monthly values of climatic conditionsobtained from G C Ms predicting a 2.9 C annual airtemperature increase and a 55 cm precipitation rise.To assess the hydrologic consequences for the threeriver basins with different conditions of runoffformation, a conceptual hydrologic model withdiurnal time increments was used. The calculationsshowed increasing annual runoff (3-10%) andchanges in the seasonal runoff pattern. The winterand spring runoff increased by 10-14%, while thesummer runoff remained unchanged of decreasedfrom 1% to 15%).

To date there have been no studies in the UK usinghydrologic models to estimate the effect of climaticchange on annual runoff. Beran and Arnell suggest,however, that it may be possible to make somegeneralised statements using a simple water-balanceapproach where changes in mean annual runoff aredetermined from changes in mean annual rainfalland mean annual evaporation (Beran and Arnell,1989). Evaporation, however, is dependent onseveral climatic attributes which vary together in a

complex manner. Thus, a simple empirical formula

developed by Turc (1954), which operates on anannual time scale, was utilised by the British.

Mean annual runoff is estimated by subtractingevaporation from precipitation, for three combinations of mean annual temperature and precipitationfound in different parts of the UK (Table 4.2).

Parts a and b in the table represent catchments withsimilar rainfall, but different mean annual temperature and, for a given change in precipitation, thereis relatively little difference in effect on runoff

between the 'cool Scottish' and 'mil d English ' sites.The data in the table imply that if temperatureswere to increase by 3C, an increase in annual

precipi tation of 10% or more would lead to anincrease in annual runoff. For comparative pur

poses, the current scenario implies an increase inannual rainfall of 140 mm (14% on an initial valueof 1000 mm) if summers are wetter (as in Scotland

and the north of England), but only 80 mm (8%) ifsummers are drier . The estimated conditionsdepicted in part c of Table 4.2 are more representative of the drier climates of southern and easternEngland. Her e annual precipitation would have toincrease by over 10% before annual runoff wouldincrease. The scenario increase in precipitation of80 mm (10% of 800 mm) is just at this thresholdvalue.

These estimates imply that mean annual runoffwould probably increase over most of the UK in a

warmer world, with the greatest increase in thenorth. The actual increase in runoff is very sensitiveto the estimated increase in precipitation (more sothan to changes in vapot ranspi ra tion) . It is important to note that alterations in annual runoff due toclimatic change may approach changes in runoff dueto major land use change. Gross et al. (1989) useda conceptual rainfall-runoff model to show, forexample, that clearing the forest in a small basin inupland Wales could produce an increase in annualrunoff of nearly 30%.

Although the evidence is limited, it is possible thatthe model-simulated increased seasonality in rainfallwill be reflected in a stronger seasonal variation instreamflow throughout the U K . Note , even underthe 'wet summer' scenario, increases in summer

precipi tation are less than increases in winter precipitat ion. However , the effect wi ll depend stronglyon catchment geology, and it is possible that summerflows in catchments with high groundwater levels wil l

be maintained or even increased by higher winterand spring rainfalls.

Saelthun et al. (1990) have produced an initialassessment of the consequences of climatic changefor water resources in Norway. Using a climate

4-9


14/46

Tab le 4.2 Extimated percentage changes in mean annual runoff at three sites in Great Britain in response to changes intemperature and precipitation*

a) Init ial mean annual precipitation:Ini t ia l mean annual temperature :Init ial estimated mean annual runoff:

Temp, change (C)

12345

b) Init ial mean annual precipitation:Ini t ia l mean annual temperature :

-10%

-21.3-25.8-30.4-35.1-39.8

Init ial estimated mean annual runoff:

Temp, change (C)

12345

c) Init ial mean annual precipitation:Ini t ia l mean annual temperature :

-10%-23.0-28.1-33.3-38.5-43.7

Init ial estimated mean annual runoff:

Temp, change (C)

12345

* Based

-10%

-26.1-32.2-38.3-44.2-49.9

on the Turc

Change0%

-4.8-9.6

-14.7-19.8-25.0

in annual precipitation10%

12.37.11.7

-3.8-9.5

Change in annual

0%-5.6

-11.2-17.0-22.8-28.7

Change

0%

-7.0-14.0-20.9-27.8-34.6

10%12.66.40.2

-6.2-12.7

20%

29.724.218.512.7

6.6

precipitation

20%31.224.718.01114.1

in annual precipitation

10%

13.35.5

-2.3-10.1-17.8

formula (Beran and Arncll,

20%

34.426.017.48.90.3

, 1989)

30%

47.441.735.829.623.2

30%50.243.436.329.021.5

30%

56.347.338.128.819.4

1000 mm8 C

530 mm

40%

65.359.453.346.840.1

1000 mm10 C

479 mm

40%69.662.555.147.439.5

800 mm10 C

311 mm

40%

78.869.359.549.539.4

change scenario of a temperature increase of1.5 3.5C (with the higher temperatures prevailing

during winter and in upland areas) and a precipitation increase of 7-8%, the following hydrologic consequence s were posited. Ann ual runoff in mountainous areas and regions of high annual precipitation will likely experience a mod erate increase. Inlowland and forest inland basins, annual runoff maydecrease in response to increased evapotranspira-tion. Th e intensity of the spring flood could decrease in most basins, while winter runoff mayincrease substantially and summer runoff may decrease. Floods are likely to occur more frequentlyin autum n and winter. The period of snow coverwill probably be shortened by one to three months.The scenarios indicate that glaciers will decrease,particularly those in inland areas. The net meltcould be lower on glaciers near the west coast, whilethe glaciers on the coast of northern Norway willmost likely rema in unchan ged. The duration of ice

cover may decrease and it is possible that many ofthe larger lakes in southern Norway may not develop

ice covers in most winters.

Erosion and sediment transport is likely to increasesignificantly during winter unless the agriculturalpractice of leaving soils bare during winter ischanged. Soil loss could bec om e a very seriousproblem.

Finally, a positive potential outcome of globalwarming is that the hydroelectric power productionof Norway could increase by 2-3% under the as

sumed climate change scena rio. This is due in partto an increase in reservoir inflows, and in part toreductions in reservoirs spillovers. Th e seasonaldistribution of runoff may more closely match thatof energy consumption, thereby increasing firmpower yield.


15/46

4.4 Union of Soviet Socialist Republics

Quantitative estimates of potential climate changeeffects on rivers in the USSR have generally been

based on palaeoclimatic analogs of the past. The proxy-based estimates were first obtained at theState Hydrological Institute in 1978 under theguidance of M. I . Budyko (Budyko et al , 1978).

Subsequently, as the climatic predictions wererefined, estimates were made for changes in annualrunoff of rivers in the USSR for the years 1990,2000, and 2020. This procedure was repeated forother regions of the Nor thern Hemisphere. Theresults of these simulations are presented in thereport Anthropogenic Climatic Changes (Budykoand Izrael, 1987) and by Shiklomanov (1988). Thesimulations were based on meteorological-runoffrelationships, or on water-balance equations, developed over long time periods. The results of thesimulations were presented in the form of runoffchange isoclines and also as averages for the main

basins for the years 2000-2005. These averagechanges were in the range of + /-12% (Shiklomanov,1988). The values are only rough approximationsdue to the uncertainty associated with the initial airtemperature and precipitation data for certain river

basins.

The mean annual runoff changes over the USSR fora 1C global warming (expected to occur between2000-2010) have been estimated. This work was

based on refined predictions of changes in summerand winter temperature and annual precipitation for

the USSR (Budyko, 1988) developed at the StateHydrological Institute under the guidance of K. Ya .Vinnikov. The comprehensive data presented inFigure 4.4 suggest the following conclusions. Themost detrimental effect on resources can be expected in the south of the forest zones of the EuropeanUSSR and west Siberia. Annual runoff could fall to10-20% of normal (20-25 mm) . This effect is due toan increase in air temperature accompanied by nochange or even a slight decrease in precipitation.

In the southernmost steppe regions, owing to a

marked increase in precipitation, the annual runoffis most likely to increase by 10-20%. In thenorthern regions of the European USSR andSiberia, runoff may increase by 280-320 km 3 /yr or7%, which may compensate for future water consumption. Estimates for the more distant future areless certain; nevertheless, there is evidence to suggest that with a 2C global warming the annualrunoff will increase by 10-20% on all the large riversof the USSR. This increase wil l provide an additional 700-800 km 3 /yr of water for the wholecountry. Although these values are very approxi

mate, they cannot be ignored in the long-range

planning and management of water resources and protection of the natural environment.

Additional effects may be caused by global warmingon seasonal and inter-annual runoff patterns. Fo r anumber of river basins in the European USSR, theinfluence of climatic change on the hydrologic cyclehas been simulated using a specially developed

water-balance model with a 10-day time increment.The model was calibrated and checked by observational data over long-term periods. The simulationswere carried out for four river basins: one smalland two middle-sized basins located in theforest-steppe and steppe zones, and the Volga RiverBasin, the largest in Eur ope. The climat ic scenariosused included changes in summer and winter airtemperature and annual precipitation in the years2000-2005, taken from the palaeoclimatic maps foreach river basin (increasing summer temperatures by0.5-1.5C and winter temperatures by 1.5-2.5Cand changing annual precipitat ion by -3% to +12%).

The model calculations show that even small changes in climatic conditions can lead to considerablealterations in the runoff regime. The primarymanifestation of such an alteration is a dramaticincrease in winter runoff, owing to more intensesnowmelt during the winter season and a corres

ponding decrease in spring runoff. The seasonalrunoff changes averaged over a long-term period forthe indicated river basins are presented in Table 4.3.This table indicates that winter runoff for the relatively dry zone (steppe) increases many-fold andconsiderably rises even for the vast Volga River

basin. Observational data show, on the r ivers of theEuropean USSR since the second half of the 1980s,that winter runoff has tended to increase noticeably,

possibly as a consequence of global climate warming.

4.5 Northern China

The analysis of hydrometeorological observationaldata for northern China (Chunzhen, 1989) showsthat the warmest period over the last 250 years

began in 1981. The mean air temperature over the

period 1981-87 was 0.5C above the normal, whilefor the same period precipitation was somewhatlower than normal (for Beijing by 4%). Studies ofnatural climate variations for the past 100 yearssuggest that the warming in northern China willcontinue up to the next century. Estimates of the

potential influence of increasing atmospheric C 0 2 onnorthern China arc not available. At the same time,it is possible that small climate changes can haveconsiderable hydrologic consequences. Simulationsusing the Hinangchzang hydrological model showthat in semi-arid regions, with a 10% increase in

precipi tation and a 4% decrease in evaporation,

4-11


16/46

Tab le 4.3 Potential changes in seasonal runoff in response to a temperature warming of 1 C for selected rivers in theUSSR

River

Volga-Volgograd

Sonsa-Yeletz

Chir-Oblivskoye

Devitza-Nizhnedeyitsk

Basin area(km')

1 360 000

16,300

8 470

76

Zone

Forest/Forest-steppe

Forest-steppe

Steppe

Forest-steppe

Annual

187165

144129

4770

127

112

Runoff (mm)

Winter

2235

3053

322

26

36

Spring

10794

10263

4345

60

41

S u m m e r-

Fall

5836

1213

13

41

35

Note: For each river the top set of runoff values refers to the long-term mean under natural conditions. Thebottom set of values is the estimated runoff for an assumed warming of C.

the runoff will increase by 27% . Increasing precipitation by 10% and evaporation by 4% results in an

18% increase in runoff. If these climatic changestake place in semi-arid regions, runoff increases by30%-50% (Chunzhen, 1989).

4.6 Japan

A number of agencies and institutes within Japanhave begun to focus on how climate warming wouldaffect the water resources and related environmentsystems of that country. To d ate, the studies haveemphasised the empirical description and characteri

sation of effects and problem areas with the aim ofidentifying how to derive quantitative estimates oflikely impa cts. To estim ate the potential hydrologicconseq uences of global warming, long-term series ofmeteorological observations in different regions ofthe country have bee n analysed. In particular,precipitation and runoff characteristics have beencompared over the coldest and the warmest 10-yearperiods with the mean temperature difference of0.074C. Precipitation over the warmest periodturned out to increase by 10%, the incidence ofheavy rains (more than 300 mm for two days)becoming much more frequent. At the same time,precipitation sums for 60- to 90-day periods withminimum rains during the warm decade have beennoticeably smaller than in the cold decade (Yamada,1989). This empirical data analysis makes it possibleto assume that with global warming in most regions

of Japan, maximum precipitation and river runoffvolume are expected to somewhat grow, especially

over the periods of rain storm s. At the same time,in dry seasons the runoff can decrease thus exacerbating the problem s of water supply. Becau se of thelack of data, it is impossible to estimate quantitatively the correspo nding values. Th e use of GC Moutput is particularly troublesome in Japan since alarge percentage of precipitation is generated bytyphoons and convectivc storms, which arc notresolved by the GCMs, as well as by frontal storms,which are. Acc urate accounting of water reso urceimpacts would, therefore, need to consider, for

example, how typhoon frequency, magnitude, andintensity would be affected.

Given the critical concern with which thegovernment of Japan views the potential threatimposed by a greenhouse-gas-induced climaticwarming, a preliminary study was undertaken withthe support of the Japanese Environment Agency(M atsu o et al., 1989). It assess ed effects on anumber of environmental systems, including waterresources, associated with a potential severe change

in climatic conditions. Th e specific wate r res ource sissues addressed included flood control, water use,and water quality.

Looking first at flood control, the report notes thatcurrent reservoir storag e capacity is limited, so muchso that the multipurpose use of reservoirs for flood


17/46

control and water use could become increasinglydifficult. An increase in rainfall intensity brought bytyphoons or fronts would give rise to serious flooding. Prolonged periods of drought, punctuated byshort bursts of intense precipitation, could lead toincreased frequency of mudslides. Since many

Japanese cities are situated in coastal areas andlowlands, any flood could produce heavy damages.Rising atmospheric temperatures could be expectedto alter the timing and volume of snowmelt generated runoff, thereby producing the potential forincreases in flood peaks and associated damages.

In terms of water use, a number of effects wereidentified. In general , water demand is expected toincrease, while water supply decreases, spreadinglocalised water shortages to broader regions or tothe nation as a whole. Reduced precipitation and theassociated reduction in discharge can be expected toreduce both the amount and regularity of hydroelectric power generation. In the particular case ofrivers dependent upon snowmelt, the impact could

be quite significant. Moreover , contamination dueto salt-water intrusion could lead to acuteagricultural and municipal water supply problems.Reduced water levels in rivers and lakes could also

produce deleterious effects on navigation and recreational activities.F inally, with respect to water quality,the expectation is that river water quality wi l l declinewith decreases in minimum flow. Lake quality could

be threatened by a prolonged state of thermal

stratification, and increasing temperatures may leadto increased eutrophication. It is also thought thatdeclines in the water level and storage capacity oflakes and dams could adversely affect water quality,although the precise mechanism behind such deterioration remains unclear.

4.7 New Zealand

In June, 1988, the New Zealand government instituted a national climate change program under thecoordination of the Ministry for the Environment.

Three working groups were formed, consideringrespectively facts, impacts, and responses, consistentwith the three working groups of the IPCC. TheFacts Working Group was formed to produce areport on the scientific basis for the predictions ofclimate change. An abridged version of this reporthas been published which contains the followingconclusions (Salinger and Hicks, 1989). Annualaverage warming in New Zealand under a doubledC0 2 climatic regime range from 1.4C to 3.5C.There is some possibility that winter temperatureswill rise faster than summer temperatures in moresoutherly parts of the country. It is possible that thesnowline could retreat between 100 m and 150 m forevery 1C rise in temperature. However, even

though New Zealand temperatures have increased by 0.5 C since the 1940s, no retreat of snowline oradvance of tree line has yet been measured.

The Impacts Working Group was formed to producea series of reports addressing the broad range of

physical, biological, economic, and social issues likelyto be affected by the climatic change estimates

provided by the Facts Wor ki ng Group. One of theseimpact reports deals with water resources (Gr iffiths ,1989). A l l the impact assessments considered twoscenarios of climate change, one based on a temporal analog (referred to as SI) and the other on thelimit ing conditions simulated by GC Ms (S2). Theanalog scenario is based on the period of maximumwarmth 8,000 to 10,000 years ago, when westerlieswere weaker and there was more airflow from thenorthwest. New Zealand temperatures were 1.5Cwarmer than at present and westerly winds werelighter, especially in winter. There was a reducedfrequency of frontal storms, and global sea-level was20 cm to 40 cm higher than at present.

The second scenario is based on themodel-simulated upper limi t to greenhouse warmingin the New Zealand area of 3 C. It is accompanied

by positive mode Southern Oscil lat ion conditions (La Nina) on average. Frequent incursions of tropicalair from the north are expected and, although theintensity of the westerlies may decrease, they willstill prevail. Moreover, global sea-level may be 30cm to 60 cm higher than at present.

Regionally, the principal likely impacts of climatechange on runoff are depicted in Figure 4.5, andmore broadly on water resources as follows. In

Nor thl and, flooding could become more frequentand more severe. Baseflows in rivers and streamsmay be enhanced, while lake levels may rise. Therecould be greater groundwater recharge and lessdemand for irrigation. Wetlands may not drain aseasily, especially in the Ruawai flats south of Darga-ville. Earthflow and other erosion in weathered

greywackes, Tangihia volcanics and podsolised sandhill country may precipitate increased suspendedsediment loadings in rivers and streams.

A number of similar effects could probably occur inthe region around Auc kland. These include morefrequent and severe flooding, enhanced baseflows inrivers and streams and increased lake levels, andgreater recharge of groundwater and less demandfor irr igat ion. There could also be changes in lakestratification patterns and a rise in the number ofdune lakes. Moreover, a number of urban-related

problems are possible. It may be that more carewill be needed in siting urban developments to

protect them from increased flooding. Hi gh intensi -

4-13


18/46

ty rainfalls may overload stormwater/sewerage systems leading to frequent discharges of contaminatedwater. Finally , the possibility of increased erosionmay result in higher suspended sediment concentrations in estuarine and harbour receiving waters.

The region encompassing Waikato Bay ofPlenty-Taupo also is likely to be more prone tofrequent and sever flooding, especially in theHauraki Plains, Bay of Plenty coastal plains, andLower Waikato River valley. Enhanced baseflows inrivers and streams and lake levels on the volcanic

plateau wi ll rise. Lakes may stratify more often anda reduction in lake carrying capacity could possiblyoccur, although this may be offset to some extent byincreased flushing rates and higher water levels.Anoxic conditions are likely in some lakes, whilethose already anoxic wi l l probably have the period

extended. This area, too, may experience greatergroundwater recharge and reduced demand forirr igation. The increased frequency of subtropicalcyclones may be compensated for by the lowerfrequency of southeasterly storms under climatescenario SI , but not under scenario S2. A numberof other unique impacts are also likely. For exam

ple, geothermal systems may receive greater groundwater recharge; Waitomo Glow Worm Cave may beclosed more frequently because of flooding, andsiltation in the cave may increase; a significantincrease in water yield from Pinus rodiafa-forested

basins can be expected with more yield in summerthan in winter; the operation of the Huntley thermal

power station may be affected by increased rivertemperatures (discharges are already limited bytemperature restrictions in the summer), althoughincreased river flows may compensate to somedegree; and sever gullying could occur in highlyerodible volcanic materials leading to increasedinfilling of hydro-dam reservoirs, thus reducingreservoir life and flood storage capacity.

In the Taranaki-Manawatu-Well ington region flooding is also likely to become more frequent andsevere, especially around Ohura, along with enhanced baseflows in rivers and streams. This areamay experience a slight increase in groundwateravailability and less need for irrigation under scenario SI and greater irr igat ion demand under S2. Anincrease in high intensity rainfalls will probably havesignificant effects on agricultural productivity inTaranaki, as well as overloading storm water/sewerage systems that could produce frequent dischargesof contaminated waters. Increased erosion here willalso probably result in higher suspended sedimentconcentrations in rivers and streams.

The southeast part of the North Island, theGisborne-Hawkes Bay-Wairarapa region may exhibit

a number of unique water resources effects. Forexample, greater incidence of drought and the morefrequent drying-up of long reaches of river mayaccompany climatic warming, as well as diminished

baseflows in rivers and streams in Hawkes Bay andeastern Wairarapa. Lake levels w i l l be low more

often and anoxic conditions could result at times,especially in Lake Tuti ra, while impedance of drainage at river mouths will occur in Hawkes Bayrequiring increased pumping capacity. Althoughlittle change in groundwater recharge is likely, theincreased use of groundwater for irr igation is. Also,under scenario S2, it is possible that devastatingimpacts could be produced by the more frequent

passage of large tropical cyclones.

On the South Island, in the Nelson-MarlboroughRegion, the likely impacts include more frequent

and severe flooding along with a small increase inthe pumping of drains; urban flooding of east coastcatchments; flooding in Tikaki Valley; enhanced

baseflows; more droughts in the drier eastern hillcountry of Marlborough; and more frequent erosionepisodes in the Mar lborough Sounds. In addition,there may be a small beneficial effect in the availability of groundwater resources, and the Takaka Rivermay run dry less often. However, salinity increasesin the Waikoropupu Springs are possible, along witha slight increase in mean discharge, and highlyerodible basins in eastern Marlborough could supplymore suspended sediment to the Wairau River.This may affect irrigation regimes and produce anincreased demand for groundwater.

Potential impacts in the Westland-Fiordland regioninclude several that are more related to mountainousconditions. For example, the meltdown of thesnowpack could significantly increase the size offlood peaks, while the drying out of regolith could

promote slope failure during high intensity rainfallsgiving increased suspended sediment loading instreams. Moreover, under scenario S2, there maybean overall tendency toward a more stable, lessextreme environment, but periods of water deficitare likely to occur.

In Canterbury, litt le change in groundwater rechargeis likely under either scenario SI or S2, but anincrease in the number of days of soil moisture

being below the wil ting point suggests that theremay be greater demands for groundwater-suppliedirrigat ion. There may also be a greater incidence ofdrought and of river reaches drying up, along withmajor aridity problems on non-irrigated downlandareas. Canterbury w i l l also probably experiencegreatly increased competition for water betweeninstream and out-of-stream uses. Ano the r potential

problem relates to snowmelt. It is probable that

4-14


19/46

there wi l l be less water in rivers in late spring andearly summer in those regions where snowmelt isnow important. Moreover, very little snow storageis likely in South Canterbury resulting in significantchanges in the tempora l pattern of runoff. Snow-melt contributions wi ll probably move forward abouta month in time.

Finally, in the Otago-Southland region, a significantreduction in the long-term availability of groundwater is likely in the eastern parts of the regions.Droughts are likely to increase and major aridity

problems may prevail in downland areas particularlyin central and eastern regions. Eastern areas mayalso experience a greater and an earlier demand forirrigation water, while forested areas there willsupply a reduced water surplus to stream flows andgroundwater recharge. There is also likely to beincreased water use by permanent grassland and ashortening of the period of soil water recharge.

5 Case studies of effects incritical or sensitive environments

5.1 Large water bodies

5.1.1 The Great Lakes Basin (US/Canada)

A detailed examination of available water supply andusers for the Great Lakes basin reveals a number ofcommonly encountered conflicts among competingwater uses, even under present climatic conditions.First, it should be recognised that there are twomajor and different hydrologic regions: the GreatLakes proper and the tributary watersheds. Mos t ofthe salient conflicting water uses for which theregulation and stabilisation of the Great Lakes isimportant is navigation and hydroelectric power

production, which favour high lake levels versusrecreation, reduction of flooding and shorelineerosion which favours low lake levels. The former(hydroelectric power and navigation) affects the

basis of an extensive industrial economy in theregion, while the latter affects millions of residents

who wish to recreate on the shores of the GreatLakes and whose houses and property are affected

by storm damage. Since the lakes themselves are avast reservoir of water, the projected typical consumptive uses of water (municipal and industrialwater supply, thermal cooling) are not expected toimpose a significant incremental adverse impact.Tributary watersheds, on the other hand, experiencehighly variable flows, with significant present constraints on in-stream and off-stream water uses.

Over 8,000 km 3 , storing 20% of the world's freshsurface water and 95% of the fresh surface water of

the US, the Great Lakes have a surface area of246,000 km 2 . The drainage area, including thesurface area of the Great Lakes encompasses nearly766,000 km 2 . Eight states of the US and two provinces of Canada (Ontario and Quebec) border theGreat Lakes (Figure 4.6). Two of the Great Lakesare regulated: Lake Superior and Lake Ontario; ietheir outflows are controlled. Lakes Michigan and

Huron are connected by the Straits of Mackinac andtheir surface water elevations respond synchronouslyto changes in water supply.

The Great Lakes Basin encompasses a population of29 million Americans (12% of the US population)and 8 million Canadians (27% of the Canadian

population) (Cohen et a l , 1989). Mill ions of people benefit directly from the water-resources related-services of the Great Lakes in the form of hydroelectric power, navigation/transportation of mineralresources including coal and iron and food. In 1975economic activity in the US portion amounted to$155 billion (1971) while that of Canada was estimated to $27 billion (1971) (International JointCommission, 1985). By 1985, the Great Lakes Basinaccounted for 37% of the US manufacturing output,consisting of transportation equipment, machinery,

primary metals, fabricated metals and food and beverage products.

The Great Lakes water levels have fluctuated over arange of 2 m during the past 150 years. Seasonallake level variations average 0.3 m in Lake Erie and0.5 m in Lake Ontario . Al so , the Great Lakesregion has experienced twelve serious droughts inthe past 60 years and six periods of extensive flooding. A high precipitation regime during the past 20years resulted in 1986 in record high lake levels forthis century.

The G CMs used to predict temperature and precipitation differ substantially in their prediction for eachof the regions under considerat ion. Figure 4.7shows, for example, that the GISS model predicts anet increase in summer precipitation in the GreatLakes of about 0.4 mm/day, whereas the G F D Lmodel shows a decrease of over 0.6 mm/day and theO S U model shows no change. Precipi tation isimportant for non-irrigated agriculture (drylandfarming). However, water availability in the streams(runoff) and in the soil is a function of vapotranspiration, which is a non-linear function of temperature. Hence, runoff computation based on G C Mmodel outputs wi l l also vary widely depending on thetemperature and precipitation conditions predicted

by each model. A comparison of the runoff computations for the doubled-C0 2 scenario based on theoutputs of three G C Ms (GISS, G F D L and OS U)

was conducted by Croley (in press). The analysis

4-15


20/46

used daily models to simulate moisture storage andrunoff for 121 wate rsheds draining into the Gr eatLakes, over-lake precipitation into each lake, and theheat storage and evaporation for each lake. Tab le4.4 shows that th ere is a 23 % to 5 1 % reduction innet basin supplies to all the Great Lakes for thethre e mo dels. Th e largest changes in net basinsupply would occur in Lake Erie, a feature that wasuniform in all three models, with the GISS modelshowing the greatest percentage change. Sanderson(1989) simulated the effects that the reduction of netbasin supply would have on the historical levels ofLake Erie from 1900-1980 (Figure 4.8).

Table 4.4 Estimated percentage ch nges in Great Lakes netbasin supplies between current climatic conditions* anddoubled C02 conditions**

l x C 02 Net basin supply and2xC02 relative changes

BASE GISS GDFL OSUBasin (mm) (%) ( ) (%)

Superior

Michigan

Huron

St.Clair

Erie

Ontario

* (1 x CO

847

698

1052

4395

592

1941

-2

-45

-46

-50

-113

-43

: or the BASE case)

** as simulated by the GISS,

-57

-65

-41

-28

-91

-29

GFDL

-19

-32

-19

-25

-56

-11

and (GCMs (Croley, undated)

Thermoelectric power cooling is projected to increase significantly, accounting also for much of theincrease in consum ptive water use. Consum ptiveuses of water, largely through evaporation, representa net loss of wate r to the hydrologic system. Thisfairly large loss was estimated by the USGS to be 85cm in 1985, rising to 170 cm in 2000. This loss,however, would only account for a reduction of afew centimetres in lake levels in comparison to aone-half metre drop in lake levels projected from areduction of net basin supplies due to increased

natu ral evapo transpira tion. If the dram atic climatechange scenario were to materialise, the largesocioeconomic impacts would stem from water useconflicts and shortages that are currently encountered primarily during naturally fluctuating low lakelevel stand s. Th at is, hydroelectric power productionwould be significantly affected, particularly in the

Lake Ontario and Lake Eric drainage areas. Th ereare substitutes for the loss of hydroelectric power,but most will add to the emission of greenhousegases (GHG) and consumption of thermoelectriccooling water.

Commercial navigation is a very important component of the Great Lakes economy that would beseriously affected by lower lake levels. Eit hercargoes would have to decrease to get through thelocks, imposing increased transportation costs, or thelocks on the Great Lakes would have to be rebuilt.In either case, the economic cost would be high.

The water resources impacts are also likely to be ofgreat socioeconomic consequence in the streams ofthe watersheds that drain into the Great Lakes.These watersheds arc largely unregulated and themunicipalities and industries depend on naturalstream flow and grou ndw ater. Althou gh the variability of weather and associated shifts in the frequencyand magnitude of climate events were not availablefrom the outputs of the GCMs, there is reason tobelieve that the increased precipitation regimepredicted by some of the GCMs will result ingreater and more frequent flooding in the tributarywatersh eds. Along with this trend, it is possible thatthe frequency, duration and magnitude of droughtsmight also increase as a consequence of the warming

tren d. In oth er wo rds, the cycles of floods anddroughts experienced in the current hydrologicrecord could become worse, exacerbating futureconditions of higher water demands.

5.1.2 T he Caspian S ea

The Caspian Sea is the world's largest enclosed lakewith a surface area of 371,000 km2. Lake-levelvariations have a very large amplitude and dependmainly on the river inflow-water surface evaporationratio (precipitation on the lake surface is small).Over the last 2000 years the amplitude of lake-levelvariation has been abou t 10 m. Instrum ental observations (since 1837) indicate that lake-level variations ranged between -25.5 m and -26.5 m absduring the period 1837 to 1932. Th e lake thenbegan to fall and reached a low of -29.10 m abs in1977. From 1978 to prese nt the lake level hasincrea sed, reac hing a level of -27.7 m in 1989(Shiklomanov, 1988).

Future lake levels depend on natural inflow variation(80% of the inflow is contributed by runoff from theVolga Rive r). Oth er controlling factors are ma n'sactivities in the river basin, anthropogenic climatechanges, and changes in precipitation and evaporation. To assess future levels of the Caspia n Sea,methods for probabilistic prediction have been


21/46

developed which take into account potential effectsof natural and anthropogenic factors, includingglobal warming. Th e meth ods are based on acomplex water balance model which uses MonteCarlo procedures to simulate an annual inflowseries.

The scenarios were based on empirical data for allthe water-balance components obtained since 1880.The mean decrease in river water inflow due topresent and projected man's activities are, accordingto SHI estimates (in km3 y r' ) : in 1989, -40; 2000,-55; 2010, -60; 2020, -65. For ev ery year the valuesof river runoff were taken separately for all the mainrivers. In the case of non-stable climatic conditions,possible climate changes for the years 2000 and 2020have been estimated for the basin and the Seaaccording to predictions by Soviet climatologists onthe basis of palaeoclimatic reconstructions (see

Section 4.4). Poss ible chang es in river wa ter inflowto the Sea have been obtained independently by twoextreme estimates derived from the river runoffchange maps for 2000 (Figure 4.9) and by the waterbalance model over decades (Section 4.4). Changingevaporation has been calculated from air temperature and precipitation data.

Table 4.5 shows calculated changes in the water-balance elements for the Caspian Sea with globalwarm ing. Acc ording to these estima tes a considerable decrease (relative to the normal) in river waterinflow to the Sea would occur by the end of thecentury, mainly due to decreased Volga RiverRunoff This would be followed by a markedincrease in inflow. Precipitation increases over theSea would be particularly significant (up to 60% inthe year 2020, from the current annual mean precipitation of 240 mm); evaporation would changeinsignificantly (up to 3%).

Table 4.5 Changes in components of the water balance ofthe Caspian Sea in response to global warming

Water balance If fcomp onent 1989 2000 2020

Total inflow flcm 3 yr ')From map analysis 0 -20 +40From model analysis 0 -30 +1 0

Precipitation (mm) 0 +1 00 +15 0Evaporation (mm) 0 +1 5 +3 0

A simulation of water-balance components undervarying climate conditions has been carried out bythe same technique, taking into account, however,the data in Table 4.5 with interpolation for each yearof the period und er cons ideration. In this case, theshape of distribution curves, variability of water-balance components, coefficients of their autoregres-

sion and intercorrelation are accepted according toobservational records. The results are presented inFigure 4.9. By the end of the century som elake-level lowering is expected to take place due toincreasing huma n activities in the basin. After theyear 2000, some stabilisation may occur followed bya significant rise in the Sea's level owing to predic tedincreases in inflow and precipitation over its surface(Shiklomanov, 1988; 1989a). Th ese results, thoughpreliminary, show the importance and necessity oftaking into account anthropogenic climate changeforecasts to estimate the fate of continental reser

voirs and to plan future management.

5.2 The arid and semi-arid zones of NorthAfrica, including the Sahel

North Africa and the Sahelian zone are both subjectto frequent and disastrous droughts, progressivearidisation, and encroa ching desertification. Th eSahel, a vast though narrow, belt stretching fromWest Africa to the Horn of Africa, is the transitionzone from the Saharan desert to the hot andsemi-arid African savannahs. Analysis of annu al

precipitation data from stations in the Sahelian zonefor the period 1968-88 indicate that this period wasexceptionally dry. Althou gh the curren t Saheliandrought appears to many scientists and policymakers to be a persistent and unprecedented phenom eno n, this drough t is far from u niqu e. Du ringthe current century, the Sahelian region experiencedseveral dry periods of varying duration, magnitude,and sp atial extent (G rove, 1973; Sircoulon, 1976).Du ring the last five centuries , historical ac counts andpalaeoclimatic evidence indicate that there havebeen rainfall fluctuations of extremely variableduration (from years to several decades) (Nicholson,1978; 1989). Demarde and Nicolis (1990) viewed theSahelian drought as a fluctuation-induced abrupttransition between a stable state of 'high' rainfalland a stable state of 'low' rainfall.

The major reason for Sahelian droughts is a decrease in annua l precip itation. Ojo (1987) analysedprecipitation data for 1901-85 using 60 stations inwes tern Africa. H e found that during 1970-79average precipitation there was 62% of normal and,during 1981-84, about 50% of normal (Figure 4.10).The Sahelian zone is characterised by a strongsensitivity to hydrometeorological conditions, especially precipitation. This is confirmed by analysis of

4-17


22/46

precipitation and runoff over the most dry and wetfive year periods of time in the Senegal, Niger andShari Rivers basins, as well as by Lake Chad levelsand areas indicated in Sircoulon (1987). Acco rdingto these data, with precipitation increasing by20-30% the runoff rises by 30-50% in the river

basins; with precipitat ion decreasing by 9-24% therunoff is reduced by 15-59%. In addition, between

the early 1960s and 1985, the area of Lake Chadshrank by more than 11 times

ipcc far wg ii chapter 04

Documents