Summary of world waterresources
This chapter presents the results of the study for the world(Annex 3 Maps 1 and 2). Chapter 5 further develops the analysis of the resultsfor ten large regions of the world showing distinct climatecharacteristics.
World water resources
Table 2 presents the results of the global water resourcesreview by region (Annex 3 Map 3). This section comments briefly on theparticularities which can be observed on a large scale across the world, as wellas difficulties which emerge from the information collected. The table in Annex2 provides details on individual countries and territories. It presentsinternal, external and total water resources computed according to the methoddescribed in Chapter 3 (differentiating between natural and actual resources).It does not indicate the exploitable resources and the non-renewable resourcesas figures are available for only a few countries. Figures for water resourcesper inhabitant are as in 2000.
The total water resources in the world are estimated in theorder of 43 750 km3/year, distributed throughout the world accordingto the patchwork of climates and physiographic structures. At the continentallevel, America has the largest share of the worlds total freshwaterresources with 45 percent, followed by Asia with 28 percent, Europe with 15.5percent and Africa with 9 percent.
In terms of resources per inhabitant in each continent,America has 24 000 m3/year, Europe 9 300 m3/year, Africa 5000 m3/year and Asia 3 400.1 m3/year.
The distribution ofwater
At a country level, there is an extreme variability in TRWR:from a minimum of 10 m3/inhabitant in Kuwait to more than 100 000m3/inhabitant in Canada, Iceland, Gabon and Suriname. For 19countries or territories, the TRWR per inhabitant are less than 500m3; and the number of countries or territories with less than 1 000m3/inhabitant is 29. The ten poorest countries in terms of waterresources per inhabitant are Bahrain, Jordan, Kuwait, Libyan Arab Jamahirya,Maldives, Malta, Qatar, Saudi Arabia, United Arab Emirates and Yemen. In thelarge countries, water resources are also distributed unevenly in relation tothe population.
In addition to spatial variability, there is a highvariability in time within the year or among different years. This study doesnot include temporal variability but Shiklomanov (2000) provides estimates on a67-year data set of the minimum and maximum internal resources for 50countries.
TABLE 2: World water resources,by region
Variability in dry years is important as it may reducesignificantly the rainfall and the volume manageable even in relatively humidareas.
Nine countries are the world giants in terms of internal waterresources, accounting for 60 percent of the worlds natural freshwater(Table 3). At the other extreme, the water poor countries are usually thesmallest (notably islands) and arid ones (Table 4). The thresholds of 1 000 and500 m3/inhabitant correspond respectively to the water stress andwater scarcity levels proposed by Falkenmark (1986). In an average year, 1 000m3 of water per inhabitant can be considered as a minimum to sustainlife and ensure agricultural production in countries with climates that requireirrigation for agriculture.
Thirty-three countries depend on other countries for over 50percent of their renewable water resources: Argentina, Azerbaijan, Bahrain,Bangladesh, Benin, Bolivia, Botswana, Cambodia, Chad, Congo, Djibouti, Egypt,Eritrea, Gambia, Iraq, Israel, Kuwait, Latvia, Mauritania, Mozambique, Namibia,Netherlands, Niger, Pakistan, Paraguay, Portugal, Republic of Moldova, Romania,Senegal, Somalia, Sudan, Syrian Arab Republic, Turkmenistan, Ukraine, Uruguay,Uzbekistan, Viet Nam and Yugoslavia.
Exploitable renewableresources
The scarcity and disparity of water resources are exacerbatedby differing levels of usability (and therefore mobilization costs) andparticularly by environmentally sustainable usability. Water quality alsodiffers. Only a part of natural water resources can be contained and utilized.Basin management is generally recommended but is not common practice; it isunsuitable for arid areas (with no functional basin), large karstic zones andhighly fragmented basins.
Scarcity and disparity are intensified by the threat andimpact of human activity that disrupts water regimes and leads to adeterioration in water quality, and also by the vulnerable nature of somechronically overutilized resources: salinization of coastal aquifers (e.g. Spainand Israel) and the disappearance of sources (e.g. Tunisia). Moreover, thepartitions between numerous countries (the Balkans, the Nile Basin) make thesituation more complex.
Therefore, the inequalities among countries in accessingfreshwater are amplified when considering the differences in development,treatment and rehabilitation works, as are the related costs required to obtainexploitable natural resources. The effort required varies significantly with theaccessibility and regularity of the resources. For example, the ratio ofexploitable water resources to total renewable resources is close to 100 percentin areas of the Mediterranean where the main source of water is groundwater(Israel, Gaza Strip (Palestinian Authority) and Libyan Arab Jamahiriya) butgenerally less than 70 percent in countries where surface water resources areimportant (Turkey, Morocco, Greece, etc.) and even lower where there are majortechnical constraints (Malta) or political restrictions (Portugal) (Table5).
Assessment of results
As the data collection was based essentially on a literaturereview both at a country level and within FAO, the quality of the results isrelated closely to national data production and reporting systems. Theconsistency of results at regional level was checked carefully.
TABLE 3: Water richcountries
TABLE 4: Water poorcountries
TABLE 5: Exploitable resourcesin Mediterranean countries
TABLE 6: Comparison of waterresources data from different sources, the United States ofAmerica
The option chosen was to rely on country information. This wasbased on the assumption that no regional information can be more accurate thanstudies carried out at country level. However, there are a number ofdifficulties when dealing with national sources:
In most cases, acritical analysis of the available information is necessary to ensureconsistency between the different data collected for a country and a riverbasin.
Gatheringdata from different sources highlights similarities between the differentsources, contradictions, and errors in data transcription. For example, in Table6 some references indicate different figures for the United States of America.Such discrepancies could probably be explained by country experts and may be dueto different aggregations and accounting methods.
Very little information existson water resources in humid Africa.
In aridand semi-arid climates, abundant literature exists as water plays an importantrole in economic development. However, access to information on water resourcesis sometimes restricted for strategic reasons.
Theaccuracy and reliability of information vary significantly between regions,countries and types of data. No consistency can be ensured at regional level onthe duration and dates of the period of reference.
Data evaluation
This section presents some of the difficulties in computingexternal water resources.
Precipitation
Precipitation is indicated in the country summary tables as areference but is not used in the data computation. The average volumes forprecipitation given by countries are generally estimated from precipitation maps(isohyets). Country values were obtained from country surveys, but the overallreliability of the results was considered low because of the lack of historicaldata sets or poor geographical coverage of the country. Therefore, theprecipitation data used in this study are based on data prepared and publishedby the Intergovernmental Panel on Climate Change (IPCC) (New et al.,1999; Mitchell et al., 2001). Box 8 provides some examples of existingmonitoring systems.
BOX 8 - HYDROLOGICAL OBSERVATIONS ANDMEASUREMENTS
The quality of knowledge in hydrology is dependent on theavailability of historical data sets and therefore on the continuity of datacollection. Hence, current efforts at observation and investigation are not ofimmediate use. The collecting of climate data started in the nineteenth centuryin some countries (Table 8). However, hydrological and hydrogeologicalmeasurements are more recent, and sensitive to political change. Therefore, longhistorical data sets are rare. However, the longest ones concern arid countrieswhere water development has long been important. An indication of the currentefforts to produce basic data is evidenced by the statistics on the measurementnetworks (precipitation and runoff) gathered by the WMO and presented in Table 7on the Mediterranean countries.
This compilationis probably not complete and may have its shortcomings. However, it doeshighlight very diverse situations:
Rainfallnetworks: their density is relatively hom*ogeneous in Europe (about 10 stationsper 1 000 km2), higher in the islands (highest in Malta with 168) andin the small countries of the Near East (Israel, Lebanon), considerably lower inthe Maghreb (fewer than 1 per 1 000 km2 in Algeria and Morocco) andin Turkey, and very low in Egypt and Libyan Arab Jamahiriya (very aridanyway).
Hydrography networks: widely varying density in Europe (2-8stations per 1 000 km2) and in the Near East (1-10 stations per 1 000km2), and lower in Africa. However, these densities are notcomparable as the level of knowledge and the hydrography structures differgreatly.
Surface water(renewable)
The following comments stem from an analysis of the surfacewater resources at river-basin level and by country:
The figures oninflow come from national reports. They may correspond either to actual inflowaffected by upstream consumption or to theoretical inflow secured throughtreaties. Therefore, the actual inflow in certain countries may be differentfrom the inflow secured by treaties indicated in the tables.
Anotherdifficulty related to the inflows or outflows secured by treaties relates to thefact that most treaties establish the exchanges between countries in terms ofpercentage of basin water resources, whose amount varies from year toyear.
For flowsnot subject to formal treaties or agreements, the values indicated in the tablesshould correspond to average annual actual flows in recent years. However, owingto data collection difficulties, the flow average was not obtainedsystematically for the same period. Therefore, the figures should be consideredas best possible estimates.
The quality of the estimationof internal surface resources depends on:
i. the density of the hydrometricstations in each country (related to the structure of the hydrography network)and the length of the observation period (Box 8);
ii. the percentage of the territory wherethe runoff is measured and where it is calculated from modelling using climatedata sets.
TABLE 7: Measurement stationsin Mediterranean countries
Source: Margat, in preparation
TABLE 8: Observation stationsin the world
Source: Shiklomanov, 1997
Groundwater (renewable)
Depending on the source, the value provided under groundwaterresources may indicate either the groundwater recharge or the groundwaterproductivity.
It is difficult to evaluate the groundwater flow entering orleaving a country. The transboundary groundwater flows are generally very smallin comparison with the surface water flows. Therefore, uncertainties about themdo not affect the results significantly. This is also generally the case for thegroundwater outflow into the sea, except for Lebanon and some small islands suchas Malta.
Total water resources
In the case of conflicting sources of information, thedifficulty lies in selecting the most reliable one. In some cases, waterresources figures vary considerably from one source to another. There may bevarious reasons for such differences:
Differences incomputation methods or definitions used in computing water resources.
Differences in the referenceperiod used to calculate water resources.
Overestimation of resourceswhere there is double counting of surface water and groundwater.
Specificproblems of transboundary rivers. Methods used by countries to computetransboundary rivers flows are not always transparent.
Misuse ofthe concept of renewable water resources. Some sources include extraction offossil water as part of water resources. Others include secondary sources ofwater such as wastewater or return flows from agriculture.
Changesin estimates, often upwards, following improvements in knowledge, methods ormeasurement networks. For example, for three Maghreb countries (Tunisia,Algeria, and Morocco) the average total flow increased by 20 percent in 20years, from 38 km3/year in 1970 to 48 km3/year in1990.
Data production
Where national data are absent or not reliable, it may benecessary to obtain estimates from models and satellite imagery. However, whilemodelled data may be useful, they cannot replace local measurements. The sourcesrarely provide information on the origin of data (meta-data: which dataproduction, monitoring and treatment schemes; when exceptions from the ruleswere applied; differences in definitions; etc.) and on data processing (how thedata were extrapolated, how data-gap situations were solved, etc.).
Comparison with previousstudies
Because of the global approaches used in all but one of theprevious studies mentioned in Chapter 1 (Table 9), comparisons between them andthe information collected in this report were possible only at continental orglobal level.
Prior to analysing the figures, some clarification isnecessary. First, this study used the figures proposed by Lvovitch (1974)for 10 out of 53 countries in Africa as no better information was found inrecent country reports. These ten countries are located in well-endowed regions,and together they account for about 54 percent of the total water resources ofthe Africa region. Thus, the comparison with Lvovitchs figures isrelatively biased.
TABLE 9: Major assessments ofthe worlds natural internal freshwater resources
The figures given by the World Resources Institute werecomputed using various data sources: FAO (Aquastat), the Institute of Geographyof the former Soviet Union, and Shiklomanov (2000). Gleick (1993, 1998 and 2000)provides an exhaustive compilation work and includes data from FAOsAquastat programme. These indications may help explain the similar resultsobtained by these different sources.
It is not the purpose of this comparison to explaindifferences in results. Computing methods and the assumptions they imply are sodifferent that such an exercise would be of limited value. Rather, this studyconfirms the relatively good knowledge of the state of the worlds waterresources at global and regional scales. This study indicates a world waterresources total of 43 764.3 km3/year, between the 42 780km3/year indicated by Shiklomanov (2000) and the 44 540km3/year indicated by Gleick (2001). The main differences relate totwo regions: Europe and Oceania. For Europe, the main reason relates toaccounting for the Russian Federation. The Aquastat programme considers all theRussian Federation to be in the Europe region whereas the other studies includedpart of it in Asia. The case of Oceania is unclear but probably relates toaccounting for the water resources of the islands (the estimates for Australiaare the same).
The potential of models for globalwater resources assessment
In this study, water resources assessment at country level wasbased mainly on hydrological information on the main rivers extrapolated toareas where direct measurements were not available. Although all efforts weremade to present a standard framework for water resources computation, themethodology used in this study (relying on country information), does not ensureconsistency in the water resources assessment methods from one country toanother.
In order to overcome this problem and to improve thecomparability of the water information at regional and global levels, FAOsAquastat programme is also working on the development of global GIS-based datasets and modelling tools.
A water balance model has been developed and implemented onAfrica. The results are presented in the CD-ROM Atlas of water resourcesand irrigation in Africa (FAO, 2001). Available information on Africa wasprocessed through a continental GIS-based model to provide a comprehensivepicture of the different elements of the water balance at continent scale. Thisapproach makes the best use of scattered information and enables extrapolationof point data or data available at country level to develop a credible pictureof the situation of the continents water use and its impact on waterresources. It also has the advantage of presenting a hom*ogenous methodology forcomputing the water balance across the continent.
Description of the model
The model used in the Africa study is simple and performedentirely within the GIS environment. It makes the best possible use of availableinformation, be it regional coverage of the main climate elements of the waterbalance (precipitation and crop water requirements), soil properties, orirrigation. The model consists of two parts. A vertical soil-water balancemodel, performed monthly for every grid cell (10 km × 10 km), computes thepart of precipitation that does not return to the atmosphere throughevapotranspiration. This water, termed surplus in the study, is then routedthrough the landscape in the rivers by the horizontal part of the model. In GIS,this is performed by generating a grid-based hydrological network based on anavailable digital elevation model.
Crop water requirements were calculated using the modifiedPenman-Monteith method as described in FAO (1998c). They were calculated foreach grid cell on a monthly time step and compared with the actualevapotranspiration, ETa(m), resulting from the soil water balance model. Thedifference was then multiplied by the cropping intensity to obtain a monthlygrid of irrigation water requirements. The model was calibrated as far aspossible against measured natural river-flow data.
Data sources used in themodel
The precipitation data used for this study are based on dataprepared and published by the International Institute for Applied SystemsAnalysis (IIASA). For each of the stations used in the gridding exercise, datawere averaged over a 30-year period from 1961 to 1990.
Data on reference evapotranspiration used for this study hadalso been prepared by the IIASA for FAO (FAO, 2000). The resolution of this dataset is equal to the resolution of the precipitation data set, 0.5 degrees oflatitude by 0.5 degrees of longitude, with mean monthly values for global landareas (excluding Antarctica) for the period 1961-1990. The data set was preparedaccording to the FAO Penman-Monteith method with limited climate data asdescribed in FAO (1998c). The input data used to calculate ETo are part of theCRU Global Climate Dataset prepared by the Climate Research Unit of theUniversity of East Anglia, the United Kingdom, and distributed through the Website of the IPCC.
The digital data layer with the drainage pattern used for thisproject was a 1:5 000 000 line coverage with the rivers of Africa (digitized in1994 for the UNEP/FAO Desertification Assessment and Mapping Project). Thecoverage with water bodies originates from the Digital Chart of the World 1:1000 000. The water bodies of Africa have been characterized (as lake, lagoon,reservoir, etc.) and named (where names were readily available). The data layeras used in this project contains all the water bodies that had a name and werenot characterized as rivers.
TABLE 10: Comparison of measuredand modelled data for African countries
Results of the model and comparisonwith country-based data
Table 10 compares the IRWR as published by FAO (Aquastat) andpresented in this report (table in Annex 2) with the values computed by themodel. The calculated values in Table 10 are generally lower than thecountry-based data. This is especially apparent in the more arid countries. Forthese countries, the model calculates hardly any runoff while the countrystatistics indicate some renewable water resources. The water balance model usedfor this study computes the internally generated water resources (IRWR) bysubtracting the total inflow to the country from the total flow accumulationleaving the country, disregarding water leaving the system by evaporation fromlarge lakes and wetlands. In arid areas, this method leads to an underestimationwhen compared with the results of conventional studies that estimate the waterpotential through the recharge of groundwater and the river discharge at thepoints where the runoff is maximum.
Another explanation for the difference with country-based datais that the time-span used for the model is a month. In arid areas, such a longtime-span tends to overestimate evaporation, thus reducing the estimate of waterresources. It is difficult to assess the reliability of the models resultsfor the semi-arid and arid countries (e.g. Egypt, Djibouti, Eritrea, Morocco andNamibia). In other cases, the model may give a higher figure than the nationalstatistics as it assesses resources before local evaporations happen. This isthe case for the Libyan Arab Jamahiriya, Algeria and Burkina Faso.
In humid areas, the comparison shows a relatively goodconcordance between country-based and modelled figures. In some countries of theintertropical humid area (Democratic Republic of Congo, Gambia, Mali, SierraLeone, Somalia and Uganda), the difference between national statistics andcalculated ones is negative. This might indicate a problem in thesecountries estimates of water resources, in particular in the distinctionbetween internal and external water resources.
This modelling exercise shows how it may be necessary toobtain estimates of water resources from models where national data are absentor unreliable. The model is a useful tool for checking the overall results ofthe study and for pinpointing possible errors. The model was used to cross-checkthe Africa data sets. Where there were clear inconsistencies, the country waterbalances were reviewed and modified as necessary. Therefore, the combined use ofcountry-based data and global water-balance modelling can enhance the overallreliability of the results.
Concluding remarks
This report presents the approach used by FAO to assessnatural and actual water resources for the world by country. It deals withrenewable freshwater resources and concentrates mainly on the physicalassessment of internal and external resources. It presents a picture of thestate of the worlds water resources that is not only the natural state butalso the current situation, taking into account existing uses of water and theirimplications for countries sharing common river basins. It is also a firstattempt, albeit still an incomplete one, to present estimates of exploitablewater resources, i.e. the part of the countries water resources that canbe put to beneficial use.
The major characteristics of this approach are:
It proposes acomprehensive way to: (i) compute surface water resources and groundwaterresources; (ii) avoid double counting; and (iii) assess resources in atransparent manner from available national information.
It is transparent in the waywater resources figures are calculated.
Itreviews in depth the water exchanges at border level in order to ensureconsistency in the results between countries.
Itintroduces the concept of exploitable resources in order to obtain a morerealistic estimate of water resources availability.
It distinguishes clearlybetween renewable and non-renewable resources.
Itenhances the overall quality of country-based data by comparing them to theresults of a global water-balance model. It highlights drawbacks andoverestimation in existing data sets. Both tools are mutuallyreinforcing.
Future improvements
There remains much to do in order to obtain sound statisticson water resources, and particularly standardized data sets, at global level.Therefore, the methodology used in this study to compute water resources isintentionally simple and based on transparent rules. However, more effort needsto be focused on the assessment of the variability of water resources in space(watershed level), in time (dry-year resources) and according to constraints(exploitable resources). Desegregated information at river-basin level isparticularly important in large countries with very diverse climate conditions(e.g. Russian Federation, Brazil, United States of America, and China). Nationalaverages hide local differences and, for large countries, are of little use forassessing the countrys water situation.
The use of global data sets (meteorological, etc.) coupledwith water-balance models can contribute to improving the assessment of waterresources as shown above for Africa. This experience should be extended to therest of the world, keeping in mind that the field-based approaches (based onmeasurements) and the modelling approaches are complementary.
The future of waterresources
This study bases the country-level estimates of naturalrenewable water resources mainly on climate and hydrological data sets for thelast decades of the twentieth century. They are representative of the averageflows of the last 25 to 50 years. These averages are considered to be stable andnot affected by change. However, there is no certainty that these averages willremain stable in the long term.
There is no certainty today about the extent, dynamics andregional distribution of climate changes forecast for the twenty-first century.However, the potential impacts on water resources cannot be ignored.
The first world projections on this subject such as those madeby the University of Kassel (Alcamo et al. 1999, for the World WaterVision) provide country projections for 2025 and 2075. Although it may be tooearly to draw a picture of the future of water resources, water developmentplanning must take into account the uncertainty related to the possible impactsof climate change on water resources within the context of growing waterdemand.
