Resources

Total water consumption in the USA in 2015 was 1218 billion litres per day, of which thermoelectric power used 503 billion litres, irrigation used 446 billion litres and 147 billion litres per day went to supply 87% of the US population with potable water13.
Data centres consume water across two main categories: indirectly through electricity generation (traditionally thermoelectric power) and directly through cooling. In 2014, a total of 626 billion litres of water use was attributable to US data centres4. This is a small proportion in the context of such high national figures, however, data centres compete with other users for access to local resources. A medium-sized data centre (15 megawatts (MW)) uses as much water as three average-sized hospitals, or more than two 18-hole golf courses14. Some progress has been made with using recycled and non-potable water, but from the limited figures available15 some data centre operators are drawing more than half of their water from potable sources (Fig. 2). This has been the source of considerable controversy in areas of water stress and highlights the importance of understanding how data centres use water.
Fig. 2: Water source by year for Digital Realty, a large global data centre operator.

Consumption from potable water was 64% (2017), 65% (2018) and 57% (2019)15.

Full size image

This section considers these two categories of data centre water consumption.
Water use in electricity generation
Water requirements are measured based on withdrawal or consumption. Consumption refers to water lost (usually through evaporation), whereas water withdrawal refers to water taken from a source such as natural surface water, underground water, reclaimed water or treated potable water, and then later returned to the source16.
Power plants generate heat using fossil fuels such as coal and gas, or nuclear fission, to convert water into steam which rotates a turbine, thereby generating electricity. Water is a key part of this process, which involves pre-treating the source water to remove corroding contaminants, and post-treatment to remove brines. Once heated into steam to rotate the turbine, water is lost through evaporation, discharged as effluent or recirculated; sometimes all three16.
The US average water intensity for electricity generation for 2015 was 2.18 litres per kilowatt hour (L/kWh)17, but fuel and generator technology type have a major impact on cooling water requirements. For example, a dry air cooling system for a natural gas combined cycle generator consumes and withdraws 0.00–0.02 L/kWh, whereas a wet cooling (open recirculating) system for a coal steam turbine consumes 0.53 L/kWh and withdraws 132.5 L/kWh. Efficiency varies significantly, with consumption ranging from 0.00 to 4.4 L/kWh and withdrawal ranging from 0.31 to 533.7 L/kWh depending on the system characteristics16.
Hydropower systems also use large volumes of water despite being considered a cleaner source of electricity. Water evaporation from open reservoirs is a major source of losses, particularly in dry regions and where water is not pumped back into the reservoir or passed onto downstream users. The US national average water consumption for hydropower is 16.8 L/kWh compared to 1.25 L/kWh for thermoelectricity17.
With the majority of generation still from fossil fuels18, the transition to renewables is important for both carbon and water intensity. Only solar and wind energy do not involve water in generation, yet both still consume water in the manufacturing and construction processes9. Estimates suggest that by 2030, moving to wind and solar energy could reduce water withdrawals by 50% in the UK, 25% in the USA, Germany and Australia and 10% in India19.
In the data centre sector, Google and Microsoft are leading the shift to renewables. Between 2010 and 2018, the number of servers increased 6 times, network traffic increased 10 times and storage capacity increased by 25 times, yet energy consumption has only grown by 6%6. A major contributor to this has been the migration to cloud computing, as of 2020 estimated to be a $236 billion market20 and responsible for managing 40% of servers4.
Due to their size, the cloud providers have been able to invest in highly efficient operations. Although often criticised as a metric of efficiency21, an indicator of this can be seen through low power usage effectiveness (PUE) ratios. PUE is a measure of how much of energy input is used by the ICT equipment as opposed to the data centre infrastructure such as cooling22, defined as follows:

$${rm{PUE}}=frac{{rm{Data}} {rm{Centre}} {rm{Total}} {rm{Energy}} {rm{Consumption}}}{{rm{ICT}} {rm{Equipment}} {rm{Energy}} {rm{Consumption}}}$$
(1)

PUE is relevant to understanding indirect water consumption because it indicates how efficient a particular facility is at its primary purpose: operating ICT equipment. This includes servers, networking and storage devices. An ideal PUE of 1.0 would mean 100% of the energy going to power useful services running on the ICT equipment rather than wasted on cooling, lighting and power distribution. Water is consumed indirectly through the power generation, so more efficient use of that power means more efficient use of water.
Traditional data centres have reported PUEs reducing from 2.23 in 2010 to 1.93 in 20206. In contrast, the largest “hyperscale” cloud providers report PUEs ranging from 1.25 to 1.18. Some report even better performance, such as Google with a Q2 2020 fleet wide trailing 12-month PUE of 1.1023.
As data centre efficiency reaches such levels, further gains are more difficult. This has already started to show up in plateauing PUE numbers24, which means the expected increase in future usage may soon be unable to be offset by efficiency improvements25. As more equipment is deployed, and more data centres are needed to house that equipment, energy demand will increase. If that energy is not sourced from renewables, indirect water consumption will increase.
Power generation source is therefore a key element in understanding data centre water consumption, with PUE an indicator of how efficiently that power is used, but it is just the first category. Direct water use is also important—all that equipment needs cooling, which in some older facilities can consume up to 30% of total data centre energy demand26,27,28.
Water use in data centre cooling
ICT equipment generates heat and so most devices must have a mechanism to manage their temperature. Drawing cool air over hot metal transfers heat energy to that air, which is then pushed out into the environment. This works because the computer temperature is usually higher than the surrounding air.
The same process occurs in data centres, just at a larger scale. ICT equipment is located within a room or hall, heat is ejected from the equipment via an exhaust and that air is then extracted, cooled and recirculated. Data centre rooms are designed to operate within temperature ranges of 20–22 °C, with a lower bound of 12 °C29. As temperatures increase, equipment failure rates also increase, although not necessarily linearly30.
There are several different mechanisms for data centre cooling27,28, but the general approach involves chillers reducing air temperature by cooling water—typically to 7–10 °C31—which is then used as a heat transfer mechanism. Some data centres use cooling towers where external air travels across a wet media so the water evaporates. Fans expel the hot, wet air and the cooled water is recirculated32. Other data centres use adiabatic economisers where water sprayed directly into the air flow, or onto a heat exchange surface, cools the air entering the data centre33. With both techniques, the evaporation results in water loss. A small 1 MW data centre using one of these types of traditional cooling can use around 25.5 million litres of water per year32.
Cooling the water is the main source of energy consumption. Raising the chiller water temperature from the usual 7–10 °C to 18–20 °C can reduce expenses by 40% due to the reduced temperature difference between the water and the air. Costs depend on the seasonal ambient temperature of the data centre location. In cooler regions, less cooling is required and instead free air cooling can draw in cold air from the external environment31. This also means smaller chillers can be used, reducing capital expenditure by up to 30%31. Both Google34 and Microsoft35 have built data centres without chillers, but this is difficult in hot regions36. More

1.
UNICEF and World Health Organization. Progress on household drinking water, sanitation and hygiene 2000-2017; Special focus on inequalities. (Joint Monitoring Programme, New York, 2019).
2.
IPCC. Climate Change 2014: Impacts, Adaptations, and Vulnerability; Part A: Global and Sectoral Aspects; Summary for Policymakers. 1-32, https://doi.org/10.1017/CBO9781107415379.003. (Cambridge University Press, Cambridge, UK and New York, 2014).

3.
Hoque, S. F. & Hope, R. The water diary method – proof-of-concept and policy implications for monitoring water use behaviour in rural Kenya. Water Policy 20, 725–743 (2018).
Article  Google Scholar 

4.
Thomson, P. et al. Rainfall and groundwater use in rural Kenya. Sci. Total Environ. 649, 722–730 (2018).
Article  Google Scholar 

5.
Thomas, E. et al. Quantifying increased groundwater demand from prolonged drought in the East African Rift Valley. Sci. Total Environ. 666, 1265–1272 (2019).
CAS  Article  Google Scholar 

6.
Foster, T. & Hope, R. A multi-decadal and social-ecological systems analysis of community waterpoint payment behaviours in rural Kenya. J. Rural Stud. 47, 85–96 (2016).
Article  Google Scholar 

7.
Foster, T. & Hope, R. Evaluating waterpoint sustainability and access implications of revenue collection approaches in rural Kenya. Water Resour. Res. 53, 1473–1490 (2017).
Article  Google Scholar 

8.
Global WASH Cluster, Sanitaiton and Water for All, UNICEF, and ICRC. COVID-19 and WASH: Mitigating the socio-economic impacts on the Water, Sanitation and Hygiene (WASH) Sector. (2020).

9.
Shannon, R., Erhardt, D. & Kolker, J. Considerations for Financial Facilities to Support Water Utilities in the COVID-19 Crisis (World Bank, Washington, DC, 2020).

10.
Diffenbaugh, N. S., Giorgi, F., Raymond, L. & Bi, X. Indicators of 21st century socioclimatic exposure. Proc. Natl Acad. Sci. USA 104, 20195–20198 (2007).
CAS  Article  Google Scholar 

11.
Choularton, R. J. & Krishnamurthy, P. K. How accurate is food security early warning? Evaluation of FEWS NET accuracy in Ethiopia. Food Security 11, 333–344 (2019).
Article  Google Scholar 

12.
Boluwade, A. Remote sensed-based rainfall estimations over the East and West Africa regions for disaster risk management. ISPRS J. Photogramm. Remote Sens. 167, 305–320 (2020).
Article  Google Scholar 

13.
Manandhar, A., Greeff, H., Thomson, P., Hope, R. & Clifton, D. A. Shallow aquifer monitoring using handpump vibration data. J. Hydrol. X 8, 100057 (2020).
Article  Google Scholar 

14.
Thomas, E. et al. Reducing drought emergencies in the Horn of Africa. Sci. Total Environ. 727, 138772 (2020).
CAS  Article  Google Scholar 

15.
World Health Organization. Coronavirus disease 2019 (COVID-19) Situation Report 42. (WHO, Geneva, 2020).

16.
World Health Organization. Coronavirus disease 2019 (COVID-19) Situation Report 71. (WHO, Geneva, 2020).

17.
McNicholl, D. et al. Performance-Based Funding for Reliable Rural Water Services in Africa. (Uptime consortium, 2019).

18.
Colman, A. W., Graham, R. J. & Davey, M. K. Direct and indirect seasonal rainfall forecasts for East Africa using global dynamical models. Int. J. Climatol. 40, 1132–1148 (2020).
Article  Google Scholar 

19.
McNicholl, D. et al. Results-Based Contracts for Rural Water Services. (Uptime consortium, 2020).

20.
Funk, C. C. et al. A quasi-global precipitation time series for drought monitoring (USGS, Reston, VA, 2014). More

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript. More

1.
Matthews, G. V. T. The Ramsar Convention on Wetlands: Its History and Development (Ramsar Convention Bureau, 1993).
2.
Ramsar Convention on Wetlands Global Wetland Outlook: State of the World’s Wetlands and Their Services to People (Ramsar Convention Secretariat, 2018).

3.
Díaz, S. et al. Summary for Policymakers of the Global Assessment Report on Biodiversity and Ecosystem Services of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES, 2019).

4.
Bowman, M. in Yearbook of International Co-operation on Environment and Development (eds Stokke, O. S. & Thommessen, Ø. B.) 61–68 (Earthscan, 2002).

5.
Morrison, T. H. et al. Nat. Sustain. 3, 947–955 (2020).
Article  Google Scholar 

6.
Jones, T. & Pritchard, D. Comprehensive Review and Analysis of Ramsar Advisory Mission (RAM) Reports (Ramsar Convention Secretariat, 2018).

7.
Bridgewater, P., Kim, R. E. & Bosselmann, K. Yearbook Int. Environ. Law 25, 61–78 (2014).
Article  Google Scholar 

8.
Millennium Ecosystem Assessment Ecosystems and Human Well-Being: Wetlands and Water – Synthesis (World Resources Institute, 2005).

9.
Green, A. J. et al. Front. Ecol. Environ. 15, 99–107 (2017).
Article  Google Scholar 

10.
Cumming, G. S. Anthropocene 13, 46–56 (2016).
Article  Google Scholar 

11.
Mauerhofer, V., Kim, R. E. & Stevens, C. Environ. Sci. Policy 51, 95–105 (2015).
Article  Google Scholar 

12.
Tanzi, A. et al. (eds) The UNECE Convention on the Protection and Use of Transboundary Watercourses and International Lakes (Brill Nijhoff, 2015)

13.
Loures, F. & Rieu-Clarke, A. (eds) The UN Watercourses Convention in Force (Routledge, 2013).

14.
Global Biodiversity Outlook 5 (Secretariat of the Convention on Biological Diversity, 2020).

15.
Heger, T. et al. BioScience 69, 888–899 (2019).
Article  Google Scholar 

16.
Mcleod, E. et al. Front. Ecol. Environ. 9, 552–560 (2011).
Article  Google Scholar 

17.
Hettiarachchi, M., Morrison, T. H. & McAlpine, C. Glob. Environ. Change 32, 57–66 (2015).
Article  Google Scholar  More

1.
Hanjra, M. A., Ferede, T. & Gutta, D. G. Reducing poverty in sub-Saharan Africa through investments in water and other priorities. Agric. Water Manag. 96, 1062–1070 (2009).
Article  Google Scholar 
2.
Moris, J. R. Irrigation Development in Africa: Lessons of Experience (Routledge, 1990).

3.
Aw, D. & Diemer, G. Making a Large Irrigation Scheme Work: A Case Study from Mali (World Bank, 2005).

4.
Bertoncin, M., Pase, A., Quatrida, D. & Turrini, S. At the junction between state, nature and capital: irrigation mega-projects in Sudan. Geoforum 106, 24–37 (2019).
Article  Google Scholar 

5.
Adams, W. M. Wasting the Rain: Rivers, People and Planning in Africa (Earthscan, 1992).

6.
Chambers, R. & Moris J. R. (eds) Mwea: An Irrigated Rice Settlement in Kenya (Weltforum Verlag, 1973).

7.
Awojobi, O. & Jenkins, G. P. Were the hydro dams financed by the World Bank from 1976 to 2005 worthwhile? Energy Policy 86, 222–232 (2015).
Article  Google Scholar 

8.
Bazin, F., Hathie, I., Skinner, J. & Koundouno, J. Irrigation, Food Security and Poverty—Lessons from Three Large Dams in West Africa (IIED and IUCN, 2017).

9.
World Commission on Dams Dams and Development: A New Framework for Decision-Making (Earthscan, 2000).

10.
Thomas, D. H. L. & Adams, W. M. Adapting to dams: agrarian change downstream of the Tiga dam, northern Nigeria. World Dev. 27, 919–935 (1999).
Article  Google Scholar 

11.
Balasubramanian, V., Sie, M., Hijmans, R. J. & Otsuka, K. Increasing rice production in sub-Saharan Africa: challenges and opportunities. Adv. Agron. 94, 55–133 (2007).
CAS  Article  Google Scholar 

12.
Alam, M. Problems and potential of irrigated agriculture in sub-Saharan Africa. J. Irrig. Drain. Eng. 117, 155–172 (1991).
Article  Google Scholar 

13.
Mold, A. Will it all end in tears? Infrastructure spending and African development in historical perspective. J. Int. Dev. 24, 237–254 (2012).
Article  Google Scholar 

14.
Veldwisch, G. J., Bolding, A. & Wester, P. Sand in the engine: the travails of an irrigated rice scheme in Bwanje Valley, Malawi. J. Dev. Stud. 45, 197–226 (2009).
Article  Google Scholar 

15.
Carney, J. Converting the wetlands, engendering the environment: the intersection of gender with agrarian change in the Gambia. Econ. Geogr. 69, 329–348 (1993).
Article  Google Scholar 

16.
Zarfl, C., Lumsdon, A. E., Berlekamp, J., Tydecks, L. & Tockner, K. A global boom in hydropower dam construction. Aquat. Sci. 77, 161–170 (2015).
Article  Google Scholar 

17.
Woodhouse, P. et al. African farmer-led irrigation development: re-framing agricultural policy and investment? J. Peasant Stud. 44, 213–233 (2017).
Article  Google Scholar 

18.
Dakar Declaration on Irrigation: Building Resilience and Accelerate Growth in Sahel and West Africa by Boosting Irrigated Agriculture (ICID, 2013).

19.
Merrey, D. J. & Sally, H. Another well-intentioned bad investment in irrigation: the Millennium Challenge Corporation’s ‘compact’ with the Republic of Niger. Water Altern. 10, 195–203 (2017).
Google Scholar 

20.
Rufin, P. et al. Global-scale patterns and determinants of cropping frequency in irrigation dam command areas. Glob. Environ. Change 50, 110–122 (2018).
Article  Google Scholar 

21.
Blanc, E. & Strobl, E. Is small better? A comparison of the effect of large and small dams on cropland productivity in South Africa. World Bank Econ. Rev. 28, 545–576 (2014).
Article  Google Scholar 

22.
Flyvbjerg, B. Policy and planning for large-infrastructure projects: problems, causes, cures. Environ. Plan. B 34, 578–597 (2007).
Article  Google Scholar 

23.
Ansar, A., Flyvbjerg, B., Budzier, A. & Lunn, D. Should we build more large dams? The actual costs of hydropower megaproject development. Energy Policy 69, 43–56 (2014).
Article  Google Scholar 

24.
Ika, L. A., Diallo, A. & Thuillier, D. Critical success factors for World Bank projects: an empirical investigation. Int. J. Proj. Manag. 30, 105–116 (2012).
Article  Google Scholar 

25.
Duponchel, M., Chauvet, L. & Collier, P. What Explains Aid Project Success in Post-Conflict Situations? Policy Research Working Paper Series 5418 (World Bank, 2010).

26.
Flyvbjerg, B. Survival of the unfittest: why the worst infrastructure gets built—and what we can do about it. Oxf. Rev. Econ. Policy 25, 344–367 (2009).
Article  Google Scholar 

27.
Scott, J. C. Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed (Yale Univ. Press, 1998).

28.
Li, T. M. Beyond ‘the state’ and failed schemes. Am. Anthropol. 107, 383–394 (2005).
Article  Google Scholar 

29.
de Bont, C., Komakech, H. C. & Veldwisch, G. J. Neither modern nor traditional: farmer-led irrigation development in Kilimanjaro region, Tanzania. World Dev. 116, 15–27 (2019).
Article  Google Scholar 

30.
Li, T. M. The Will to Improve: Governmentality, Development, and the Practice of Politics (Duke Univ. Press, 2007).

31.
Project Performance Audit Report—Chad Lake Chad Polders Project (Credit 592-CD) Report No. 6751 (World Bank, 1987).

32.
Chad: Appraisal of Sategui-Deressia Irrigation Project Report No. 145a-CD (World Bank, 1974).

33.
Adams, W. M. Large scale irrigation in northern Nigeria: performance and ideology. Trans. Inst. Br. Geogr. 16, 287–300 (1991).
Article  Google Scholar 

34.
Biswas, A. K. Irrigation in Africa. Land Use Policy 3, 269–285 (1986).
Article  Google Scholar 

35.
Ansar, A., Flyvbjerg, B., Budzier, A. & Lunn, D. in The Oxford Handbook of Megaproject Management (ed. Flyvbjerg, B.) Ch. 4 (Oxford Univ. Press, 2017).

36.
Adams, W. M. The downstream impacts of dam construction: a case study from Nigeria. Trans Inst. Br. Geogr. 10, 292–302 (1985).
Article  Google Scholar 

37.
Hirschman, A. O. Development Projects Observed (Brookings, 1967).

38.
de Bont, C. The continuous quest for control by African irrigation planners in the face of farmer-led irrigation development: the case of the Lower Moshi area, Tanzania (1935–2017). Water Altern. 11, 893–915 (2018).
Google Scholar 

39.
Bertoncin, M. & Pase, A. Interpreting mega-development projects as territorial traps: the case of irrigation schemes on the shores of Lake Chad (Borno State, Nigeria). Geogr. Helv. 72, 243–254 (2017).
Article  Google Scholar 

40.
Schumacher, E. F. Small Is Beautiful: A Study of Economics as If People Mattered (Vintage, 1973).

41.
Jones, B. Desiccation and the West African colonies. Geogr. J. 91, 401–423 (1938).
Article  Google Scholar 

42.
Adams, W. M. How beautiful is small? Scale, control and success in Kenyan irrigation. World Dev. 18, 1309–1323 (1990).
Article  Google Scholar 

43.
Ahlers, R., Brandimarte, L., Kleemans, I. & Sadat, S. H. Ambitious development on fragile foundations: criticalities of current large dam construction in Afghanistan. Geoforum 54, 49–58 (2014).
Article  Google Scholar 

44.
Green, N., Sovacool, B. K. & Hancock, K. Grand designs: assessing the African energy security implications of the Grand Inga dam. Afr. Stud. Rev. 58, 133–158 (2015).
Article  Google Scholar 

45.
Mbara, C. J., Gadain, H. M. & Muthusi, F. M. Status of Medium to Large Irrigation Schemes in Southern Somalia Technical Report No. W-05 (FAO-SWALIM, 2007).

46.
Flyvbjerg, B. Policy and planning for large-infrastructure projects: problems, causes, cures. Environ. Plan. B 34, 578–597 (2007).
Article  Google Scholar 

47.
Burney, J., Woltering, L., Burke, M., Naylor, R. & Pasternak, D. Solar-powered drip irrigation enhances food security in the Sudano–Sahel. Proc. Natl Acad. Sci. USA 107, 1848–1853 (2010).
CAS  Article  Google Scholar 

48.
Higginbottom, T. P., Symeonakis, E., Meyer, H. & van der Linden, S. Mapping fractional woody cover in semi-arid savannahs using multi-seasonal composites from Landsat data. ISPRS J. Photogramm. Remote Sens. 139, 88–102 (2018).
Article  Google Scholar 

49.
Müller, H., Rufin, P., Griffiths, P., Siqueira, A. J. & Hostert, P. Mining dense Landsat time series for separating cropland and pasture in a heterogeneous Brazilian savanna landscape. Remote Sens. Environ. 156, 490–499 (2015).
Article  Google Scholar 

50.
Pal, M. Random forest classifier for remote sensing classification. Int. J. Remote Sens. 26, 217–222 (2005).
Article  Google Scholar 

51.
Gorelick, N. et al. Google Earth Engine: planetary-scale geospatial analysis for everyone. Remote Sens. Environ. 202, 18–27 (2017).
Article  Google Scholar 

52.
Abatzoglou, J. T., Dobrowski, S. Z., Parks, S. A. & Hegewisch, K. C. Terraclimate, a high-resolution global dataset of monthly climate and climatic water balance from 1958–2015. Sci. Data 5, 170191 (2018).
Article  Google Scholar 

53.
Weiss, D. J. et al. A global map of travel time to cities to assess inequalities in accessibility in 2015. Nature 553, 333–336 (2018).
CAS  Article  Google Scholar 

54.
Lloyd, C. T., Sorichetta, A. & Tatem, A. J. High resolution global gridded data for use in population studies. Sci. Data 4, 170001 (2017).
Article  Google Scholar 

55.
Kaufmann, D., Kraay, A. & Mastruzzi, M. The worldwide governance indicators: methodology and analytical issues. Hague J. Rule Law 3, 220–246 (2011).
Article  Google Scholar 

56.
Walsh, R. P. D. & Lawler, D. M. Rainfall seasonality: description, spatial patterns and change through time. Weather 36, 201–208 (1981).
Article  Google Scholar 

57.
Wood, S. N. Generalized Additive Models: an Introduction with R 2nd edn (Chapman and Hall/CRC, 2017).

58.
Wood, S. N. Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models. J. R. Stat. Soc. B 73, 3–36 (2011).
Article  Google Scholar  More

To compute the water footprint of global energy trade, we rely upon data originating from the UN Commodity Trade (UN Comtrade) database, sourced through an Application Program Interface (API) in the R coding language. Trade data are then cleaned to eliminate outliers and erroneous values. The International Energy Agency (IEA) provides information on electricity generation portfolios, retrieved using an API in Python. Water intensity factors are sourced through various literature based on the energy type; see Table 1. Figure 1 illustrates the process of generating the database. Each step is described thoroughly below. All referenced scripts, input files, and output files are accessible via the study’s accompanying Zenodo database22, https://doi.org/10.5281/zenodo.3891722. These methods are expanded versions of descriptions in our related work21 and offer additional insights to related discussions on water footprints of electricity. This data descriptor facilitates the reproducibility of results and provides the scripts needed to add additional years to the database should the study need to be revisited. Therefore, this data descriptor goes beyond the complementary manuscript by providing greater insights, assumptions, and opportunities on the methods and resultant datasets.
Table 1 Water intensity factors and ranges for each energy commodity considered in the database.
Full size table

Fig. 1

There are five general steps in the creation of the databases, requiring the integration of data from the United Nations, International Energy Agency, and literature sources.

Full size image

Determining electricity water footprints
There is no comprehensive database associated with water footprints of electricity across the globe. Therefore, many studies rely on data based in the United States to inform global estimations23. For thermoelectric power plants, water footprints vary based on cooling system and fuel type24. Water footprint values for thermoelectric power generation were obtained from Macknick et al.3, a widely utilized reference in the literature. Additionally, Davies et al.25 provides estimates of cooling technology, globally, by region. The range and expected values of water consumption intensity for each of these generation technologies and fuel types were utilized. Finally, country-specific water footprint values of hydroelectricity were gathered from Mekonnen et al.26; no uncertainty or range for these values was available. The water intensity values for electricity generation were static and did not vary interannually.
Creating country-specific water footprints of electricity
Here, we determine the water footprint of electricity for each country based on their electricity generation portfolio. The portfolios are gathered from the IEA for each year27. In several instances, specific country values are not available and generation portfolios were manually determined. The database contains the Python script used to interface with the API and download electricity portfolios from 2010-2017 (IEA-webscraping.ipynb). At the time of writing, the IEA values were incomplete for 2018 and generation profiles in this year were assumed to be identical to 2017 absent any other data. We also provide the cleaned outputs of this script (IEA-electricity-mix-20XX-GWh.csv).
In this study, we consider the water footprint of renewable electricity technologies such as solar or wind power as negligible with respect to its operational stage and assign a water consumption rate of 0 m3/MWh to these electricity resources. Utilizing the water intensity factors completed in the previous step, we determine the virtual water footprint of each country, weighted by generation; see Eq. 1. Therefore, interannual variations in a countries’ water footprint are dictated solely by changing electricity generation portfolios.

$$VW{F}_{i}=frac{sum _{g}{w}_{g}times {e}_{i,g}}{sum _{g}{e}_{i,g}}$$
(1)

Where i is the country of origin, w is the water footprint of each electricity generation technology, g, and e is the electricity generation in each country by technology. This calculation is completed using R and the script, getElectricityWF.R. The resultant database contains estimates of water footporint in m3/MWh for each country from 2010–2018 (ElectricityWaterIntensity.csv).
Water footprints of other energy sources
Fossil fuels
Only static values of water footprint were available for the energy resources of coal, lignite, and oil resources, which did not vary temporally or spatially. Table 1 shows the assumed range of water footprints and the literature sources for these values23,26,28,29,30. All water intensities are provided in m3/kg to be consistent with the reported values of the UN Comtrade data. For coal, we assume a conversion value of 36.04 kg/MMBtu to convert between literature estimates. Similarly, we assume a conversion value of 45 MJ/kg for crude oil.
The water consumption of natural gas varies widely depending on the method of extraction. However, there are only four countries that produce shale gas commercially: the United States, Canada, China, and Argentina. Therefore, for exports from these four countries, we define a weighted average of conventional and shale gas extraction water intensity, based on data from the Energy Information Administration31. In the process of extracting and processing natural gas, other hydrocarbons such as ethane, propane, butane, or pentanes are produced. Using production factors from the United States, we estimate the ratio of butane and propane production versus natural gas production32. Using these ratios, we estimate water footprints of these fuels assuming similar water footprints to natural gas. Approximately 8.1 MMBtu of propane was produced per MMBtu of natural gas from 2010 to 2018. The ratio of butane to natural gas was lower, at 2.6 MMBtu butane per MMBtu natural gas.
Biodiesel
Biofuel trade data are only available from 2012–2018. To calculate the temporal and spatial differences in water footprints for biodiesel, we combine water footprint estimates from Gerbens-Leenes et al.33 and Mekonnen and Hoekstra34 with biofuel reports from the US Department of Agriculture (USDA) and Energy Information Administration (EIA) that detail foodstock inputs by weight for biodiesel production. These reports provide a significant amount of data to capture the major countries and foodstock sources, but it is still necessary to create assumptions for the remainder of the countries and foodstocks. Animal fat was a common source for biodiesel production. We estimate that one liter of biodiesel requires 0.88 kg of animal fat and has a yield of 90%35, averaging the water footprint of pork, chicken, and beef fat based on global meat production estimates from the UN Food and Agriculture Organization and water footprints of meat36. The resultant estimate of water intensity for animal fat-derived biodiesel is 217 m3/GJ. Another common component of biodiesel production is used cooking oil, which was assigned a water footprint of zero as it is a waste product. Equation 2 describes the process of generating country-specific biodiesel water footprints.

$$W{F}_{C}=frac{mathop{sum }limits_{C}^{f}{w}_{f}times {y}_{f}times {m}_{f}}{sum {y}_{f}times {m}_{f}}$$
(2)

Where, C is a country, f is a specific feedstock, w is the water intensity of each feedstock, y is the yield ratio of each feedstock to biodiesel, and m is the mass of feedstock used in each country.
Per the UN trade definition, the biodiesel category must contain less than 70% petroleum based fuels. This creates a wide range of uncertainty in the actual biodiesel content of the trade. To account for this uncertainty, we set mean, maximum, and minimum thresholds of the fuel mix. We assume the mean to be 50% biodiesel, the minimum to be 30% biodiesel, and the maximum to be 70% biodiesel. The resulting water footprints of biodiesel for each exporting country are provided in BiodieselWF.csv.
Firewood and charcoal
Schyns et al.37 provide globally gridded blue and green water footprints of roundwood production, attributing the water consumption of forests based on an economic evaluation of the wood product relative to other forest values. Blue water footprints refer to water consumed from surface or groundwater sources, whereas green water footprints are driven by rainfall38. The globally gridded values were aggregated by country with minimum, maximum, and average values reported in m3water/m3wood. We assume a specific volume of firewood to be 2.08 × 10−3 m3/kg and 5.92 × 10−3 m3/kg of charcoal39. For countries that did not have gridded values within the dataset, we average values from neighboring countries. No interannual variation in water footprint was available. The final water footprints by country for firewood and charcoal are provided in FuelwoodWF.csv and CharcoalWF.csv, respectively.
Trade data download and cleaning
The UN Comtrade data provide the basis for the analysis40. These data were downloaded using the comtradr package in R, which interfaces with the Comtrade API. Both import and export data were downloaded for all countries from 2010-2018 across eleven different energy commodities. These trade statistics provide the value (USD) of the economic transfer, direction of trade (import or export), trade partners, commodity traded, and the amount of the good transferred. Electricity trade is reported in 1000 kWh (1 MWh); all other energy commodities report trade in kilograms. The script to download trade data is included in the database; see DownloadComtradeData.R. Querying the Comtrade API is limited by the number of qualifiers; therefore, the queries are broken down by energy commodity and combined using the CompileTradeData.R script provided in the database.
Upon investigation of the data, we identify four areas of data cleaning: (i) resolve differences in imports versus export data, (ii) address discrepancies in electricity trade, (iii) fill data gaps, and (iv) remove outliers.
i
To be conservative, in cases where the import and export data were different (or one was not available), the largest traded volume was kept. This assumption was made in absence of any other estimate acknowledging the potential for overestimation. This conservative approach for estimating the water footprint of energy is consistent with similar studies19.

ii
In some instances, there was reported electricity trade between two non-neighboring countries (i.e., European Countries and the United States). To resolve these potential concerns, we created a database of geographically neighboring countries and inventoried a list of undersea connections and eliminate trades occuring outside these connections. While this assumption would negate potential agreements of electricity trade through countries, we assumed that this proportion of electricity trade is relatively small. Additionally, these assumptions reflect the constraints of electricity trade with infrastructure. Removal of these links is completed in CompileTradeData.R using the ElectricConnection.R function and the accompanying database of country neighbors, ElectricityConnections.csv.

iii
For values that were reported as zero but had a monetary value, we determined a unit value ($/kg or $/MWh) as the median of a commodities’ trade between the two countries in the preceding year, following year, and the overall unit value of the commodity originating from the country in the current year. Data gaps are filled using the filldatagaps.R script.

iv
There were some instances of extreme trade values, particularly in the year 2017. These quantities were often reported two orders of magnitude greater than similar trade links in other years. To manage these errors, we took an objective approach to all reported quantities and identified any values that met all of the following three criteria:

reported quantities greater than 5 times the median value from other years on the same link,

reported quantities with a unit value greater than 5 times the median unit value from other years on the same link, and

reported quantities with a unit value greater than 5 times the median unit value for all exports from the originating country in that year

The above assumptions allowed us to remove extreme values and replace them with estimations based on the reported monetary trade value of the commodity, as above. The script for removing and correcting outliers is provided in the database; see reviseOutliers.R.
Creating the virtual water trade network
Following a cleaned and formatted version of the Comtrade data from Step 4, the virtual water trade network was created by multiplying a water footprint of each energy source, f, based on its country of origin i and year y and the reported trade volume from country i to j:

$$VW{T}_{i,j}^{f,y}={e}_{i,j}^{f,y}times {w}_{i}^{f,y}$$
(3)

For each link in the network, a mean, minimum, and maximum value of trade is calculated based on the ranges of water footprint calculated in Steps 1-3. This step is completed using the DetermineWF.R script and results in the main output of the database EnergyWF_Trade.csv. The final dataset maintains information on export/import, countries in the trade link, quantity traded, trade value, and associated water footprints. ‘Reporter’ columns refer to the country of origin and ‘partner’ columns are the destination country. The final database includes virtual water exports from 215 countries. Of these countries, only 25 (12%) did not feature exports for all nine years of the analysis. Figure 2 illustrates the global extent of the database with many countries exporting all eleven commodities for at least one year during the study period.
Fig. 2

Many countries had export data for all 11 commodities for at least one year from 2010–2018.

Full size image More