ISSN 1391-9407 ISBN 92-9090-542-5

Postal Address: IWMI, P O Box 2075, Colombo, Sri Lanka Location: 127 Sunil Mawatha, Pelawatte, Battaramulla, Sri Lanka Telephone: +94-11 2787404, 2784080 Fax: +94-11 2786854

Email: comp.assessment@cgiar.org Website: www.iwmi.org/assessment

Taking into Account Environmental Water Requirements in Global-scale Water Resources Assessments

Vladimir Smakhtin, Carmen Revenga and Petra Döll

Research Report 2

I n t e r n a t i o n a lWater ManagementI n s t i t u t e

The Comprehensive Assessment of Water Management in Agriculture takes stockof the costs, benefits, and impacts of the past 50 years of water development foragriculture, the water management challenges communities are facing today, andsolutions people have developed. The results of the Assessment will enable farmingcommunities, governments, and donors to make better-quality investment andmanagement decisions to meet food and environmental security objectives in the nearfuture and over the next 25 years.

The Research Report Series captures results of collaborative research conducted underthe Assessment. It also includes reports contributed by individual scientists andorganizations that significantly advance knowledge on key Assessment questions. Eachreport undergoes a rigorous peer-review process. The research presented in the seriesfeeds into the Assessment’s primary output—a “State of the World” report and set ofoptions backed by hundreds of leading water and development professionals and waterusers.

Reports in this series may be copied freely and cited with due acknowledgement.Electronic copies of reports can be downloaded from the Assessment website(www.iwmi.org/assessment).

If you are interested in submitting a report for inclusion in the series, please seethe submission guidelines available on the Assessment website or through writtenrequest to: Sepali Goonaratne, P.O. Box 2075, Colombo, Sri Lanka.

Comprehensive Assessment outputs contribute to the Dialogueon Water, Food and Environment Knowledge Base.

ii

Comprehensive Assessment Research Report 2

Taking into Account Environmental WaterRequirements in Global-scale WaterResources Assessments

Vladimir SmakhtinCarmen Revengaand Petra Döll

With contributions fromRebecca TharmeJanet Nackoneyand Yumiko Kura

Comprehensive Assessment of Water Management in Agriculture

ii

The authors: Vladimir Smakhtin is Principal Scientist in Eco-hydrology and WaterResources at the International Water Management Institute (IWMI) Colombo,Sri Lanka. Carmen Revenga is a Senior Associate in the Information Program of theWorld Resources Institute (WRI) Washington D.C., USA. Petra Döll is a Professor ofHydrology at the Institute of Physical Geography, University of Frankfurt, Germany(Former address: Center for Environmental System Research, University of Kassel,Germany). Contributors to map design and analysis: Rebecca Tharme is a researcherin aquatic ecology at IWMI, Colombo, Sri Lanka. Janet Nackoney and Yumiko Kuraare GIS specialists at WRI, Washington D.C., USA.

The authors gratefully acknowledge the financial contributions from the Government ofThe Netherlands to the CGIAR research programme on Comprehensive Assessmentof Water Management in Agriculture, from which this study was partially funded. TheMinistry of Foreign Affairs of the Netherlands supported the involvement of the WorldResources Institute in this study. Mr Michael Märker and Mrs Martina Flörke of KasselUniversity contributed towards the provision of hydrological data for this research. Thiscontribution was partially supported by the IUCN (Switzerland). Thanks are also dueto Dr Ger Berkamp and Mr Elroy Bos (both of IUCN, Switzerland), Dr David Moldenof the International Water Management Institute (Colombo, Sri Lanka) and Dr JuergUtzinger of Princeton University (USA) for their detailed and helpful comments on andsuggestions to the multiple early drafts of this report. Multiple comments by threeanonymous reviewers are also gratefully acknowledged.

Smakhtin, V.; Revenga, C.; and Döll, P. 2004. Taking into account environmental waterrequirements in global-scale water resources assessments. ComprehensiveAssessment Research Report 2. Colombo, Sri Lanka: Comprehensive AssessmentSecretariat.

/ water availability / water stress / water use / water scarcity / water requirements /ecosystems / water resource management / hydrology / water allocation / naturalresources / food security / environmental effects /

ISSN 1391-9407ISBN 92-9090-542-5

Copyright © 2004, by the Comprehensive Assessment Secretariat.

Please direct inquiries and comments to: comp.assessment@cgiar.org

The Comprehensive Assessment is organized through the CGIAR’s System-Wide Initiative on WaterManagement (SWIM), which is convened by the International Water Management Institute. TheAssessment is carried out with inputs from over 90 national and international development andresearch organizations—including CGIAR Centers and FAO. Financial support for the Assessmentcomes from a range of donors, including the Governments of the Netherlands, Switzerland, Japan,

Taiwan, and Austria; the OPEC Fund; FAO; and the Rockefeller Foundation.

iiiiii

Contents

Summary ............................................................................................................ v

Introduction ........................................................................................................ 1

Data and Methodology ....................................................................................... 3

Discussion.......................................................................................................... 10

Literature Cited .................................................................................................. 22

v

Summary

Assessment of water availability, water use andwater stress at the global scale has been thesubject of increasingly intensive research overthe course of the past 10 years. However, therequirements of aquatic ecosystems for waterhave not been considered explicitly in suchassessments. It is, however, critically importantthat in global studies a certain volume of wateris planned for the maintenance of freshwaterecosystem functions and the services theyprovide to humans. This report summarizes theresults of the first pilot global assessment ofthe total volumes of water required for suchpurposes in world river basins. These volumesare referred to in this report as EnvironmentalWater Requirements (EWR). The total EWR areassumed to consist of ecologically relevant low-flow and high-flow components. Bothcomponents are related to river flow variability,and estimated by conceptual rules from thedischarge time series simulated by the global

hydrology model. The concept of environmentalwater scarcity is then introduced and analyzedusing a water stress indicator, which showswhat proportion of the utilisable water in worldriver basins is currently withdrawn for directhuman use and where this use is in conflictwith EWR. The results are presented on globalmaps. EWR required to maintain a faircondition of freshwater ecosystems rangeglobally from 20 to 50 percent of the meanannual river flow in a basin. It is shown thateven at estimated modest levels of EWR, partsof the world are already or soon will beclassified as environmentally water scarce orenvironmentally water stressed. The totalpopulation living in basins where modest EWRlevels are already in conflict with current wateruse, is over 1.4 billion, and this number isgrowing. The necessity of further research inthis field is advocated and the directions forsuch research are discussed.

1

Taking into Account Environmental WaterRequirements in Global-scale Water ResourcesAssessments

Vladimir Smakhtin, Carmen Revenga and Petra Döll

With contributions from Rebecca Tharme, Janet Nackoney and Yumiko Kura

drastically degraded. Many freshwater speciesare facing rapid population decline or extinction,and yields from many wild fisheries havedwindled as a result of flow regulation, habitatdegradation and pollution.

Every inland or coastal, fresh or brackishwater ecosystem has specific water requirementsfor the maintenance of the ecosystem structure,functioning and the dependent species. For riverdependent aquatic ecosystems, these arenormally referred to as environmental flowrequirements, or environmental flows. There isno universally agreed definition of environmentalflows. On the one hand, they may be defined inrather general terms as an allocation of water forconservation purposes. On the other hand, thedefinition may be very specific as a suite of flowdischarges of certain magnitude, timing,frequency and duration, all of which jointlyensure a holistic flow regime capable ofsustaining a complex set of aquatic habitats andecosystem processes (Knights 2002).

In order to sustain the ability of freshwaterdependent ecosystems to support foodproduction and biodiversity, environmental flowsmust be established scientifically, madelegitimate and maintained. This poses a numberof challenges. The physical processes andinteractions between freshwater ecosystemcomponents are extremely complex, and their

Introduction

Emerging freshwater scarcity has beenrecognized as a global issue of the utmostimportance. There is a growing awareness thatincreased water use by humans does not onlyreduce the amount of water available for futureindustrial and agricultural development but alsohas a profound effect on aquatic ecosystems andtheir dependent species. Balancing the needs ofthe aquatic environment and other uses isbecoming critical in many of the worlds’ riverbasins as population and associated waterdemands increase (Postel et al. 1996;Vörösmarty et al. 2000; Naiman et al. 2002). Inthis context, what is often lacking is theunderstanding that planning environmental waterallocation means striking the right balancebetween allocating water for direct human use(e.g. for agriculture, power generation, domesticpurposes and industry) and indirect human use(maintenance of ecosystem goods and services,e.g. Acreman 1998).

The services that freshwater ecosystemsprovide to humans such as fisheries, floodprotection, recreation and wildlife are estimatedto be worth trillions of US dollars annually(Constanza et al. 1997; Postel and Carperter1997). A recent global assessment of the statusof freshwater ecosystems (Revenga et al. 2000)showed that their capacity to provide the fullrange of such goods and services appears to be

2

understanding and quantification in most of theworld remains poor. The impacts of changinghydrological regimes on ecosystems are notproperly understood. A satisfactory level ofconfidence in environmental flow estimates maybe achieved by the application of time-consuming methods, which involvemultidisciplinary expert judgment and site-specific information. This information and theexpertise required to apply such methods arelimited. As a result, in most of the world,environmental flows have never been estimated.The allocation of water for environmentalpurposes is still assigned a low priority in waterresources management, and the condition offreshwater ecosystems worldwide continues todeteriorate (Revenga et al. 2000; Rosenberg etal. 2000). The current situation therefore pointsto the need for an intensive, large-scale researchprogram on environmental flow assessment.

At the same time, there are areas of waterresearch and planning, where at least crude,pilot estimates of environmental water needs areurgently required at the global scale. In recentyears, a number of global assessments of waterresources, current and future water use, waterand food security projections and water povertyanalyses have been completed (Alcamo et al.2000; FAO 2000; IWMI 2000; Shiklomanov 2000;Sullivan et al. 2003; Vörösmarty et al. 2000;UNESCO 2003). Global approaches help identifythe areas of present and future water scarcity,the areas of potential water related conflicts, andalso set priorities for international financing ofwater projects. However, as a rule, global studiesundertaken to date were limited to assessingwhether the human water needs (domestic,industrial and agricultural) can be satisfied by thetotal renewable water resources in a country, ariver basin or a grid cell. These studies did notexplicitly consider the water requirements ofecosystems and the needs of people, whodepend upon them. It is therefore criticallyimportant to assess environmental water

requirements in the context of global waterresources assessment and to respond to theneed for incorporating these requirements inwater-food-environment projection models(Comprehensive Assessment of WaterManagement for Agriculture 2002: http://www.cgiar.org/iwmi/Assessment/Index.htm).

This report presents the first attempt toinclude environmental flow requirements in aglobal water resources assessment. Due to thelack of river basin specific ecological informationand also the scale of this assessment, weattempt to estimate only the mean total annualvolume of water that should be allocated forenvironmental purposes. This total annualvolume is hereafter referred to as EnvironmentalWater Requirements (EWR), and is computed forall world river basins. No attempt was made inthis study to determine environmental flowregimes which would include seasonal low flows,peak flows, their timing, frequency and duration.The computed EWR values are based on thetime series of monthly river flows that aremodeled by a state-of-the-art global hydrologicalmodel.

We then use the EWR values to calculate anew basin-specific water scarcity index thattakes into account environmental requirements.This index combines three components: totalavailable water resources, total actual water useand EWR.

In the following sections, we first describethe global hydrology and the water use models,followed by the description of the conceptualrules for the estimation of EWR. We thenpresent global maps of the estimated EWR andof the new water scarcity index that takes intoaccount environmental requirements, identifyingareas where current water use is in conflict withEWR. After a discussion of the results, thecurrent trends, problems and initiatives in theglobal environmental water management areexplored and steps for future research in thisfield, in a global scale, are suggested.

3

Data and Methodology

area, crop type, climate variables and water useefficiency. Livestock water use is calculated bymultiplying livestock numbers by livestock-specific water use. Household water use by gridcell is computed by downscaling publishedcountry values (Shiklomanov 1997) based onpopulation density, urban population and accessto safe drinking water. Water use by thermalpower plants is derived from the capacity andcooling technology of more than 60,000 powerplants worldwide. Manufacturing waterwithdrawals are estimated using country data(Shiklomanov 1997) on the main water-usingindustries and the distribution of the urbanpopulation. The total water use is calculated asthe sum of water withdrawals for all sectors.

Conceptual Rules for a Pilot GlobalAssessment of EWR

Existing methods for the estimation of EWRdiffer in input information requirements, types ofecosystems they are designed for, time which isneeded for their application, and the level ofconfidence in the final estimates. They rangefrom purely hydrological methods, which deriveenvironmentally acceptable flows from flow dataand use limited ecological information or eco-hydrological hypotheses (e.g. Richter et al. 1997;Hughes and Münster 2000), to multidisciplinary,comprehensive methods, which involve expertpanel discussions and collection of significantamounts of geo-morphological and ecologicaldata (e.g. Arthington et al. 1998; King and Louw1998). Reviews of these methods may be foundin multiple sources, including Tharme 1996,Arthington et al. 1998 and Dunbar et al. 1998.The necessary quantitative information on theecological functioning of aquatic ecosystems andtheir dependent species in many parts of theworld is currently lacking and in many regions,

Global Hydrology and Water UseModeling

Basin-specific information on river dischargesand water use used in this study were simulatedby the global water model WaterGAP 2 (Alcamoet al. 2003; Döll et al. 2003), which has a spatialresolution of 0.5 by 0.5 (approximately 67,000grid cells worldwide). The model consists of twomain parts; the Global Water Use Model and theGlobal Hydrology Model.

The physically based global hydrology modelcomputes the time series of monthly runoff foreach cell (as the sum of surface runoff andgroundwater recharge) and river discharge. Thecalculation of the latter takes into account thestorage capacity of aquifers, lakes, wetlands andrivers, and routes river discharge through eachriver basin according to a global drainagedirection map. Computations are based on thetime series of monthly climate variables for theperiod 1961-1990. The model is calibratedagainst the measured river discharge at 724gauging stations worldwide. During thecalibration, the simulated values take intoaccount the reduction of river discharge due tototal withdrawals (for agriculture, industry etc.).However, for the computation of the long-termmean annual runoff (MAR) and EWR, it isnecessary to compute a reference condition —the natural river discharge that would haveoccurred in the absence of human impacts inriver basins. The global hydrology model setupallows a pseudo-natural river discharge to becalculated: the discharge that would occurwithout withdrawals but with the reservoirs thatexisted in the world around 1995.

The global water use model includes sub-models for irrigation, livestock, households,thermal power plants and manufacturing industry.Irrigation water requirements (withdrawals) aresimulated as a function of a cell-specific irrigated

4

even basic inventories for freshwater species arenon-existent. Therefore, for this first global-scaleassessment of EWR, it was not feasible to relyon data intensive methods, and hence, it wasconsidered logical to use and/or develophydrology-based conceptual rules.

It is now an accepted concept in aquaticecology that the conservation of aquaticecosystems should be considered in the contextof the natural variability of flow regime (Poff etal. 1997; Richter et al. 1997; Hughes andMünster 2000). The two primary geneticcomponents of any flow regime are baseflow andquickflow. Baseflow represents that part of theriver flow, which originates from an aquiferhydraulically connected with the river or fromother delayed sources such as perchedsubsurface storage or lakes. In perennial rivers,throughout most of the dry season of the year,the discharge is composed entirely of baseflow.In intermittent and ephemeral rivers, baseflowduring the dry season is zero. Quickflowrepresents the immediate response of acatchment to rainfall or snowmelt events and iscomposed primarily of overland flow or interflowin the topsoil. During the wet season, dischargein most of the rivers is composed of bothbaseflow and quickflow, but is dominated by thelatter. Baseflow and quickflow components maybe separated from each other by digital filteringapplied to observed or simulated flow time series(Smakhtin 2001). Both components can beexpressed as a proportion of the long-term MARin a river.

A hydrology-based approximation of EWRshould therefore endeavor to incorporate portionsof both baseflow and quickflow that wouldcontribute to maintaining freshwater ecosystemproductivity and dynamics. To be consistent withthe emerging terminology, we further refer tothese components as the environmental low-flowrequirement (LFR) and the environmental high-flow requirement (HFR). LFR is believed toapproximate the minimum requirement of waterof the fish and other aquatic species throughout

the year. HFR is important for river channelmaintenance, as a stimulus for processes suchas migration and spawning, for wetland floodingand recruitment of riparian vegetation. The sumof LFR and HFR forms the total EWR.

The principles, which were used indeveloping the conceptual rules for this pilotglobal assessment, were similar to those whichunderpin the method for the preliminaryestimation of the water quantity component ofecological reserve for rivers currently in use inSouth Africa. This method is described in detailby Hughes and Münster (2000) and Hughes andHannart (2003). It was developed through theanalysis of quantitative results from a number ofspecialist workshops on the estimation ofenvironmental flows held in South Africa in thelate 1990s. Each such workshop focused onspecific rivers and produced estimates ofenvironmental flows for each month of the yearand for several components of the river flowregime (high and low flows required for years ofnormal wetness, and high and low-flows requiredfor dry years). Hughes and Münster (2000)related these environmental recommendations toa combined non-dimensional hydrologicalvariability index. The latter accommodated ameasure of long-term flow variability (acoefficient of variation of seasonal flow) and ameasure of flow ‘stability’ (base flow index,defined in hydrology as a proportion ofsubsurface flow in the total river flow (e.g.Smakhtin 2001)). Hughes and Hannart (2003)further suggested that in basins with highlyvariable flow regimes, a larger proportion of thetotal annual flow occurs as “short periods of highflow interspersed with low or no flow”. Theassumption was then made that “the biota willbe adjusted to a relative scarcity of water andtherefore requires a lower proportion of the long-term mean”. For less variable flow regimes itwas assumed that “the biota is more sensitive toreductions in flow and that a larger proportion ofthe long-term mean will be required to achievethe ecological objectives.”

5

The concepts suggested by Hughes andMünster (2000) and Hughes and Hannart (2003)may be reinterpreted in the global context asfollows. In basins with highly variable flowregimes, a larger proportion of the total annualflow occurs during the wet period, which (in suchbasins) usually lasts for one to three months.During a dry period of the year, such rivers mayeither go completely dry, or have very lowdischarges. Estimates of the total EWR for suchbasins or regions will most likely be dominatedby the estimates of the environmental HFRcomponent. The annual total flow in basins withstable flow regimes is made of a significant LFR,which continues through most of the year and ofrelatively small flow increases during the wetterperiod. Estimates of the total EWR for suchbasins will most likely be dominated by theestimates of the environmental LFR component.

To illustrate the differences in flow regimes,we used several observed data sets availablefrom the Internet at http://webworld.unesco.org/water/ihp/db/shiklomanov/index.shtml - a jointweb site of Russian State Hydrological Institute

and International Hydrological Programme ofUNESCO. The Krishna River in India is a typicalexample of a monsoon-driven flow regime, wheremost of the annual flow (up to 80% of MAR)occurs in the 3-4 wettest months of the year (figure1). During the rest of the year, the river carries arelatively low proportion of the total flow. Small andmedium-sized rivers in arid and semi-arid regions,and even some large rivers like the Limpopo insouthern Africa, on the other hand, may gocompletely dry for several months during the dryseason. These are referred to as intermittentrivers. The extreme case is the ephemeral rivers,which flow only after infrequent rainfall events. Onthe opposite end of the spectrum are rivers withvery stable flow regimes (figure 1). These normallyhave stable groundwater inflow and/ or arenaturally regulated by lakes in the catchment. Thetypes of flow regimes in the world are numerous,and consequently, the proportion of environmentalLFR and HFR components in total EWR vary. Thequantitative, hydrology-based rules, whichapproximate these two components, are describedbelow.

FIGURE 1.Examples of two contrasting seasonal flow distributions.

% o

f the

mea

n an

nual

flow

35

30

25

20

15

10

5

01 2 3 4 5 6 7 8 9 10 11 12

Krishna, India Amazon

Months of the year

6

Environmental LFR was assumed to beequal to the monthly flow, which is exceeded90 percent of the time on average throughout ayear (Q90). This flow may be interpreted simplyas the discharge that is exceeded 9 out of 10months. It represents one of the low-flowcharacteristics widely used in hydrology andwater resources assessments (e.g. Smakhtin2001) and is normally estimated from a flowduration curve (FDC). FDC is a cumulativedistribution of flows recorded at a site in a riverbasin over a certain period of observations.Alternatively, a FDC may be constructed from asimulated flow record (e.g., such as simulated byWaterGAP 2 model). A FDC is constructed byranking all the flows in a record from the highestto the lowest and assigning the probability (%time of exceedence) to each flow in a rankedseries.

The area under the curve represents MAR.The area under the threshold of the median flow(Q50) may approximate the total annualbaseflow. A “steep” FDC is a reflection of a very

variable flow regime, and a FDC with a smallslope is the indication of a stable flow regime.For rivers with highly variable flow regimes (likethe Krishna or the Limpopo in figure 2), Q90may be equal to zero, or be very small. Forbasins with a stable flow regime, it mayconstitute a larger proportion of MAR. Manystudies on low-flow hydrology, reviewed bySmakhtin (2001) suggest that Q90 variesprimarily in the range of 0 to 50 percent ofMAR. Q90 accounts only for a smaller part ofthe total annual baseflow (Smakhtin 2001). Byusing Q90 as a measure of LFR, we implicitlysuggest that only part of the river baseflow couldbe allocated to the environment.

Some existing experience with setting HFR(e.g. Hughes and Hannart 2003) suggests thatHFR may vary in the approximate range of 5 to20 percent of MAR, depending on the type offlow regime and the objective of theenvironmental flow management. Following theprinciples of highly variable and stable flowregimes described above, it was decided to

FIGURE 2.Examples of non-dimensional flow duration curves for rivers with different flow regimes.

Mon

thly

flow

(pro

port

ion

of th

e m

ean)

100

10

1

0.1

0.01

1.001

0.01 0.1 1 5 10 20 40 50 60 80 90 95 99 99.9 100

Time for which flow is exceeded (%)

30 70

Limpopo Krishna

Thames

7

approximate HFR by a set of thresholds linked tothe different LFR levels (table 1). For basins withhighly variable flow, where Q90 < 10 percent ofMAR, HFR was set at 20 percent of MAR. Forrivers with stable flow, where Q90 is higher than30 percent of MAR, HFR was considered to beequal to zero. Finally, for rivers with Q90ranging from 10 to 20 percent and from 20 to30 percent of MAR, HFR levels were set at15 percent and 7 percent of MAR, respectively.These “rules of thumb” effectively represent astepwise substitute for the method suggested byHughes and Münster (2000) and Hughes andHannart (2003) as the two components of EWR–LFR (Q90) and HFR –complement each other.In reliably flowing rivers with high baseflowcontribution (and consequently high LFR), HFRis low. In highly variable rivers, baseflowcontributions are normally low (and consequentlyLFR is low), and the total environmentalrequirement is dominated by high HFR.

The total annual EWR was calculated as asum of two estimates: LFR and HFR. As bothcomponents varied between river basins, thetotal EWR reflected the differences in flowregimes globally. The main restriction of thedeveloped rule in its present form was that it didnot explicitly specify a desired conservationstatus or environmental management class inwhich an ecosystem needed to be maintained.Therefore the estimation rules effectivelyproduced one possible “scenario” of EWR,

which is interpreted in one of the followingsections.

Environmental Water Scarcity Indicesand Mapping

Every river may be characterized by its totalresource capacity. The measure of this capacityis MAR, which is the average of total annualvolumes of water, recorded or calculated at aparticular point in a river over a long period. Theconcept of long-term average water resource isequally applicable at the scale of a specificaquatic ecosystem, a country, a geographicalregion or the entire world. For example, the sumof all the world rivers’ long-term average annualflow at their outlets would represent MAR of theworld. The difference between total resource andEWR represents utilizable capacity of theresource. Ideally, it is only the surplus water, inexcess of these requirements, that should bemade available for human use (Molden andSakthivadivel 1998; Smakhtin 2002).

By comparing estimates of total resourcecapacity with estimates of EWR and actual totalwater use, it is possible to identify the regionswhere the human water use and themaintenance of functional ecosystems are inconflict. Cases where the total actual water useexceeds the difference between the total wateravailable and EWR may be referred to as cases

TABLE 1.A conceptual rule for the estimation of environmental high-flow requirement.

Low Flow Requirement (Q90) High Flow Requirement Comment(HFR)

If Q90 < 10% MAR Then HFR = 20%MAR Basins with very variable flow regimes. Most of the flowoccurs as flood events during the short wet season.

If 10%MAR � Q90 < 20%MAR Then HFR = 15%MAR

If 20%MAR � Q90 < 30%MAR Then HFR = 7%MAR

If Q90 � 30%MAR Then HFR = 0 Very stable flow regimes (e.g. groundwater dominatedrivers). Flow is consistent throughout the year. Low-flowrequirement is the primary component.

Note: MAR= Mean annual runoff.

8

of ‘environmental water scarcity’. Depending onthe spatial resolution used in such calculations,the regions, countries, watersheds or partsthereof may be declared as environmentallywater scarce.

The concept of environmental water scarcityis illustrated graphically in figure 3. The entirebox in these pictures represents the averagetotal volume of water available in a basin (e.g.MAR). The bottom portion of the box representsthe ecological water needs; that is the amountof water needed to sustain a functioning

ecosystem. The rest of the box represents theamount of water that can potentially be put toother uses: agriculture, industry etc. – utilizablewater. The actual water use can berepresented either by the consumptive wateruse or by the total water withdrawals. Theconsumptive water use is the complete removalof water for use from the system. Withdrawal isthe water that is taken out of the system, b•Butpart of this water, may return to the system afteruse. Withdrawal is a better indication of impactson available water resource and ecosystems,

FIGURE 3.A schematic representation of the relationships between total water resources, total present water withdrawals andEWR in (a) environmentally safe, (b) environmentally water scarce, and (c) environmentally water stressed river basins.

9

because it is normally not clear how muchwater returns to the system, in what conditionor where in the basin. The present actual wateruse (withdrawals) is represented in figure 3 bylight gray shading.

Basins where the use is not tapping into ornot even close to the water reserved for theenvironment were considered to be ecologically“safe” (at least in the present water use terms,and irrespective of possible degradation due topollution, figure 3a). Basins where the use istapping into EWR were “environmentally waterscarce” (figure 3b). In basins like this, thedischarge has already been reduced by totalwithdrawals to such levels that the amount ofwater left in the basin is less than estimatedEWR. Basins in which the total water use issuch that the remaining amount of water isclose to the estimated EWR, are at risk ofbecoming “environmentally water scarce” i.e.highly likely to be degraded and threatened(figure 3c). The latter may also be referred to as“environmentally water stressed”. It should berecognized that many of the basins that appearto be on the “safer side” in our approach may,however, already be environmentally stressedfrom other causes (pollution, siltation, introducedspecies etc). Water withdrawals at that point cancause ecological damage to them much quicker.

There are several possible ways ofnumerically categorizing environmental waterscarcity. The option we selected for this pilotassessment focuses exclusively on that portion ofwater, which is available in a basin on top of theestimated EWR, that is the utilizable water sectionin figure 3. We then define a water stressindicator (WSI), which reflects the scarcity ofwater for human use by taking into account EWR.

assumed that water had been reserved forecological purposes and estimated a ratio (or apercentage) of total withdrawals to utilizablewater. If the index exceeded 1, the basin wasclassified as water scarce (table 2). Smallerindex values indicated progressively lower waterresources exploitation, and consequently, lowerrisk of environmental water scarcity. Waterstress indicators of a similar form are commonlyused in human water stress assessments(Alcamo et al. 2002; Oki et al. 2001). However,EWR have not been considered in human waterstress assessments before, and only the ratio oftotal use to total water available was calculated.The previous expression for WSI was therefore:

The severity of water scarcity in thesepublications was ranked as follows: WSI < 0.1no water stress; 0.1 < WSI < 0.2 low waterstress; 0.2 < WSI < 0.4 moderate water stress;WSI > 0.4 high water stress and WSI >0.8 veryhigh water stress.

These categories, as well as the categoriessuggested in table 2, were rather arbitrary.However, the following should be taken intoaccount while interpreting WSI values. Waterproblems usually aggravate during the dryseason of the year or in drought years. Themethods at present do not suggest means ofaddressing this seasonal issue as averageannual values are used throughout the study. Itis however very likely that in basins with a highWSI, EWR are tapped into during dry periodsand during dry years. This is why the basinswith WSI values of 0.6 may be classified aswater stressed. In the context of figure 3, thetotal water (entire box size) available during adrought year is much smaller than the averageMAR. The actual water use in a dry year doesnot change much. Even if EWR are smallerduring a drought, the buffer between the wateruse and the water available gets smaller or

This index may not be negative, as EWR isalways less than total water available (MAR).WSI calculated by this method implicitly

10

disappears completely. Therefore, basins withhigher WSI values have a higher risk of being“environmentally water scarce” during droughts.It has to be understood however that athreshold of 0.6 is rather arbitrary. Without

explicitly introducing the concepts of frequencyand assurance into the assessment ofenvironmental flows, it is impossible to predictwhether the estimated EWR will or will not besatisfied in any particular year.

WSI (proportion) Degrees of Environmental Water Scarcity of River Basins

WSI > 1 Overexploited (current water use is tapping into EWR)— environmentally water scarce basins.

0.6 < WSI < 1 Heavily exploited (0 to 40% of the utilizable water is still available in a basin before EWR are inconflict with other uses) – environmentally water stressed basins.

0.3 < WSI < 0.6 Moderately exploited (40% to 70% of the utilizable water is still available in a basin before EWRare in conflict with other uses).

WSI < 0.3 Slightly exploited.

TABLE 2.Categorization of environmental water scarcity.

Notes: WSI= Water stress indicator; EWR= Environmental water requirements.

Discussion

map in figure 4 reproduces well the knownpattern of water availability distribution over theglobe.

The estimates of EWR based on thesimulated monthly time series, range from 20 to50 percent of MAR (figure 5). At the lower endof EWR (20%), there are river basins withextremely variable flow regimes, where EWR aredominated exclusively by HFR (table 1). At thetop end of EWR (50%) there are river basinswith stable flow regimes where EWR aredominated exclusively by LFR (represented byQ90). While the calculated values of Q90 rangeglobally mostly from 0 to 50 percent of MAR, inNorth America, a number of river basins werefound to have unrealistically high Q90 values (upto 83 % of MAR). It is unlikely that suchestimates can be correct in basins with apronounced annual snowmelt-generated rise inflow, which should increase flow variability,ensure a large slope in the shape of FDCs andconsequently result in smaller low-flow proportion

Global Mapping of EnvironmentalWater Requirements and WaterScarcity

The estimates of the long-term total annual waterresources (MAR) calculated by WaterGAP 2model are presented in figure 4. The detailedanalysis of this map and other model outputs arepresented in Döll et al. (2003) and are notrepeated here. It may be noted that the model islikely to underestimate the water availability inthose snow-dominated basins where nodischarge measurements are available forcalibration, due to the fact that snow precipitationis strongly underestimated by precipitationgauges. In some semi-arid and arid basins,water availability is likely to be overestimated, asthe model in its present form does not capturesome of the complex processes in these areas,like transmission losses. Similar problems areencountered in other global water modelingstudies (e.g. Fekete et al. 1999). In general, the

11

FIGURE 4.A map of long-term average annual water resources (MAR) by the basin, calculated by the WaterGAP 2 model.

12

in total annual runoff. The maximum value of50 percent of MAR has been assigned in thisstudy to LFR for all those basins, where Q90exceeded the threshold of 50 percent of MAR.This rule of thumb, albeit arbitrary, ensured thatthe total EWR in such basins were coherent withthe range of Q90 values normally referred to inthe literature (Smakhtin 2001). One possiblereason for the calculated high Q90 values inNorth America was that the hydrological modelwas calibrated against observed flow recordsaffected by regulation for hydropower. Suchimpacts are typical in Canada, for example. Theymay decrease flow variability and escalate Q90(and subsequently LFR and total EWR). Thisissue needs to be examined more closely in thefuture in a broader context of simulating naturalflow sequences throughout the world.

As the map in figure 5 shows, EWR are thehighest (normally over 40% of MAR) for therivers of the equatorial belt (e.g. parts of Amazonand Congo basins) where there is a stablerainfall input throughout the year, and for somelake-regulated rivers (in Canada, Finland etc).Most of the river flow regimes in northern andcentral Europe are characterized by the highproportion of groundwater generated baseflow.The plains of the western Siberia are dominatedby the vast stretches of swamps, which performthe natural flow regulation function. In suchcases, the flow variability is relatively low, whichleads to higher EWR (figure 5). In highly variablemonsoon-driven rivers (e.g. in India), the riversof the arid areas that flow after infrequent rains(e.g. Limpopo basin, North Africa), and also inmost of the East Siberian rivers with highsnowmelt flows, the estimates of EWR are lower(20-30% of MAR).

EWR estimates obtained by any method maynot be considered without a reference to somenegotiated or prescribed ecosystem conservationstatus (Petts 1996; Durban et al. 1998; DWAF1997). The higher this status, the higher therequired EWR will be. Some earlier studies(Tennant 1976) suggested that 10 percent of

MAR is the lowest feasible limit for EWR as itcorresponds to a severe degradation of anecosystem, while 60 to 100 percent of MARrepresents an “optimal range”. Other sourcessuggest that “the probability of having a healthyriver falls from high to moderate when thehydrological regime is less than two-thirds of thenatural” (Jones 2002). In this context, theestimates in figure 5 appear to be on the lowside of the plausible EWR. In Tennant’s (1976)terminology, they may be interpreted as EWR,which are required to ensure a “fair” condition ofecosystems (in terms of water volume only andwithout taking into account water pollution). Onthe other hand, most estimates of the totalannual EWR obtained by comprehensive expert-panel methods for rivers in South Africa andrange between approximately 15 to 55 percent ofMAR, correspond to ecosystems that are“moderately modified” from the natural state(Hughes and Hannart 2003). This is the thirdconservation status (environmental managementclass) in the South African category system(DWAF 1997), after “largely natural” and “slightlymodified” ecosystems. “Moderately modified”ecosystems are interpreted as those where“sensitive biota is reduced in numbers andextent”. This category is followed only by “highlymodified” ecosystems with a very high degree ofdegradation. EWR estimates obtained in ourglobal scenario vary from 20 to 50 percent ofMAR, and therefore are in the “moderatelymodified” ecosystem range at least in terms ofthe South African system, which we haveeffectively used at this stage. The discussionabove may suggest that, to avoid furtherdegradation of freshwater dependentecosystems, attempts should be made toachieve a compromise between the allocation ofwater for environment and other uses (e.g.agriculture), at least in the calculated range(20 to 50% of MAR should be allocated forenvironmental purposes). It does not howeverimply that more detailed ecosystem-specificstudies may not indicate higher or lower EWR.

13

FIGURE 5.A global distribution of estimated total EWR expressed as a percentage of long-term mean annual river runoff.

14

The higher EWR could be established forecosystems with high conservation importanceand / or sensitivity. The lower EWR may benegotiated in river basins with alreadyoversubscribed water resources.

The distribution of WSI values over the globe(water withdrawal as a proportion of wateravailable for human use) is presented on themap in figure 6. This map was developed basedon the estimated EWR illustrated by figure 5. Ithighlights basins where there is insufficient waterto meet EWR and therefore it may beinterpreted as a first global picture ofenvironmental water scarcity by the basin. Incontrast to figure 6, figure 7 reflects thetraditional way in which water scarcity isassessed (as expressed by (2) above). Itcompares water withdrawals with the wateravailability, without taking into account EWR.This is the approach used in most of the currentwater resources models and scenarios (Alcamoet al. 2002; Oki et al. 2001). The comparison ofmaps in figures 6 and 7 illustrates that when theecosystem’s water requirements are taken intoaccount, more basins show a higher degree ofwater stress. This is not surprising as, if water isreserved for the environmental purposes, itsavailability for other human uses naturallydecreases. This seemly straightforward fact ishardly taken into account in most current waterscarcity assessments and projections. And, giventhat many livelihoods, especially those of thepoor, depend on productive freshwater-dependent ecosystems, these assessmentscurrently overestimate the amount of waterdirectly available for people.

In this context, the basins in the yellow andred bands are those, where humans are at ahigher water stress if water allocations are madefor the maintenance of freshwater dependentecosystems in moderately modified conditions.On the other hand, if environmental waterallocations are not maintained at the estimatedEWR levels over a long term, the ecosystems ofthe basins in the yellow and red bands may not

be preserved even in moderately modified stateand are likely to deteriorate further, consequentlyaffecting local livelihoods. The circled basins infigure 6 are a few example basins where over-abstraction of water is causing problems to theecosystems and to the people that depend ontheir environmental services. These and someother basins “move” into the higher water stresscategory, if EWR are “reserved”.

The Murray–Darling Basin in Australia, with awater stress indicator greater than 1 (figure 7), isan example of an environmentally water scarcebasin. This basin, the largest in Australia, has atotal area of just over 1 million km2 with highlyuneven distribution of flow (both spatially and intime). Throughout Australia’s history, the riversin the basin have been severely modified andregulated. The main economic activity, whichuses 95 percent of the total water withdrawal inthe basin, is irrigated agriculture. This sustainedover-abstraction of water has negativelyimpacted agricultural production and has causedsevere environmental problems in the system.These impacts include high salinity levels thataffect soil productivity, massive algae booms,nutrient pollution, and the consequent loss ofnative species, floodplain areas and wetlands.Maintaining EWR at even a modest level of 30percent of the total flow to ensure a fair conditionof the river (as estimated and shown in figure 5)is a big challenge, as water consumption abovethe mouth of Murray has reduced the mean flowby nearly 80 percent.

The Orange River in southern Africa isanother example of a river with a high waterstress indicator (0.8-0.9). The basin, with anarea of just over 1 million km2 and a total flow of12 km3 , crosses the boundaries of four southernAfrican countries and carries approximately 25percent of the total surface water resources ofsemi-arid South Africa. The upstream reachesof the basin have been so severely modified andregulated that the combined reservoir storage inthe basin has already exceeded the total annualflow. This degree of modification will increase

15

FIGURE 6.A map of a water stress indicator which takes into account EWR. Areas shown in red are those where EWR presented in fig. 5 may not be satisfied under currentwater use. Most of the areas with variable flow regimes (and consequently the modest EWR of 20-30% of MAR) fall into the areas of environmental waterscarcity. The circles include example river basins which can move into a higher category of human water scacrity, if EWR are to be satisfied. The risk of notmeeting EWR will remain high in these basins, particularly as water withdrawals grow.

16

FIGURE 7.A map of the “traditional” water stress indicator (water withdrawals as a proportion of the mean annual river runoff).

17

even more with the development of LesothoHighland Water Scheme. In addition to theimpacts of dams and canals on migratory andother native freshwater species, otherenvironmental problems in the basin includedeteriorating water quality, the occurrence ofblack fly pest populations, and the proliferation ofreed beds in river channels that affect nativespecies.

The Huang-Ho River (Yellow River) Basin inChina is one of the major river basins of thecountry, with an area of almost 800,000 km2, apopulation of over 100 million, and an annualflow of 70 km3. The river has nearly reached thelevel of complete water resources exploitationand can be defined as an area in crisis both forpeople and nature. The water utilization rate inthe Huang-Ho River is about 90 percent of theavailable water. The duration of low flow periodsin the river has increased from forty days inearly 1990s to two hundred days in 1997. Asreflected in figures 6 and 7, reserving even abare minimum of 25 percent of the total flow forthe environment brings the basin to the level ofabsolute water scarcity with a water stressindicator exceeding 1.

We have overlaid the country boundaries(with country population figures at the 1995level) on the major river basin boundaries toquantify the extent of environmental waterscarcity. Basins where the current water use isalready in conflict with EWR, cover over15 percent of the world land surface and arepopulated by over 1.4 billion people in total. Aswater withdrawals increase, more river basinswill “move” from the “environmentally safe” to the“environmentally stressed” and further into the“environmentally scarce” categories. It is highlyunlikely that any transition in the reverse order ispossible if food production is not intensified inthe agricultural sector (which currently accountsfor approximately 70 percent of the total waterwithdrawals in the world), and if environmentalwater allocation is not made a common practice

and an integral part of every river basinmanagement plan.

Improving Methodology and DataCollection Efforts

This first attempt to consider ecosystem waterrequirements as a factor in the assessment ofglobal water resources used an estimate ofthese requirements based exclusively onhydrological data and simple conceptual rules.No ecological information is currently explicitlypresent in the approach.

It is clearly recognized that such quickmethods to establish EWR will not provideinformation with the level of confidence neededto guide day to day water management decisionsin individual river basins. These methods arealso unlikely to provide anything more thanestimates of total environmental water volumes,as opposed to the full environmental flow regime,which could be established using more detailedassessment approaches. However, suchestimates are vital to raise the awareness of theneed to conduct more detailed assessment onenvironmental flows taking into considerationecological variables. Global estimates, evencrude, may also positively influence waterallocation policies worldwide, by highlighting thelimitations of current water scarcity assessmentsthat do not consider aquatic ecosystems aslegitimate users of water. Increased awarenessof the need to establish environmentalallocations, and the need for more detailed dataand understanding of the relationships between aproductive and healthy aquatic ecosystem andthe flows required to sustain its functioning,should direct funding and effort into theseresearch areas. This will be a step forward inachieving sustainable use of water resources forboth people and ecosystems. The preliminaryEWR estimates presented here may and shouldbe further refined through the testing of methods

18

with local data and dialogues with stakeholderson the ground. Such refinement would effectivelyconstitute the “groundtruthing” of any desktopassessment and would provide the feedback,which is necessary to improve it.

The future development of the work reportedhere should focus on the development of a moreecologically relevant method for the assessmentof EWR. The latter would require that locallyand/or regionally available information onfreshwater biodiversity and also on the sensitivityand conservation importance of aquaticecosystems is collated and analyzed in thecontext of hydrological variability. This wouldallow a better understanding and quantification ofhydrology and aquatic ecology links to beachieved, and the realistic environmental watermanagement targets to be formulated fordifferent aquatic ecosystems, basins and regions.Some useful indices of biodiversity (fish familyrichness) and stress on water resources(wilderness measure, water resourcesvulnerability) have been suggested andcalculated for major river basins by variousdevelopment and conservation organizationssuch as WRI, WWF-US, and UNEP-WCMC(Revenga et al. 1998; 2000; Abell et al. 2000;Groombridge and Jenkins 1998). These indices,however, are not related to the productivity offreshwater dependent ecosystems, which is amore important parameter for considering thelinks between environmental flows andlivelihoods. The relevant indices of productivityshould be suggested/developed and used toimprove the existing crude estimates.

As was already mentioned, estimates ofEWR for an individual aquatic ecosystem dependupon and should be related to the environmentalmanagement target set for this system. Thestudy in its present form does not explicitlyconsider such targets. It is possible that theenvironmental management class or the desiredfuture state of a basin may be (at least at thepreliminary level of global assessment)determined using these or similar indices, where

highly diverse and/or highly stressed riversystems have the highest conservation priorityand basins with low diversity and/or low stresshave the lowest priority. Information onfreshwater biodiversity and its condition,however, is sparse worldwide, even in thosedeveloped nations that have considerablefinancial and technical resources. With a fewexceptions, such as Australia and the UnitedStates, most countries have a large informationgap regarding freshwater species, especially atlower taxonomic orders. This makes theestablishment of EWR an even greaterchallenge. Fortunately there are currently a fewongoing freshwater assessments and initiativesaround the world that focus on inland waterspecies, their status and conservation. As partof IUCN’s large-scale Water and Nature Initiative(WANI), the IUCN’s Species SurvivalCommission (SSC) and the Ramsar Bureau areengaged in synthesizing data on the status offreshwater biodiversity. In addition, the IUCN/SSC has a broader Freshwater Initiative, whichintends to identify a set of species groups toserve as indicators of freshwater biodiversity andecosystem integrity. Also, the IUCN/SSC iscompiling baseline species data includingspecies distribution maps, population trends,species’ ecological requirements (e.g., habitatpreferences, altitude ranges), the degree andtypes of threat, conservation actions (taken andproposed) and key information on the use ofeach particular species. These programs aim toaddress the gaps and the lack of globalcoverage in current information on freshwaterecosystems. It is envisaged that informationgenerated through these initiatives will enablethe development of a more ecologically soundmethod for EWR at the global scale. At thesame time, while it is logical to build on theseinitiatives, it must be noted that biodiversity asdefined in terms of species diversity is only apartial indicator of the ecosystem importance forhumans. It can omit many of the naturalresources that are important for supporting food

19

security and livelihoods. Other goods andservices of freshwater dependent ecosystemsshould also be quantified in different hydrologicalsettings and used to obtain more robust andecologically sound desktop methods for theassessment of EWR.

The study at present deals exclusively withthe river flow (discharge). Changes in flow mayimpact ecosystem processes indirectly throughwater temperature, flow velocity, water depth,etc. The sensitivity of river ecosystems to waterabstractions may be determined by channelgeometry (M.Acreman, CEH, pers.comm). Forexample, wide shallow rivers may have atendency to be sensitive to water abstractions,as any reduction in flow reduces availablehabitat (e.g. represented by wetted perimeter),while narrow deep rivers tend to be lesssensitive to water abstraction. Rivers in uplandareas with wide and shallow channels may thusbe sensitive to flow reduction, and have higherEWR, while still having flashy, variablehydrological regimes. Similarly, narrow deeprivers in lowland areas may appear to be lesssensitive to flow reduction and thus have lowerEWR, while still having invariant hydrologicalregimes. These considerations may also need tobe taken into account in the future in refining theEWR desktop assessment methods.

Another issue that may need addressingwhen discussing the establishment of EWR, iswhether the water left in the system is ofsufficient quality to sustain ecosystems and theirdependent species. Unfortunately, as in manyother aspects of freshwater resources, theinformation and data on the quality of rivers,lakes, streams and groundwater resources arelacking in most countries. Information aboutwater quality at the global level is difficult toobtain for a number of reasons. Water qualityproblems are often local, and natural waterquality is highly variable depending on the

location and season. There have been fewsustained programs for the global monitoring ofthe quality of water, and the information receivedis highly localized and far from complete. One ofthe global attempts at water quality monitoringhas been the UNEP GEMS/WATER program thatexamined data from 82 major river basinsworldwide over a period of a decade and a half.This program gathered data on a variety of waterquality issues, including nutrients, oxygenbalance, suspended sediments, salinization,microbial pollution, and acidification. Yet, thenumber of monitored watersheds was too smalland the frequency and type of measurementswere too inconsistent to paint a comprehensivepicture of the global water quality trends. Furtherstudies of this kind need more comprehensiveand systematic data collection, and monitoringshould be carried out indefinitely so that long-termtrends can be analyzed. Data needs are especiallycritical for developing countries, which often do nothave strong national monitoring programs, yet, faceserious water quality problems.

The hydrological data will remain animportant integral part of EWR assessments atthe global scale. Therefore, subsequent researchshould also focus on improved estimates of theflow time series and water use characteristics.As has already been mentioned above, flowsimulations in several basins in North Americahave produced unrealistically high values of Q90.In some other areas (e.g. upstream Nile basin,where most of the Nile water is generated andlow flows maintained partially due to lakeregulation), Q90 values were found to be closeto zero. Such problem areas will need to berevisited and model simulations, adjusted. It mayalso be useful and necessary to compare theperformance of other global hydrology models(Fekete et al. 1999; Oki et al. 2001) in thisregard.

20

Taking into Account the Spatial andTemporal Dimensions

Our study at present does not explicitly considerthe differences in the availability of waterbetween the different seasons of the year. Nordid we attempt to analyze the conditions whichoccur during droughts. However, it is clear thata number of the water-related problems,including the provision of water for ecosystems,occur during the dry season of the year andduring prolonged multi-annual droughts. Themonthly temporal resolution of simulatedhydrological data, however, allows theseanalyses to be carried out in the future. Acombination of the improved estimates of EWRand water resources availability during droughtsand dry seasons should allow deeper insightsinto water scarcity.

One of the possible spin-offs of the globalassessment of EWR could be the developmentof more detailed, regionally or basin levelfocused methods and guidelines. While a fewcountries already have expertise and a goodtrack record of developing and applying differentenvironmental water assessment methodologies,many developing countries have neither of these.For countries, where experience and relevantlegislation already exist (e.g. South Africa,Australia, New Zealand, UK), the estimates ofEWR, which emerge from this pilot globalassessment are too coarse. However, for thosecountries that do not have such experience,legislation and expertise, the global estimatesmay represent a starting point to help setpriorities for action, or at least developawareness of the possible consequences of notaccounting for environmental water allocations.Global estimates may be “downscaled” to thelevel of a particular country, basin, or ageographical region and interpreted/developedfurther, using more detailed, region/basin-specificinformation. While every attempt should bemade to make use of ecological information,

there is still an urgent need to improve thehydrologically based methodologies as well. Thework in this direction would include a moreexplicit estimation of high and low-flowrequirements for rivers with different hydrologicalregimes and/or ecological characteristics. Morespecifically, it could include baseflow filteringfrom the available flow records, or theincorporation of river classification studies (e.g.such as Haines et al. 1988). As global studyprogresses, it could be possible to design“environmentally acceptable modified flowregimes” and eventually convert them intooperational rules.

This study deals with the “current” water use(withdrawals) at the 1995 reference level. Byusing various projections of water use for thefuture, it would be possible to determine whichbasins may become environmentally waterscarce in the future and when. It could also bepossible to identify the risks of further ecologicaldegradation if an ecosystem’s EWR are not met.Many environmental NGOs, conservationorganizations and funding agencies may be ableto use such projections to aid in their prioritysetting exercises.

The current study does not distinguishbetween different types of water uses. Bycomparing the estimates of water use in aspecific sector (e.g. agriculture) with EWR, itcould be possible to establish whereenvironmental degradation is caused by a specificsector, and identify policies and managementchanges that will help improve the situation.

Finally, because many rivers crossinternational boundaries, political conflicts andnational priorities need to be considered. Thenecessity and commitment to satisfy EWR inone country, for example, may cause oraggravate a water scarcity problem in another.The lack of commitment in one of theneighboring countries to satisfy environmentalwater demand may cause similar problemswithin the basin. Global assessments, such as

21

the one presented here, may therefore point tothe regions where conflicts may occur due toenvironmental water scarcity. This may be ofparticular interest to many developmentagencies that find it difficult to decide as to howto allocate aid. A global assessment highlightingproblem areas and potential future conflictsamong water users and countries can be usedby these agencies as a tool in their resourceallocation process.

Policy Considerations and InstitutionalChallenges

This pilot study attempts to highlight the urgentneed to set and implement EWR for river basinsglobally, and suggests some directions for futureresearch. It has been largely technical. There ishowever a number of non-technical issues whichpertain to EWR, scarcity and security. It is nolonger proper to manage water in a fragmentedand sectoral manner, as has hitherto prevailed inthe world. To achieve sustainable development,future water management will require anecosystem approach, that recognizes theimportance of the functioning and integrity of theecosystem for an adequate water supply and the

attainment of development objectives (Poff et al.1997; Naiman et al. 2002).

One of the major challenges in adopting andimplementing an ecosystem approach to waterresources management is overcominginstitutional barriers. A first step in this regardwould be the establishment of basin-leveldialogues among different users to negotiate andagree on the allocation of water resources.These dialogues, in combination with improveddata on water availability, use and quality, aswell as improved information on ecosystemrequirements, can lead to a range of measuresthat would prevent future water scarcity whilemeeting development needs and maintainingfunctional ecosystems.

The incorporation of EWR in the picture ofglobal water resources assessment may changethe existing estimates of water availability indifferent regions and lead to re-evaluation of theconcepts of water scarcity. The results ofimproved global water analyses would showwhere populations and ecosystems are more atrisk of water scarcity and help formulateenvironmentally relevant policy options and waterresource management strategies for selectedcountries, basins and other levels of spatialresolution.

22

Literature Cited

Acreman, M.C. 1998. Principles of water management for people and the environment. In Water and PopulationDynamics, eds. A. de Shirbinin and V. Dompka. American Association for the Advancement of Science. pp: 321.

Abell, R.A.; D.M. Olson; E. Dinerstein; P.T. Hurley; J.T. Diggs; W. Eichbaum; S. Walters; W. Wettengel; T. Allnutt; C.J.Loucks; and P. Hedao. 2000. Freshwater Ecoregions of North America: A Conservation Assessment. WashingtonDC, USA: World Wildlife Fund. pp: 319.

Alcamo, J.; T. Henrichs; and T. Rösch. 2000. World water in 2025: Global modeling and scenario analysis. In WorldWater Scenarios Analyses, ed. F.R. Rijsberman. London: Earthscan Publications. pp: 396.

Alcamo, J.; P. Döll; T. Henrichs; F. Kaspar; B. Lehner; T. Rösch; and S. Siebert. 2003. Development and testing of theWaterGAP 2 global model of water use and availability. Hydrological Sciences Journal, 48(3): 317-338.

Arthington A.H.; S.O. Brizga; and M.J. Kennard. 1998. Comparative Evaluation of Environmental Flow AssessmentTechniques: Best Practice Framework. LWRRDC Occasional Paper 25/98. Canberra: Land and Water ResourcesResearch and Development Corporation.

Constanza, R; Ralph d’Arge; Rudolf de Groot; Stephen Farber; Monica Grasso; Bruce Hannon; Karin Limburg; ShahidNaeem; Robert V. O’Neill; Jose Paruelo; Robert G. Raskin; Paul Sutton; and Marjan van den Belt. 1997. The valueof the world’s ecosystem services and natural capital. Nature 387: 253–260.

Döll, P.; F. Kaspar; and B. Lehner. 2003. A global hydrological model for deriving water availability indicators: Modeltuning and validation. Journal of Hydrology, 270: 105-134

Dunbar, M.J.; A. Gustard; M.C. Acreman; and C.R.N. Elliott. 1998. Review of overseas approaches to setting riverflow objectives. Environmental Agency R&D Technical Report W6B (96)4. Wallingford, UK: Institute of Hydrology.pp: 61.

DWAF ( Department of Water Affairs and Forestry). 1997. White paper on a National Water Policy for South Africa.Pretoria, South Africa: Department of Water Affairs and Forestry.

FAO ( Food and Agriculture Organization). 2000. Agriculture: Towards 2015 and 2030. Technical Interim Report. Rome,Italy: Food and Agriculture Organization.

Fekete, B.M.; C.J. Vörösmarty; and W. Grabs. 1999. Global, composite runoff fields based on observed river dischargeand simulated water balance. Report No 22. Koblenz, Germany: World Meteorological Organization, Global RunoffData Centre (WMO-GRDC). pp. 41.

Groombridge, B. and M. Jenkins. 1998. Freshwater Biodiversity: A Preliminary Global Assessment. World ConservationMonitoring Centre, Cambridge, UK: World Conservation Press. pp: 104.

Haines, A.T.; B.L. Finlayson; and T.A. McMahon. 1988. A global classification of river regimes. Applied Geography 8:255-272.

Hughes, D.A. and F. Münster. 2000. Hydrological information and techniques to support the determination of waterquantity component of the ecological reserve for rivers. Water Research Commission Report N TT 137/00. Pretoria,South Africa. pp: 91.

Hughes, D.A. and P. Hannart. 2003. A desktop model used to provide an initial estimate of the ecological instreamflow requirements of rivers in South Africa. Journal of Hydrology 270: 167-181.

IWMI ( International Water Management Institute). 2000. World water supply and demand in 2025. In World WaterScenarios Analyses, ed. F.R. Rijsberman. London: Earthscan Publications. pp: 396.

Jones, G. 2002. Setting environmental flows to sustain a healthy working river. Watershed, February 2002. Canaberra:Cooperative Research Centre for Freshwater Ecology. (http://freshwater.canberra.edu.au).

23

King, J. and D. Louw. 1998. Instream flow assessments for regulated rivers in South Africa using Building BlockMethodology. Aquatic Ecosystems Health and Management 1: 109-124.

Knights, P. 2002. Environmental flows: lessons from an Australian experience. Proceedings of International Conference:Dialog on Water, Food and Environment. October 2002. Hanoi, Vietnam. pp: 18.

Molden, D.J. and R. Sakhthivadivel. 1998. Water accounting to assess use and productivity of water. InternationalJournal of Water Resources Development 15: 55-71.

Naiman R.J.; S.E. Bunn; C. Nilsson; G.E. Petts; G. Pinay; and L.C. Thompson. 2002. Legitimizing fluvial ecosystemsas users of water. Environmental Management 30: 455-467.

Oki, T.; Y. Agata; S. Kanae; T. Saruhashi; D. Yang; and K. Musuake. 2001. Global assessment of current waterresources using total runoff integrating pathways. Hydrological Sciences Journal. 46: 983-995.

Petts, G.E. 1996. Water allocation to protect river ecosystems. Regulated Rivers: Research and management12: 353-365.

Poff, N.L.; J.D. Allan; M.B. Bain; J.R. Karr; K.L. Prestegaard; B.D. Richter; R.E. Sparks; and J.C. Stromberg. 1997.The natural flow regime: A paradigm for river conservation and restoration. Bioscience 47: 769-784.

Postel, S. L.; G.C. Daily; and P.R. Ehrlich. 1996. Human appropriation of renewable fresh water. Science271: 785-788.

Postel, S. and S. Carpenter. 1997. Freshwater Ecosystem Services. In Nature’s Services: Societal Dependence onNatural Ecosystems, ed. G. C. Daily. Washington DC: Island Press. pp: 195-214.

Revenga, C.; J. Brunner; N. Henninger; K. Kassem; and R. Payne. 2000. Pilot Analysis of Freshwater Ecosystems:Freshwater Systems. Washington DC, USA; World Resources Institute. pp 83.

Revenga, C.; S. Murray; J. Abramovitz; and A. Hammond. 1998. Watersheds of the World: Ecological Value andVulnerability. Washington DC, USA: World Resources Institute and WorldWatch Institute. pp: 197.

Richter, B.D.; J.V. Baumgartner; R. Wigington; and D.P. Braun. 1997. How much water does a river need ? FreshwaterBiology 37: 231-249.

Rosenberg, D.M.; P. McCully; and C.M. Pringle. 2000. Global-scale environmental effects of hydrological alterations:Introduction. Bioscience 50: 746-751.

Shiklomanov, I.A. 1997. Assessment of water resources and water availability in the world: Comprehensive assessmentof the freshwater resources of the world. Stockholm, Sweden: Stockholm Environment Institute.

Shiklomanov, L.A. 2000. World water resources and water use: Present assessment and outlook for 2025. In WorldWater Scenarios Analyses, ed. F.R. Rijsberman. London: Earthscan Publications. pp 396.

Smakhtin, V.U. 2001. Low flow hydrology: A review. Journal of Hydrology 240: 147-186.

Smakhtin, V.U. 2002. Environmental water needs and impacts of irrigated agriculture in river basins. IWMI workingpaper N 42. Colombo, Sri Lanka: International Water Management Institute. pp: 20.

Sullivan, C.A.; J.R. Meigh; A.M. Giacomello; T. Fediw; P. Lawrence; M. Samad; S. Mlote; C. Hutton; J.A. Allan; R.E.Schulze; D.J.M. Dlamini; W. Cosgrove; J. Delli Priscoli; P. Gleick; I. Smout; J. Cobbing; R. Calow; C. Hunt; A.Hussain; M.C. Acreman; J. King; S. Malomo; E.L. Tate; D. O’Regan; S. Milner; and I. Steyl. 2003. The WaterPoverty Index: Development and application at the community scale. Natural Resources Forum, 27: 3, 25-41.

Tennant, D.L. 1976. Instream flow regimens for fish, wildlife, recreation and related environmental resources. Fisheries1: 6-10.

Tharme, R.E. 1996. Review of international methodologies for the quantification of the instream flow requirements ofrivers. Pretoria, South Africa: Department of Water Affairs and Forestry. pp: 116.

24

Vörösmarty, C.J.; P. Green; J. Salisbury; and R.B. Lammers. 2000. Global water resources: Vulnerability from climatechange and population growth. Science 289: 284-288.

UNESCO (United Nations Educational Scientific and Cultural Organization). 2003. World Water Development Report.Paris: United Nations Educational Scientific and Cultural Organization.

Research Reports

1. Integrated Land and Water Management for Food and Environmental Security.F.W.T. Penning de Vries, H. Acquay, D. Molden, S.J. Scherr, C. Valentin, andO. Cofie. 2004.

2. Taking into Account Environmental Water Requirements in Global-scale WaterResources Assessments. Vladimir Smakhtin, Carmen Revenga and Petra Döll.2004.

ISSN 1391-9407 ISBN 92-9090-542-5

Postal Address: IWMI, P O Box 2075, Colombo, Sri Lanka Location: 127 Sunil Mawatha, Pelawatte, Battaramulla, Sri Lanka Telephone: +94-11 2787404, 2784080 Fax: +94-11 2786854

Email: comp.assessment@cgiar.org Website: www.iwmi.org/assessment

Taking into Account Environmental Water Requirements in Global-scale Water Resources Assessments

Vladimir Smakhtin, Carmen Revenga and Petra Döll

Research Report 2

I n t e r n a t i o n a lWater ManagementI n s t i t u t e