Downloaded by guest on September 26, 2021 eltd ihosre rudae-ee elnso . to 0.9 of rapidly declines being groundwater-level is observed groundwater with country, Syria depleted, upstream the by primary Throughout captured Jordan’s largely (3–6). currently also is tributary, source, River surface-water Israel Yarmouk by of Carrier diversion The Water National the (2). the into to since River 90% Jordan due upper nearly mostly the by Jordan (1), declined the conditions have lower and to predevelopment Israel the estimated with are in border Bank, western Flows West Jordan’s transboundary 1A). marks on which (Fig. River, dependence groundwater and and supply rivers new alternatives few generating with for availability water natural limited extremely J modeling hydroeconomic security water and supply- of coordination the measures. through demand-side security improve- water exponential in revealing ments results water model comprehensive with interventions reform, and ambitious sector desalinization of large-scale portfolio prevalent include sweeping Jor- that future, a are freshwater enact its welfare must on dan foothold economic short- a gain lengthening declining To narratives. use, across and freshwater durations, in receiving age century, disparity the Widening of inse- day. end critical the experiences by population curity measures, household intervening low-income Without by century. the 50% the approximately of by end decreases the security. availability freshwater water indicate short- capita in results Per declines equity, Model destabilizing, (vulnerability, well-being). potentially economic severe, metrics and water-security duration, into evaluation age of translated the rep- suite outcomes of distributional for a and thousands allowing systemwide of both sectors, level of across the enables at agents approach engineered impacts resentative and modeling of natural quantification multiagent the the A and environment. actors of water hierarchy interactions dynamic a representing trajec- between system, socioeco- the water simulates Jordan’s and model of coupled climate systems tory complex nationwide under The eval- changes. security to a nomic freshwater used Model, Jordan’s is Water that uate model Jordan systems the human–natural-engineered economics, and present geography, clima- we science, hydrology, political pop- across Jordan’s agriculture, knowledge conflict-induced tology, countries. Integrating by many shocks. compounded in ulation emblematic, crises is freshwater situation in resulted change climate have and growth, population availability, water Limited 2020) 1, October Rinaldo, Andrea by Edited and 78412; TX Christi, Corpus University, A&M Texas Engineering, nvriyo aionaMre,Mre,C 95343; CA Merced, Merced, California of University Kingdom; United 2LS, Jordan; WC2R 22110, London London, College King’s Geography, Kingdom; United 9PL, M13 Manchester Germany; Leipzig, 04318 Research–UFZ, Environmental a Tilmant Amaury ou Medell Josue Yoon Jim change population and climate under security freshwater of analysis system human–natural coupled A PNAS nryadEvrnetDrcoae aicNrhetNtoa aoaoy ihad A99352; WA Richland, Laboratory, National Northwest Pacific Directorate, Environment and Energy e uue h onr’ ae hlegsse rmhaving from freshwa- stem tenuous challenges water a country’s has The nation, future. land-locked ter nearly a ordan, 01Vl 1 o 4e2020431118 14 No. 118 Vol. 2021 a,1 hita Klassert Christian , | g aut fEooisadBsns aaeet epi nvriy 40 epi,Germany; Leipzig, 04109 University, Leipzig Management, Business and Economics of Faculty utaetmodel multiagent ´ ın-Azuara d en Klauer Bernd , cl oyehiu F Polytechnique Ecole ´ h uhaBataineh Bushra , | Jordan b d D hlpSelby Philip , preetd G de epartement ´ b | utsco dynamics multisector ans Mustafa Daanish , ed ´ 0Lprcpt per capita per L <40 rl eLuan,Luan,Sizrad n prvdFbur 9 01(eevdfrreview for (received 2021 19, February approved and Switzerland, Lausanne, Lausanne, de erale ´ i ae nrsrcueDvlpet ehe oprto,Rso,V 20190; VA Reston, Corporation, Bechtel Development, Infrastructure Water c i eateto ehncl eopc,adCvlEgneig nvriyo Manchester, of University Engineering, Civil and Aerospace, Mechanical, of Department c u Zhang Hua , neCvle eG de et Civil enie ´ hbu Lachaut Thibaut , 0 of >90% k eateto at ytmSine tnodUiest,Safr,C 94305 CA Stanford, University, Stanford Science, System Earth of Department | e f aut fCvlEgneig odnUiest fSineadTcnlg,Irbid Technology, and Science of University Jordan Engineering, Civil of Faculty aj Sigel Katja , j nedsEu,Universit Eaux, des enie ´ n tvnM Gorelick M. Steven and , ilo,goigt vr1. ilo y22,apro hnat when period a 2020, 7.2 by was million population 10.8 Jordan’s over 2010, to In growing large 14). million, sudden, (10, by influxes punctuated Jor- been refugee (13). has 2100 population by the growing droughts in dan’s of doubling intensity with and is duration, temperatures frequency, increased Jordan mod- climate further in while predict 12), (11, decline els century past Rainfall the over (10). evident already situation tenuous Jordan’s stood have cooperation 1960s, the international way. in the fragile of in saline conceived and the first costs of Although dispose Sea. project and Dead trans- the Amman, water, to to brine Sea north Red freshwater desalinate the con- Sea for port would Sea–Dead which Arabia Red the project, Saudi of veyance construction with Jor- sought (9). long competes aquifer has regional Jordan dan Disi fossil south, the from the groundwater shared aquifer To productive highly 8). most country’s (7, the in 1995 since m/y 3.5 doi:10.1073/pnas.2020431118/-/DCSupplemental at online information supporting contains article This 1 ulse ac 9 2021. 29, March Published BY-NC-ND) (CC 4.0 NoDerivatives License distributed under is article access open This Submission.y Direct PNAS a is article This paper.y interest.y the competing wrote no S.M.G. and declare and S.M.G. data; authors S.T., analyzed The and D.M., H.Z. B.K., H.Z., A.T., and K.S., J.H., B.B., S.K., N.A., K.S., T.L., P.S., N.A., S.K., C.K., S.K., T.L., J.Y., T.L., P.S., P.S., C.K., C.K., J.Y., J.Y., research; J.M.-A., research; E.G., performed S.T., designed D.M., S.M.G. B.K., A.T., and J.H., N.A., B.B., S.K., T.L., P.S., C.K., J.Y., contributions: Author owo orsodnemyb drse.Eal [email protected] Email: addressed. be may correspondence whom To h n ftecnuy ogi otodo t water includ- its measures, on reform. water-sector demand-side comprehensive foothold and and desalinization large-scale supply- a ing interven- span of gain portfolio that To ambitious tions an century. enact by must the insecurity Jordan future, of water critical end experiencing low-income the Jordan’s be of will 90% population With- over security. measures, potentially freshwater intervening severe, Jordan’s out to in points declines analysis destabilizing, Our decisions. use allocation water and determining agents thousands human with representative of environment of simulation water combines built and that natural model Jordan’s We systems integrated influxes. enabled by an country, the refugee by exacerbated for analysis security conflict-induced crisis, freshwater a water and present unfolding change an climate facing is Jordan Significance d lmt hneadpplto rwhfrhrthreaten further growth population and change Climate tpe Knox Stephen , b ae Talozi Samer , b aa,Qu Laval, e ´ eateto cnmc,HlhlzCnr for Centre Helmholtz Economics, of Department c ioa Avisse Nicolas , https://doi.org/10.1073/pnas.2020431118 f h k ii n niomna Engineering, Environmental and Civil bc CGV06 Canada; 0A6, G1V QC ebec, ´ rkGawel Erik , . y raieCmosAttribution-NonCommercial- Commons Creative . y d https://www.pnas.org/lookup/suppl/ b,g uinHarou Julien , j olg fSineand Science of College , e eatetof Department c | , f12 of 1

ENVIRONMENTAL SUSTAINABILITY SCIENCES SCIENCE Fig. 1. Map of Jordan and conceptual model. (A) Jordan relies on surface-water sources (primarily the Yarmouk River) and groundwater wells for water supply. The most recent record of total freshwater use from Jordan’s MWI is 1,054 MCM for 2017, with groundwater contributing 59% of the total supply, surface water 27%, and treated wastewater 14%. The domestic sector uses 45% of all water, agriculture 52%, and industry 3%. Groundwater is the primary source for the domestic sector, constituting over 70% of its supply. For the agricultural sector, groundwater constitutes 46% of the supply for irrigation, surface water 28%, and treated wastewater 26%. (B) The JWM consists of two types of modules: human modules (white rectangles) and biophysical modules (gray rectangles) of natural and engineered physical phenomena. The systems model includes interactions between modules, distinguished by endogenous human decisions (blue lines), endogenous physical flows and production (green lines), and exogenous scenarios (pink lines) and human interventions (yellow lines).

least 1.1 million Syrian refugees fled Syria’s 2011 war to Jordan and socioeconomic facets of such a challenge, there has been (2, 15). In response to water shortage, Jordan has implemented a growing call for analytic frameworks to evaluate freshwater significant water-supply efficiencies. In Amman, the largest city systems that account for both the physical processes that gov- and capital, over 95% of wastewater is treated and recycled. ern freshwater supply and the human institutions and behaviors However, Jordan’s water-distribution system is inefficient and that influence the management, allocation, and consumption of intermittent. Approximately 50% of Jordan’s piped supply is lost water (30–36). Here, we present such a coupled human–natural- as “nonrevenue water” (NRW), due to either physical factors engineered system framework to explore the long-term impacts (e.g., pipeline leaks) or administrative issues (e.g., water theft, of a suite of policy interventions aimed at achieving freshwater incorrect meter readings, or underbilling). On average, house- security in Jordan in the face of anticipated changes in climate, holds in the capital of Amman receive piped water for only 36 population, and the economy. h per week (16), with lower-income neighborhoods receiving as low as 24 h of municipal supply, while higher-income house- Coupled Human–Natural Systems Modeling Approach holds receive up to 5 d of uninterrupted supply per week (17). We introduce the Jordan Water Model (JWM), a nation- As a result, urban users purchase expensive water delivered by wide, modular, multiagent water policy-evaluation model (Fig. tanker trucks that obtain water from private agricultural wells 1B). This tool integrates biophysical modules that simulate through both formal and informal tanker-water markets (17, 18). natural and engineered phenomena with human modules Ecological impacts related to both groundwater and surface- that represent behavior at multiple levels of decision-making. water withdrawals by Jordan and upstream riparian nations in The hydrologic and agricultural modules are developed by the Jordan River Basin have been severe, with notable exam- using spatially distributed, process-based, groundwater, surface- ples including the drying of the Azraq Oasis, a Ramsar wetland water, piped-network conveyance, and crop water-use mod- (19, 20), and the shrinking of the Dead Sea, whose shoreline is els, capturing water-resource depletion and impacts of climate receding by ∼1 m/y (21). change. The human modules comprise agents, here defined The situation in Jordan is emblematic of water crises around as autonomous entities that make decisions in relation to one the world, in which rapid population growth, intensifying water another and in response to hydrologic and socioeconomic con- use, sudden demographic shocks, climate change, transbound- ditions in the system. The model has 1,923 agents representing ary water competition, and institutional challenges pose serious water managers, suppliers, and water-user groups across sec- threats to freshwater security (22–28). In the face of such global tors. A simplified conceptual diagram of the JWM is shown changes, an overarching sustainability goal is the long-term pro- in Fig. 1B. vision of freshwater as formalized in the 2015 Sustainable Devel- The biophysical and human modules coevolve in dynamic fash- opment Goals (29). Given the complex and interacting physical ion, with endogenized human decisions impacting the state of

2 of 12 | PNAS Yoon et al. https://doi.org/10.1073/pnas.2020431118 A coupled human–natural system analysis of freshwater security under climate and population change Downloaded by guest on September 26, 2021 Downloaded by guest on September 26, 2021 nya 00U olr(S)vle.Scocnmcgrowth Socioeconomic values. presented in (USD) and further and dollar described inflation are projections US for assumptions 2010 scenario adjusted year the are in con- in results historical model values to which future relative monetary increase in All prices a one ditions. crop and SSPs, conditions, which historical in and to second relative domestic RCPs steady remain also gross the prices are crop of SSP scenarios independently crop-price the Two defined to com- projections. and related (GDP) incomes product is household establishments of socioeco- for growth mercial For (SSPs) the (40). pathways conditions, projections nomic socioeconomic socioeconomic shared and and projec- rep- population 39) climate Change’s (38, adapted for Climate (RCPs) tions are on pathways sets concentration Panel These resentative Intergovernmental conditions. the crisis from and pop- growth, climate, drier ulation optimistic, and representing sets, scenario assumption four of combination the unique in with a or “narratives,” as assumptions. on intervention now defined based managers, narrative of results model a or purview compare the makers We within future. policy are water that Jordanian infrastructure etc.). of influxes, and policies development change, government refugee climate of implementations change, (e.g., are manager “Interventions” population or maker prices, policy commodity water influ- a of of scope ence the and outside physical considered those scenarios conditions as categories: socioeconomic defined via two are into represented “Scenarios” interventions. fall quantitatively and that plau- are inputs represent model and narratives exogenous The conditions century. future change the system sible of of end narratives future the simulate through to used is JWM future The simulate to Narratives Future 2100 to increments. time 2016 monthly on from proceeding conditions, his- and represent Validation) to in 2014 included Model to are results 2006 (validation from conditions run torical is of model component The specific Methods). ( a system represents water module architecture Jordan’s each software which object-oriented structure in an modular (37), in a implemented adopts is model country’s and integrated entire The an sector. across and interactions water physical their significant and all features capture human secu- to freshwater attempting nationwide rigorously of rity, evaluation for model multiagent agent–environment and model, agent–agent integrated of the interactions. multitude throughout a subsequent prevalent including for coupled are distributed, state system spatially interactions hydrologic multiscale, groundwater These updated the decisions. agent the in establishes responses spatially and head simulates hydraulic which unified a explicit module, into feed groundwater and decisions conditions, physics-based pumping socioeconomic These extraction, objectives. ground- pumping of human local their cost availability, basing levels, surface-water water each changing wells, on local decisions agents at farmer individual pumping pump- time while determine wellfields, groundwater model municipal given coordinates large a for agent ing for institutional subprocess an illustrative period, an and In space across time. through environment to the over environment to modulated back from conditions, agents rippling from are and future impacts agents that with changing system, perturbations complex represent the external to as used acting are prices) commod- or decision-making precipitation resolved ity (e.g., hierarchical further factors driving and are Exogenous pro- scales levels. interactions model spatial dynamic multiple system the Such across natural as time. the in decisions of ceeds state human altered influencing the subsequently and system, natural the ope ua–aua ytmaayi ffehae euiyudrclimate under security change freshwater population of and analysis system human–natural coupled A al. et Yoon The integrated an into components these combines JWM The cnrocomponent scenario aeil n Methods and Materials ftenraie sdsigihdby distinguished is narratives the of aeil n Methods and Materials and IAppendix, SI IAppendix, SI and • • • described is run test • of system category complex each below. for into method and implementation and insight definition of reasonability and The interdependencies. the behavior and evaluating interactions model runs of of test means of stability a suite provides This outcomes. further water-security signif- sce- that impact parameters sensitivity, and icantly stress, inputs, conducted features, of is structural identify tests set to model a counterfactual narratives, and modeled nario/intervention, the Tests. to Counterfactual tion and Intervention, Sensitivity, Stress, 2. Table in further described The are population. metrics household security Jordan’s freshwater of use attributable water sec- changing well-being the economic domestic to and vulnerability, the quantifying duration, metrics shortage on water-security equity, four focusing of water-resources set and, a capita For tor, sectors per industry). across estimate or we availability categories dwellers, impacts, different urban system to farmers, human available (e.g., water consumers as of well groundwa- and as surface-water resources, system available ter country’s human the in and the changes in natural narratives. represented future scenarios consider and 20 interventions we policy of Jordan, impacts in security Metrics. Security Freshwater Jordanian a of controllable perspective maker. both policy the involving water from conditions factors future uncontrollable of and range cover broad Narratives trajectory. a sto- system plausible freshwater represent Jordan’s that of narratives rylines 20 generate we 1), (Table system the the of as which conditions 2019). infrastructure in year portfolios (current portfolio quo intervention status baseline at four a remains The against compared implementation interventions. are for water various timeline in Jordan’s the a stakeholders of on including with sector, reports discussions water and of Jordan’s plans variety supply were and a interventions resource Potential upon theft. based enforcement water identified and eliminating reallocation, reduc- to water at distribution, sectoral adjustments aimed supply leaks, desali- pipe piped desalination), major in of Sea tion as equalization Red prices, such (e.g., water infrastructure, projects modified nation include or They new security. that freshwater the major Jordan’s strategies have improve that of to levers potential represent range Interventions reallocation. a sectoral water and management, characterizing demand enhancement, supply portfolios include four in by river case the transboundary been impacts (4). has Jordan into significantly as flows Yarmouk water which low, the remains years, of or recent portion recovers Syrian basin the River in production agricultural Appendix, SI ie hticuewtrdlvr n rnfraons h state the amounts, transfer and delivery water include that tified eue gn ouainfo h model). the the from removing population (e.g., agent situation refugee hypothetical a represent modifi- to model cation structural a is Implementing that – assumptions. test intervention project) Counterfactual or desalinization scenario Sea inclu- other from Sea–Dead (e.g., unpackaged Red intervention the or of 8.5) sion sce- RCP individual (e.g., an assumption nario Implementing groundwater – (e.g., test Scenario/intervention percentage a endogenous drawdown). by or inputs/outputs parameters model exogenous module Adjusting an – to test change Sensitivity population). percentage (e.g., a input model Applying – test Stress obnn orseai esadfieitreto portfolios intervention five and sets scenario four Combining distinguished is narratives the of component intervention The ocmaeipcsars h et,asto niaosi iden- is indicators of set a tests, the across impacts compare To h cnro lodsigihwhether distinguish also scenarios The Methods. aua ytmimpacts system Natural ocaatrz uuefreshwater future characterize To https://doi.org/10.1073/pnas.2020431118 ou ntemporal on focus PNAS naddi- In | f12 of 3

ENVIRONMENTAL SUSTAINABILITY SCIENCES SCIENCE Table 1. Twenty future model narratives consisting of four ulated northern governorates, containing the largest cities of scenario sets and five intervention portfolios Amman, Zarqa, and Irbid, as well as some regions in the south. Description In over 20% of the region where groundwater is exploited, there is groundwater decline of more than 50% of saturated aquifer Scenario sets thickness by end of the century relative to year 2016. Declin- 1) Optimistic Moderate climate change (RCP 4.5), ing levels result in reduced groundwater-production capacity, higher transboundary flows with a loss in production capacity of 78 million cubic meters (agriculture in the Syrian Yarmouk (MCM)/year by the end of the century compared to conditions in remains low), moderate socioeconomic 2016 (a 12% reduction in groundwater production capacity over growth (SSP 2), crop prices steady this time period). 2) Drier climate Adverse climate change (RCP 8.5), lower For transboundary surface water flows into Jordan from Syria, transboundary flows (agriculture the country’s most significant source of surface water in recent in the Syrian Yarmouk recovers), SSP 2, history, the scenarios cover two trajectories of climate change crop prices steady and two trajectories of Syria’s utilization of the Yarmouk River 3) Population growth RCP 4.5, higher transboundary flows, (13). Surface-water availability decreases across all four combi- aggressive population growth (SSP 3) nations of these scenario conditions, with the most substantial 4) Crisis RCP 8.5, lower transboundary flows, flow decrease due to adverse climate-change impacts (RCP 8.5) SSP 3, crop prices increase, refugee and intensive irrigated agriculture upstream of Jordan in the Syr- wave in 2030 ian Yarmouk basin. Under the most optimistic conditions (RCP 4.5 with low water use in Syria), average total surface-water Intervention portfolios inflow during the last two decades of the century decreases by a) Baseline No additional interventions 25% compared to the 2016–2020 period (from 232 MCM to 173 b) Supply enhancement Red Sea–Dead Sea desalinization phase 1 MCM). (80 MCM/year), Red Sea–Dead The combined effects of growing population and declining Sea desalinization phase 2 (150 MCM/ water availability result in marked declines in per capita water year), all other planned water supply use on a sectoral basis. In the optimistic-baseline narrative, projects (132 MCM/year), physical NRW annual per capita water use across all sectors decreases by over reduction (50% of current losses) 50% from nearly 100 m3 per person in 2020 to 43 m3 per person c) Demand management Doubling of piped water tariffs for by the end of the century. In the household sector, annual per higher tiers, equalization of piped capita water use decreases from 31 to 14 m3. Agricultural water supply availability for all household use on a per capita basis decreases from 59 to 24 m3 annually. users on a per capita basis, By the end of the century, the combined effects of population administrative NRW reduction growth and shrinking water availability cut per capita water use (50% of current losses) roughly in half across all sectors. The sectoral water availabil- d) Agriculture to urban 25% transfer of groundwater production ity estimates are further extended across all future narratives transfer capacity from agricultural (Fig. 3). Under worst-case conditions (baseline-crisis), annual to municipal sector 3 e) Aggressive action Supply enhancement plus demand per capita water use across all sectors drops to nearly 30 m per management plus agricultural to person by the end of the century. urban water transfer Widescale Decline of Household Water Security. We further evalu- ate water-security impacts on end consumers, translating results at the agent level to a set of metrics quantifying water availability, of the hydrologic system, agricultural sector performance, and vulnerability, equity, economic well-being, and shortage dura- urban water-security metrics. Changes across this set of metrics tion that can be attributed to the changing water use of Jordan’s are determined for each model test, recorded as a percentage population. For the current study, we focus our water-security difference relative to a baseline model run. A complete list of analysis on the household population, though we note that com- conducted test runs is included in SI Appendix, Sensitivity Analysis mercial and agricultural end users are also represented in the and Fig. S5. model. Results from the optimistic-baseline narrative reveal widescale Results: Jordan’s Evolving Water-Security Problem decline in water security across all four metrics for Jordan’s Growing Population, Dwindling Water Resources. Across all sce- household sector, even under relatively favorable future condi- nario conditions, Jordan experiences significant growth in popu- tions. The percentage of water-vulnerable population increases lation and decrease in freshwater resources availability, resulting from 16 to 58% for normal-income households, while the lower- in marked declines in water use on a per capita basis (Fig. income households experience a drastic increase from 52 to 91% 2). Under moderate population growth assumed in SSP 2, the (Fig. 4). The Gini coefficient of water use, which provides a mea- total population grows from 9.7 million to 21.4 million people sure of water-use equity across the population, reveals widening by the end of the century, a 121% increase and 1.4% average disparity in water use, with the Gini coefficient increasing by annual growth rate. Under more intensive population-growth 31% by the end of the century compared to conditions at the assumptions in SSP 3, the total population grows to 28.8 mil- start of the model run (Gini coefficient value of 0.23 at baseline lion, a 197% increase and 2.3% average annual growth rate. starting conditions) (Fig. 5). Shortage durations indicate that by With another refugee wave of similar magnitude to the recent the end of the century, higher-income households fall below the influx due to the Syrian war and underlying SSP 3 conditions, critical water-use threshold (40 L per capita per day [LPCD]) the total population grows to 32 million by the end of the for 7 consecutive months per year on average, while lower- century. income households fall below this threshold for 11 consecutive As population trends upward, available water resources months on average (Fig. 6). Economic well-being attributable shrink. Even under optimistic scenario conditions, groundwater to changing water use declines significantly, with higher-income levels decline by approximately 60% in Jordan’s highly pop- households experiencing an annual consumer surplus loss

4 of 12 | PNAS Yoon et al. https://doi.org/10.1073/pnas.2020431118 A coupled human–natural system analysis of freshwater security under climate and population change Downloaded by guest on September 26, 2021 Downloaded by guest on September 26, 2021 ope ua–aua ytmaayi ffehae euiyudrclimate under security change freshwater population of and analysis system human–natural coupled A equally. al. et and Yoon efficiently more water existing the distributing at and an lower-income the population. demand-management curbing for household particularly at vulnerability, (pre-2040), water comparable in increase horizon are portfolios portion time earlier supply-enhancement the During the still household century. lower-income the baseline, population of the of end for the the higher water-vulnerable by or population to 50% of least compared at percentage reaches security the modest water though in in result improvements portfolios intervention higher-income agricultural-to-urban the population. for lower-income the 80% for 96% interven- reaches and no population vulnerability with water (crisis) conditions tions, interventions. all worst-case-scenario supply-enhancement for the the 80% In over exclude popu- reaches that vulnerability lower-income narratives water the trends household For lation, vulnerability category. narrative the household-income on water depending and 98% to household up increasing upward, instituted, steeply is portfolio conditions. a scenario across of security range water interven- Jordan’s the various enhancing of from toward efficacy portfolios results the tion evaluate Utilizing we run 1). narratives, Table model is 20 in model detailed intervention the (as and scenario assumptions analysis, identifying under various combining our security for narratives 20 Extending water across need future. Jordanian critical precarious bolster a sce- the can optimistic to that widespread despite interventions point observed which assumptions, are in security nario water narrative, in optimistic-baseline declines the Sector. Water Jordan’s from Bolster to Needed Action Aggressive to attributable solely of use. 11% water households to reduced lowest-income and equivalent the loss (Fig. conditions, surplus for consumer loss income a starting USD to equates to $245 This a 7). relative experiencing USD households lowest-income $340 to equivalent indicator of Description Calculation impacts system Natural impacts future evaluate to used Metric metrics security Freshwater 2. Table ua ytmimpacts system Human rudae elto nofie qie saturated aquifer Unconfined depletion Groundwater oshl ae euiy osmrsrlscagsascae hnei oshl cnmcwl-en soitdwt changes with associated well-being economic household in Change associated changes surplus Consumer security: water Household water capita per falling of households coefficient lower-income Gini and higher-income of Number security: water Household less receiving households of Percentage Vulnerability security: water Household water Sectoral to relative inflows reservoir Total availability Surface-water oshl ae euiy aiu ubro osctv otsaDrto vrwihahueodeprecswtruebelow use water experiences household a which over Duration a months consecutive of number Maximum duration Shortage security: water Household h eadmngmn otoi otisplce aimed policies contains portfolio demand-management The and demand-management, supply-enhancement, The aggressive-action the which in those except narratives all For cnmcwl-en ihwtruse water with well-being Economic Equity availability oa ae vial o ahsco oa ae viaiiyo e aiabssfrec sector, each for basis capita per a on availability water Total sector each for available water Total s esrdars h oshl gn ouain esrda ecnaecag nteGini the in change percentage a as measured population, agent household the across measured use basis capita per a on hn4 PDo naeaeannual average an on LPCD 40 than oshl eevsls hn4 PDteciia ae-s hehl f4 LPCD. 40 of threshold water-use critical the LPCD 40 than less receives household rudae wells at groundwater levels 2016 to relative thickness cpe t12) at (capped population basis vrg itrclflows historical average Results niaigwdnn iprt nwtruecmae ocur- in to results compared portfolio use demand-management water coefficient, The Gini in conditions. the rent disparity in increase widening an indicating see narratives demand-management all coeffi- include optimistic-baseline measures, that Gini the narratives the in Besides in conditions narrative. difference starting percentage to in compared a shown cient as are presented narratives all 5, for Fig. population agent conditions household eled all crisis under nearly population under low-income century the of conditions). the exception of the (with end the 20% below through remains vulnerability water combined household these interventions, of Under effects combined interventions. redistributive demand-management the the and the by supply total compounded increased are of improvements impacts security water portfolio, all household aggressive-action across to effects ameliorative the mitigative With all significant scenarios. The in combines vulnerability. is results which water interventions, portfolio rising portfolio, mitigating supply-enhancement at aggressive-action the effective most action), the aggressive ing the to relative century the of portfolio. end (no-intervention) the baseline increases by durations water-vulnerability shortage exacerbates absolute and interventions and of demand-management off levels the tapers reaches quickly of available efficacy the supply the raising water scarcity, and total population worsen the conditions the scenario and As of consumption. dire share water mitigating daily large average in effect a beneficial for a conditions have even still portfolio, policies its demand-management LPCD). though the 30 cri- under below LPCD increase 40 (e.g., the terion conditions by measured dire durations more shortage be Consequently, even 40 can actually of consumption the out households’ water other lifted below In that to consumption LPCD. so access threshold, 40 households’ LPCD least equal some systems at reducing more everyone water-supply entails providing give urban situation, to water the this little scenarios, too crisis receive and population- the in growth 2100 approaching conditions severe the Under h acltdGn ofcet fwtruears h mod- the across use water of coefficients Gini calculated The exclud- (i.e., portfolios intervention individual the of Out rudae viaiiyfrwtrspl.As nindicator an Also supply. water for availability Groundwater cnmcdsaiyi e aiawtruears h household the across use water capita per in disparity Economic indicator an Also supply. water for availability Surface-water ofcetrltv o21 odtosfrteoptimistic-baseline the for conditions 2016 to relative coefficient 90% top the as defined Higher-income households are basis. annual ecosystems. baseflow-dependent fautceoytmhealth. ecosystem aquatic of narrative. 10%. bottom the lower-income and earners, income household of needs. human eaiet 06cniin o h piitcbsln narrative. optimistic-baseline the for conditions 2016 to relative nwtrue esrda ifrnei osmrsrlsloss surplus consumer in difference a as measured use, water in for availability water of assessment aggregate an providing eo rtclwtruetrsodo 0LC,aeae nan on averaged LPCD, 40 of threshold water-use critical a below feooia elh ssrasadwtad r primarily are wetlands and streams as health, ecological of https://doi.org/10.1073/pnas.2020431118 PNAS | f12 of 5

ENVIRONMENTAL SUSTAINABILITY SCIENCES SCIENCE AB

Fig. 2. Population growth and water availability in Jordan. Increasing population and decreasing water resources availability will place increasing strain on Jordan’s water system. (A) Jordan’s pop- ulation increases from 9.7 million to 21.4 million people by the end of the century under moderate growth assumptions, while under crisis conditions, the total population reaches 32 million by the end of the century. (B) Surface-water availability decreases across all four combinations of scenario conditions, with the most substantial flow decrease due to adverse climate-change impacts (RCP 8.5) CD and intensive irrigated agriculture upstream of Jor- dan in the Syrian Yarmouk basin (“prewar” condi- tions). Under the most optimistic conditions, aver- age total surface-water inflow during the last two decades of the century decreases by 25% com- pared to current conditions. (C) Even under opti- mistic scenario conditions, declining groundwater levels result in reduced groundwater production capacity, with a loss in production capacity of 78 MCM/year by the end of the century compared to current conditions. (D) In the optimistic-baseline narrative, annual per capita water use across all sec- tors decreases in half from nearly 100 m3 per person in 2020 to 43 m3 per person by the end of the century.

an improvement in equity over the entire modeled time hori- the most substantial mitigation of consumer surplus loss is zon, though equity gains gradually decline as scenario conditions observed for the supply-enhancement portfolio for lower-income worsen. By the end of the century, the Gini coefficient for households. For the early part of the century, lower-income users the demand-management narratives ranges from 29 to 34% experience an increase in consumer surplus relative to starting lower (higher equity) compared to starting conditions in the baseline conditions, though these gains dissipate post-2040. The baseline. The aggressive-action portfolio sees improvements in aggressive-action portfolio significantly curbs household eco- water-use equity that are largely sustained across the entire nomic loss, though higher-income households still experience an time horizon, with a 42 to 45% decrease in the Gini coefficient annual loss ranging from $109 to $131 USD, and lower-income value (depending on scenario conditions) compared to current households experience a loss of $34 to $56 USD, depending conditions. on the scenario conditions. In the most pessimistic narrative The shortage-duration metric indicates the number of con- (baseline-crisis), population-corrected total consumer surplus secutive months that households fall below a critical water-use loss across the entire population reaches an annual value of $691 threshold of 40 LPCD in any given year. Prior to midcen- million USD by the end of the century. tury, baseline scenarios show similar shortage-duration values, because climate and socioeconomic impacts have yet to diverge Socioeconomic versus Climate-Induced Impacts. Supplementing the across scenarios. By 2100, the baseline-policy scenarios exhibit narrative runs, scenario and stress tests indicate the relative a marked difference in shortage durations for 90% of the pop- importance of socioeconomic versus climate-change drivers on ulation, ranging from 6 to 10 mo between optimistic and crisis Jordan’s future water security. In scenario tests, the implemen- scenarios. For lower-income households, the average shortage tation of SSP 3 population, GDP, and income assumptions durations asymptotically approach 12 of 12 mo per year in the (relative to SSP 2) result in a 10% increase in household water latter part of the century, so differences between scenarios are vulnerability, whereas implementation of RCP 8.5 (relative to muted. Most lower-income agents persistently experience a crit- RCP 4.5) increases household water vulnerability by 5%. In ical water shortage, while some agents remain shielded from related stress-test runs, a 10% increase in the household pop- this situation by better local supply conditions. The supply- ulation leads to an 11% increase in household water vulnera- enhancement portfolio results in significant reductions in short- bility. In contrast, a climate-related stress test imposing a 50% age duration, though low-income households still experience decrease in surface-water inflows to reservoirs only leads to a 2% up to 9 mo of consecutive shortage by the end of the century increase in household water vulnerability compared to baseline under crisis-scenario conditions. In the aggressive-action port- conditions. folio, shortage durations are curbed to between 0 and 4 mo, with the longest experienced by low-income households under Complex System Interactions and Intersectoral Dynamics. Sensitiv- crisis-scenario conditions. The aggressive-action portfolio also ity tests and counterfactual simulations further highlight system largely closes the disparity gap in shortage durations experienced interactions and reveal complex interdependencies between sec- between higher- and lower-income households. tors and system components. Notable results (the full set of test Household water users experience significant economic losses run results is in SI Appendix, Sensitivity Analysis and Fig. S5) across all narratives, measured in terms of the change in con- related to the household sector include: sumer surplus (the difference between the price consumers pay and what they would be willing to pay) attributable to changes 1. A 24% increase in household water vulnerability and 18% in water use relative to starting conditions for the optimistic- decrease in farmer profits, if water supply via tanker markets baseline narrative. Apart from the aggressive-action portfolio, is eliminated.

6 of 12 | PNAS Yoon et al. https://doi.org/10.1073/pnas.2020431118 A coupled human–natural system analysis of freshwater security under climate and population change Downloaded by guest on September 26, 2021 Downloaded by guest on September 26, 2021 ope ua–aua ytmaayi ffehae euiyudrclimate under security change freshwater population of and analysis system human–natural coupled A al. et Yoon approach multiagent distributional and outcomes. The agents, systemwide water-security both individual of of century. evaluation level the socioeconomic the the for at allowing impacts of and of diagnosis end population, the enables under the climate, interventions to demand-side of change and scenarios supply- of future for range impacts wide water-security a distributional and sectoral, temwide, hydrologic built and natural the that of dynam- modules system. state changing actors, biophysical the process-based water-sector simulate with real-world modeled interacting onto of and ically map cast natural diverse water-allocation that a shared by Modeled agents performed a are environment. and decisions water water-use decision-making built of and network interconnected temporally tanker a and informal spatially through and institu- are suppliers informal that water markets, and water government formal as alongside including such users, agents, tions, demonstrate consumer urban We of and and thousands scale. farms simulate management a country to water ability the of the inform at example decision-making to an policy deployed and is biophysical modules JWM, coupled human with the model here, multiagent human–natural-engineered comprehensive described Jordan’s system of water model integrated The Conclusion and Discussion a (or project Sea Sea–Dead Red the of Implementation in increase 24% 6. a increase in results 376% income household in a increase 50% A in 5. results prices crop of 10% a doubling in A results 4. rates drawdown groundwater water of household Doubling reduces 3. test counterfactual no-refugee A 2.

Wastewater Irrigation Jordan Valley Irrigation Sector Supply (cubic meters/year/capita) sn h utsaeitgae oe,w vlaesys- evaluate we model, integrated multiscale the Using % hl qaiighueodspl uain reduces 35%. durations by disparity supply by water-use and disparity household 6% by water-use vulnerability equalizing and while 60% house- by 8%, reduces vulnerability capacity) water equivalent hold of project desalination market. tanker the through transferred water of volume the water in household in increase 13% a and vulnerability. profits farmer total vulnerability. water household in increase 14%. by vulnerability Population Drier Crisis Optimistic Growth Climate 100 100 100 100 50 50 50 50 0 0 0 0

2020 2030 2040 Baseline 2050 2060 2070 Industrial Highland Irrigation 2080 2090 2100 2020 2030 Management 2040 Demand 2050 2060 2070 2080 2090 2100 2020 Commercial Piped Commercial Tanker Agriculture to 2030

2040 Urban 2050 2060 2070 2080 2090 2100 2020 Enhancement 2030

2040 Supply 2050 2060 Household Piped Household Tanker 2070 2080

2090 higher- both dramatic for a century scenarios, the all of across households end remains lower-income the vulnerability across and through water observed 33% Household and are below metrics. enhancements improvements infrastructure evaluated implementa- of marked all With set distribu- measures, users. sweeping the urban policy this for equalizing of supply and water tion sector, piped of urban from tion water to reallocating agricultural piped- tiers, desalination raising the higher-water-use Sea losses, on physical Red tariffs and planned water theft full-scale water currently the reducing all project, including implementing aggressive projects, entail take water must would partners This international action. its and Jordan that and threshold water-use lower- critical all nearly the recover. that never cross indicating century, households the income of by average year critical on closing the months household under the consecutive of fall 12 nearly and 12% threshold conditions water-use to current to equivalent relative absence losses house- income economic the Lower-income suffer in interventions. holds particularly signif- reveals equity, demand-management which water-use use, of of in water calculation of house- declines coefficient through icant the Gini quantified a across in formally use changes is water population in hold disparity intervening critical Widening without measures. a conditions crisis-scenario below Jordan’s under pop- of the segment population drop of low-income 91% most-vulnerable, that to the up for reaching ulation households LPCD, 40 of of threshold water-use percentage measured as households the the vulnerability, water for narratives, by in severe all increases drastic particularly Across experience are households. impacts lowest-income these population, the shortage loss. equity, economic vulnerability, and of duration, metrics impacts, via water-security quantified consumer latter the and capita per availability, depletion, supply water-resource water of terms in well- interventions, human measured potential and being environmental of on impacts suite severe broad suggest results socioe- a and and climate scenarios of range conomic a Considering security. water future 2100 ogi otodo t ae uue u nlsssuggests analysis our future, water its on foothold a gain To of segment every upon touch declines water-security While Jordan’s for outlook sobering a paint narratives model Our Aggressive Action 2020 2030 2040 2050 2060 2070 2080 2090 2100 yteedo h century. the m of 58 end to the by declines still analysis, supply the the water in total of considered interventions sce- implementation of optimistic suite the full under with Even conditions, century. the nario nearly of end from the 50% over totaled m by supply 100 decreases water sectors capita across per annual baseline-optimistic the narrative, In narratives. signifi- all drops across supply cantly water capita avail- per resources annual water increas- ability, shrinking of and combination population a ing to Due narratives. future 3. Fig. 3 e esni 00t 3m 43 to 2020 in person per etrlprcpt ae upyacross supply water capita per Sectoral https://doi.org/10.1073/pnas.2020431118 3 PNAS e esnby person per 3 e person per | f12 of 7

ENVIRONMENTAL SUSTAINABILITY SCIENCES SCIENCE Fig. 4. Household water vulnerability across future narratives. Results for household water vulnerabil- ity across the 20 model narratives, measured as the percentage of the population that uses less than 40 LPCD averaged over a given year, are shown. House- hold water vulnerability is further distinguished between a low-income household category (bot- tom 10% of households by household income) and the rest of the population. Under optimistic con- ditions and no new interventions, the percentage of water-vulnerable population increases from 16 to 58% for normal-income households, while low- income households experience an increase from 52 to 91%. For all narratives except those in which the aggressive-action portfolio is instituted, household water vulnerability trends steeply upward. In the worst-case narrative (crisis scenario conditions with no interventions), water vulnerability reaches 80% for the high-income population and 96% for the low-income population. With the aggressive-action portfolio, improvements to household water secu- rity are compounded by the combined impacts of increased total supply and the redistributive effects of the demand-management interventions. With these combined interventions, household water vul- nerability remains below 20% under all narra- tives besides for lower-income households in crisis conditions.

improvement over the baseline no-intervention portfolio. The critically needed in tandem to bolster water security for Jordan’s aggressive-action portfolio further results in a substantial nar- population. rowing of the water-use disparity gap, with an improvement in Scenario tests further distinguish the individual contributions water-use equity compared to current conditions that is sus- of climate and socioeconomic changes on water-security out- tained through the end of the century (as measured through comes, indicating that socioeconomic drivers outweigh climate the Gini coefficient of water use). In the earlier portions of the drivers in exacerbating water-security declines. While many stud- model horizon, the aggressive-action portfolio in fact results in ies evaluating future water security focus exclusively on climate- water-security improvements compared to current conditions. change impacts, our results indicate that socioeconomic changes As scenario conditions worsen, water security across all metrics will likely outweigh climate change in terms of detrimental degrades, though this decline pales in comparison to interven- impact on Jordan’s water security. More rapid population growth tion portfolios that focus singularly on supply enhancement or (as represented in the SSP 3 socioeconomic scenario) results demand management. in a 10% increase in household water vulnerability relative to The multiagent approach adopted in the JWM further the baseline (RCP 4.5 and SSP 2), whereas more adverse cli- enables the evaluation of aggregate and distributional mea- mate conditions (as represented in the RCP 8.5 climate scenario) sures of water security and the comparative evaluation of only result in a 5% increase. An additional refugee wave in 2030 both supply- and demand-side interventions. Results from of equal magnitude to the recent Syrian refugee wave likewise the aggressive-action portfolio reveal the compounding bene- results in a 10% increase in water-insecure households relative fits of packaging demand and supply measures alongside one to the baseline. another. Across all water-security metrics, the improvements Similar outcomes are observed in the stress-test runs com- observed in the aggressive-action portfolio cannot merely be paring socioeconomic versus climate drivers. When population ascribed to additive benefits from the individual interventions. is increased by 10%, the percentage of vulnerable households Rather, the interventions work in synergistic fashion, with increases by 11%. In contrast, a climate-focused stress test beneficial impacts increasing nonlinearly as interventions are involving a 50% decrease in surface-water inflows to reser- combined. voirs results in a mere 2% increase in water-vulnerable house- A notable example of this can be observed in the crisis nar- holds. These results point to the need to carefully consider ratives, in which the demand-management portfolio results in a future socioeconomic changes in future assessments of water worsening of water vulnerability by the end of the century when security, as these may overshadow water-supply impacts from implemented independent of supply-enhancement measures. climate change, especially for regions undergoing substantial However, when packaged with the supply-enhancement and demographic shifts such as Jordan. sectoral-reallocation interventions, we see a dramatic decrease Structural-model tests further reveal unexpected interactions in the percentage of lower-income population that is water- and interdependencies across sectors and stakeholders, illustrat- vulnerable at the end of the century from 97 to 33%. With the ing the importance of adequately representing dynamic agent supply-enhancement intervention alone, the percentage is only interactions, including noncentralized supply features, such as reduced to 71%, while the agriculture-to-urban intervention only informal water-transfer mechanisms. In Jordan, one such inter- sees a reduction to 93%. Similar, nonlinear effects are observed connection is notably observed between the agricultural and with the equity, shortage-duration, and economic-loss metrics. urban sectors, in which governmental and private agricultural Although supply- and demand-management interventions are actors access a shared common-pool groundwater resource. Due often presented as opposing alternatives in water-policy debates to municipal water-supply shortages, urban consumers supple- in Jordan and beyond, our analysis suggests that the two modes ment piped water with groundwater purchased from farmers of system interventions have a mutually enhancing effect and are via informal tanker markets. In counterfactual simulations, the

8 of 12 | PNAS Yoon et al. https://doi.org/10.1073/pnas.2020431118 A coupled human–natural system analysis of freshwater security under climate and population change Downloaded by guest on September 26, 2021 Downloaded by guest on September 26, 2021 ope ua–aua ytmaayi ffehae euiyudrclimate under security change freshwater population of and analysis system human–natural coupled A al. et Yoon decision-making consumer agents. and called subnational, cap- national, entities simulate approach, model agents and autonomous These simulation use through multiagent decisions water a human crop adopt turing and modules wastewater, human treated The and yields. raw network of water-supply tracking changes, routing, storage and reservoir historical conditions, under climate rainfall–runoff future watershed depletion, groundwater on of levels multiple at systems; agency physical human decision-making. represent built which modules, and human phenomena 2) and natural encompassing of modules simulators biophysical gener- 1) two quantitative modules: of of consists types model distinguishable feedbacks integrated ally and The components. interactions sub- system significant and between identifying independently, by calibrated integrated and sequently The developed (37). are architecture software modules object-oriented The individual an system. in water Jordan’s implemented of is which component model in specific approach, a modular represents multiagent, module each systems-based, a adopts JWM The Methods and Materials international monu- its a and ahead. Jordan presents decades for the undertaking in effort partners an imperative, Such yet mental, change. climate growth to socioeconomic and challenging measures by driven demand-side politically declines and water-security ambitious, mitigate supply- an coordinated Jor- of implement future, portfolio on water to its impacts on needs foothold dan a water-security point gain To destabilizing results population. Jordan’s model potentially for- multiagent severe, carries integrated to Jordan of our coun- If impacts passively, the mining. the ward Meanwhile, groundwater under and further century. change dwindle the climate will of twofold resources grow end water to the try’s expected by yet is threefold influxes, to refugee of from reeling impacts already the population, Jordan’s future. water its secure water of misdiagnosis such severe ignore in security. resulting commonly Jor- and potentially in approaches markets dynamics, security modeling water tanker household Traditional role for dan. play critical sector the agricultural prices the illustrating crop 24% when doubled, 13% by by and are restricted increases are markets households tanker when water-vulnerable of percentage h ipyia oue nld ioossmlto fpmigimpacts pumping of simulation rigorous include modules biophysical The to challenge grand the faces Jordan ahead, century the In ouefrdsrpieproe,teruigmdl shnldinternally handled is module routing the independent an purposes, maintained as descriptive is here for presented balance module Though mass network. water the the that throughout ensuring system, water-transmission other module. and equation. Routing balance evaporation, mass modules), a human in losses module), various watershed by the and (determined (from Mujib, inflows releases Wala, reservoir Kafrein, for Al-Wehdah, Shueib, accounting (including Karameh, by Talal, Tannour) country King the Ziglab, in Arab, reservoirs Wadi provided surface-water SWAT main data the The of streamflow 13. monthly module. ref. Reservoir upon in Mujib/Walah based MWI. detail the and by calibrated further Valleys, are in the the Side models described throughout Zarqa, utilizing watersheds (43). basins, Yarmouk, (SWAT) main country surface-water the Tool the the for Assessment including developed Water country, throughout were and watersheds models Soil SWAT the main model, the rainfall-runoff well of approach linearization each a module. using (42). Watershed matrix literature groundwater-modeling response the in a established into model MODFLOW translated compu- three-dimensional reduce is the To that complexity, and supply. and aquifers groundwater layers intensity for major 12 tational country all the of for within consists Irriga- accessed accounting model are and cells, MODFLOW Water model the original 350,000 of by The approximately Ministry developed (41). Jordanian dis- (MWI) the region spatially tion for the existing Survey an of Geological utilizing model French country, groundwater the numerical across tributed aquifers all physical in engineered levels and natural module. major Groundwater the system. water simulate Jordan’s to comprising features used mod- are water-use that crop and ules wastewater, reservoir, routing, watershed, water, Modules. Biophysical in presented are modules individual integrated the Appendix, the of SI for each details and framework implementation modeling Further sections. following the modules. across communication for total) in (89 Jordan’s unit administrative to subdistrict translated output with resolution, individual spatial Each specific agents. a farmer adopts institutional module 16 84 and plants, agents, industrial large commercial-establishment 14 of agents, 1,005 consisting country, real- agents, the onto household mapped across are 804 groups agents consumer environment 1,923 and of socioeconomic organizations total and world A other. hydrologic each the with interact in and changes to response in h niiulbohscladhmnmdlsaedsrbdfrhrin further described are modules human and biophysical individual The Methods. h otn ouehnlsruigtruhtemain the through routing handles module routing The h eevi ouedtrie h trg volumes storage the determines module reservoir The h ipyia oue oss fitrcigground- interacting of consist modules biophysical The h aese ouecpue ufc osfor flows surface captures module watershed The artvsaeaot3%lwr(ihrequity) (higher baseline. the lower century, in conditions 30% starting the to about compared of demand-management are the end for narratives the coefficients Gini scenario By horizon, the as time worsen. decline modeled gradually conditions gains entire equity the improvement though over an equity in demand- Gini in results The the conditions. portfolio current in management water to increase in compared disparity an use widening see indicating narratives coefficient, all mea- sures, demand-management include that Besides conditions). narratives starting coefficient baseline (Gini at run 0.23 con- model of to value the compared of century start the the at of ditions increasing end the use, by water 31% by in disparity widening In reveals coefficient narrative. starting Gini the to optimistic-baseline narrative, optimistic-baseline the compared differ- the coefficient percentage in Gini a conditions the as in agents). presented ence household are all capita across results per same The the (i.e., is use indi- use water 0 water in of equity coefficient the complete a between cates while use indicates agents, water 1.0 household capita all per of in max- value disparity A by greatest coefficient use. water measured Gini of is imum coefficient Gini equity a water calculating Household ratives. 5. Fig. h rudae ouesmltsgroundwater simulates module groundwater The oshl ae qiyars uuenar- future across equity water Household https://doi.org/10.1073/pnas.2020431118 PNAS | f12 of 9

ENVIRONMENTAL SUSTAINABILITY SCIENCES SCIENCE Fig. 6. Household water-shortage durations across future narratives. Household water-shortage dura- tion is measured by the average number of con- secutive months that households fall below a crit- ical water-use threshold of 40 LPCD in any given year. For the baseline, demand-management, and agriculture-to-urban intervention portfolios, the average number of consecutive months ranges from 8 to 12 mo by the end of the century for the lower-income-household agent population and 5 to 11 mo for the rest of the population. The highest shortage durations are observed in the demand-management portfolio, with lower-income- household users reaching 12 consecutive months of water use below 40 LPCD on average by the end of the century (indicating that households per- sistently experience critical water shortage). The supply-enhancement portfolio results in significant reductions in shortage duration, though low-income households still experience up to 9 mo of consec- utive shortage by the end of the century under crisis-scenario conditions. Shortage durations are strongly reduced in the aggressive-action portfolio to between 0 and 4 mo, with the longest shortage durations experienced by lower-income households under crisis-scenario conditions. The aggressive- action portfolio also largely closes the shortage duration gap between higher- and lower-income households.

within the Jordan Valley Authority (JVA) and Water Authority of Jordan use module, determine the groundwater abstractions for irrigation and (WAJ) modules via mass balance constraints defined in the JVA and WAJ minor uses of supplementary surface water. The module covers all agri- optimization procedures. The routing module also accounts for physical cultural areas in Jordan that are mainly irrigated with groundwater or NRW losses in the water-distribution network. rainfed, while the rest is covered by the JVA Module. Farmers’ crop choices Wastewater module. The wastewater module is used to simulate wastew- are determined by applying a Positive Mathematical Programing approach ater flows from urban users (connected to the sewer system) to wastewater (49) of profit maximization. They respond to simulated temperature and treatment plants and the treated effluent flows from the wastewater treat- precipitation developments via the crop water-use module, as well as sce- ment plants to various reservoirs and irrigation water users throughout nario inputs capturing changes in prices, costs, or regulatory constraints on the country. For urban users connected to the sewer system, the module land and water use. Farmers can sell unused well-water quantities to the assumes that a fixed percentage of total water use is nonconsumptive and tanker-water market module. enters the wastewater system. Information on Jordan’s wastewater system, Central Water Authority. The Central Water Authority (CWA) is an insti- including treatment plants, treatment-plant capacities, and treated wastew- tutional agent that is used to represent decision-making at the national ater destinations, is based on information from Jordan’s National Strategic level (e.g., management of transboundary water transfers, allocations Wastewater Master Plan (44). between the JVA and WAJ). The primary role of the CWA module is Crop water-use module. The crop water-use module determines the net to determine water releases from the Al-Wehdah Dam (Jordan’s largest irrigation requirements necessary to sustain historical crop yields under surface-water reservoir along the Yarmouk River on the border with Syria), increasing climate-change impacts. The relationship between crop yields water transfers with Israel via Lake Tiberias, diversions from the Yarmouk and irrigation requirements is captured by the relationship defined in River and Lake Tiberias into the King Abdullah Canal (KAC), and trans- the Food and Agricultural Organization of the United Nations 33 equa- fers from the KAC to Amman at Deir-Allah. The CWA adopts a hybrid tion (45), which has been widely used in integrated water-resources optimization and rule-based approach to determine the set of decisions studies (46). described above. JVA. The JVA is the institutional authority that supplies surface water to Human Modules. The human modules simulate the allocation and consump- farms in the Jordan Valley via the KAC system. The primary role of the tion of water from the piped network, tanker-water markets, groundwater JVA module is to minimize Jordan Valley farmers’ supply–demand deficit, wells, and surface-water abstractions in response to changing biophysi- while also maintaining target water levels in its reservoirs, accounting for cal, socioeconomic, and policy conditions. The modules allow for economic projected inflows into the KAC. assessments of the value of water to the water users. They consist of WAJ. The WAJ is a government body responsible for water allocation in submodules of urban and agricultural water-user agents and allocation the country, excluding the Jordan Valley. The module determines cost- institutions for piped and tanker water. minimizing groundwater abstractions and transfers between governorates Urban water-user agents. A total of 1,823 representative urban water- to pursue per-capita supply targets for each of Jordan’s 12 governorates. user agents capture various types of households, refugee households, The WAJ module uses inputs on population from the urban water-user and commercial establishments across all 89 subdistricts of Jordan. These module, on network topology and capacity from the water-conveyance net- water-user agents determine their demand for piped and tanker water work module, on well-field yields and costs from the groundwater module, and their utility derived from water consumption by applying a tiered and on surface-water availability from transboundary sources from the CWA supply-curve approach (18, 47), based on econometric demand-function module. estimates (48) parameterized with census and survey data for house- Local piped-water-allocation institutions. Local piped-water-allocation holds and establishments (n = 16,153). Piped-water demands account institutions in each of Jordan’s 12 governorates receive bulk water quan- for local supply intermittency and storage constraints. Tanker-water tities allocated to their governorate by the WAJ water-conveyance net- demand is the residual demand remaining after the consumption of work. They then distribute this bulk water quantity across all subdistrict- piped water. Projections of future agent parameters are based on the level urban water-user agents of the various sectors and supply duration SSPs (38, 39) and regressions of income and establishment numbers classes within their governorate. We use an algorithm employing observed onto the GDP. subdistrict-level piped supply durations, reflecting access to the public water Agricultural water users. The agricultural water-user module simulates network, as distribution weights in an iterative allocation of piped water to farmers’ crop choices in 84 subdistricts that, together with the crop water- users.

10 of 12 | PNAS Yoon et al. https://doi.org/10.1073/pnas.2020431118 A coupled human–natural system analysis of freshwater security under climate and population change Downloaded by guest on September 26, 2021 Downloaded by guest on September 26, 2021 ope ua–aua ytmaayi ffehae euiyudrclimate under security change freshwater population of and analysis system human–natural coupled A al. et Yoon oe,i rhvda h tnodDgtlRpstr n vial for available and Repository Digital Stanford the at (https://purl.stanford.edu/zw908ds8394 ). download archived is model, Availability. are Data tests, a counterfactual to and metrics intervention, model sensitivity, in various stress, described of of responses set showing broad con- analyses, water Sensitivity agricultural S4. and in urban presented model observed are the sumption, to whether outputs model evaluate tests, Validation effectively. comparing to patterns water- and etc.) consumers, dynamics system to historical levels, deliveries captures groundwater (e.g., rather module prices, modules single multiple a market of of behavior interactions the that dynamic than observations the as upon defined dependent are are which data, Analyses. system-level observed Sensitivity against and Validation from water the get to other. required the costs for to demand one transportation agents’ the water-user and source, water, rural its tanker agricultural at the water on of cost based the opportunity transfers determines tanker-water approach of This set (50). Jordan welfare-maximizing approach across price-equilibrium markets private spatial on a trucks via tanker via transfers water all resents institution. market Tanker-water .Mnsr fWtradIrgto (MWI), Irrigation and Water of Ministry contested of 7. assessment Quantitative Talozi, S. Rosenberg, D. basins,” Tilmant, river A. Avisse, transboundary N. non-cooperative 6. of analysis “Quantitative Avisse, N. 5. River. Yarmouk M F. the M. of management 4. on thereof lack and Cooperation Haddadin, M. 3. transformations “Historical Jridi, A. Suleiman, R. Molle, F. Venot, J. the Courcier, in resources water R. of variability 2. natural the into insights New Lange, J. Gunkel, A. 1. ae ssadmngmn ntecnittr amu ie Basin. 2019). River Jordan, Yarmouk conflict-torn the in Manag. Plann. management Resour. and uses water (2018). Canada Quebec, Laval, Universite Dissertation, (2016). 14932–14937 resources. freshwater 113, transboundary and use land on crisis Int. projec- Water and use Lanka, Sri water Colombo, in Institute, Management Changes Water 2005). Jordan): International 9, Agriculture (in in Report Management Research Basin Water of River Assessment (Comprehensive Jordan (1950-2025)” tions Lower the of Basin. River Jordan lower le,J on .M oeik .Ais,A imn,Ipc fteSra refugee Syrian the of Impact Tilmant, A. Avisse, N. Gorelick, M. S. Yoon, J. uller, ¨ 2–3 (2009). 420–431 34, , Appendix SI h W oe nldn l aarqie ornthe run to required data all including code, JWM The 5200(2020). 05020010 146, ae eor Manag. Resour. Water estvt Analyses Sensitivity IAppendix, SI h akrwtrmre nttto rep- institution market tanker-water The h nertdmdli validated is model integrated The ae erok2016-17 Yearbook Water oe Validation Model 6–8 (2011). 963–980 26, n i.S5. Fig. and rc al cd c.U.S.A. Sci. Acad. Natl. Proc. n is 3and S3 Figs. and MI Amman, (MWI, .Water J. 4 .El-Naser, H. cascad- and 14. change Climate Jordan: in drought Increasing Gorelick, S. rain- Rajsekhar, Declining D. Rajaratnam, 13. B. Yoon, J. Dennedy-Frank, P. Gorelick, S. Rahman, K. drought. Mediterranean 12. of dry. frequency running increased from the On Jordan al., stop et Horeling to M. scramble The 11. water: without land A Whitman, E. 10. oeytoeo h uhr n ontncsaiyrflc h iw fthe of views the data. reflect or necessarily funding provided not that do agencies and other authors or NSF the are material of this those in expressed solely of recommendations University or opinions, conclusions The Any and (033WU002). acknowledged. findings, is UFZ Facility to Educa- Shared funding Computational of Manchester’s provided Ministry and Federal Research German 2764/1-1) and the tion (KL and (UFZ) 506/4-1); Research (GA Environmental University Leipzig for Center to the Helmholtz funding of provided the part Forschungsgemeinschaft Deutsche As the (NE/L009285/1). Research Forum, funding Environment Belmont UK Natural provided The Forum Initiative. sup- Belmont in Freshwater Council Environment Global the the for of Institute Woods port Stanford’s Bel- by and the theme; of supported Initiative/Food-Water-EnergyNexus conclu- part was Global as the Urbanization work ICER/EAR-1829999 Forum–Sustainable for This and mont data. GEO/OAD-1342869 responsible available Grants solely NSF upon confidentiality by are drawn the inferences researchers data, preserve and The to relevant sions data. aiming to individual processing access Inter- to Jordanian of researchers for data the the Agency subjecting and US granted after Forum information the Statistics Research and and of Economic Survey data Department The Geological Additional US Development. project. the national the Khalaf by their of Bani for provided El-Din Refaat course were Badr and the Amin pro- MWI and over support Bataineh at Mohammad support for Subah thank Ali grateful also We and particularly WAJ. data El-Naser at are of Hazim We Dr. provision analysis. by for the vided Statistics for of reports Department and and Agriculture, of Ministry ACKNOWLEDGMENTS. .M .M F. M. 9. forecasts, and trends “Groundwater-level Jaber, A. Subah, A. Senior, A. L. Goode, J. D. 8. WTPes otapo,U,2009). UK, Southampton, Press, (WIT flow. transboundary reducing on (2017). impacts land-use Syrian ing Jordan. in changes variability (2015). regional and fall (2012). 2146–2161 Nature groundwater. shared over commons (2017). the 5451–5468 53, of tragedy a avoiding n aq rudae ais odn Oe-ieRpr 0316,U Geological US Yarmouk, 2013-1061, Valleys, 2013). Report VA, Side Reston, (Open-File Jordan Survey, Jordan” Hammad, Basins, Sea, groundwater Dead Zarqa Azraq, and the in trends, salinity and 02 (2019). 20–23 573, le,M .M C. M. uller, ¨ aaeeto creWtrRsucs ideEsenExperience Eastern Middle A Resources: Water Scarce of Management le-te,S .Grlc,HwJra n ad rbaare Arabia Saudi and Jordan How Gorelick, M. S. uller-Itten, ¨ 3 o$6UDls,dpnigo h scenario the on depending loss, USD to conditions. $56 $109 to experience from households $34 ranging lower-income and loss USD $131 annual still households an higher-income experience eco- though household loss, aggressive- curbs nomic The significantly portfolio households. portfo- action lower-income supply-enhancement for the lio the with observed portfolio, loss is pay consumer-surplus aggressive-action of to mitigation substantial the most willingness from reduced losses. consumer-surplus Apart in those offsetting result water, for 3 associated incomes SSP lower with However, capita losses. per lower surplus consumer- adopt to potential pop- lead exacerbating which availability, scenarios the water latter scenarios, to The crisis 3. 2, SSP and SSP opti- growth adopts the ulation which comparing scenario, particularly mistic muted, narra- narra- are optimistic-baseline between tives start- differences the to Consumer-surplus relative for of tive. surplus terms conditions consumer in ing in measured change face. narratives, economic the they all significant costs across experience losses the users and household from Household curves between calculated demand integrating agents’ is by baseline-optimistic results surplus the model compared in Consumer use) conditions narrative. water starting changes changing surplus to consumer to of terms (attributable in well-being measured economic is Household narratives. future 7. Fig. etaksafi h odna W,WJ JVA, WAJ, MWI, Jordanian the in staff thank We oshl cnmcwl-en across well-being economic Household https://doi.org/10.1073/pnas.2020431118 ae eor Res. Resour. Water c.Adv. Sci. PNAS ae eor Res. Resour. Water 3828–3835 51, e1700581 3, | .Clim. J. 1o 12 of 11 25,

ENVIRONMENTAL SUSTAINABILITY SCIENCES SCIENCE 15. Central Intelligence Agency (CIA), World Fact Book: Middle East–Jordan (CIA, 35. J. Padowski, S. Gorelick, B. Thompson, S. Rozelle, S. Fendorf, Assessment of human– Washington, DC, 2019). natural system characteristics influencing global freshwater supply vulnerability. 16. D. Humpal et al., “A review of water policies in Jordan and recommendations Environ. Res. Lett. 10, 104014 (2015). for strategic priorities” (Technical report, United States Agency for International 36. V. Srinivasan, M. Konar, M. Sivapalan, A dynamic framework for water security. Water Development (USAID), Washington, DC, 2012). Security 1, 12–20 (2017). 17. D. Mustafa, S. Talozi, Tankers, wells, pipes and pumps: Agents and mediators of water 37. S. Knox, P. Meier, J. Yoon, J. Harou, A python framework for multi-agent simulation geographies in Amman, Jordan. Water Altern. 11, 916–932 (2018). of networked resource systems. Environ. Model. Softw. 103, 16–28 (2018). 18. C. Klassert, K. Sigel, E. Gawel, B. Klauer, Modeling residential water consumption in 38. K. Riahi, A. Grubler,¨ N. Nakicenovic, Scenarios of long-term socio-economic and envi- Amman: The role of intermittency, storage, and pricing for piped and tanker water. ronmental development under climate stabilization. Technol. Forecast. Soc. Change Water 7, 3643–3670 (2015). 74, 887–935 (2007). 19. A. Al-Kharabsheh, Ground-water modelling and long-term management of the Azraq 39. M. Wise et al., Implications of limiting CO2 concentrations for land use and energy. basin as an example of arid area conditions (Jordan). J. Arid Environ. 44, 143–153 Science 324, 1183–1186 (2009). (2000). 40. K. Riahi et al., The Shared Socioeconomic Pathways and their energy, land use, and 20. D. Mustafa, M. Tillotson, The topologies and topographies of hydro-social territorial- greenhouse gas emissions implications: An overview. Global Environ. Change 42, 153– isation in Jordan. Polit. Geogr. 70, 74–82 (2019). 168 (2017). 21. E. Salameh, M. Alraggad, M. Amaireh, Degradation processes along the new 41. Y. Barthelemy, et al., “Jordan deep aquifers modelling project: Final report” (Tech- northeastern shores of the Dead Sea. Environ. Earth Sci. 78 (2019). nical Report BRGM/RC-59281-FR, French Geological Survey, Orleans´ Cedex, France, 22. R. Jackson et al., Water in a changing world. Ecol. Appl. 11, 1027–1045 (2001). 2010). 23. S. Postel, A. Wolf, Dehydrating conflict. Foreign Pol. 126, 60 (2001). 42. S. Gorelick, A review of distributed parameter groundwater management modeling 24. C. King, A. Holman, M. Webber, Thirst for energy. Nat. Geosci. 1, 283–286 (2008). methods. Water Resour. Res. 19, 305–319 (1983). 25. P. Milly et al., Stationarity is dead: Whither water management?. Science 319, 573–574 43. S. Neitsch, J. Williams, J. Arnold, J. Kiniry, “Soil and water assessment tool: Theoretical (2008). documentation version 2009” (Technical Report 406, Texas Water Resources Institute, 26. C. Vor¨ osmarty¨ et al., Global threats to human water security and river biodiversity. Texas A&M University, College Station, TX, 2011). Nature 467, 555–561 (2010). 44. International Resources Group (IRG), “Institutional Support and Strengthen- 27. P. Gleick, M. Palaniappan, Peak water limits to freshwater withdrawal and use. Proc. ing Program (ISSP) national strategic wastewater master plan” (Technical Natl. Acad. Sci. U.S.A. 107, 11155–11162 (2010). report, US Agency for International Development (USAID), Washington, DC, 28. P. Gleick, Transitions to freshwater sustainability. Proc. Natl. Acad. Sci. U.S.A. 115, 2013). 8863–8871 (2018). 45. J. Doorenbos, A. Kassam, “Yield response to water” (FAO Irrigation and Drainage 29. United Nations General Assembly, “Transforming our world: The 2030 agenda for Paper 33, Food and Agricultural Organization of the United Nations, Rome, Italy, sustainable development” (Technical Report A/RES/70/1, United Nations, New York, 1986). NY, 2015). 46. X. Cai, D. Mckinney, L. Lasdon, Integrated hydrologic-agronomic-economic model for 30. T. Wagener et al., The future of hydrology: An evolving science for a changing world. river basin management. J. Water Resour. Plann. Manag. 129, 4–17 (2003). Water Resour. Res. 46, W05301 (2010). 47. V. Srinivasan, S. Gorelick, L. Goulder, Sustainable urban water supply in south India: 31. R. Vogel, Hydromorphology. J. Water Resour. Plann. Manag. 137, 147–149 Desalination, efficiency improvement, or rainwater harvesting? Water Resour. Res. (2011). 46, W10504 (2010). 32. V. Srinivasan, E. Lambin, S. Gorelick, B. Thompson, S. Rozelle, The nature and causes 48. C. Klassert, K. Sigel, B. Klauer, E. Gawel, Increasing block tariffs in an arid developing of the global water crisis: Syndromes from a meta-analysis of coupled human-water country: A discrete/continuous choice model of residential water demand in Jordan. studies. Water Resour. Res. 48, W10516 (2012). Water 10, 248 (2018). 33. M. Sivapalan, H. Savenije, G. Bloschl,¨ Socio-hydrology, A new science of people and 49. T. Heckelei, W. Britz, Y. Zhang, Positive mathematical programming approaches— water. Hydrol. Process. 26, 1270–1276 (2012). Recent developments in literature and applied modelling. Bio Base Appl. Econ. 1, 34. S. Thompson et al., Developing predictive insight into changing water systems: Use- 109–124 (2012). inspired hydrologic science for the Anthropocene. Hydrol. Earth Syst. Sci. 17, 5013– 50. M. Florian, M. Los, A new look at static spatial price equilibrium models. Reg. Sci. 5039 (2013). Urban Econ. 12, 579–597 (1982).

12 of 12 | PNAS Yoon et al. https://doi.org/10.1073/pnas.2020431118 A coupled human–natural system analysis of freshwater security under climate and population change Downloaded by guest on September 26, 2021