Quantifying the Urban Food−Energy−Water Nexus: The Case of theDetroit Metropolitan AreaSai Liang,† Shen Qu,‡ Qiaoting Zhao,‡ Xilin Zhang,‡ Glen T. Daigger,§ Joshua P. Newell,‡

Shelie A. Miller,‡ Jeremiah X. Johnson,∥ Nancy G. Love,§ Lixiao Zhang,† Zhifeng Yang,†

and Ming Xu*,‡,§

†State Key Joint Laboratory of Environment Simulation and Pollution Control, School of Environment, Beijing Normal University,Beijing 100875, People’s Republic of China‡School for Environment and Sustainability, University of Michigan, Ann Arbor, Michigan 48109-1041, United States§Department of Civil and Environmental Engineering, University of Michigan, Ann Arbor, Michigan 48109-2125, United States∥Department of Civil, Construction, and Environmental Engineering, North Carolina State University, Raleigh, North Carolina27695-7908, United States

*S Supporting Information

ABSTRACT: The efficient provision of food, energy, and water (FEW)resources to cities is challenging around the world. Because of the complexinterdependence of urban FEW systems, changing components of onesystem may lead to ripple effects on other systems. However, the inputs,intersectoral flows, stocks, and outputs of these FEW resources from theperspective of an integrated urban FEW system have not beensynthetically characterized. Therefore, a standardized and specificaccounting method to describe this system is needed to sustainablymanage these FEW resources. Using the Detroit Metropolitan Area(DMA) as a case, this study developed such an accounting method byusing material and energy flow analysis to quantify this urban FEW nexus.Our results help identify key processes for improving FEW resourceefficiencies of the DMA. These include (1) optimizing the dietary habitsof households to improve phosphorus use efficiency, (2) improving effluent-disposal standards for nitrogen removal to reducenitrogen emission levels, (3) promoting adequate fertilization, and (4) enhancing the maintenance of wastewater collectionpipelines. With respect to water use, better efficiency of thermoelectric power plants can help reduce water withdrawals. Themethod used in this study lays the ground for future urban FEW analyses and modeling.

■ INTRODUCTION

Food, energy, and water are three essential resources for meetingbasic human needs. The efficient provision of food, energy, andwater is important but challenging due to the complexinterdependence of these three systems.1,2 For example, wateris required for food and energy production; energy is needed forfood and water production, as well as wastewater treatment; andindustrial agriculture and food production lead to the oversupplyof nitrogen and phosphorus in the water system from fertilizeruses. Such complex interdependence makes the food−energy−water (FEW) nexus a complex system of subsystems. Changingcomponents of one system may lead to ripple effects (desired orundesired) on other systems. Therefore, policy and technologysolutions addressing challenges in individual FEW systems needto be evaluated through the lens of the FEW nexus to identifycobenefits and avoid unintended consequences. Ultimately,instead of examining FEW subsystems individually, we need toexamine them simultaneouslyas an integrated wholewhendeveloping policy and technology solutions.

As of 2017, 55% of the world’s population lives in urbanareas,3 a proportion projected to increase to 66% by 2050, withthe expected addition of 2.5 billion people to cities.4 Cities havebecome the primary users of FEW resources. Managing FEWresources wisely in cities can contribute significantly to thesustainable management of FEW resources at the global scale.To help a city better manage its FEW resources, the first andforemost step is to understand how FEW resources flow acrossand within a city. Accounting for FEW resource flows and stocksof a city provides a quantitative wiring diagram of the city’s FEWnexus, allowing the city tomeasure the efficiency of its utilizationof FEW resources, to identify critical processes that are key forthe demand of FEW resources, and to develop policy andtechnology solutions accordingly.Previous studies at the urban scale either analyze FEW

systems individually (e.g., nutrient flows,5−17 energy flows,18−25

Received: November 5, 2018Accepted: December 12, 2018Published: December 12, 2018

Article

pubs.acs.org/estCite This: Environ. Sci. Technol. 2019, 53, 779−788

© 2018 American Chemical Society 779 DOI: 10.1021/acs.est.8b06240Environ. Sci. Technol. 2019, 53, 779−788

Dow

nloa

ded

via

UN

IV O

F M

ICH

IGA

N A

NN

AR

BO

R o

n Fe

brua

ry 6

, 201

9 at

20:

34:1

0 (U

TC

).

See

http

s://p

ubs.

acs.

org/

shar

ingg

uide

lines

for

opt

ions

on

how

to le

gitim

atel

y sh

are

publ

ishe

d ar

ticle

s.

and water flows26−28), or examine two of the three FEW systems(e.g., water−energy nexus29−41 and food−water nexus42−44).Only a few studies consider all three FEW systems, and then onlyto evaluate the impact of one system on the other two, such ashow innovations in the urban water sector change food andenergy flows.45−47 Studies have also examined environmentalfootprints of an urban system by life-cycle-based ap-proaches.48,49 However, the inputs, intersectoral flows, stocks,and outputs of FEW resources within the integrated urban FEWsystem have not been synthetically characterized. This synthesiscan lay a solid ground for subsequent studies on the investigationof structural characteristics of the urban FEW nexus and toevaluate the consequences of specific policies or technologysolutions. Despite the importance of measuring urban FEWflows and stocks, we lack case studies and, more importantly, astandardized specific accounting method to describe theintegrated FEW system of cities.This study addresses these gaps by developing a method to

quantify these urban FEW flows and stocks and theirinteractions. We develop this method on the basis of materialand energy flow analysis (MEFA), which quantifies flows andstocks of a substance, a group of substances, bulk materials, orenergy within a system during a given period of time.50−52

Applying this method enables a mathematic representation ofthe urban FEW nexus, which can be the foundation of furtheranalyses and modeling of these systems. We demonstrate thismethod using the Detroit Metropolitan Area (DMA) in 2012 asa case study. The DMA grew rapidly in the 1950s and thendeclined significantly. Now it is experiencing a rebirth, withredevelopment plans and projects that are reshaping flows of

food, energy, and water. Thus, the DMA typifies a city-regionthat has changed significantly over the past decades and offers adynamic case to demonstrate the method’s flexibility. We alsodeveloped an Excel-based template to streamline the datacompilation process, largely using publicly available sources, foreasy adoption by other cities, particularly cities in the UnitedStates.

■ MATERIALS AND METHODSMEFA quantifies the amounts of materials and energy flowing into and out from an economic system as well as the flows andstocks of materials and energy within the economic system. Inother words, MEFA traces the flows and stocks of materials andenergy of an economic system,50−52 based on mass and energybalance principles. In this study, we use MEFA to characterizethe urban FEW nexus by tracking and quantifying the flows andstocks of food (represented by the amounts of nitrogen andphosphorus embedded in food), water, and energy in an urbanarea during a certain period of time.In this study, we focus on the interconnections among

processes of FEW subsystems rather than just interconnectionsamong different flows. For example, the energy−food nexusrefers to energy flows from processes of the energy subsystem toprocesses of the food subsystem (e.g., energy use by foodproduction) and food flows from processes of the foodsubsystem to processes of the energy subsystem (e.g., electricitygeneration from food wastes). As such, FEW subsystemprocesses are interconnected through FEW flows, andinterventions in a process of a subsystem can have ripple effectsthrough other internal and external processes.

Figure 1. System boundaries and processes of (a) food system, (b) energy system, (c) water system, and (d) FEW nexus of a simplified urban system.These graphs illustrate the simplified interactions of urban FEW systems. Panels a, b, and c describe inputs, interprocess flows, and outputs of food,energy , and water systems, respectively. Panel d describes the integrated urban FEW system. Dashed arrows indicate intersystem flows, which are flowsfrom processes of one system to processes of another system.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b06240Environ. Sci. Technol. 2019, 53, 779−788

780

Urban Food−Energy−Water Flows. Figure 1 panels a−cconceptualize the flows and stocks of food, energy, and watersystems, respectively. The food system consists of eightprocesses: fertilizer production, plantation, feed production,husbandry, food processing, food retail, domestic consumption,and solid waste management (Figure 1a). Given the variety offood crops and products, we use nitrogen (N) and phosphorus(P) flows as the currency to represent the flows of materials in thefood system. Plantation and husbandry processes fix N and Pfrom the environment and discharge N and P to theenvironment through N/P loss, N/P embedded in wastewater,and N/P embedded in landfilled solid wastes.The energy system (Figure 1b) extracts fossil fuels or utilizes

renewable resources (e.g., hydrological, solar, and wind power)to produce electricity. After production, waste heat, airpollutants, and solid wastes are discharged. Solid wastes areeither reused or landfilled.The water system primarily consists of water treatment and

supply, wastewater collection and treatment, and residualprocessing (Figure 1c). Water treatment processes withdrawsurface and groundwater and supply treated water to businessesand households for use. Wastewater is collected from businessesand households, treated, and discharged to natural water bodies.There are water and wastewater losses in water distribution andwastewater collection processes. Sludge from water andwastewater treatment facilities is treated by chemical-facilitatedresiduals processing before discharge.Figure 1d shows the flows and stocks of the integrated FEW

system. The food system uses energy produced by the energysystem; the energy system uses solid wastes and landfill gas fromthe food system to produce electricity. The food system also useswater from the water system and generates wastewater collectedand treated by the water system. The energy system providesenergy for the water system to use and generates wastewatercollected and treated by the water system. The urban FEWsystems are also connected with the outside regions throughimports and exports of products.In this study, we first quantify the flows and stocks of

individual FEW systems for major processes (Figure 1) acrossthe urban system. Second, we identify and measure the flowsconnecting the individual FEW systems. Finally, we characterizeeach individual FEW system as a network of processesconnected by resource and energy flows. We integrate thethree individual FEW flow networks as a network of threesubnetworks. Subsequently, we get an integrated FEW flownetwork that provides a topological representation of the urbanFEW nexus.This method does not consider material and energy flows

embodied in imports and exports, because there are alreadyestablished methods53 and abundant case studies.54,55 Suchmethods can be easily integrated with our method in the future.Moreover, stakeholders in an urban area have limited control ofmaterial and energy flows in other regions. Thus, examiningmaterial and energy flows closely associated with an urbansystem is more likely to provide information to guide actionablepolicies for cities.Data for Detroit Metropolitan Area Food−Energy−

Water Nexus. We use the city-region concept to define thesystem boundary for the urban FEW nexus. A city-regiondenotes a metropolitan area and hinterland, typically an urbanarea with multiple administrative districts but sharing resourceslike water and energy infrastructures.56 The system boundary ofthe city-region in this studythe Detroit Metropolitan Area

(DMA)is the Metropolitan Statistical Area (MSA) ofDetroit−Warren−Dearborn and comprises six counties: Lapeer,Livingston, Macomb, Oakland, St. Clair, and Wayne. Wecharacterize the urban FEW nexus for the DMA in 2012, themost recent year for which data are available.In 2012, the DMAwas ranked 14th in population (4.3 million

people), 13th in gross domestic product (GDP, $218.5 billion),109th in per capita income ($42 168/capita), and 95th in percapita GDP ($50 893/capita) among all MSAs in the UnitedStates.57 Data used to construct the DMA FEW nexus networkare primarily from publicly available sources. Estimations weremade when data were unavailable. Table S1 summarizes datasources and estimationmethods for constructing theDMAFEWnexus network in 2012.Data for food flows of the DMA FEW nexus network are

mostly from the U.S. Department of Agriculture (USDA).58,59

Given the variety of food crops and products, we use theamounts of nitrogen (N) and phosphorus (P) contained in thesecrops and products as the currency to represent the food flows.Flows of the crops and products are converted into N and Pflows by multiplying their weight by N/P content parame-ters.60−62 In particular, food imports and exports of DMA arefrom the Commodity Flow Survey.63 Data for crop straws,fertilizers, and biological fixation of N/P are estimated on thebasis of agricultural activity levels.64−69 Stock changes of N andP related to animal and human body weight changes areestimated by mass balance.Data for water withdrawal and supply are from the U.S.

Geological Survey (USGS).70 In particular, USGSwater data areupdated every 5 years, and the latest data are for 2010. We usedthe 2010 USGS data for DMA as a proxy for 2012. We assumethat water used by each process mostly became wastewater. Theamounts of treated wastewater and generated sludge wereobtained from the Detroit Water and Sewerage Department(DWSD).71 Water losses from water distribution and waste-water collection are estimated by water loss rates.72

Data for fossil fuel extraction are from the USDA EconomicsResearch Service.73 Data for electricity generation and fuel usesare from the U.S. Energy Information Administration(EIA).74,75 Data for oil refinery in the DMA come fromMarathon Petroleum.76 The USDA Census of Agriculture58

only has aggregated energy use data for agricultural activities.We estimated energy use of each agricultural process bymultiplying agricultural activity data by corresponding energyintensities.77,78 DWSD provides data for gasoline used tosupport water and wastewater treatment.71 Data for electricityuses for water treatment, wastewater treatment, and residualprocessing were estimated by activity data and electricity useintensities.79 Data for electricity use by households is from theEIA.80 Data for energy imports and exports are from theCommodity Flow Survey.63

This study uses the DMA as a case to illustrate the processesfor data collection and estimation, but the method and data usedare not unique to the DMA. For example, most of the datasources used in this study (e.g., USDA, USGS, Commodity FlowSurvey, and EIA) contain not only the data for the DMA but alsothe data for eachMSA in the United States. Moreover, there maybe differences in the statistics of different countries. However, byproperly adjusting certain data estimation methods andparameters, the proposed methods and templates can also beused in FEW nexus analyses of other countries.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b06240Environ. Sci. Technol. 2019, 53, 779−788

781

■ RESULTSFood−Energy−Water Flows in Detroit Metropolitan

Area.Table 1 shows the inputs and outputs of FEW resources in

the DMA in 2012. Household consumption and related servicesare the main drivers within the city-region. Upstream processesincluding extraction, plantation, processing, and manufacturingare primarily located outside the city-region for food and energy.In particular, total inputs of N and P were 119 Kt (83%imported) and 16 Kt (94% imported), respectively. Energyinputs were 1270 PJ, of which 99% was imported. In contrast,water inputs (5 billion t) were totally withdrawn from local waterbodies.Total N output of DMAwas 114 Kt (73 Kt discharged to local

environment and 41 Kt exported to other regions), and N stockin DMA increased by 5 Kt. Total P output of DMA was 16 Kt(12 Kt to local and 5 Kt exported), and P stock in DMAincreased by 0.1 Kt. Used water was discharged into naturalwater bodies in the form of treated wastewater (4342Mt), waterloss from distribution (140 Mt), and wastewater loss fromcollection (193 Mt). Energy outputs included heat loss duringenergy use (916 PJ) and energy product exports to other regions(354.7 PJ).Figure 2 compares per capita N and P intakes of the DMA

with those of other cities reported in the literature. Per capita Nintake of the DMA (7.3 kg/capita) is at the medium level amongstudied cities: lower than Beijing, Paris, and Vienna and higher

than Linkoping and Toronto. Per capita P intake of DMA (0.9kg/capita) is higher than most of the studied cities (e.g., Harare,Chaohu Watershed in 1995 and 2012, Busia, Linkoping, andThachin Basin). However, it is lower than the ChaohuWatershed in 1978 (1.1 kg/capita) and Phoenix during 2005−2010 (1.0 kg/capita). These comparisons indicate that there arestill potentials to reduce the pressures of N/P demands in theDMA.Figure 3 compares per capita N and P emissions to water

bodies in DMA with those in other cities reported in theliterature. Per capita N emissions to water bodies in DMA were4.0 kg/capita, the highest among investigated cities (e.g., Beijing,Paris, and Linkoping). Per capita P emissions to water bodies inDMA were 0.4 kg/capita, at the medium level of investigatedcities: lower than the Chaohu Watershed, Beijing in 1998 and2008, and Tianjin but higher than Beijing in 1978 and 1988. Pemission level of the DMA was much better than other citiesreported in the literature. However, N emission level of theDMA was much worse than other investigated cities. Thus,reducing per capita N emissions should be of particular concernin the DMA.Figure 4 shows N, P, energy, and water flows of the integrated

FEW system in the DMA. N and P flows (Figure 4a,b) aremainly attributable to five processes: plantation, food process-ing, food retail, domestic consumption, and solid wastemanagement. The largest N and P flows are those embeddedin fly ash from solid waste incineration, 53 and 9 Kt, respectively.The second largest N flow is embedded in imports of processedfoods from other regions (42 Kt), while the largest P flow isembedded in imports of agricultural products from other regions(8 Kt).Energy flows (Figure 4c) are mainly related to liquid fuel/

coke production. Top three largest energy flows are imports ofliquid fuel/coke (513 PJ), liquid fuel/coke use by other sectors(452 PJ), and heat release from other sectors to the environment(769 PJ).Water flows (Figure 4d) are mainly related to electricity

generation, water treatment and supply, and wastewatercollection and treatment processes. Electricity generationwithdrew 3256 Mt of water from local water bodies for coolingpurposes. The used water was then returned to surface waterbodies or evaporated to the atmosphere. Other major waterflows are water withdrawal and supply (872 Mt), wastewatercollection (1209 Mt), and wastewater treatment (1016 Mt). Inparticular, water flows related to the food system are relativelysmall in the DMA, because agricultural activities mainly occuroutside of it.

Food−Energy−Water Nexus in Detroit MetropolitanArea. Figure 4 also shows the nexus among individual FEWsystems. In particular, the food system used 425Mt of water, andit discharged 424 Mt of wastewater, 4 Kt of N, and 1 Kt of P tothe water system through wastewater flows. Food wastes, sludge,and landfill gas were in part used to generate 8 PJ of electricity.On the other hand, the food and water systems used 63 and 4 PJof energy, respectively. Moreover, the energy system withdrew3262 Mt of water from the environment and returned the sameamount of used water to the environment.Compared with other states in the United Staters, water

withdrawal for generating unitary thermoelectric power in DMA[34.8 gallons·(kW·h)−1] was at a relatively higher level (Figure5). In contrast, water withdrawals for unitary N and P in DMAcrops (0.03 and 0.21 gallon/g, respectively) were lower thanthose in most states (Figure 5).

Table 1. Food−Energy−Water Resource Inputs and Outputsof the Detroit Metropolitan Area, 2012

domestic inputs domestic outputs

item quantity item quantity

N from biological fixation 18.3 Kt N in wastewater tonature

10.6 Kt

N in water withdrawals 1.9 Kt N in fertilizer loss 4.4 KtP from biological fixation 0.6 Kt N in wasted crop

straws2.3 Kt

P in water withdrawals 0.4 Kt N in water loss 0.1 Ktwater withdrawals 4675

MtN in wastewater loss 2.0 Kt

energy extraction/capture fromnature

14.5 PJ N in landfilled solidwastes

0.3 Kt

N in landfilled fly ash 53.0 KtP in wastewater tonature

0.5 Kt

P in fertilizer loss 0.6 KtP in wasted cropstraws

0.2 Kt

P in water loss 0.04 KtP in wastewater loss 0.5 KtP in landfilled solidwastes

0.6 Kt

P in landfilled fly ash 9.2 Ktwastewater to nature 4342

Mtwater/wastewater loss 333 Mtheat loss/release 916 PJ

imports exports

item quantity item quantity

N 98.5 Kt N 41.0 KtP 15.4 Kt P 4.6 Ktwater 0 Mt water 0 Mtenergy 1256 PH energy 355 PJ

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b06240Environ. Sci. Technol. 2019, 53, 779−788

782

■ DISCUSSIONThis study developed a standardized and specific accountingmethod to synthetically characterize the inputs, intersectoralflows, stocks, and outputs of FEW resources from theperspective of an integrated urban FEW system. This methodcan help identify key processes for improving FEW resourceefficiencies of urban FEW systems. It also lays the ground forfuture urban FEW analyses and modeling.Policy Implications for the Detroit Metropolitan Area.

We quantified the FEW nexus of the DMA and compared itsFEW resource efficiencies at certain processes with those inother cities reported in literature. Our results help identify keyprocesses for improving FEW resource efficiencies of the DMA.The DMA has a relatively high per capita P intake, indicating

that domestic consumption is a key process for improving P useefficiency. Thus, it is necessary to optimize the dietary habits ofhouseholds in the DMA: for instance, using a certificationscheme (e.g., green labeling of goods and services) to educatehouseholds about P footprints of their purchased products.Moreover, the DMAhas relatively high per capita N emissions

to water bodies. Wastewater treatment, plantation, andwastewater collection are three major sources of N emissionsin the DMA. Potential solutions to reduce N emission levels canbe improving effluent-disposal standards for N removal,promoting adequate fertilization, and enhancing the main-tenance of wastewater collection pipelines.Electricity generation is the largest water user in the DMA.

The DMA has a relatively high level of water withdrawal forunitary thermoelectric power. Thus, improving water use

efficiency of thermoelectric power plants in DMA can helpreduce water withdrawals.Since food, energy, and water systems are interconnected with

one another, changing components of one system may lead toripple effects (desired or undesired) on other systems.81 Forexample, using air-cooled units in place of water-cooled units inpower plants can reduce water use but would increase thedemand for energy materials and subsequent emissions ofcarbon dioxide.82 Thus, the effectiveness of potential solutionsmust be assessed in the context of the urban FEW nexus.Currently, government departments of most countries make

policy decisions individually. The nexus of FEW subsystems isnot fully taken into account. The findings in this study informgovernments to coordinate FEW-related policies of differentdepartments to maximize cobenefits and avoid unintendedconsequences. The method developed in this study can serve asa basic tool for FEW resource management of urbangovernments.

Data Availability, Applicability for Other Regions, andUncertainties.We proposed a method to quantify urban FEWnexus based onMEFA. This method uses publicly available dataincluding government statistics and peer-reviewed estimationmethods. This makes it easy to adopt this method to quantifyurban FEW nexus across multiple years and multiple cities,although this study only used the DMA in 2012 as ademonstration. In particular, this method is more adoptableby cities in the United States, because data sources andparameters for estimations are more relevant to the U.S. context.

Figure 2.Comparisons of per capita N and P intakes of DMAwith those of other cities. Detailed data sources for N and P are listed in Tables S2 and S3,respectively.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b06240Environ. Sci. Technol. 2019, 53, 779−788

783

Some data required for quantifying FEW nexus are not readilyavailable for urban areas. We need to estimate those data bypeer-reviewed estimationmethods, energy andmass balances, orproxies of other cities or national averages. These estimationscan bring uncertainties to the results. Improving urban statisticalsystems in the future can help reduce these uncertainties.

Templates for Urban Food−Energy−Water NexusAccounting. To encourage the adoption of the proposedmethod for urban FEW nexus accounting, we developed a set ofExcel-based templates to streamline the data compilationprocess (Supporting Information). These templates consist ofthree Excel files for N and P flows, energy flows, and water flows,respectively, and one Excel file for final balanced tables andsummarized results. Each of the former three Excel files includesseveral sheets for data inputs, predefined parameters,intermediate calculations, and initial result outputs. The latterExcel file mainly balances initial results on the basis of mass andenergy balances and then summarizes the final results. Users caninput data for a particular city or region, adjust parameters by useof region-specific data if available, and then get the final resultsquantifying the FEW nexus.We present the final results in the format of physical input−

output tables (PIOTs)53 to show the mass and energy flows forN, P, energy, and water. There are four main matrices in thePIOTs. The core matrix shows material and energy flows amongvarious processes within the integrated urban FEW system,including various processes related to FEW and “other sectors”processes. The final demand matrix shows FEW stock changeswithin the system and material and energy flows between thelocal FEW system and the outside (i.e., FEW imports andexports). The From-Nature and To-Nature matrixes describeFEW flows between the local FEW system and the environment.

Figure 3. Comparison of per capita N and P emissions to water bodiesin the DMAwith those of other cities. Detailed data sources are listed inTables S2 and S3, respectively.

Figure 4. FEW nexus of the DMA, 2012.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b06240Environ. Sci. Technol. 2019, 53, 779−788

784

Particularly, the From-Nature matrix shows the biologicalfixation of N/P, N/P embedded in water withdrawal, waterwithdrawal, and energy extraction/capture from the environ-ment. The To-Nature matrix shows N/P emissions, wastewaterdischarge, water loss, and energy loss to the environment. Thereare row and column balances for each process. Total outputsrepresent the sum of each row, indicating the total FEW outflowof each process. Total inputs indicate the sum of each column,representing the total inflow of each process. Total output ofeach process equals its total input, indicating mass or energybalance of each process. The PIOT layout clearly showstransactions and stock changes within the urban FEW system, itsinteractions with the outside through imports/exports, and itsinteractions with the environment. The summary sheet showsgeneral inputs and outputs of individual FEW systems and thewhole integrated FEW system.These templates are developed in the basis of U.S. statistical

systems for metropolitan statistical areas. Thus, application toother U.S. urban systems is relatively simple. In addition, the

templates can also be adapted for urban areas in other countrieswith necessary but minimal modifications to accommodatespecific data sources.

Future Research. The proposed method quantifies theurban FEW nexus, which provides the basis for a variety offurther investigations of this nexus. First, we can evaluate theholistic features of the quantified urban FEW nexus. Thequantified urban FEW nexus can be treated as a network, withprocesses as nodes and mass/energy flows among processes aslinks. We can then apply network analysis tools to evaluate theefficiency and resilience of the urban FEW network.83 We canalso detect the community structure84,85 of this network, whichshows the clustering characteristics of the urban FEW network.The community structure can help policy makers to identifyprocesses that will be strongly influenced by interventions at aspecific process, because processes falling into the same clusterare more closely interconnected with one another than withprocesses outside this cluster.84

Figure 5.Comparison of FEW nexus of the DMAwith those of other states in the United States. Data for water withdrawals and thermoelectric powerare from the U.S. Geological Survey (USGS). N and P in crops are calculated by crop yields multiplied by N and P content parameters. Data for cropyields are from the Census of Agriculture of the USDA.66 N and P content parameters come from literature.60

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b06240Environ. Sci. Technol. 2019, 53, 779−788

785

Second, we can identify critical processes and supply-chainpaths within the urban FEW nexus. Applying input−outputanalysis techniques53 to the quantified urban FEWnexus, we canidentify critical processes from multiple viewpoints (e.g.,primary suppliers,86−88 transmission centers,84,89 final pro-ducers,54,90 and final consumers91−94) and critical supply-chain paths95,96 that strongly influence FEW demands andwastes/emissions of the integrated FEW system. These resultsidentify hotspots for different types of policy decisions (e.g.,supply-side measures,86,87 demand-side measures,90,92,93 andproduction efficiency improvement measures54,89,90).Third, we can evaluate the consequences of specific policy or

technology solutions within the FEW system on the basis of thequantified urban FEWnexus. Since processes are interconnectedwith one another, interventions at a specific process can lead tocobenefits or unintended consequences at other processes.81

The urban FEW nexus accounting quantifies these intercon-nections among processes and describes the equilibrium state ofthe integrated FEW system by using mass and energy balances.When there is an intervention in a certain process, the wholeFEW system will change and reach a new equilibrium state. Wecan quantify the consequences of specific policy or technologysolutions by combining urban FEW nexus accounting withmethods like system dynamics modeling.Finally, current statistics on cities are not detailed enough to

characterize the FEW nexus for specific products. For example,the USGS has only aggregated water use data for irrigation at thecounty level, instead of water use data for each type ofagricultural product.70 Thus, the method used in this study canreveal the FEWnexus at themacro (i.e., regional) andmeso (i.e.,sectoral) levels, but it cannot yet reveal micro-level processes(i.e., product level). Applying this method at the microlevel willrequire more detailed statistical data.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.est.8b06240.

Three tables listing data sources and estimation methodsfor constructing DMA FEW nexus network and datasources for per capita N and P flows (PDF)Three Excel files for N and P flows, energy flows, andwater flows and one Excel file for final balanced tables andsummarized results (ZIP)

■ AUTHOR INFORMATIONCorresponding Author*E-mail: mingxu@umich.edu; phone: +1-734-763-8644; fax:+1-734-936-2195.ORCIDSai Liang: 0000-0002-6306-5800Shelie A. Miller: 0000-0003-0379-3993Ming Xu: 0000-0002-7106-8390NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSWe appreciate the financial support of the National NaturalScience Foundation of China (51661125010, 51721093, and71874014), the U.S. National Science Foundation (1605202),the Fundamental Research Funds for the Central Universities,

and Interdiscipline Research Funds of Beijing NormalUniversity.

■ REFERENCES(1) Bazilian, M.; Rogner, H.; Howells, M.; Hermann, S.; Arent, D.;Gielen, D.; Steduto, P.; Mueller, A.; Komor, P.; Tol, R. S. J.; Yumkella,K. K. Considering the energy, water and food nexus: towards anintegrated modelling approach. Energy Policy 2011, 39 (12), 7896−7906.(2) Logan, B. Urgency at the nexus of food, energy, and water systems.Environ. Sci. Technol. Lett. 2015, 2 (6), 149−150.(3) World development indicators (http://data.worldbank.org/data-catalog/world-development-indicators); The World Bank Group,Washington, DC, 2017.(4) 2014 Revision of World Urbanization Prospects; United Nations,New York, 2014; https://esa.un.org/unpd/wup/publications/files/wup2014-report.pdf.(5) Barles, S. Feeding the city: Food consumption and flow ofnitrogen, Paris, 1801−1914. Sci. Total Environ. 2007, 375 (1−3), 48−58.(6) Billen, G.; Barles, S.; Garnier, J.; Rouillard, J.; Benoit, P. The food-print of Paris: long-term reconstruction of the nitrogen flows importedinto the city from its rural hinterland. Regional Environmental Change2009, 9 (1), 13−24.(7) Forkes, J. Nitrogen balance for the urban food metabolism ofToronto, Canada. Resour. Conserv. Recycl. 2007, 52 (1), 74−94.(8) Ma, L.; Guo, J. H.; Velthof, G. L.; Li, Y. M.; Chen, Q.; Ma, W. Q.;Oenema, O.; Zhang, F. S. Impacts of urban expansion on nitrogen andphosphorus flows in the food system of Beijing from 1978 to 2008.Glob. Environ. Change-Human Policy Dimens. 2014, 28, 192−204.(9)Ma, L.; Velthof, G. L.; Wang, F. H.; Qin, W.; Zhang, W. F.; Liu, Z.;Zhang, Y.; Wei, J.; Lesschen, J. P.; Ma, W. Q.; Oenema, O.; Zhang, F. S.Nitrogen and phosphorus use efficiencies and losses in the food chain inChina at regional scales in 1980 and 2005. Sci. Total Environ. 2012, 434,51−61.(10) Metson, G. S.; Bennett, E. M. Phosphorus cycling in Montreal’sfood and urban agriculture systems. PLoS One 2015, 10 (3),No. e0120726.(11) Neset, T. S. S.; Bader, H. P.; Scheidegger, R. Food consumptionand nutrient flows - nitrogen in Sweden since the 1870s. J. Ind. Ecol.2006, 10 (4), 61−75.(12) Neset, T. S. S.; Bader, H. P.; Scheidegger, R.; Lohm, U. The flowof phosphorus in food production and consumption - Linkoping,Sweden, 1870−2000. Sci. Total Environ. 2008, 396 (2−3), 111−120.(13) Qiao, M.; Zheng, Y. M.; Zhu, Y. G. Material flow analysis ofphosphorus through food consumption in two megacities in northernChina. Chemosphere 2011, 84 (6), 773−778.(14) Wu, H.; Zhang, Y.; Yuan, Z.; Gao, L. Phosphorus flowmanagement of cropping system in Huainan, China, 1990−2012. J.Cleaner Prod. 2016, 112, 39−48.(15) Yuan, Z. W.; Shi, J. K.; Wu, H. J.; Zhang, L.; Bi, J. Understandingthe anthropogenic phosphorus pathway with substance flow analysis atthe city level. J. Environ. Manage. 2011, 92 (8), 2021−2028.(16) Gierlinger, S. Food and feed supply and waste disposal in theindustrialising city of Vienna (1830−1913): a special focus on urbannitrogen flows. Regional Environmental Change 2015, 15 (2), 317−327.(17) Zhang, Y.; Lu, H.; Fath, B. D.; Zheng, H.; Sun, X.; Li, Y. Anetwork flow analysis of the nitrogen metabolism in Beijing, China.Environ. Sci. Technol. 2016, 50 (16), 8558−8567.(18) Chen, S.; Chen, B. Network environ perspective for urbanmetabolism and carbon emissions: a case study of Vienna, Austria.Environ. Sci. Technol. 2012, 46 (8), 4498−4506.(19) Chen, S.; Chen, B.; Su, M. Nonzero-sum relationships inmitigating urban carbon emissions: a dynamic network simulation.Environ. Sci. Technol. 2015, 49 (19), 11594−11603.(20) Zhang, L.; Hu, Q.; Zhang, F. Input-output modeling for urbanenergy consumption in Beijing: dynamics and comparison. PLoS One2014, 9 (3), No. e89850.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b06240Environ. Sci. Technol. 2019, 53, 779−788

786

(21) Zhang, Y.; Li, S.; Fath, B. D.; Yang, Z.; Yang, N. Analysis of anurban energy metabolic system: comparison of simple and complexmodel results. Ecol. Modell. 2011, 223 (1), 14−19.(22) Zhang, Y.; Yang, Z.; Fath, B. D.; Li, S. Ecological network analysisof an urban energy metabolic system: Model development, and a casestudy of four Chinese cities. Ecol. Modell. 2010, 221 (16), 1865−1879.(23) Liang, S.;Wang, C.; Zhang, T. An improved input−output modelfor energy analysis: a case study of Suzhou. Ecological Economics 2010,69 (9), 1805−1813.(24) Liang, S.; Zhang, T. Managing urban energy system: a case ofSuzhou in China. Energy Policy 2011, 39 (5), 2910−2918.(25) Chen, S.; Chen, B. Coupling of carbon and energy flows in cities:A meta-analysis and nexus modelling. Appl. Energy 2017, 194, 774−783.(26) Zhang, Y.; Yang, Z.; Fath, B. D. Ecological network analysis of anurban water metabolic system:model development, and a case study forBeijing. Sci. Total Environ. 2010, 408 (20), 4702−4711.(27) Bai, H.; Zeng, S. Y.; Dong, X.; Chen, J. N. Substance flow analysisfor an urban drainage system of a representative hypothetical city inChina. Front. Environ. Sci. Eng. 2013, 7 (5), 746−755.(28) Chevre, N.; Guignard, C.; Rossi, L.; Pfeifer, H. R.; Bader, H. P.;Scheidegger, R. Substance flow analysis as a tool for urban watermanagement. Water Sci. Technol. 2011, 63 (7), 1341−1348.(29) Ahjum, F.; Stewart, T. J. A systems approach to urban waterservices in the context of integrated energy andwater planning: a City ofCape Town case study. J. Energy South. Afr. 2014, 25 (4), 59−70 (http://www.scielo.org.za/scielo.php?script=sci_arttext&pid=S1021-447X2014000400006) .(30) Carmin, J.; Carrington, T.; Ebinger, J.; Evans, B.; Gerner, F.;Hamso, B.; Lim, A.; Nakat, Z.; Plecas, A.;Webster, M.; Zhang, Y. F. Thebuilt environment: cities, water systems, energy, and transport.Adapting to Climate Change in Eastern Europe and Central Asia 2010,121−137.(31) Cohen, E.; Ramaswami, A. The water withdrawal footprint ofenergy supply to cities conceptual development and application toDenver, Colorado, USA. J. Ind. Ecol. 2014, 18 (1), 26−39.(32) Kenway, S.; McMahon, J.; Elmer, V.; Conrad, S.; Rosenblum, J.Managing water-related energy in future cities - a research and policyroadmap. J. Water Clim. Change 2013, 4 (3), 161−175.(33) Kenway, S. J.; Lant, P.; Priestley, T. Quantifying water-energylinks and related carbon emissions in cities. J. Water Clim. Change 2011,2 (4), 247−259.(34) Kenway, S. J.; Lant, P. A.; Priestley, A.; Daniels, P. Theconnection between water and energy in cities: a review. Water Sci.Technol. 2011, 63 (9), 1983−1990.(35) Miller, L. A.; Ramaswami, A.; Ranjan, R. Contribution of waterand wastewater infrastructures to urban energy metabolism andgreenhouse gas emissions in cities in India. J. Environ. Eng. 2013, 139(5), 738−745.(36) Nair, S.; George, B.; Malano, H. M.; Arora, M.; Nawarathna, B.Water−energy−greenhouse gas nexus of urban water systems: reviewof concepts, state-of-art and methods. Resour. Conserv. Recycl. 2014, 89,1−10.(37) Novotny, V. Water and energy link in the cities of the future -achieving net zero carbon and pollution emissions footprint.Water Sci.Technol. 2011, 63 (1), 184−190.(38) Ruddell, D. M.; Dixon, P. G. The energy-water nexus: are theretradeoffs between residential energy and water consumption in aridcities? Int. J. Biometeorol. 2014, 58 (7), 1421−1431.(39) Salazar, R.; Rojano, A.; Lopez, I. Energy and environmental costsrelated to water supply in Mexico City. Water Sci. Technol.: WaterSupply 2012, 12 (6), 768−772.(40) Venkatesh, G.; Chan, A.; Brattebø, H. Understanding the water-energy-carbon nexus in urban water utilities: comparison of four citycase studies and the relevant influencing factors. Energy 2014, 75, 153−166.(41) Chen, S.; Chen, B. Urban energy−water nexus: A networkperspective. Appl. Energy 2016, 184, 905−914.

(42) Barthel, S.; Isendahl, C. Urban gardens, agriculture, and watermanagement: sources of resilience for long-term food security in cities.Ecological Economics 2013, 86, 224−234.(43) Siegfried, T.; Sobolowski, S.; Raj, P.; Fishman, R.; Vasquez, V.;Narula, K.; Lall, U.; Modi, V. Modeling irrigated area to increase water,energy, and food security in semiarid India. Weather, Climate, andSociety 2010, 2 (4), 255−270.(44) Swaney, D. P.; Santoro, R. L.; Howarth, R. W.; Hong, B.;Donaghy, K. P. Historical changes in the food and water supply systemsof the New York City Metropolitan Area. Regional EnvironmentalChange 2012, 12 (2), 363−380.(45) VillarroelWalker, R.; Beck,M.; Hall, J. Water and nutrient andenergy systems in urbanizing watersheds. Front. Environ. Sci. Eng.2012, 6 (5), 596−611.(46) Villarroel Walker, R.; Beck, M. B.; Hall, J. W.; Dawson, R. J.;Heidrich, O. The energy-water-food nexus: Strategic analysis oftechnologies for transforming the urban metabolism. J. Environ.Manage. 2014, 141, 104−115.(47)Water Security: The Water−Food−Energy−Climate Nexus. WorldEconomic Forum: Washington, DC, 2011; DOI: 10.5822/978-1-61091-026-2.(48) Das, T.; Cabezas, H. Tools and concepts for environmentalsustainability in the food-energy-water nexus: Chemical engineeringperspective. Environ. Prog. Sustainable Energy 2018, 37 (1), 73−81.(49) Sherwood, J.; Clabeaux, R.; Carbajales-Dale, M. An extendedenvironmental input−output lifecycle assessment model to study theurban food−energy−water nexus. Environ. Res. Lett. 2017, 12 (10),105003.(50) Brunner, P. H. Substance flow analysis: a key tool for effectiveresource management. J. Ind. Ecol. 2012, 16 (3), 293−295.(51) Brunner, P. H.; Ma, H.-W. Substance flow analysis: anindispensable tool for goal-oriented waste management. J. Ind. Ecol.2009, 13 (1), 11−14.(52) van der Voet, E. Substance flow analysis methodology. In AHandbook of Industrial Ecology, Ayres, R. U., Ayres, L. W., Eds. EdwardElgar: Cheltenham, UK, and Northampton, MA, 2002.(53) Miller, R. E.; Blair, P. D. Input−Output Analysis: Foundations andExtensions; Cambridge University Press: New York, 2009;DOI: 10.1017/CBO9780511626982.(54) Liang, S.; Wang, Y.; Zhang, C.; Xu, M.; Yang, Z.; Liu, W.; Liu, H.;Chiu, A. S. F. Final production-based emissions of regions in China.Economic Systems Research 2018, 30 (1), 18−36.(55) Liang, S.; Zhang, C.; Wang, Y.; Xu, M.; Liu, W. Virtualatmospheric mercury emission network in China. Environ. Sci. Technol.2014, 48 (5), 2807−2815.(56) Demographia Definition of urban terms; http://demographia.com/db-define.pdf (Accessed October 15, 2015).(57) Interactive Data of Bureau of Economic Analysis, http://www.bea.gov/itable; Bureau of Economic Analysis, U.S. Department ofCommerce, 2012.(58) 2012 Census of Agriculture, Michigan State and County Data;Department of Agriculture and National Agricultural Statistics Service,2014; Vol. 1; https://www.nass.usda.gov/Publications/AgCensus/2012/.(59) Loss-Adjusted Food Availability. In Economics Research Service,U.S. Department of Agriculture, 2015; https://www.ers.usda.gov/data-products/food-availability-per-capita-data-system/loss-adjusted-food-availability-documentation/.(60) Warncke, D.; Dahl, J.; Jacobs, L. Nutrient Recommendations forField Crops in Michigan; Department of Crop and Soil Sciences,Michigan State University Extension, 2009; p 8; https://soil.msu.edu/wp-content/uploads/2014/06/MSU-Nutrient-recomdns-field-crops-E-2904.pdf.(61) Analytical Methods Committee. Meat and poultry nitrogenfactors. Anal. Methods 2014, 6 (13), 4493−4495.(62) USDA Food Composition Databases. Agricultural ResearchService, U.S. Department of Agriculture, 2015; https://ndb.nal.usda.gov/ndb/.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b06240Environ. Sci. Technol. 2019, 53, 779−788

787

(63) Geographic Area Series: Shipment Characteristics by OriginGeography by Commodity: 2012 and 2007 Commodity Flow Survey.U.S. Census Bureau, 2014; https://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?src=bkmk.(64) Lal, R.World crop residues production and implications of its useas a biofuel. Environ. Int. 2005, 31 (4), 575−584.(65) U.S. Fertilizer Consumption and Use - By Year 1960−2011, 7/12/2013 ed.; Economics Research Service, U.S. Department of Agriculture,2013.(66) 2012 Census of Agriculture, United States Summary and State Data;National Agricultural Statistics Service, U.S. Department of Agriculture,2014; Vol. 1, p 50; https://www.nass.usda.gov/Publications/AgCensus/2012/.(67) Kucharik, C. J.; Brye, K. R. Integrated BIosphere Simulator(IBIS) yield and nitrate loss predictions for Wisconsin maize receivingvaried amounts of nitrogen fertilizer. J. Environ. Qual. 2003, 32 (1),247−68.(68) Tabbara, H. Phosphorus loss to runoff water twenty-four hoursafter application of liquid swine manure or fertilizer. J. Environ. Qual.2003, 32 (3), 1044−52.(69) Salvagiotti, F.; Cassman, K. G.; Specht, J. E.; Walters, D. T.;Weiss, A.; Dobermann, A. Nitrogen uptake, fixation and response tofertilizer N in soybeans: A review. Field Crops Research 2008, 108 (1),1−13.(70) USGS Water Use Data for Michigan; U.S. Department of theInterior, U.S. Geological Survey, 2010; https://waterdata.usgs.gov/mi/nwis/wu.(71) DWSD Summary of Operating Statistics, Fiscal Year 2011−2012;Detroit Water and Sewerage Department, 2013; p 77.(72) Water Audits and Water Loss Control for Public Water Systems;EPA 816-F-13-002; Office of Water, U.S. Environmental ProtectionAgency: 2013; https://www.epa.gov/sites/production/files/2015-04/documents/epa816f13002.pdf.(73) County-level Oil and Gas Production in the U.S.; EconomicsResearch Service, U.S. Department of Agriculture, 2000−2011;https://www.ers.usda.gov/data-products/county-level-oil-and-gas-production-in-the-us/.(74) Electric generator capacity data; U.S. Energy InformationAdministration: 1990−2015.(75) Electric power data; U.S. Energy Information Administration:2011−2016.(76) Marathon Detroit Refinery Fact Sheet; https://www.marathonpetroleum.com/Operations/Refining_and_Marketing/Refining/Detroit_Refinery/; Marathon Petroleum Corporation, 2016.(77) Camargo, G. G. T.; Ryan, M. R.; Richard, T. L. Energy Use andGreenhouse Gas Emissions from Crop Production Using the FarmEnergy Analysis Tool. BioScience 2013, 63 (4), 263−273.(78) Karimi, M.; Moghaddam, H. On-farm energy flow in grapeorchards. J. Saudi Soc. Agric. Sci. 2018, 17, 191−194.(79) Pabi, S.; Amarnath, A.; Goldstein, R.; Reekie, L. Electricity Useand Management in the Municipal Water Supply and WastewaterIndustries; Water Research Foundation, Electric Power ResearchInstitute, 2013; http://www.waterrf.org/Pages/Projects.aspx?PID=4454.(80) Residential Energy Consumption Survey (RECS); U.S. EnergyInformation Administration, 2010; https://www.eia.gov/consumption/residential.(81) Liang, S.; Xu, M.; Suh, S.; Tan, R. R. Unintended environmentalconsequences and co-benefits of economic restructuring. Environ. Sci.Technol. 2013, 47 (22), 12894−12902.(82) Zhang, C.; Anadon, L. D.; Mo, H.; Zhao, Z.; Liu, Z. Water−carbon trade-off in China’s coal power industry. Environ. Sci. Technol.2014, 48 (19), 11082−11089.(83)Goerner, S. J.; Lietaer, B.; Ulanowicz, R. E. Quantifying economicsustainability: Implications for free-enterprise theory, policy andpractice. Ecological Economics 2009, 69, 76−81.(84) Liang, S.; Feng, Y.; Xu, M. Structure of the global virtual carbonnetwork: Revealing important sectors and communities for emissionreduction. J. Ind. Ecol. 2015, 19 (2), 307−320.

(85) Newman, M. E. J. Modularity and community structure innetworks. Proc. Natl. Acad. Sci. U. S. A. 2006, 103 (23), 8577−8582.(86) Liang, S.; Qu, S.; Zhu, Z.; Guan, D.; Xu, M. Income-basedgreenhouse gas emissions of nations. Environ. Sci. Technol. 2017, 51 (1),346−355.(87) Liang, S.; Wang, H.; Qu, S.; Feng, T.; Guan, D.; Fang, H.; Xu, M.Socioeconomic drivers of greenhouse gas emissions in the UnitedStates. Environ. Sci. Technol. 2016, 50 (14), 7535−7545.(88) Marques, A.; Rodrigues, J.; Lenzen, M.; Domingos, T. Income-based environmental responsibility. Ecological Economics 2012, 84, 57−65.(89) Liang, S.; Qu, S.; Xu, M. Betweenness-based method to identifycritical transmission sectors for supply chain environmental pressuremitigation. Environ. Sci. Technol. 2016, 50 (3), 1330−1337.(90) Liang, S.; Wang, Y.; Cinnirella, S.; Pirrone, N. Atmosphericmercury footprints of nations. Environ. Sci. Technol. 2015, 49 (6),3566−3574.(91) Hui, M.; Wu, Q.; Wang, S.; Liang, S.; Zhang, L.; Wang, F.;Lenzen, M.; Wang, Y.; Xu, L.; Lin, Z.; Yang, H.; Lin, Y.; Larssen, T.; Xu,M.; Hao, J. Mercury flows in China and global drivers. Environ. Sci.Technol. 2017, 51 (1), 222−231.(92) Liang, S.; Guo, S.; Newell, J. P.; Qu, S.; Feng, Y.; Chiu, A. S. F.;Xu, M. Global drivers of Russian timber harvest. J. Ind. Ecol. 2016, 20(3), 515−525.(93) Liang, S.; Stylianou, K.; Jolliet, O.; Supekar, S.; Qu, S.; Skerlos, S.J.; Xu, M. Consumption-based human health impacts of primaryPM2.5: The hidden burden of international trade. J. Cleaner Prod. 2017,167, 133−139.(94) Chen, L.; Meng, J.; Liang, S.; Zhang, H.; Zhang, W.; Liu, M.;Tong, Y.; Wang, H.; Wang, W.; Wang, X.; Shu, J. Trade-InducedAtmospheric Mercury Deposition over China and Implications forDemand-Side Controls. Environ. Sci. Technol. 2018, 52 (4), 2036−2045.(95) Lenzen, M. Structural path analysis of ecosystem networks. Ecol.Modell. 2007, 200 (3−4), 334−342.(96) Liang, S.; Wang, Y.; Zhang, T.; Yang, Z. Structural analysis ofmaterial flows in China based on physical and monetary input-outputmodels. J. Cleaner Prod. 2017, 158, 209−217.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b06240Environ. Sci. Technol. 2019, 53, 779−788

788