Markus Andrews

COMMENT
08 December 2021

Brazil is in water crisis — it needs a drought plan

To avoid crop failures and soaring power costs, Brazil needs to diversify sources, monitor soil moisture, model local hydroclimate dynamics and treat water as a national security priority.

Augusto Getirana

0
,

Renata Libonati

1
&

Marcio Cataldi

2

Augusto Getirana

Augusto Getirana is a principal research scientist at Science Applications International Corporation, Greenbelt, Maryland, USA, and at the Hydrological Sciences Laboratory, NASA Goddard Space Flight Center, Greenbelt, Maryland, USA.

View author publications

You can also search for this author in PubMed
 Google Scholar

Renata Libonati

Renata Libonati is an assistant professor at the Federal University of Rio de Janeiro, Rio de Janeiro, Brazil, and a research scientist at the Dom Luiz Institute, University of Lisbon, Lisbon, Portugal.

View author publications

You can also search for this author in PubMed
 Google Scholar

Marcio Cataldi

Marcio Cataldi is an associate professor at the Fluminense Federal University, Niteroi, Brazil.

View author publications

You can also search for this author in PubMed
 Google Scholar

Twitter

Facebook

Email

Download PDF

Jaguari dam is part of the Cantareira system that supplies water to São Paulo, Brazil.Credit: Paulo Fridman/Bloomberg/Getty

Brazil has the largest amount of fresh water in the world. Two-thirds of what flows in the Amazon River alone could supply the world’s demand. Yet much of the nation now faces drought.It’s the worst in many decades in a nation that grows more than one-third of the world’s sugar crops and produces almost 15% of the world’s beef.This year, between March and May, dry weather in Brazil’s south-central region led to a 267 km3 shortage of water held in rivers, lakes, soil and aquifers, compared with the seasonal average for the past 20 years (see ‘Brazil dries out’ and ‘Low water mark’). The result? Many major reservoirs have reached less than 20% capacity. Farming and energy generation have been hit. Since July, coffee prices have risen by 30% — Brazil accounts for one-third of global exports. Soya bean prices rose by 67% from June 2020 to May this year. And electricity bills have soared by 130%. Many cities face imminent water rationing.

Source: H. Save et al. J. Geophys. Res. Solid Earth 121, 7547–7569 (2016)

How has this happened? And what needs to be done?Worldwide climate change is making droughts more intense and more frequent. Deforestation in the Amazon is a contributor locally and globally. The hydroclimate in the south-central region — the engine of 70% of Brazil’s gross domestic product (GDP) — is partly controlled by moisture transfer from the rainforest. Atmospheric fluxes caused by tree transpiration — also known as ‘flying rivers’ — might contribute as much water per day in rainfall as the Amazon River itself carries. Cutting down these trees reduces precipitation over those areas, as well as eroding a crucial global carbon sink.For decades there has been a governmental failure to recognize drought as a matter of national and international security. Brazil’s water crisis is a world crisis. What’s needed is a coordinated nationwide drought-mitigation plan crafted by researchers, policymakers, industry, the public sector and civil society. Here are some key points that such a plan should address; these points are supported by 95 Brazilian and international water and climate scientists (see Supplementary information for list of co-signatories).Vast reservesAbout 20% of all global inland water flowing to the oceans is generated in Brazilian territory1. This fuels the country’s welfare and economic growth. About 85% of the nation’s fresh water needs are supplied by surface waters — rivers and lakes2. In the United States, that figure is 75%; in India, it is 60%.Brazil has the world’s second-largest installed hydropower capacity, at 107.4 gigawatts (GW); it produces 65% of the country’s electricity. Two-fifths of this is generated in the Paraná River Basin, where river discharges have fallen to their lowest levels in 91 years. The country has had to revert to burning fossil fuels and biofuel, passing the higher costs onto consumers. Thermal power produced 13.2% of the nation’s electricity in July 2021, the highest in its history.In a nation dependent on agriculture for almost one-quarter of its GDP, crops such as soya, coffee and sugar cane, and livestock use much of the water. Irrigation feeds about 13% of the cultivated land3, drawing down 68% of total water consumption — some 68.4 billion litres per day4.But water is not equally available across the country, nor over time.

Source: H. Save et al. J. Geophys. Res. Solid Earth 121, 7547–7569 (2016)

Different droughtsWater crises can originate from many types of drought: meteorological, hydrological, agricultural and socio-economic.Meteorological droughts are dry weather patterns due to periods of little rainfall or high temperatures, which increase evaporation rates. These can cause hydrological droughts, water shortages on land surfaces such as rivers and lakes.Agricultural droughts — a decline in soil moisture levels — can result. These can jeopardize crop yield and increase food insecurity. Shortages to the domestic and industrial supply — socio-economic droughts — can also follow. This might lead to rationing, disease, conflict and migration. It could also bring water-intensive processes such as concrete and steel production to a halt.These different droughts can interact in complex and non-linear ways. Hydrological droughts, for example, are intensified when prolonged periods of low soil moisture begin to dry out shallow aquifers. This can drop their levels below riverbed elevations, interrupting river–groundwater connectivity. Depleted rivers or lakes can then have a knock-on effect on reservoir levels, triggering a socio-economic drought.Human fingerprintThe 2021 Intergovernmental Panel on Climate Change (IPCC) report highlighted that unabated regional land-cover change and global warming are causing a cascade of persistent dry conditions around the globe5. Studies suggest an extended dry season in most of central South America under an extreme, but not unlikely, scenario6.Decades of deforestation of the Amazon has led to vast knock-on effects. Cutting down trees, as well as slashing the amount of moisture transported from the rainforest towards central-southern Brazil7, is the main driver of fire8. The particulate matter released into the upper air alters the formation of rain clouds9.
End the drought in drought research
Improper land use can worsen droughts, too, and even cause rivers to run dry. Intensive cattle farming leads to unvegetated land and compacted soils. As well as decreasing the amount of moisture given off by plants, it limits the soil’s capacity to retain water and recharge aquifers.But droughts alone don’t explain the recurrence of water crises in Brazil. Failure to treat water as an essential national resource has led Brazil to a long history of mismanagement. Science denialism is now promoted at the highest levels around the country10,11. And national policies have driven increasingly erratic land occupation by agribusiness and mining interests, increasing deforestation and wildfires and undermining climate mitigation12–14.As the country plunged into severe water shortages in 2014 and 2015, the Brazilian Academy of Sciences upbraided state authorities for failure to take swift, bold actions and for a lack of transparency about the gravity of the situation15.Six years have passed and not much has changed. This time around, the country’s economy is recovering to pre-pandemic levels. Economic growth requires extra energy to power production. With the current hydropower situation, this demand might have to be met by burning biofuel or fossil fuel.Research prioritiesThe nation’s groundwater and meteorological monitoring is sparse and insufficient to properly track water variability and availability across the country. Brazil monitors groundwater at 409 sites nationwide; to put that into perspective, the North American and Indian networks have more than 16,000 and 22,000 sites, respectively. There are no nationwide systems in place to track soil moisture in Brazil, and monitoring of water use is patchy.
Rescue Brazil’s burning Pantanal wetlands
Governance of these networks must be strengthened, and more effective guidance on how to respond to future crises is needed. Monitoring networks are currently operated across different national agencies and departments, often leading to duplicated efforts and inefficient data access. Drought monitoring initiatives developed in Brazil through international partnerships, such as the Monitor de Secas, have been emerging in recent years. However, reducing delays to the availability of data, and improving accuracy and inaccessibility for end-users, such as farmers and local water departments, would make these initiatives more useful.There needs to be more investment into high-quality, readily available data and computing power — the key ingredients for multidisciplinary drought research. Tupã — Brazil’s most powerful supercomputer at the Brazilian National Institute for Space Research (INPE) is nearing the end of its life. Funds from the United Nations have provided temporary access to alternative computers, but these are not powerful enough to perform hydrometeorological forecasts and climate predictions. US$20 million of federal funds should be put aside for a new supercomputer. Instead, the science and technology ministry’s budget for 2022 has been reduced by 87% ( see Nature https://doi.org/g77w; 2021).Many processes that affect south-central Brazil’s water availability are not well understood. These need more research to best inform policy. They include:Climate feedbacks. Deforestation, land use, biomass burning and global warming interact to determine water availability. Fresh approaches should exploit emerging knowledge and computational tools to better incorporate small-scale and fast processes, such as vegetation, land cover, clouds and aerosol feedback effects in climate models. This will need higher-resolution simulations, more computational power and reliable in situ and satellite-based observations.
Policy, drought and fires combine to affect biodiversity in the Amazon basin
Compound events. Hazards such as droughts, heatwaves and fires can have devastating impacts beyond an area related to an isolated event. Risk-assessment approaches should consider how the co-occurrence of multiple and dependent hazards affect models. Climate, health and social scientists, as well as engineers and modellers, should work to improve predictions.Groundwater. Intensive pumping, especially combined with droughts, has led to severe depletion in regions such as the western and central United States, northern India and the Middle East16. More research, along with groundwater and soil-moisture monitoring, is needed to understand how Brazilian aquifers respond to pumping, as well as climate variability and change.Migration and health. Climate change could intensify migration from the northeast, Brazil’s driest and poorest region, to the southeast. Other movements of people could be triggered across the country as longer, more frequent and severe droughts arise. Massive climate migrations might result in an increase of water insecurity, as well as unemployment and poverty in major Brazilian cities. Social, political and economic scientists should work to identify the drivers of climate migration to guide policymaking. Research initiatives should also consider the long-term effects of drought on human health, such as malnutrition and mental health.Diversify sourcesStable, long-term investment is needed to upgrade the nation’s water and power system. Hydropower has a small carbon footprint once installed, despite its initial high environmental and social impacts. When there isn’t enough water to generate electricity, however, expensive and more-polluting fossil-fuel-based thermal power currently picks up the slack.Instead, Brazil could diversify by amplifying wind and solar capacity. This could be supported by an existing system of contract auctions, providing a mechanism to gather funds for clean energy. The success of such a mechanism in Brazil is demonstrated by recent investments totalling nearly $8 billion over the past 5 years, mostly from the private sector. An estimated 300 GW could be generated from clean energy sources by 2050 — 3 times the nation’s current demand17.Brazil lies on major aquifers — valuable and underused resources. The agricultural sector should build climate resilience by using this groundwater, especially during extreme hydrological droughts. This needs to be done sustainably, to avoid the depletion experienced by other countries16. A clear picture of the spatial distribution and temporal variability of aquifers could guide farmers towards appropriate locations and rates of extraction.In November, Brazil promised to end illegal deforestation and cut emissions from 2005 levels by 50% by 2030 at the 2021 United Nations Climate Change Conference (COP26) in Glasgow, UK. However, such measures are not ambitious enough and would not bring the country in line with green policies, such as the European Green Deal and the US Green New Deal.There might be short-term economic harm from stemming deforestation, especially among farmers and landowners. But the costs of doing nothing are too extreme to ignore. The World Economic Forum has classed water crises as a top global risk, owing to their impact on food production, human health, conflict, ecosystem function and extreme weather (see go.nature.com/3lwow7x).Brazil has the expertise and motivation to mitigate this risk. The research community must work with governments to craft laws, policies and investments that enforce optimal water practice — preventive and adaptive. With political willpower, funding and infrastructure to match, the country could become a world leader in hydroclimate resilience.

Nature 600, 218-220 (2021)
doi: https://doi.org/10.1038/d41586-021-03625-wA.G. writes in their personal capacity and not on behalf of NASA Goddard Space Flight Center or Science Applications International Corporation.

References1.Getirana, A. J. Hydrometeorol. 17, 591–599 (2016).Article 

Google Scholar 
2.National Water and Basic Sanitation Agency. Report on the Situation of Water Resources in Brazil 2020 (ANA, 2020).
Google Scholar 
3.National Water and Basic Sanitation Agency. Atlas Irrigation 2021: Water Use in Irrigated Agriculture 2nd edn (ANA, 2021).
Google Scholar 
4.National Water and Basic Sanitation Agency. Manual of Consumptive Uses of Water in Brazil (ANA, 2019).
Google Scholar 
5.Intergovernmental Panel on Climate Change. Assessment Report 6 Climate Change 2021: The Physical Science Basis (IPCC, 2021).
Google Scholar 
6.Gomes, G. D., Nunes, A. M. B., Libonati, R. & Ambrizzi, T. Clim. Dyn. https://doi.org/10.1007/s00382-021-05955-x (2021).Article 

Google Scholar 
7.Khanna, J., Medvigy, D., Fueglistaler, S. & Walko, R. Nature Clim. Change 7, 200–204 (2017).Article 

Google Scholar 
8.Libonati, R. et al. Sci. Rep. 11, 4400 (2021).PubMed 
Article 

Google Scholar 
9.Correia, A. L., Sena, E. T., Silva Dias, M. A. F. & Koren, I. Commun. Earth Environ. 2, 168 (2021).Article 

Google Scholar 
10.Escobar, H. Science https://doi.org/10.1126/science.aay9857 (2019).Article 

Google Scholar 
11.Diele-Viegas, L. M., Hipólito, J. & Ferrante, L. Science 374, 948–949 (2021).PubMed 
Article 

Google Scholar 
12.Feng, X. et al. Nature 597, 516–521 (2021).PubMed 
Article 

Google Scholar 
13.da Silva, C. A. et al. Sci. Rep. 10, 16246 (2020).PubMed 
Article 

Google Scholar 
14.Abessa, D., Famá, A. & Buruaem, L. Nature Ecol. Evol. 3, 510–511 (2019).PubMed 
Article 

Google Scholar 
15.de Mattos Bicudo, C. E. et al. Rev. USP 106, 11–20 (2015).Article 

Google Scholar 
16.Rodell, M. et al. Nature 557, 651–659 (2018).PubMed 
Article 

Google Scholar 
17.Ministry of Mines and Energy. National Energy Plan 2050 (MME, 2020).
Google Scholar 
Download references

Supplementary Information

List of co-signatories

Competing Interests
The authors declare no competing interests.

Related Articles

End the drought in drought research

Policy, drought and fires combine to affect biodiversity in the Amazon basin

Rescue Brazil’s burning Pantanal wetlands

Heed blame for extreme weather

Subjects

Policy

Environmental sciences

Climate change

Water resources

Latest on:

Policy

Are female science leaders judged more harshly than men? Study it
Correspondence 07 DEC 21

Portugal: female science leaders could speed up change
Correspondence 07 DEC 21

Omicron: the global response is making it worse
Editorial 07 DEC 21

Environmental sciences

Surging plastic use is fed by coal power — with deadly results
Research Highlight 02 DEC 21

Shutting ‘super-polluters’ slashes greenhouse gases — and deaths
Research Highlight 01 DEC 21

Earth’s eccentric orbit paced the evolution of marine phytoplankton
News & Views 01 DEC 21

Climate change

An IPCC reviewer shares his thoughts on the climate debate
Career Q&A 08 DEC 21

The UN must get on with appointing its new science board
Editorial 08 DEC 21

Build solar-energy systems to last — save billions
Comment 07 DEC 21

Jobs

Postdoc in Formulation Development for Gene Delivery Therapies

Technical University of Denmark (DTU)
2800 Kgs. Lyngby, Denmark

​​​​​​​Postdoc in Molecular Biology for Gene Delivery Project

Technical University of Denmark (DTU)
2800 Kgs. Lyngby, Denmark

Post-doctoral Research Fellows

Brigham and Women’s Hospital (BWH)
Boston, MA, United States

HPC/Research Computing Engineer

Francis Crick Institute
London, United Kingdom More

Kidney progression projectThe Kidney Progression Project was initiated in 2017 in the Wilgamuwa Divisional Secretariat, a highly endemic CKDu area of 40,000 people in the lowland dry zone area of the Central Province (Supplementary Fig. 1). All protocols were reviewed and approved by review boards at the University of Connecticut in the US and National Hospital in Kandy, Sri Lanka. The detailed methodological approach including a description of behavioral and clinical and expanded environmental variables is described in Vlahos et al. (2018)13. Briefly, in 2016, the Ministry of Health conducted a screening of urine and blood in Wilgamuwa for residents 11 years and older to identify those with CKDu. Using the resulting serum creatinine values obtained during this screening effort, the KiPP team calculated CKD-EPI eGFR23, which resulted in a total of 330 people at Stage 3 and 4 of CKDu (eGFR in the range of 20-60 ml/min/1.73 m2), who did not have identifiable cause for CKD with evidence of chronic interstitial nephritis in renal biopsies or small echogenic kidney. Of these, 304 agreed to participate but ultimately 293 answered the baseline questionnaire and came for at least one serum creatinine measurement and were included in this analysis.Baseline survey componentsAll participants were administered a baseline survey that focused on environmental exposure, behavioral and occupational factors, and clinical values as described in the KiPP protocol13. We probed water sources in detail. Water sources in the study area and the dry zone in general include household wells dug by hand that are 10 meters deep or shallower, tube wells dug to a depth of 20–30 m with drilling equipment, and lesser-used sources including surface water (tanks, channels and river water), rainwater collection, natural spring water, publicly supplied pipe water, and public water delivered to individual houses by truck (bowsers) and stored in large roof containers. The rise in CKDu cases led the government to invest heavily in reverse osmosis (RO) units and nanofiltration membrane technology for many dry zone villages14. These were installed at the end of 2017 and early 2018 to provide rationed, free drinking water.Baseline water samples and analysisThe wells of each participant household were sampled once for target agrochemicals as described in Shipley et al.24. In all, 272 household wells were sampled with 31 households sharing wells.Agrochemical analysesAgrochemical analyses follow methods of Shipley et al., (2022)24 and EPA (2018)25. Briefly, 1 L well water samples were collected at each participant’s home and pre-filtered through a 0.45 µm nominal GFF to remove particulates. The sample was then extracted using 3 mL Chromabond C-18 SPE cartridges and a Supelco Visiprep SPE vacuum manifold. Three deuterated surrogate standards (chrysense d12, acenaphthene d10, and 1,4-dichlorobenzene d4) were loaded onto the cartridge before elution with 5 ml of acetonitrile and nitrogen reduction to 1 ml. Recoveries ranged from 70 to 101%.An initial non-targeted analysis was run on samples in scan mode which identified over 100 compounds, including pyrolytic compounds that are likely the result of field burning practices in preparation for the new season. We supplemented these analyses with data from a local list of agrochemicals for the year 2017–2018 supplied by the Sri Lankan Ministry of Environment. Based on these data, targeted analyses were performed for 30 agrochemicals using selective ion mode.Inorganic analysesPhosphate in samples was measured with an Ion Chromatograph (Thermo Dionex ICS-1100). For repeated analyses of selected samples, an analytical precision better than ±5% of relative standard deviations was achieved. Total hardness was determined by EDTA titration method (APHA 2012)26.Follow up: From December 2017 to the beginning of 2020, study participants had quarterly follow-up visits assessing behavioral changes including water consumption and serum creatinine testing. Serum creatinine was tested using an IDMS-calibrated enzymatic assay and converted to estimated glomerular filtration using the CKD-EPI equation.GIS Analysis: Using GPS coordinates recorded by the field team for the domestic wells of each participant, individual eGFR at baseline and eGFR slopes over the study period were plotted over the ArcMap World Topographic map. For the baseline eGFR map, values were separated into five categories using Jenks Natural breaks provided by the ArcGIS software. The uppermost category was manually set to 65 mL/min/1.73 m2 and points with null or More

1.Prüss-Ustün, A. et al. Burden of disease from inadequate water, sanitation and hygiene in low- and middle-income settings: a retrospective analysis of data from 145 countries. Trop. Med. Int. Heal. 19, 894–905 (2014).Article 

Google Scholar 
2.WHO & UNICEF. Progress on Drinking Water, Sanitation and Hygiene in Households 2000-2020: Five Years into the SDGs (WHO & UNICEF, 2021).3.World Health Organization. Guidelines for Drinking-water Quality 4th edn. (WHO, 2011) https://doi.org/10.1016/S1462-0758(00)00006-6.4.Gil, M. I., Gómez-López, V. M., Hung, Y.-C. & Allende, A. Potential of electrolyzed water as an alternative disinfectant agent in the fresh-cut industry. Food Bioprocess Technol. 8, 1336–1348 (2015).CAS 
Article 

Google Scholar 
5.Drinking Water Inspectorate. Guidance on the implementation of the water supply (water quality) regulations 2000 (as amended) in England. Drinking Water Inspectorate vol. 2000 (Drinking Water Inspectorate, 2012).6.Chowdhury, S. Trihalomethanes in drinking water: effect of natural organic matter distribution. Water SA 39, 1–8 (2013).CAS 

Google Scholar 
7.Grunwald, A., Nikolaou, A. D., Golfinopoulos, S. K. & Lekkas, T. D. Formation of organic by-products during chlorination of natural waters. J. Environ. Monit. 4, 910–916 (2002).Article 

Google Scholar 
8.Clayton, G. E., Thorn, R. M. S. & Reynolds, D. M. Comparison of trihalomethane formation using chlorine-based disinfectants within a model system; applications within point-of-use drinking water treatment. Front. Environ. Sci. 7, 35 (2019).Article 

Google Scholar 
9.Malliarou, E., Collins, C., Graham, N. & Nieuwenhuijsen, M. J. Haloacetic acids in drinking water in the United Kingdom. Water Res. 39, 2722–2730 (2005).CAS 
Article 

Google Scholar 
10.World Health Organization. Trihalomethanes in Drinking-water (World Health Organization, 2005).11.Fawell, J. & Nieuwenhuijsen, M. J. Contaminants in drinking water. Br. Med. Bull. 68, 199–208 (2003).CAS 
Article 

Google Scholar 
12.Carratalà, A. et al. Solar disinfection of viruses in polyethylene terephthalate bottles. Appl. Environ. Microbiol. 82, 279–288 (2016).Article 
CAS 

Google Scholar 
13.Zhu, J., Fan, X. J., Tao, Y., Wei, D. Q. & Zhang, X. H. Study on an integrated process combining ozonation with ceramic ultra-filtration for decentralized supply of drinking water. J. Environ. Sci. Heal. 49, 1296–1303 (2014).CAS 
Article 

Google Scholar 
14.Glaze, W. H., Kang, J.-W. & Chapin, D. H. The chemistry of water treatment processes involving ozone, hydrogen peroxide and ultraviolet radiation. Ozone Sci. Eng. 9, 335–352 (1987).CAS 
Article 

Google Scholar 
15.McGuire, M. J. Drinking Water Chlorination (American Chemistry Council, 2016). https://chlorine.americanchemistry.com/Chlorine-Benefits/Safe-Water/Disinfection-Practices.pdf 10.1002/(SICI)1521-401X(199902)27:23.3.CO;2-1.16.Han, Q. et al. Removal of foodborne pathogen biofilms by acidic electrolyzed water. Front. Microbiol. 8, 1–12 (2017).
Google Scholar 
17.Thorn, R. M. S., Pendred, J. & Reynolds, D. M. Assessing the antimicrobial potential of aerosolised electrochemically activated solutions (ECAS) for reducing the microbial bio-burden on fresh food produce held under cooled or cold storage conditions. Food Microbiol. 68, 41–50 (2017).CAS 
Article 

Google Scholar 
18.Kirkpatrick, R. D. The mechanism of antimicrobial action of Electro-Chemically Activated (ECA) water and its healthcare applications (University of Pretoria, 2009).19.Thorn, R. M. S., Lee, S. W. H., Robinson, G. M., Greenman, J. & Reynolds, D. M. Electrochemically activated solutions: evidence for antimicrobial efficacy and applications in healthcare environments. Eur. J. Clin. Microbiol. Infect. Dis. 31, 641–653 (2012).CAS 
Article 

Google Scholar 
20.Ghebremichael, K., Muchelemba, E., Petrusevski, B. & Amy, G. Electrochemically activated water as an alternative to chlorine for decentralized disinfection. J. Water Supply.: Res. Technol.—Aqua 60, 210–218 (2011).CAS 
Article 

Google Scholar 
21.Venczel, L. V., Likirdopulos, C. A., Robinson, C. E. & Sobsey, M. D. Inactivation of enteric microbes in water by electro-chemical oxidant from brine (NaCl) and free chlorine. Water Sci. Technol. 50, 141–146 (2004).CAS 
Article 

Google Scholar 
22.Kerwick, M. I., Reddy, S. M., Chamberlain, A. H. L. & Holt, D. M. Electrochemical disinfection, an environmentally acceptable method of drinking water disinfection? Electrochim. Acta 50, 5270–5277 (2005).CAS 
Article 

Google Scholar 
23.Liao, L. B., Chen, W. M. & Xiao, X. M. The generation and inactivation mechanism of oxidation–reduction potential of electrolyzed oxidizing water. J. Food Eng. 78, 1326–1332 (2007).CAS 
Article 

Google Scholar 
24.Robinson, G. M., Lee, S. W.-H., Greenman, J., Salisbury, V. C. & Reynolds, D. M. Evaluation of the efficacy of electrochemically activated solutions against nosocomial pathogens and bacterial endospores. Lett. Appl. Microbiol. 50, 289–294 (2010).CAS 
Article 

Google Scholar 
25.Cherney, D. P., Duirk, S. E., Tarr, J. C. & Collette, T. W. Monitoring the speciation of aqueous free chlorine from pH 1 to 12 with Raman spectroscopy to determine the identity of the potent low-pH oxidant. Appl. Spectrosc. 60, 764–772 (2006).CAS 
Article 

Google Scholar 
26.Nakagawara, S. et al. Spectroscopic characterization and the pH dependence of bactericidal activity of the aqueous chlorine solution. Jpn. Soc. Anal. Sci. 14, 691–698 (1998).CAS 
Article 

Google Scholar 
27.Jeong, J., Kim, J. Y. & Yoon, J. The role of reactive oxygen species in the electrochemical inactivation of microorganisms. Environ. Sci. Technol. 40, 3–4 (2006).Article 

Google Scholar 
28.Martínez-Huitle, C. A. A., Brillas, E., Martinez-Huitle, C. A. & Brillas, E. Electrochemical alternatives for drinking water disinfection. Angew. Chem. Int. Ed. 47, 1998–2005 (2008).Article 
CAS 

Google Scholar 
29.Inoue, Y. et al. Trial of electrolyzed strong acid aqueous solution lavage in the treatment of peritonitis and intraperitoneal abscess. Artif. Organs 21, 28–31 (1997).CAS 
Article 

Google Scholar 
30.Bernstein, R. et al. ‘Should I stay or should I go?’ Bacterial attachment vs biofilm formation on surface-modified membranes. Biofouling 30, 367–376 (2014).CAS 
Article 

Google Scholar 
31.Schwering, M., Song, J., Louie, M., Turner, R. J. & Ceri, H. Multi-species biofilms defined from drinking water microorganisms provide increased protection against chlorine disinfection. Biofouling 29, 917–928 (2013).CAS 
Article 

Google Scholar 
32.O’Toole, G., Kaplan, H. B. & Kolter, R. Biofilm formation as microbial development. Annu. Rev. Microbiol. 54, 49–79 (2000).Article 

Google Scholar 
33.Flemming, H.-C. C. et al. Biofilms: an emergent form of bacterial life. Nat. Rev. Microbiol. 14, 563–575 (2016).CAS 
Article 

Google Scholar 
34.Ashbolt, N. J. Microbial contamination of drinking water and human health from community water systems. Curr. Environ. Heal. Rep. 2, 95–106 (2015).CAS 
Article 

Google Scholar 
35.Skraber, S., Schijven, J., Gantzer, C. & de Roda Husman, A. M. Pathogenic viruses in drinking-water biofilms: a public health risk? Biofilms 2, 105–117 (2005).Article 

Google Scholar 
36.Crozes, G. F., Jacangelo, J. G., Anselme, C. & Laîné, J. M. Impact of ultrafiltration operating conditions on membrane irreversible fouling. J. Memb. Sci. 124, 63–76 (1997).CAS 
Article 

Google Scholar 
37.Sillanpää, M. In Natural Organic Matter in Water 1–15 (Butterworth-Heinemann, 2015). https://doi.org/10.1016/B978-0-12-801503-2.00001-X.38.Wingender, J. & Flemming, H.-C. Biofilms in drinking water and their role as reservoir for pathogens. Int. J. Hyg. Environ. Health 214, 417–423 (2011).Article 

Google Scholar 
39.De Beer, D., Srinivasan, R. & Stewart, P. S. Direct measurement of chlorine penetration into biofilms during disinfection. Appl. Environ. Microbiol. 60, 4339–4344 (1994).Article 

Google Scholar 
40.Stewart, P. S., Rayner, J., Roe, F. & Rees, W. M. Biofilm penetration and disinfection efficacy of alkaline hypochlorite and chlorosulfamates. J. Appl. Microbiol. 91, 525–532 (2001).CAS 
Article 

Google Scholar 
41.British Standards Institution. Chemical disinfectants and antiseptics—quantitative suspension test for the evaluation of basic bactericidal activity of chemical disinfectants and antiseptics—test method and requirements (phase 1). European Committee for Standardization vol. 3 http://www.cen.eu/cen/Sectors/TechnicalCommitteesWorkshops/CENTechnicalCommittees/Pages/Standards.aspx?param=6197&title=Chemical disinfectants and antiseptics (2005).42.British Standards Institution. Chemical disinfectants and antiseptics—Quantitative suspension test for the evaluation of bactericidal activity of chemical disinfectants and antiseptics used in food, industrial, domestic and institutional areas—Test method and requirements (phase 2, European Committee for Standardization vol. 3 http://www.cen.eu/cen/Sectors/TechnicalCommitteesWorkshops/CENTechnicalCommittees/Pages/Standards.aspx?param=6197&title=Chemical disinfectants and antiseptics (2009).43.Clayton, G. E., Thorn, R. M. S. & Reynolds, D. M. Development of a novel off-grid drinking water production system integrating electrochemically activated solutions and ultrafiltration membranes. J. Water Process Eng. 30, (2019).44.Loret, J. F. et al. Comparison of disinfectants for biofilm, protozoa and Legionella control. J. Water Health 3, 423–433 (2005).CAS 
Article 

Google Scholar 
45.Diao, H., Li, X., Gu, J., Shi, H. & Xie, Z. Electron microscopic investigation of the bactericidal action of electrochemical disinfection in comparison with chlorination, ozonation and Fenton reaction. Process Biochem. 39, 1421–1426 (2004).CAS 
Article 

Google Scholar 
46.Clasen, T. & Edmondson, P. Sodium dichloroisocyanurate (NaDCC) tablets as an alternative to sodium hypochlorite for the routine treatment of drinking water at the household level. Int. J. Hyg. Environ. Health 209, 173–181 (2006).CAS 
Article 

Google Scholar 
47.Fukuzaki, S. Mechanisms of actions of sodium hypochlorite in cleaning and disinfection processes. Biocontrol Sci. 11, 147–157 (2006).CAS 
Article 

Google Scholar 
48.Bloomfield, S. F., Arthur, M., Looney, E., Begun, K. & Patel, H. Comparative testing of disinfectant and antiseptic products using proposed European suspension testing methods. Lett. Appl. Microbiol. 13, 233–237 (1991).CAS 
Article 

Google Scholar 
49.European Chemicals Agency. Regulation (EU) No 528/2012 concerning the making available on the market and use of biocidal products. Active chlorine released from sodium hypochloriteProduct-type 4 (Food and feed area). https://echa.europa.eu/documents/10162/3b7a78a9-9bda-f684-a088-418dc4a56adb (2017).50.Oomori, T., Oka, T., Inuta, T. & Arata, Y. The efficiency of disinfection of acidic electrolyzed water in the presence of organic materials. Anal. Sci. 16, 365–369 (2005).Article 

Google Scholar 
51.Ayebah, B., Hung, Y.-C., Kim, C. & Frank, J. F. Efficacy of electrolyzed water in the inactivation of planktonic and biofilm Listeria monocytogenes in the presence of organic matter. J. Food Prot. 69, 2143–2150 (2006).Article 

Google Scholar 
52.Robinson, G., Thorn, R. & Reynolds, D. The effect of long-term storage on the physiochemical and bactericidal properties of electrochemically activated solutions. Int. J. Mol. Sci. 14, 457–469 (2013).CAS 
Article 

Google Scholar 
53.Ignatov, I. et al. The evaluation of the mathematical model of interaction of electrochemically activated water solutions (anolyte and catholyte) with water. Eur. Rev. Chem. Res. 4, 72–86 (2015).Article 

Google Scholar 
54.Cotruvo, J., Giddings, M., Jackson, P., Magara, Y. & Ohanian, E. Sodium Dichloroisocyanurate in Drinking-water (2007).55.Xuan, X. et al. Storage stability of slightly acidic electrolyzed water and circulating electrolyzed water and their property changes after application. J. Food Sci. 81, E610–E617 (2016).CAS 
Article 

Google Scholar 
56.Richards, J. J. & Melander, C. Controlling bacterial biofilms. ChemBioChem 10, 2287–2294 (2009).CAS 
Article 

Google Scholar 
57.Stewart, P. S. In Microbial Biofilms (eds. Mukherjee, P. K., Ghannoum, M., Whiteley, M. & Parsek, M.) 269–286 (American Society of Microbiology, 2015). https://doi.org/10.1128/9781555817466.58.Kim, C., Hung, Y.-C., Bracket, R. E. & Frank, J. F. Inactivation of listeria monocytogenes biofilms by electrolyzed oxidizing water. J. Food Process. Preserv. 25, 91–100 (2011).Article 

Google Scholar 
59.Flemming, H. C. & Wingender, J. The biofilm matrix. Nat. Rev. Microbiol. 8, 623–633 (2010).CAS 
Article 

Google Scholar 
60.Zinkevich, V., Beech, I. B., Tapper, R. & Bogdarina, I. The effect of super-oxidized water on Escherichia coli. J. Hosp. Infect. 46, 153–156 (2000).CAS 
Article 

Google Scholar 
61.Cloete, T. E., Thantsha, M. S., Maluleke, M. R. & Kirkpatrick, R. The antimicrobial mechanism of electrochemically activated water against Pseudomonas aeruginosa and Escherichia coli as determined by SDS-PAGE analysis. J. Appl. Microbiol. 107, 379–384 (2009).CAS 
Article 

Google Scholar 
62.Ding, T., Oh, D. H. & Liu, D. Electrolyzed Water in Food: Fundamentals and Applications (2019). https://doi.org/10.1007/978-981-13-3807-6.63.Hall-Stoodley, L., Costerton, J. W. & Stoodley, P. Bacterial biofilms: from the natural environment to infectious diseases. Nat. Rev. Microbiol. 2, 95–108 (2004).CAS 
Article 

Google Scholar 
64.BioSurface Technologies Corp. CDC Biofilm Reactor Operator’ s Manual (BioSurface Technologies Corp.) 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

Water access data processingData on drinking water coverage by region was acquired from the WHO/UNICEF JMP. The JMP acts as official custodian of global data on water supply, sanitation and hygiene2 and assimilates data from administrative data, national census and surveys for individual countries, and maintains a database that can be accessed online through their website. We accessed data tables for national and subnational drinking water service levels from https://washdata.org.JMP datasets are not geographically linked to official boundary files. We joined the tables to GIS boundaries obtained from the following open-source collections: GADM (https://gadm.org), the Spatial Data Repository of the Demographic and Health Surveys Program of USAID (DHS) and the Global Data Lab of Radboud University (GDL)2,50,51,52,53. Subnational regions reported by the JMP are unstructured, representing various regional administrative levels (province, state, district and others).The JMP national and subnational data were joined to GIS boundaries using a custom geoprocessing tool built in Python and ArcGIS 10. The tool joins the available JMP subnational-level survey data to the closest name match of regional boundary names from a merged stack of GADM (admin1, admin2 and admin3), DHS and GDL boundaries worldwide. The JMP national-level survey data is then joined to GADM national (admin0) boundaries for countries which have no subnational data available. Finally, the two boundary-joined datasets (national and subnational) are merged, processed and exported as a seamless global fabric of water-stressed-population data at the highest respective spatial resolutions available (Fig. 1a).JMP does not report the breakdown between the SMDW and basic service level within subnational regions, and instead reports a combined category called ‘at least basic’ (ALB). To estimate the SMDW values in subnational regions, a simple cross-multiplication was performed using the splits at the national level:$${{rm{SMDW}}}_{{rm{subnational}}}=frac{{{rm{SMDW}}}_{{rm{national}}}}{{{rm{ALB}}}_{{rm{national}}}}{times {rm{ALB}}}_{{rm{subnational}}},$$where ALBnational, ALBsubnational and SMDWnational are known values.Validation of the cross-estimation of share of SMDW from ALB for subnational regions was conducted on a reference dataset of nationally representative household surveys that collected data on all criteria for SMDW54, shown in Extended Data Fig. 2. We report regression results of R2 = 0.87 and a standard error of 3.67, indicating a bias which over-reports SMDW share and a probable underestimate of people living without SMDW in our study. This discrepancy comes from JMP calculations of SMDW that rely on the minimum value of multiple drinking water service criteria (free from contamination, available when needed and accessible on premise) rather than considering whether individual households meet all criteria for SMDW55.The fraction of population without SMDW was multiplied by residential population values in the WorldPop top-down unconstrained global mosaic population count of 2017 at 1 km spatial resolution56 (https://www.worldpop.org). WorldPop was accessed online as a TIF image and imported to Google Earth Engine. The year 2017 was chosen to more closely match water access data from JMP. The percentages reported by JMP are probably not uniform within most regions57, introducing an unknown error to Fig. 1b, but represent the best estimate available to us given the limitations of these regionally reported data.Climate input and conversion approximationsGHI and reference planeWe used GHI (in W m−2) as solar energy input data. GHI has good availability in climate datasets and introduces the fewest number of assumptions. Since GHI describes the irradiance in a locally horizontal reference plane, this approximation is only exact for devices having a horizontally oriented solar harvesting area. Annually averaged comparisons between horizontal and optimal fixed-tilt panels show negligible differences in direct plus diffuse radiation in tropical latitudes, and ratios below 25% in locations within 50° north and south latitudes58. Those seeking precise absolute predictions for tilted devices or higher latitudes are encouraged to adapt the provided code to their specific assumptions.Conversion from SY to AWH outputAs discussed in the main text, solar-driven AWH devices typically have one of two predominant energy inputs: thermal (converted directly from incident sunlight on the device) or electrical (from PV). Here, the energy units used to calculate yield in l kWh−1 are incident solar energy directly from GHI. The various assumptions are made in relation to the reported values based on their source. The thermal limits33, target curves, and experimental results reported by TRP15 and MOFs were assumed to have direct (100%) conversion from sunlight to heat. For the ZMW device, the table provided by the manufacturer accounts for system losses, so the table values were directly converted in our model35. For ref. 34 and the cooler–condenser limits from ref. 32, which both assume work input instead of heat, we applied a typical PV conversion efficiency of 20% to convert from sunlight kWh (GHI) to kWhPV (electrical work) input to the device59.Sufficiently short sorbent cycling timesAWH-Geo assumes continuous or quasi-continuous AWH. AWH-Geo considers each 1-h timestep independently and is thus stateless. Aside from edge cases, this is a safe assumption for mass efficient SC-AWH devices, which typically have time constants shorter than 1 h, both for sorbent cycling and for most of the thermal time constants. For devices with longer time constants, batch devices or processes with slow (de)sorption kinetics, this assumption may introduce increased error, and may require further adaptation of the provided code.Climate time-series calculationAWH-Geo is a resource-assessment tool for AWH. It consists of a geospatial processing pipeline for mapping water production (in litres per unit time) around the world of any solar-driven continuous AWH device that can be characterized by an output table of the form output = f(RH, T, GHI).Output tables show AWH output values in l h−1 or l h−1 m−2 across permutations of the 3 main climate variables in the following ranges: RH between 0 and 100 % in intervals of 10%, GHI between 0 and 1,300 W m−2 in intervals of 100 W m−2, and T between 0 and 45 °C in intervals of 2.5 °C (2,145 total output values). The tables are converted into a 3D array image in Google Earth Engine and processed across the climate time-series image collection for the period of interest. Finally, these AWH output values are composited (reduced) to a single time-averaged statistic of interest as an image.Climate time-series data was acquired from the ERA5-Land climate reanalysis from the European Centre for Medium-Range Weather Forecasts (ECMWF)60, accessed from the Google Earth Engine data catalogue. ERA5-Land surface variables were used in 1-h intervals and 0.1°× 0.1° (nominal 9 km). The 10-year analysis period (2010–2019, inclusive) was used for this work, and represents a period long enough to provide a reasonable correction for medium-term interannual climatic variability.Climate variables GHI and T were matched to ERA5-Land parameters ‘Surface solar radiation downwards’ (converted from cumulative to mean hourly) and ‘2 metre temperature’ (converted from K to °C), respectively. RH was calculated from the ambient and dew point temperature parameters in a relationship derived from the August–Roche–Magnus approximation61 rearranged as:$${rm{RH}}=100 % times frac{{{rm{e}}}^{left(frac{a{T}_{{rm{d}}}}{b+{T}_{{rm{d}}}}right)}}{{{rm{e}}}^{left(frac{{aT}}{b+T}right)}}$$where a is 17.625 (constant), b is 243.04 (constant), T is the ERA5-Land parameter ‘2 metre temperature’ converted from K to °C, and Td is the ERA5-Land parameter ‘2 metre dewpoint temperature’ converted from K to °C.Spot validation of the climate parameters and the mapped output was performed manually in Google Earth Engine across several timesteps in 2016 in Ames, Iowa (using the Iowa Environmental Mesonet AMES-8-WSW station62) and showed insignificant error (99.99 to 95.80% for thermal absorbers, depending on the level of angular selectivity63.Rearranged, Kim’s model yields$$frac{{dot{V}}_{{rm{water}},{rm{out}}}}{A}le {E}_{{rm{GHI}}}times left(1-frac{{T}_{{rm{ambient}}}}{{T}_{{rm{hot}}}}right)times {left[frac{1}{{omega }_{{rm{air}},{rm{in}}}-{omega }_{{rm{air}},{rm{out}}}}({e}_{{rm{air}},{rm{out}}}-{e}_{{rm{air}},{rm{in}}})+{e}_{{rm{water}},{rm{out}}}right]}^{-1}times frac{1}{{rho }_{{rm{water}}}}$$where, in addition, ({dot{V}}_{{rm{water}},{rm{out}}}) is the production rate of liquid water by volume, ({A}) is the area harvesting sunlight (see approximation section below), ({E}_{{rm{GHI}}}) is GHI in Wsun m−2, and ({rho }_{{rm{water}}}) is the density of water.This is now a function of the three key climate variables: GHI (in the first term), ambient temperature (in the second and hidden in the third term) and RH (entering the third term). This was converted to an output table and processed through the AWH-Geo pipeline and presented in Fig. 3a. While this can be run for any choice of parameter ({T}_{{rm{hot}}}), we present figures here for ({T}_{{rm{hot}}}) = 100 °C, a temperature still achievable in low-cost (non-vacuum) practical devices without tracking or sunlight concentration. Higher driving temperatures increase the upper bound for water output. For the limits analysis, values of RH above 90% are clamped to prevent unrealistically high theoretical outputs as Kim’s equation goes to infinity at 100% RH. A further assumption is made that new ambient air is efficiently refreshed.Figure 3b maps the maximum yield for active cooler–condensers without recuperation of sensible heat—all given work input and an optimum coefficient of performance of the cooling unit at a condenser temperature that maximizes specific yield as modelled by Peeters32, which we digitized from their fig. 11. Peeters chose to set yield to zero whenever frost formation would be expected on the condenser. Since Peeters assumes work input, we convert from solar energy (GHI) to kWhPV as discussed above.Figure 3c maps Zhao’s experimental results from a TRP using a logistic regression curve fit to their reported SYs of 0.21, 3.71 and 9.28 l kWh−1 at 30, 60 and 90% RH, respectively15. The terms of the curve fit are reported in the next section.Custom yellow to blue map colours are based on www.ColorBrewer.org, by C. A. Brewer, Penn State64.Specific yield and target curvesTwo simple characteristic equations, linear and logistic, were used to fit a limited set of SY and RH pairs from laboratory experiments or reported values and plotted through AWH-Geo using calculated output tables. Hypothetical curves of similar form whose terms were adjusted iteratively in AWH-Geo to goal-seek a target output (5 l d−1) and user base, and are reported here (for 1-m2 devices). In the following equations, RH in % is taken as a fraction (for example 55% is equivalent to 0.55).The linear target curve is a simple linear function which crosses the y-axis at zero:$${rm{SY}}({rm{RH}})=atimes {rm{RH}}$$where a is set to 1.60, 1.86 and 2.60 L/kWh to reach targets of 0.5, 1.0, and 2.0 billion people without SMDW, respectively, and RH is input RH (fractional).The logistic target curve is a logistic function:$${rm{SY}}({rm{RH}})=frac{L}{1+{{rm{e}}}^{-k({rm{RH}}-{{rm{RH}}}_{0})}}$$where L is set to 1.80, 2.40 and 4.80 L kWh−1 to reach targets of 0.5, 1.0 and 2.0 billion people without SMDW, respectively, k is the growth rate set to 10.0, and ({rm{RH}}) and ({{rm{RH}}}_{0}) are input RH (fractional), and 0.60, respectively.The SY values reported by Zhao for TRPs (which they term ‘SMAG’) were fit to a logistic function of the same form with the following parameters: L set to 9.81 L kWh−1, k set to 11.25 and RH0 set to 0.645.The resulting fitted SY profile is expanded into an output table. As with all reports providing SY values instead of full output tables, this forces an assumption of linearity in heat rate (approximately equal to GHI), which may introduce error at lower GHI levels. Zhao reports SY of the TRP material is consistent across temperature below 40 °C—the material’s lower critical solution temperature—above which its performance drops precipitously. Accordingly, we set the SY to 0 l kWh−1 for temperatures ≥40 °C in the output table.Bagheri reported performance of three existing AWH devices across several climate conditions using an ‘energy consumption rate’ in kWh/L, which can be considered to be the SEC, and the simple reciprocal of SY. Instead of fitting a logistic curve to the reciprocals, we fit an exponential function to the average SEC of the three devices in conditions above 20 °C of the equation:$${rm{SEC}}({rm{RH}})=9.03{{rm{e}}}^{-2.99{rm{RH}}}$$where SEC is specific energy consumption in kWhPV l−1 and RH is fractional.This was applied to RH and taken as reciprocal in an output table and run through AWH-Geo. Since Bagheri reports the equivalent of kWhPV, we scale to adapt to GHI input with a photovoltaic conversion efficiency as discussed above.For performance of the ZMW device (the company’s ~3 m2 SOURCE Hydropanel), we used values from the panel production contour plot in the technical specification sheet available from the manufacturer’s website35. The decision for inclusion was made owing to the importance as an early example of a SC-AWH product with commercial intent. Values in l per panel per day were taken at each 10% RH step at 5 kWh m−2, assumed to represent kWh m−2 d−1, and divided by 15 kWh (~3 m2 × 5 kWh m−2) to convert to SY in l kWh−1. From the resulting SY curve, an output table was generated and processed with AWH-Geo.Coincidence analysis and population sumsThe coincidence analysis was run through AWH-Geo across 70 threshold pairs given the full permutation set of RH from 10 to 100% and GHI from 400 to 700 W m−2 threshold intervals, using binary image time series. The resulting mean multiplied by 24 represents average hours per day thresholds are met simultaneously, giving ophd. Below is a functional representation of this time-series calculation:$${langle ({{rm{RH}}}_{t,{rm{px}}} > {{rm{RH}}}_{{rm{threshold}}}){{rm{& & }}}_{{rm{simultaneous}}}({{rm{GHI}}}_{t,{rm{px}}} > {{rm{GHI}}}_{{rm{threshold}}})rangle }_{{rm{time; average}}}$$where ({{rm{RH}}}_{t,{rm{px}}}) is the RH in the map pixel ({rm{px}}) at time (t), ({{rm{RH}}}_{{rm{threshold}}}) is the threshold of RH above which the device is assumed to operate, ({{rm{GHI}}}_{t,{rm{px}}}) is the GHI in the map pixel ({rm{px}}) at time (t), and ({{rm{GHI}}}_{{rm{threshold}}}) is the threshold of GHI above which the device is assumed to operate.The population calculation was then conducted on these images in Google Earth Engine.Zonal statistics were performed on the mean ophd images as integers (0–24) using a grouped image reduction (at 1,000-m scale) summing the population integer counts on the population without SMDW distribution image created previously (derived from WorldPop). This reduction was performed at 1,000 m. Validation was performed in Google Earth Engine on single countries within single ophd zones and showed insignificant error ( More

General description of a geoscientific model and parameter calibrationA model for both non-dynamical and dynamical systems can be generically written for site i as$${left{{{{{{{bf{y}}}}}}}_{t}^{i}right}}_{tin T}=fleft({left{{{{{{{bf{x}}}}}}}_{t}^{i}right}}_{tin T},{{{{{{boldsymbol{varphi }}}}}}}^{{{{{{boldsymbol{i}}}}}}},{{{{{{boldsymbol{theta }}}}}}}^{{{{{{boldsymbol{i}}}}}}}right)$$
(1)
where output physical predictions (({{{{{{bf{y}}}}}}}_{t}^{i}={left[{y}_{1,t}^{i},{y}_{2,t}^{i},cdots right]}^{T}), with the first subscript denoting variable type) vary with time (t) and location (i), and are functions of time- and location-specific inputs (({{{{{{bf{x}}}}}}}_{t}^{i}={left[{x}_{1,t}^{i},{x}_{2,t}^{i},cdots right]}^{T})), location-specific observable attributes (({{{{{{boldsymbol{varphi }}}}}}}^{i}={left[{varphi }_{1}^{i}{{{{{boldsymbol{,}}}}}}{varphi }_{2}^{i}{{{{{boldsymbol{,}}}}}}{{{{{boldsymbol{cdots }}}}}}right]}^{T})), and location-specific unobserved parameters that need to be separately determined (({{{{{{boldsymbol{theta }}}}}}}^{i}={left[{theta }_{1}^{i}{{{{{boldsymbol{,}}}}}}{theta }_{2}^{i}{{{{{boldsymbol{,}}}}}}{{{{{boldsymbol{cdots }}}}}}right]}^{T})). θ may be unobservable, or it may be too expensive or difficult to observe at the needed accuracy, resolution, or coverage. This formulation also applies to dynamical systems if ({{{{{{bf{x}}}}}}}_{{{{{{boldsymbol{t}}}}}}}^{{{{{{boldsymbol{i}}}}}}}) includes previous system states ({{{{{{bf{y}}}}}}}_{t-1}^{i}) (i.e. ({{{{{{bf{y}}}}}}}_{t-1}^{i}{{{{{boldsymbol{subset }}}}}}{{{{{{bf{x}}}}}}}_{t}^{i})), and the rest of the inputs are independent (e.g. meteorological) forcing data. In a non-dynamical system, ({{{{{{bf{x}}}}}}}_{t}^{i}) is independent of ({{{{{{bf{y}}}}}}}_{t-1}^{i}).Given some observations$${{{{{{bf{z}}}}}}}_{t}^{i}=hleft({{{{{{bf{y}}}}}}}_{{t}}^{{i}}right)+{{{{{{boldsymbol{varepsilon }}}}}}}_{t}^{i}$$
(2)
where h(·) relates model outputs to observations and ({{{{{{boldsymbol{varepsilon }}}}}}}_{{{boldsymbol{t}}}}^{{{boldsymbol{i}}}}={big[{varepsilon }_{1,t}^{i},{varepsilon }_{2,t}^{i},cdots big]}^{T}) is the error between the observations (Big({{{{{{boldsymbol{z}}}}}}}_{{{{{{boldsymbol{t}}}}}}}^{{{{{{boldsymbol{i}}}}}}} = {big[{z}_{1,t}^{i},{z}_{2,t}^{i},cdots big]}^{T}Big)) and the model predictions (({{{{{{bf{y}}}}}}}_{{{{{{boldsymbol{t}}}}}}}^{{{{{{boldsymbol{i}}}}}}})), we adjust the model parameters so that the predictions best match the observations. This is traditionally done individually for each location (here generically referring to a gridcell, basin, site, river reach, agricultural plot, etc., depending on the model):$${hat{theta }}^{i}={{arg }},{{{min }}}_{{{{{{{boldsymbol{theta }}}}}}}^{i}}mathop{sum }limits_{tin T}{Vert {{{{{{boldsymbol{varepsilon }}}}}}}_{t}^{i}Vert }^{2}={{arg }},{{{min }}}_{{{{{{{boldsymbol{theta }}}}}}}^{i}}mathop{sum}limits_{tin T}{Vert h(f({{{x}_{t}^{i}}}_{tin T},{varphi }^{i},{theta }^{i}))-{z}_{t}^{i}Vert }^{2}$$
(3)
where i ∈ I and where (I=left{{1,2},ldots ,{N}_{I}right}). Note that the superscript i suggests that this optimization is done for each site independently.The process-based hydrologic model and its surrogateThe Variable Infiltration Capacity (VIC) hydrologic model has been widely used for simulating the water and energy exchanges between the land surface and atmosphere, along with related applications in climate, water resources (e.g., flood, drought, hydropower), agriculture, and others. The model simulates evapotranspiration, runoff, soil moisture, and baseflow based on conceptualized bucket formulations. Inputs to the model include daily meteorological forcings, non-meteorological data, and the parameters to be determined. Meteorological forcing data include time series of precipitation, air temperature, wind speed, atmospheric pressure, vapor pressure, and longwave and shortwave radiation. More details about VIC can be found in Liang et al.39.LSTM was trained to reproduce the behavior of VIC as closely as possible while also allowing for gradient tracking. In theory, if a hydrologic model can be written into a machine learning platform (as in our HBV case), this step is not needed, but training a surrogate model is more convenient when the model is complex. To ensure the surrogate model had high fidelity in the parameter space where the search algorithms want to explore, we iterated the training procedure multiple times. We first trained an LSTM surrogate for VIC using the forcings, attributes, and parameters from NLDAS-2 as inputs, and the VIC-simulated surface soil moisture (variable name: SOILM_lev1) and evapotranspiration (ET, variable name: EVP) as the targets of emulation. Then, as the search algorithms (SCE-UA or dPL) went near an optimum, we took the calibrated parameter sets, made perturbations of them by adding random noise to these parameters, and retrained the network with added data. The perturbation was done to better represent the parameter space close to optimal solutions. We repeated this procedure four times so that the NSEs of the parameters, obtained from the CPU-based VIC model, converged. At 1/82 sampling density (sampling one gridcell from each 8 × 8 patch), this results in fewer overall forward runs than a 1/8-degree NLDAS-2 simulation. Also note that this effort is needed similarly for both dPL and SCE-UA. If we did not use the surrogate model, SCE-UA would also have needed to employ the O(102) more expensive CPU-based VIC model. We evaluated the accuracy of the surrogate model, and the median correlations between VIC and the surrogate simulation were 0.91 and 0.92 for soil moisture and ET, respectively (Supplementary Fig. S2). When we connected the trained surrogate model to the parameter estimation network, the weights of the surrogate model were frozen and prevented from updating by backpropagation, but the gradient information could pass through. This was implemented in the PyTorch deep learning framework36.The long short-term memory networkThe long short-term memory network (LSTM) was originally developed in the artificial intelligence field for learning sequential data, but has recently become a popular choice for hydrologic time series data26. As compared to a vanilla recurrent neural network with only one state, LSTM has two states (cell state, hidden state) and three gates (input gate, forget gate, and output gate). The cell state enables long-term memory, and the gates are trained to determine which information to carry across time steps and which information to forget. These units were collectively designed to address the notorious DL issue of the vanishing gradient, where the accumulated gradients would decrease exponentially along time steps and eventually be too small to allow effective learning48. Given inputs I, our LSTM can be written as the following:$${{{{{mathrm{Input}}}}}}; {{{{{mathrm{transformation:}}}}}}quad {x}^{t}={{ReLU}}({W}_{I}{I}^{t}+{b}_{I})$$
(4)
$${{{{{mathrm{Input}}}}}}; {{{{{mathrm{node:}}}}}}quad {g}^{t}={tanh }({{{{{mathscr{D}}}}}}({W}_{{gx}}{x}^{t})+{{{{{mathscr{D}}}}}}({W}_{{gh}}{h}^{t-1})+{b}_{g})$$
(5)
$${{{{{mathrm{Input}}}}}}; {{{{{mathrm{gate:}}}}}}quad {i}^{t}=sigma ({{{{{mathscr{D}}}}}}({W}_{{ix}}{x}^{t})+{{{{{mathscr{D}}}}}}({W}_{{ih}}{h}^{t-1})+{b}_{i})$$
(6)
$${{{{{mathrm{Forget}}}}}}; {{{{{mathrm{gate:}}}}}}quad {f}^{t}=sigma ({{{{{mathscr{D}}}}}}({W}_{{fx}}{x}^{t})+{{{{{mathscr{D}}}}}}({W}_{{fh}}{h}^{t-1})+{b}_{f})$$
(7)
$${{{{{mathrm{Output}}}}}}; {{{{{mathrm{gate:}}}}}}quad {o}^{t}=sigma ({{{{{mathscr{D}}}}}}({W}_{{ox}}{x}^{t})+{{{{{mathscr{D}}}}}}({W}_{{oh}}{h}^{t-1})+{b}_{o})$$
(8)
$${{{{{mathrm{Cell}}}}}}; {{{{{mathrm{state:}}}}}}quad {s}^{t}={g}^{t}odot {i}^{t}+{s}^{t-1}odot {f}^{t}$$
(9)
$${{{{{mathrm{Hidden}}}}}}; {{{{{mathrm{state:}}}}}}quad {h}^{t}={tanh }({s}^{t})odot {o}^{t}$$
(10)
$${{{{{mathrm{Output:}}}}}}quad {y}^{t}={W}_{{hy}}{h}^{t}+{b}_{y}$$
(11)
where W and b are the network weights and bias parameters, respectively, and ({{{{{mathscr{D}}}}}}) is the dropout operator, which randomly sets some of the connections to zero. The LSTM network and our whole workflow31 were implemented in PyTorch36, an open source machine learning framework.Here we do not use LSTM to predict the target variable. Rather, LSTM is used to (optionally) map from time series information to the parameters in our gz network as described below.The parameter estimation networkWe present two versions of the dPL framework. The first version allows us to train a parameter estimation network over selected training locations Itrain where some ancillary information A (potentially including but not limited to attributes in φi used in the model) is available, for training period Ttrain (illustrated in Fig. 1b):$${hat{{{{{{boldsymbol{theta }}}}}}}}^{i}={g}_{A}left({{{{{{bf{A}}}}}}}^{i}right){{{{{{rm{for}}}}}}; {{{{{rm{all}}}}}}; iin I}_{{{{{{{mathrm{train}}}}}}}}$$
(12a)
$${hat{g}}_{A}(cdot )={{{{{rm{arg }}}}}},{{{{{{rm{min }}}}}}}_{{g}_{A}(cdot )}mathop{sum}limits_{tin T,iin {I}_{{{{{{rm{train}}}}}}}}{Vert h(f({x}_{t}^{i},{varphi }^{i},{g}_{A}({{{{{{bf{A}}}}}}}^{i})))-{z}_{t}^{i}Vert }^{2}$$
(12b)
Essentially, we train a network (gA) mapping from raw data (A) to parameters (({{{{{boldsymbol{theta }}}}}})) such that the PBM output (f) using ({{{{{boldsymbol{theta }}}}}}) best matches the observed target (({{{{{bf{z}}}}}})). We are not training to predict the observations – rather, we train gA on how to best help the PBM to achieve its goal. The difference between Eq. 12 and Eq. 3 highlights that the loss function combines the sum of squared differences for all sites at once.The second version is applicable where some observations ({left{{{{{{{bf{z}}}}}}}_{t}^{i}right}}_{tin T}) are also available as inputs at the test locations:$${hat{{{{{{boldsymbol{theta }}}}}}}}^{i}={g}_{z}left({left{{{{{{{bf{x}}}}}}}_{t}^{i}right}}_{tin T},{{{{{{bf{A}}}}}}}^{{prime} ,i},{left{{{{{{{bf{z}}}}}}}_{t}^{i}right}}_{tin T}right){{{{{rm{for}}}}}}{,,}{{{{{rm{all}}}}}}{,,}{iin I}_{{{{{{{mathrm{train}}}}}}}}$$
(13a)
$$widehat{{g}_{z}}(cdot )={{{{{rm{arg }}}}}},{{{{{{rm{min }}}}}}}_{{g}_{z}(cdot )}mathop{sum }limits_{tin {T}_{{{{{{rm{train}}}}}}},iin {I}_{{{{{{rm{train}}}}}}}}{Vert h(f({x}_{t}^{i},{varphi }^{i},{g}_{z}({{{{{{{{bf{x}}}}}}}_{t}^{i}}}_{tin T},{{{{{{bf{A}}}}}}}^{{prime} ,i},{{{z}_{t}^{i}}}_{tin T})))-{z}_{t}^{i}Vert }^{2}$$
(13b)
Essentially, we train a network (({g}_{z})) that maps from attributes (A′), historical forcings (x), and historical observations (({left{{{{{{{bf{z}}}}}}}_{t}^{i}right}}_{tin T})) to a suitable parameter set (({{{{{boldsymbol{theta }}}}}})) with which the PBM output best matches the observed target (({{{{{bf{z}}}}}})) across all sites in the domain. Ancillary attributes A′ may be the same as or different from A used in gA, and in the extreme case may be empty. Succinctly, they can be written as two mappings, gA: A → θ and gZ: (A′,x,z) → θ. gZ can accept time series data as inputs and here we choose LSTM as the network structure for this unit. There is no circular logic or information leak because the historical observations (({left{{{{{{{bf{z}}}}}}}_{t}^{i}right}}_{tin T})) are for a different period (T) than the main training period (Ttrain). In practice, this distinction may not be so crucial as the PBM acts as an information barrier such that only values suitable as parameters (({{{{{boldsymbol{theta }}}}}})) can produce a reasonable loss. As LSTM can output a time series, the parameters were extracted only at the last time step. For gA, only static attributes were employed, and so the network structure amounts to a multilayer perceptron network. After some investigation of training and test metrics, we set the hidden size of g to be the same as for the surrogate model.The whole network is trained using gradient descent, which is a first-order optimization scheme. Some second-order schemes like Levenberg–Marquardt often have large computational demand and are thus rarely used in modern DL49. To allow gradient accumulation and efficient gradient-based optimization and to further reduce the computational cost, we can either implement the PBM directly into a differentiable form, as described in the global PUB case below, or first train a DL-based surrogate model (f^{prime} left(bullet right)simeq fleft(bullet right)) and use it in the loss function instead of f(·),$$g(cdot )={{{{{rm{arg }}}}}},{{{{{{rm{min }}}}}}}_{g(cdot )}mathop{sum}limits_{tin {T}_{{{{{{rm{train}}}}}}},iin {I}_{{{{{{rm{train}}}}}}}}{Vert h(f{{mbox{‘}}}({x}_{t}^{i},{varphi }^{i},g(cdot )))-{z}_{t}^{i}Vert }^{2}$$
(14)
where (g(bullet )) generically refers to either gA or gZ with their corresponding inputs. gA can be applied wherever we can have the ancillary inputs A, while gZ can be applied over areas where forcings and observed responses (x, z) are also available, without additional training:$${hat{{{{{{boldsymbol{theta }}}}}}}}^{i}={g}_{z}({{{{{{{{bf{X}}}}}}}_{t}^{i}}}_{tin T}{{{{{boldsymbol{,}}}}}}{{{{{{boldsymbol{varphi }}}}}}}^{i},{{{{{{{{bf{Z}}}}}}}_{t}^{i}}}_{tin T}){,,}{{{{{rm{or}}}}}},,{hat{{{{{{boldsymbol{theta }}}}}}}}^{i}={g}_{A}({{{{{{boldsymbol{varphi }}}}}}}^{i}); {{{{{rm{for}}}}}},{{{{{rm{any}}}}}}, i,{{{{{rm{and}}}}}},{{{{{rm{any}}}}}},{{{{{rm{reasonable}}}}}},T$$
(15)
We tested both gA and gZ, which work with and without forcing-observation (x-z) pairs among the inputs, respectively. Since SMAP observations have an irregular revisit schedule of 2–3 days and neural networks cannot accept NaN inputs, we have to fill in the gaps, but simple interpolations do not consider the effects of rainfall. Here we used the near-real-time forecast method that we developed earlier30. Essentially, this forecast method uses forcings and integrates recently available observations to forecast the observed variable for the future time steps, achieving very high forecast accuracy (ubRMSE  More

NATURE PODCAST
03 October 2021

Audio long-read: How dangerous is Africa’s explosive Lake Kivu?

A lake in central Africa could one day release a huge amount of greenhouse gases, threatening the lives of millions.

Nicola Jones

0
&

Benjamin Thompson

Nicola Jones

Nicola Jones is a science journalist based in Pemberton, Canada.

View author publications

You can also search for this author in PubMed
 Google Scholar

Benjamin Thompson

View author publications

You can also search for this author in PubMed
 Google Scholar

Share on Twitter
Share on Twitter

Share on Facebook
Share on Facebook

Share via E-Mail
Share via E-Mail

Subscribe
Subscribe

iTunes
Google Podcast
acast
RSS

How dangerous is Africa’s explosive Lake Kivu? Read by Benjamin Thompson

Your browser does not support the audio element.

Download MP3

Lake Kivu, nestled between the Democratic Republic of the Congo and Rwanda, is a geological anomaly that holds 300 cubic kilometres of dissolved carbon dioxide and 60 cubic kilometres of methane.The lake has the potential to explosively release these gases, which could fill the surrounding valley, potentially killing millions of people.Researchers are trying to establish the likelihood of such an event happening, and the best way to safely siphon the gases from the lake.This is an audio version of our feature: How dangerous is Africa’s explosive Lake Kivu?Never miss an episode: Subscribe to the Nature Podcast on Apple Podcasts, Google Podcasts, Spotify or your favourite podcast app. Head here for the Nature Podcast RSS feed

doi: https://doi.org/10.1038/d41586-021-02984-8

Related Articles

How dangerous is Africa’s explosive Lake Kivu?

Subjects

Water resources

Latest on:

Water resources

Global potential for harvesting drinking water from air using solar energy
Article 27 OCT 21

Webcast: how to green your lab
Career Column 25 AUG 21

A staggering store of water is revealed in Earth’s crust
Research Highlight 17 AUG 21

Jobs

Staff Scientist – RNA Biology

Baylor College of Medicine (BCM)
Houston, TX, United States

Postdoctoral Associate-RNA Biology

Baylor College of Medicine (BCM)
Houston, TX, United States

Postdoctoral Associate-RNA Biology

Baylor College of Medicine (BCM)
Houston, TX, United States

Postdoctoral Associate-RNA Biology

Baylor College of Medicine (BCM)
Houston, TX, United States More

1.Siebert, S. & Döll, P. J. Hydrol. 384, 198–217 (2010).Article 

Google Scholar 
2.Davidson, N. C. in The Wetland Book (eds Finlayson, C.M. et al.) https://doi.org/10.1007/978-94-007-6173-5_197-1 (Springer, 2016).3.Micklin, P. Environ. Earth Sci. 75 (2016).4.Evans, A. E., Mateo-Sagasta, J., Qadir, M., Boelee, E. & Ippolito, A. Curr. Opin. Environ. Sustain. 36, 20–27 (2019).Article 

Google Scholar 
5.Kanter, D. R., Chodos, O., Nordland, O., Rutigliano, M. & Winiwarter, W. Nat. Sustain. 3, 956–963 (2020).Article 

Google Scholar 
6.Vushe, A. in Climate Change Management (eds Bamutaze, Y. et al.) 99–128 (Springer, 2019); https://doi.org/10.1007/978-3-030-12974-3_57.Singh, A. Ecol. Indic. 57, 128–130 (2015).Article 

Google Scholar 
8.Vercauteren, K. & Hygnstrom, S. E. in Biology and Management of White-Tailed Deer (ed. Hewitt, D. G.) 501–535 (CRC Press, 2011); http://digitalcommons.unl.edu/natrespapers/3809.Zhang, T., Mahmood, R., Lin, X. & Pielke, R. A. Weather Clim. Extrem. 23, 100197 (2019).Article 

Google Scholar 
10.Nocco, M. A., Smail, R. A. & Kucharik, C. J. Glob. Change Biol. 25, 3472–3484 (2019).Article 

Google Scholar 
11.de Vrese, P. & Stacke, T. Clim. Dyn. 55, 1521–1537 (2020).Article 

Google Scholar 
12.Pei, L. et al. J. Clim. 29, 3541–3558 (2016).Article 

Google Scholar 
13.Cook, B. I. et al. J. Geophys. Res. Atmos. 125, https://doi.org/10.1029/2019JD031814 (2020).14.Sarofim, M.C. et al. (eds) Ch. 2: Temperature-related death and illness (US Global Change Research Program, Washington, DC., 2016); https://health2016.globalchange.gov/downloads#temperature-related-death-and-illness15.Mishra, V. et al. Nat. Geosci. 13, 722–728 (2020).Article 

Google Scholar 
16.Szilagyi, J. & Franz, T. E. Sustain. Water Resour. Manag. 6, https://doi.org/10.1007/s40899-020-00368-w (2020).17.Levia, D. F. et al. Nat. Geosci. 13, 656–658 (2020).Article 

Google Scholar 
18.Strum, K. M. et al. Agric. Ecosyst. Environ. 179, 116–124 (2013).Article 

Google Scholar 
19.Devanand, A., Huang, M., Ashfaq, M., Barik, B. & Ghosh, S. Geophys. Res. Lett. 46 (2019).20.Rosa, L. et al. Proc. Natl Acad. Sci. USA 117, 29526–29534 (2020).Article 

Google Scholar 
21.Pastor, A., Biemans, H. & Gerten, D. Nat. Commun. 8 (2017).22.MacDonald, A. M. et al. Environ. Res. Lett. 16, https://doi.org/10.1088/1748-9326/abd661 (2021).23.Davis, K. F. et al. Sci. Adv. 4, eaao1108 (2018).Article 

Google Scholar 
24.Gleeson, T. et al. One Earth. 2, 223–234 (2020).Article 

Google Scholar 
25.Scheierling, S. M. & Tréguer, D. O. Beyond Crop per Drop: Assessing Agricultural Water Productivity and Efficiency in a Maturing Water Economy (International Bank for Reconstruction and Development / The World Bank, 2018) https://documents1.worldbank.org/curated/en/352321530075399351/pdf/127625-PUB-Date-6-28-2018-PUBLIC-Beyond-Crop-per-Drop.pdf26.Mishra, V., Asoka, A., Vatta, K. & Lall, U. Earths Future 6, 1672–1681 (2018).Article 

Google Scholar 
27.Gleeson, T., Wada, Y., Bierkens, M. F. P. & van Beek, L. P. H. Nature 488, 197–200 (2012).Article 

Google Scholar 
28.State of Indian Agriculture (Government of India, 2013); http://agricoop.nic.in/sia111213312.pdf29.Rahman, M. A. & Hasegawa, H. Sci. Total Environ. 409, 4645–4655 (2011).Article 

Google Scholar 
30.Singh, D. et al. Journal of Geophysical Research: Atmospheres, 123 (2018).31.Wada, Y. et al. One Earth. 1, 185–194 (2019).Article 

Google Scholar 
32.Ambika, A.K., Wardlow, B. & Mishra, V. Remotely sensed high resolution irrigated area mapping in India for 2000 to 2015. Sci. Data, 3 (2016). More