Working Paper

Aqueduct METADATA document GLOBAL MAPS 2.0

Francis Gassert, Matt Landis, Matt Luck, Paul Reig, and Tien Shiao

Executive Summary CONTENTS This document describes the specific characteristics of the Executive Summary...... 1 indicator data and calculations for the Aqueduct Water Risk Total water withdrawal...... 2 Atlas Global Maps. Complete guidelines and processes for Consumptive and non-consumptive use...... 5 data collection, calculations, and mapping techniques are Total blue water (Bt)...... 6 described fully in the Aqueduct Water Risk Framework.1 The Aqueduct Water Risk Atlas makes use of a Water Risk Available blue water (Ba)...... 7 Framework (Figure 1), that includes 12 global indicators Baseline water stress...... 8 grouped into three categories of risk and one overall score. Inter-annual variability...... 9 Seasonal variability...... 10 occurrence...... 11 Aqueduct Water Risk Framework Figure 1 | Drought severity...... 12 Upstream storage...... 13 Overall Water Risk Groundwater stress ...... 14 Return flow ratio...... 15 Upstream protected land...... 16 Physical Risk Physical Risk Regulatory and Media coverage...... 17 QUANTITY QUALITY Reputational Risk Access to water...... 18 Baseline Return Flow Media Threatened amphibians...... 19 Water Stress Ratio Coverage Endnotes...... 20 Inter-annual Upstream Access Variability Protected to Water Seasonal Land Threatened Disclaimer: Working Papers contain preliminary Variability Amphibians research, analysis, findings, and recommendations. They Flood Occurence are circulated to stimulate timely discussion and critical Drought feedback and to influence ongoing debate on emerging Severity issues. Most working papers are eventually published in Upstream another form and their content may be revised. Storage Groundwater Stress Suggested Citation: Gassert, F., M. Landis, M. Luck, P. Reig, and T. Shiao. 2013. “Aqueduct Global Maps 2.0.” Working Paper. Washington, DC: World Resources Institute. Available online at http://www.wri.org/publication/aqueduct-metadata-global.

WORKING PAPER | January 2013 | 1 The data selection and validation process involves three TOTAL WITHDRAWAL steps: (1) a literature review, (2) identification of data sources in the public domain, and (3) the compilation and Description: Total withdrawal is the total amount of expert review of the selected data sources. Calculation of 6 water removed from freshwater sources for human use. of the 12 indicators required the creation of original data- sets to estimate water availability and use. The hydrologi- Calculation: Withdrawals were estimated in a two-step cal catchments were based on the Global process. First, national-level withdrawals were estimated Database developed by Masutomi et al.2 Computation of for the year 2010 using multiple regression time-series the original datasets was completed by ISciences, L.L.C. models of withdrawals as a function of annually measured indicators such as GDP, population, irrigated area, or Two measures of water use are required: water withdrawal, electrical power production. Regressions were performed the total amount of water abstracted from freshwater separately for each sector (domestic, industrial, and agri- sources for human use; and consumptive use, the portion cultural) and were used to predict withdrawals for 2010, of water that evaporates or is incorporated into a product, whenever FAO Aquastat values were older than 2008. thus no longer available for downstream use. Withdrawals Where FAO Aquastat reported withdrawals for the year for the global basins are spatially disaggregated by sector 2008 or more recent, reported values were used. based on regressions with spatial datasets to maximize the correlation with the reported withdrawals (i.e. irrigated Second, these withdrawal estimates were then spatially areas for agriculture, nighttime lights for industrial, and disaggregated by sector based on regressions with spatial population for domestic withdrawals). Consumptive use is datasets selected to maximize the correlation with the derived from total withdrawals based on ratios of consump- reported withdrawals (irrigated areas for agricultural, tive use to withdrawals by Shiklomanov and Rodda3 and nighttime lights and power plants for industrial, and Flörke et al.4 Both withdrawals and consumptive use are population for domestic withdrawals). coded at the hydrological catchment scale. Data Sources Two metrics of water supply were computed: total blue water and available blue water. Total blue water approxi- Variable Basin delineations mates natural and does not account for Y. Masutomi, Y. Inui, K. Takahashi, and Y. Authors withdrawals or consumptive use. Available blue water is Matsuoka an estimate of availability minus upstream Development of Highly Accurate Global consumptive use. Modeled estimates of water supply are Title Polygonal Drainage Basin Data calculated using a catchment-to-catchment flow accu- mulation approach developed by ISciences, L.L.C., which Year of publication 2009 aggregates water by catchment and transports it to the http://www.cger.nies.go.jp/db/gdbd/ URL next downstream catchment. Water supply is computed gdbd_index_e.html from runoff (R), the water available to flow across the landscape from a particular location, and is calculated as Resolution 1 sq. km the remainder of precipitation (P) after evapotranspira- tion (ET) and change in soil moisture storage (∆S) are accounted for (i.e., R = P – ET – ∆S). The runoff data is Freshwater withdrawal Variable courtesy of NASA Goddard Earth Sciences Data and Infor- by country mation Services Center’s Global Land Data Assimilation Author P. H. Gleick System Version 2 NOAH land surface model for the years 1950 to 2008.5 The World's Water Volume 7: Title The Biennial Report on Freshwater The remainder of this document contains the definitions, Resources, Island Press formulas, and data specifications for the Aqueduct Water Year of publication 2011 Risk Atlas global maps. URL http://www.worldwater.org/data.html

Resolution Country

2 | Aqueduct Global Maps 2.0

Data Sources

Variable Gridded population Variable Irrigated agricultural areas

Food and Agriculture Organization of the Center for International Earth Science Author Information Network (CIESIN), Colum- United Nations (FAO) Authors bia University; United Nations Food and Agriculture Organization (FAO); and Centro Title FAOSTAT Internacional de Agricultura Tropical (CIAT) Year of publication 2009 Gridded Population of the World Version Title 3 (GPWv3): Population Count Grid, Future Time covered in analysis 2009 Estimates URL http://faostat3.fao.org/home/index.html Year of publication 2005 Resolution Country Time covered in analysis 2005

URL http://sedac.ciesin.columbia.edu/gpw

Resolution 2.5 arc minute raster Variable Area equipped for

Authors K. Freydank and S. Siebert

Towards Mapping the Extent of Irriga- Variable Nighttime lights Title tion in the Last Century: Time Series of NOAA National Geophysical Data Center Irrigated Area per Country Author (NGDC) Year of publication 2008 Version 4 DMSP-OLS Nighttime Lights Title Time Series Time covered in analysis 1990– 2003

Year of publication 2010 http://publikationen.ub.uni-frankfurt.de/ URL frontdoor/index/index/docId/5916 Time covered in analysis 2000 Resolution Country http://www.ngdc.noaa.gov/dmsp/down- URL loadV4composites.html

Resolution 30 arc second raster GDP, population, agricultural Variable land, urban population,

CO2 emissions

Variable Global irrigation areas Author World S. Siebert, P. Döll, S. Feick, J. Hoogeveen, Authors and K. Frenken Title World Development Indicators

Title Global Map of Irrigation Areas Version 4.0.1 Year of publication 2011

http://data.worldbank.org/data-catalog/ Year of publication 2007 URL world-development-indicators

Time covered in analysis 2000 Resolution Country http://www.fao.org/nr/water/aquastat/ir- URL rigationmap/index60.stm

Resolution 5 arc minute raster

WORKING PAPER | January 2013 | 3 Data Sources

Electricity, total net genera- Withdrawals, precipitation, Variable tion, refinery processing gain, Variable total renewable water supply, coal production irrigated area U.S. Energy Information Administration Food and Agriculture Organization of the Author Author (EIA) United Nations (FAO)

Title International Energy Statistics Title FAO AQUASTAT http://www.fao.org/nr/water/aquastat/ URL Year of publication 2011 dbase/index.stm

http://www.eia.gov/cfapps/ipdbproject/ URL Date Accessed May 24, 2012 IEDIndex3.cfm Resolution Country Resolution Country

Total Withdrawal

4 | Aqueduct Global Maps 2.0

CONSUMPTIVE AND Data Sources NON-CONSUMPTIVE USE Variable Consumptive use ratios Description: Consumptive use is the portion of all water withdrawn that is consumed through evaporation, Authors I.A. Shiklomanov and John C. Rodda eds. incorporation into a product, or pollution, such that it is no longer available for reuse. Non-consumptive use is World at the Beginning of the remainder of withdrawals that is not consumed and Title the Twenty-First Century, International Hy- instead returns to ground or surface water bodies. drology Series, Cambridge University Press Year of publication 2004 Calculation: Consumptive use by sector is estimated from total withdrawal using consumptive use ratios by Resolution Major regions Shiklomanov and Rodda and Flörke et al. Data Sources Variable Global water use

M. Flörke, E. Kynast, I. Bärlund, S. Eisner, Variable Withdrawals Authors F. Wimmer, and J. Alcamo Comments See Total Withdrawal “Domestic and Industrial Water Uses of the Past 60 Years as a Mirror of Socio-Economic Title Development: A Global Simulation Study,” Global Environmental Change in press, 2012.

Year of publication 2012

Resolution Country

Consumptive and Non-Consumptive Use

WORKING PAPER | January 2013 | 5 TOTAL BLUE WATER (Bt) Data Sources Description: Total blue water (Bt) for each catchment is the accumulated runoff upstream of the catchment plus Variable Runoff the runoff in the catchment. National Aeronautics and Space Author Administration (NASA) Calculation: Bt(i) = Rup(i) + R(i) where Rup(i) = ∑ Global Land Data Assimilation System Title Bt(iup), iup is the set of catchments immediately upstream Version 2 (GLDAS-2) of catchment i that flow into catchment i, and Rup(i) is Year of publication 2012 the summed runoff in all upstream catchments. For first- order catchments (those without upstream catchments, Time covered in analysis 1950–2008 e.g., headwater catchments), Rup(i) is zero, and total blue http://disc.sci.gsfc.nasa.gov/hydrology/ water is simply the volume of runoff in the catchment. URL data-holdings

Data Sources Resolution 1 degree raster

Variable Basin delineations

Comments See Total Withdrawal

Total Blue Water

6 | Aqueduct Global Maps 2.0

AVAILABLE BLUE WATER (Ba) Data Sources Description: Available blue water (Ba) is the total amount of water available to a catchment before any uses Variable Runoff are satisfied. It is calculated as all water flowing into the Comments See Total Blue Water catchment from upstream catchments plus any imports of water to the catchment (Eim(i)) minus upstream con- sumptive use plus runoff in the catchment. Variable Consumptive use Calculation: Ba(i) = R(i) + Eim(i) + ∑ Qout(iup) where Qout is defined as the volume of water exiting a catch- Comments See Consumptive and Non-consumptive Use ment to its downstream neighbor: Qout(i) = max(0, Ba(i) – Uc(i) – L(i) – Ex(i)), Uc(i) are the consumptive uses, L(i) are the in- losses due to reservoirs and other infrastructure, and Ex(i) are the exports of water from catchment i. Negative values of Qout are set to zero. In first-order catchments ∑Qout(j) is zero, so available blue water is runoff plus imports.

Available Blue Water

WORKING PAPER | January 2013 | 7 Baseline water stress Data Sources Description: Baseline water stress measures total annual water withdrawals (municipal, industrial, and agri- Variable Withdrawals cultural) expressed as a percent of the total annual avail- Comments See Total Withdrawal able flow. Higher values indicate more competition among users. Arid areas with low water use are shown in gray, but scored as high stress when calculating aggregated scores. Variable Available blue water Calculation: Water withdrawals (2010) divided by mean available blue water (1950–2008). Areas with available Comments See Available Blue Water blue water and water withdrawal less than 0.03 and 0.012 m/m2 respectively are coded as “arid and low water use”.

Baseline Water Stress

8 | Aqueduct Global Maps 2.0

Inter-annual Variability Data Sources Description: Inter-annual variability measures the variation in water supply between years. Variable Total blue water Comments See Total Blue Water Calculation: Standard deviation of annual total blue water divided by the mean of total blue water (1950–2008).

Variable Consumptive use

Comments See Consumptive and Non-consumptive Use

Inter-annual Variability

WORKING PAPER | January 2013 | 9 Seasonal Variability Data Sources Description: Seasonal variability measures variation in water supply between months of the year. Variable Total blue water

Comments See Total Blue Water Calculation: Standard deviation of monthly total blue water divided by the mean of monthly total blue water (1950–2008). The means of total blue water for each of the 12 months of the year were calculated, and the vari- ances estimated between the mean monthly values.

Seasonal Variability

10 | Aqueduct Global Maps 2.0

Flood Occurrence Data Sources Description: Flood occurrence is the number of recorded from 1985 to 2011. Variable Large Flood Events G.R. Brakenridge, Dartmouth Flood Obser- Authors Calculation: Number of flood occurrences (1985-2011). vatory, University of Colorado Flood counts were calculated by intersecting hydrological Title Global Active Archive of Large Flood Events units with estimated flood extent polygons. Time covered in analysis 1985 – October 2011

http://floodobservatory.colorado.edu/ URL Archives/index.html

Date accessed October 15, 2011

Resolution Flood extent polygons (multiple scales)

The Global Active Archive of Major Flood Events aggregates flood events from news, Comments governmental, instrumental, and remote sensing sources and estimates the extent of flooding based on reports of affected regions.

Flood Occurrence

WORKING PAPER | January 2013 | 11 Drought Severity Data Sources Description: Drought severity measures the average length of droughts times the dryness of the droughts from Variable Drought severity 1901 to 2008. Authors J. Sheffield and E.F. Wood

Calculation: Drought severity is the mean of the lengths Projected Changes in Drought Occurrence un- times the dryness of all droughts occurring in an area. Title der Future Global Warming from Multi-Model, Drought is defined as a contiguous period when soil mois- Multi-Scenario, IPCC AR4 Simulations ture remains below the 20th percentile. Length is mea- Year of publication 2007 sured in months, and dryness is the average number of percentage points by which soil moisture drops below the Time covered in analysis 1901–2008 20th percentile. Drought data is resampled from original http://ruby.fgcu.edu/courses/twimberley/ URL raster form into hydrological catchments. EnviroPhilo/Drought.pdf

Resolution 1 degree raster

Sheffield and Wood’s drought dataset com- bines a suite of global observation-based datasets with the National Centers for Environmental Prediction–National Center Comments for Atmospheric Research (NCEP-NCAR) reanalysis, and creates a global drought event occurrence dataset with a spatial resolution of 1 degree.

Drought Severity

12 | Aqueduct Global Maps 2.0

Upstream Storage Data Sources Description: Upstream storage measures the water storage capacity available upstream of a location relative Variable Major and reservoirs to the total water supply at that location. Higher values B. Lehner, C. R-Liermann, C. Revenga, indicate areas more capable of buffering variations in Authors C. Vörösmarty, B. Fekete, P. Crouzet, water supply (i.e. droughts and floods) because they have P. Döll, et al. more water storage capacity upstream. Global Reservoir and (GRanD) Data- Title base Version 1.1 Calculation: Upstream storage capacity divided by the Year of publication 2011 mean of total blue water (1950–2008). Time covered in analysis 2010

Data Sources http://atlas.gwsp.org/index. URL php?option=com_content

Variable Total blue water Resolution Dams (point)

GRanD database includes reservoirs with Comments See Total Blue Water a storage capacity of more than 0.1 cubic km although many smaller reservoirs were Comments included. The database includes approxi- mately 6,862 dams and reservoirs around the world.

Upstream Storage

WORKING PAPER | January 2013 | 13 Groundwater Stress Data Sources Description: Groundwater stress measures the ratio of groundwater withdrawal relative to its recharge rate over a Variable Groundwater footprint given . Values above one indicate where unsustain- T. Gleeson, Y. Wada, M.F. Bierkens, and Authors able groundwater consumption could affect groundwater L.P. van Beek availability and groundwater-dependent ecosystems. Water Balance of Global Revealed Title by Groundwater Footprint Calculation: Groundwater footprint divided by the aqui- Year of publication 2012 fer area. Groundwater footprint is defined as A[C/(R − E)], where C, R, and E are respectively the area-averaged Time covered in analysis 1958–2000 annual abstraction of groundwater, recharge rate, and the http://www.nature.com/nature/journal/ groundwater contribution to environmental stream flow. URL v488/n7410/full/nature11295.html A is the areal extent of any region of interest where C, R, and E can be defined. Resolution Polygons

Groundwater Stress

14 | Aqueduct Global Maps 2.0

Return Flow Ratio Data Sources Description: Return flow ratio measures the percent of available water previously used and discharged upstream Variable Non-consumptive use as wastewater. Higher values indicate higher dependence Comments See Consumptive and Non-consumptive Use on treatment plants and potentially lower water quality in areas that lack sufficient treatment infrastructure and pol- icies. Arid areas with low water use are shown in gray, and scored as low stress when calculating aggregated scores. Variable Available blue water

Calculation: Upstream non-consumptive use divided by Comments See Available Blue Water the mean of available blue water (1950–2008). Areas with available blue water and accumulated upstream non-con- sumptive use less than 0.03 and 0.012 m/m2 respectively are coded as “arid and low water use”.

Return Flow Ratio

WORKING PAPER | January 2013 | 15 Upstream Protected Land Data Sources Description: Upstream protected land measures the percentage of total water supply that originates from pro- Variable Protected areas tected ecosystems. Modified land use can affect the health International Union for Conservation of of freshwater ecosystems and have severe downstream Nature (IUCN) and United Nations Environ- Authors impacts on both water quality and quantity. ment Programme World Conservation Monitoring Centre (UNEP–WCMC)

Calculation: Percentage of total blue water that origi- Title The World Database on Protected Areas nates in protected areas. IUCN category V protected lands, as well as a large number of unclassified proposed lands, URL http://protectedplanet.net/ breeding centers, municipal parks, cultural and historic sites, and exclusively marine areas, are excluded. Date accessed June 14, 2012 Resolution Protected areas (multiple scales) Data Sources

Variable Total blue water

Comments See Total Blue Water

Upstream Protected Land

16 | Aqueduct Global Maps 2.0

Media Coverage Data Sources Description: Media coverage measures the percentage of all media articles in an area on water-related issues. Variable Media coverage Higher values indicate areas with higher public awareness Author Google about water issues, and consequently higher reputational risks to those not sustainably managing water. Title Google News

Calculation: Percentage of all media articles on water Time covered in analysis 2002–2012 scarcity and/or pollution. Google Archives was used to http://news.google.com/news/advanced_ URL search a string of keywords including a river name, “water news_search?as_drrb=a shortage” or “water pollution,” and an administrative unit, e.g. “River+ water shortage + Country.” The time frame Date accessed September 26, 2012 was limited to the past 10 years from January 1, 2002 to Resolution Country December 31, 2011. For each country, the number of articles Media articles are limited to on water shortage and water pollution was summed and Comments divided by the total number of articles on any topic found English articles. when searching for the administrative unit.

Media Coverage

WORKING PAPER | January 2013 | 17 Access to Water Data Sources Description: Access to water measures the percentage of population without access to improved drinking water Variable Access to water sources. Higher values indicate areas where people have less World Health Organization (WHO) and the Authors access to safe drinking water, and consequently higher repu- United Nations Children’s Fund (UNICEF) tational risks to those not using water in an equitable way. WHO / UNICEF Joint Monitoring Title Programme (JMP) for Water Supply Calculation: Percentage of population without access to and Sanitation improved drinking-water sources. An improved drinking- Year of publication 2012 water source is defined as one that, by nature of its con- struction or through active intervention, is protected from Time covered in analysis 2010 outside contamination, in particular from contamination with fecal matter. URL http://www.wssinfo.org/

Resolution Country

Access to Water

18 | Aqueduct Global Maps 2.0

Threatened Amphibians Data Sources Description: Threatened amphibians measures the percentage of freshwater amphibian species classified by Variable Threatened amphibians IUCN as threatened. Higher values indicate more fragile International Union for Conservation of Author freshwater ecosystems and may be more likely to be sub- Nature (IUCN) ject to water withdrawal and discharge regulations. Title The IUCN Red List of Threatened Species

Calculation: The percentage of amphibian species clas- Year of publication 2010 sified by IUCN as threatened in a particular area. For each catchment, the total number of threatened freshwater Time covered in analysis 2010 amphibian species was counted and divided by the total http://www.iucnredlist.org/technical-docu- URL number of freshwater amphibian species whose ranges ments/spatial-data#amphibians overlap the catchment. Catchments with fewer than two amphibian species were excluded. Resolution Polygons Freshwater amphibian species status database is joined to the known species Comments range spatial data. Several name correc- tions were made in joining the data.

Threatened Amphibians

WORKING PAPER | January 2013 | 19 Endnotes About WRI 1. Paul Reig, Tien Shiao, and Francis Gassert. Aqueduct Water Risk Frame- WRI focuses on the intersection of the environment and socio-economic work, WRI Working Paper, Washington DC: World Resources Institute, development. We go beyond research to put ideas into action, working forthcoming. globally with governments, business, and civil society to build transformative solutions that protect the earth and improve people’s lives. 2. Yuji Masutomi, Yusuke Inui, Kiyoshi Takahashi, and Yuzuru Matsuoka. “Development of Highly Accurate Global Polygonal Drainage Basin Data,” Hydrological Processes 23: 572-84, DOI: 10.1002/hyp.7186, 2009. About the Authors 3. I.A. Shiklomanov and John C. Rodda, eds. World Water Resources at the Beginning of the Twenty-First Century, International Hydrology Series, Francis Gassert is a research assistant with the Markets and Enterprise Cambridge University Press, 2004. Program at WRI, where he manages the data collection and GIS analysis of the Aqueduct project. 4. M. Flörke, E. Kynast, I. Bärlund, S. Eisner, F. Wimmer, J. Alcamo, “Domestic Contact: [email protected]. and Industrial Water Uses of the Past 60 Years as a Mirror of Socio- Economic Development: A Global Simulation Study,” Global Environmental Matt Landis is a research scientist at ISciences, L.L.C., where he develops Change, in press, 2012. and applies hydrological algorithms and models.

5. National Aeronautics and Space Administration (NASA). Global Land Data Matt Luck is a research scientist at ISciences, L.L.C., where he develops and Assimilation System Version 2 (GLDAS-2), Goddard Earth Sciences Data applies hydrological algorithms and models. Information Services Center, 2012. Paul Reig is an associate with the Markets and Enterprise Program at WRI, where he leads the design and development of the Aqueduct project. Acknowledgments Contact: [email protected]. This publication was made possible thanks to the ongoing support of the Tien Shiao is a Senior Associate with the Markets and Enterprise Program World Resources Institute Markets and Enterprise Program and the Aqueduct at WRI, where she manages the application and road testing of the Aqueduct Alliance. The authors would like to thank the following people for providing project for companies and investors. invaluable insight and assistance: Nicole Grohoski, Thomas Parris, Pragya- Contact: [email protected]. jan Rai, Tianyi Luo, Robert Kimball, Betsy Otto, Charles Iceland, and Kirsty Jenkinson as well as Nick Price and Hyacinth Billings for graphic support and final editing. For their extensive technical guidance and feedback during the development of the Aqueduct Water Risk Atlas Global Maps, the authors WITH SUPPORT FROM would also like to thank: The Aqueduct Alliance:

 Robin Abell, World Wildlife Fund  Goldman Sachs  David Cooper, World Resources Institute  General Electric  Martina Flörke, University of Kassel  Skoll Global Threats Fund  Tom Gleeson, McGill University  Bloomberg  Cy Jones, World Resources Institute  Talisman Energy Inc.  Mindy Selman, World Resources Institute  Dow Chemical Company  Justin Sheffield, Princeton University  Royal Dutch Shell  Richard Vogel, Tufts University  Dutch Government  Yoshihide Wada, Utrecht University  United Corporation  DuPont  John Deere  Procter & Gamble Company

Copyright 2013 World Resources Institute. This work is licensed under the Creative Commons Attribution 3.0 License. To view a copy of the license, visit http://creativecommons.org/licenses/by/3.0/

10 G Street, NE | Washington, DC 20002 | www.WRI.org