Water Security 1 (2017) 12–20

Contents lists available at ScienceDirect

Water Security

journal homepage: www.elsevier .com/ locate/wasec

A dynamic framework for water security

http://dx.doi.org/10.1016/j.wasec.2017.03.0012468-3124/� 2017 Elsevier B.V. All rights reserved.

⇑ Corresponding author.E-mail address: veena.srinivasan@gmail.com (V. Srinivasan).

V. Srinivasan a,⇑, M. Konar b, M. Sivapalan b,c

aCentre for Environment and Development, Ashoka Trust for Research in Ecology and the Environment, Bangalore, KA, IndiabDepartment of Civil and Environmental Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, USAcDepartment of Geography and Geographic Information Science, University of Illinois at Urbana-Champaign, Champaign, IL, USA

a r t i c l e i n f o a b s t r a c t

Article history:Received 20 December 2016Accepted 2 March 2017Available online 23 March 2017

Keywords:Water securityHuman-water systemsSocio-hydrologyRegime shiftsDynamic frameworkIWRMSustainable development goals

Water security is a multi-faceted problem, going beyond mere balancing of supply and demand. Earlyattempts to quantify water security relied on static index based approaches that failed to acknowledgethat human action is intrinsic to the water cycle.Human adaptation to environmental change and increasing spatial specialization in the modern world

necessitate a more flexible and dynamic view of water security.Starting from first principles, and through application of simple water balance concepts to human-

impacted water systems, we first develop a set of indicators for water insecurity. We then offer anapproach to model these indicators as outcomes of coupled human-water systems to anticipate water-shed trajectories under human impacts, predict water insecurity and inform appropriate action. In thisway, far from being a static index, water security signifies a ‘‘safe operating subspace” within a threedimensional space that maps physical resource availability, infrastructure and economic choices.

� 2017 Elsevier B.V. All rights reserved.

1. Defining water security

Water security poses one of the biggest challenges of the 21stcentury. As populations grow, humanity faces the prospect ofuncertain future water supplies both due to climate change andincreasing demands on water [52]. On the one hand, too many peo-ple still do not have access to water to meet basic hygiene andlivelihood needs. On the other hand, as human demands for waterincrease and ecosystem services become more valued, there isincreasing competition for scarce water resources; not onlybetween humans and the environment, but also between differentsectors of the economy and different sections of society [24]. Theconcept of water security has arisen in response to this multi-faceted nature of the global water crisis.

The core idea underpinning water security is the need to bal-ance human and environmental water needs. Indeed, the UnitedNations Sustainable Development Goals for water (SDG6) explicitlysets targets for each. The problem is that there is no single workingdefinition of water security; there are many [9] and the framingsveer between ‘‘broad” versus ‘‘narrow” and ‘‘academic” versus‘‘applied” [2].

This makes the field of water security an evolving one. Thelaunch of a new journal on Water Security is timely and presents

an opportunity to reflect on how water security might be betterquantified and measured. Moreover, no matter which definitionwe pick, at the core of the endeavor is a simple question. Whatmakes some places (by whatever definition) ‘‘water secure”, whileothers suffer from water insecurity? Is it environmental determin-ism; some places are hydrologically well endowed while others arenot? Is it cultural choices; some cultures have learnt to use theavailable water resources wisely? Is it economic choices; someeconomies are dependent on ecotourism and therefore on a pris-tine environment, while others are agricultural ‘‘breadbaskets”and are therefore inherently more water intensive? Is it infrastruc-ture; some places have engineered their way out of scarcity, whileothers have not? Or is it governance; some places are able toenforce controls on abstractions, whereas the absence of enforce-able limits, or inability to enforce them, causes other regions tooverexploit water resources?

We briefly review the literature on water security with a viewto understanding the framing of the problem. We then presentan approach to water security that can address the multi-facetednature of the challenge. In particular, we make the case that watersecurity is a dynamic concept and must be defined, measured andunderstood as such. The review of water security presented herehas many parallels with a companion review paper in this inaugu-ral issue of the Water Security journal that similarly proposes adynamic framework for the analysis and prediction of flood risk[3].

V. Srinivasan et al. /Water Security 1 (2017) 12–20 13

2. Water security as an index, a snapshot at a place and point intime

The conventional approach to water security as conceptualizedand implemented by international agencies (e.g., United NationalsDevelopment Program, UNDP; Global Water Partnership, GWP) isto set standards on how much water needs to be available to meethealth, livelihood and ecosystem needs, come up with appropriateindices, and then track progress against them. This approach buildson a long history of developing indices to track imbalancesbetween supply and demand. While the term ‘‘water security”itself has gained prominence only recently, the concept of measur-ing and quantifying risks/vulnerability/stress in water resourceshas been around for a long time. Early definitions of water securitywere human-centric. Several excellent reviews on measures andindices for water already exist [5,7,31].

The Falkenmark water stress index, defined as the per capitaannual renewable freshwater available, was an early attempt tounderstand the relationship between human needs and environ-mental constraints; regions with less than 1000 m3 per capita peryear were defined as ‘‘water scarce” [12]. The Falkenmark Indexcaptured hydrological constraints on water supply but demandwas based on what humans need rather than what they actuallywithdraw. Another index called the Water Resources VulnerabilityIndex, defined as the ratio of total annual withdrawals to availablewater resources, accounted for withdrawals. A country is consid-ered severely water scarce if the ratio of withdrawals to availablewater exceeds 40% [31].

The above approaches were critiqued for only accounting for‘‘physical water scarcity”; for example, they did not account forinadequacies in water infrastructure. By these measures, a countryin sub-Saharan Africa, which is entirely populated by rain-fed cul-tivators and herders with renewable water resources above1700 m3 per capita per year (in the form of green water) wouldnot be considered water insecure, even if, due to the absence ofany water infrastructure, the population barely had access to30 L (of blue water) per capita per day. Indeed, typically in theworld’s poorest countries, due to lack of infrastructure, populationsare at risk because they have very little access and control overwater.

To account for the key role of infrastructure, InternationalWater Management Institute (IWMI) introduced the concept of‘‘economic water scarcity” to identify countries, where less than25% of water from rivers was withdrawn for human purposes,and where significant improvements in water infrastructure areneeded to meet human needs [36]. The ‘‘Water Poverty” index[45] is another similar attempt to account for the role ofinfrastructure.

In most of these early indices, the emphasis was on developmentof water resources to meet human needs. As it became clear thatindiscriminate development was destroying critical ecosystem ser-vices and biodiversity, the need to account for ecosystems, andhence a new definition of water security, emerged. One populardefinition by the Global Water Partnership defines water securityas follows: ‘‘Water security, at any level from the household to theglobal, means that every person has access to enough safe water ataffordable cost to lead a clean, healthy and productive life, whileensuring that the natural environment is protected and enhanced”[17].

Operationalizing this definition necessitates a departure fromthe conventional maps of ‘‘place based indices” because theimpacts of human water uses on the environment are often dis-placed in space and time. Frequently, it is dams or cities upstreamthat cause the drying of wetlands or pollution downstream. A sem-inal paper on water security by Vörösmarty et al. [51] addressed

this challenge. The study used river networks to redistributehuman and ecological stressors from headwaters to the ocean,exploiting the spatial connectedness of river systems across theworld [51]. The results showed how rising ambient loads duringearly-to-middle stages of economic growth were followed by envi-ronmental controls and decreased threats to ecosystems as soci-eties became wealthy.

Despite the increasingly nuanced handling of human and envi-ronmental water needs many definitions and attempts at quantifi-cation have been inadequate in several ways. First, most of theseindices focus on specific spatial scales and largely ignore cross-scale feedbacks across people, groundwater and surface water.Researchers and practitioners work at different scales ranging fromthe household to community to basin to nation. Because water is amobile, common pool resource, achieving water security at onescale may jeopardize water security at a different scale. A sizeablebody of literature provides evidence for these cross-scale feedbacksand trade-offs. One study of urban water resilience in Chennai [44]showed that as urban households invested in wells, they becomeindividually more water secure in the short-term; but the regionas a whole became less resilient and more water insecure in themedium to long term. The irrigation-efficiency paradox is anotherexample [30,35,38,49]. Investments in drip irrigation improveincomes and allow farmers to expand irrigation at the farm scale,but often make the whole less resilient. Similarly, many communi-ties in India and Africa have constructed soil and water conserva-tion structures to boost artificial recharge. While these have beendemonstrated to improve agricultural productivity and incomeslocally, they have been shown to reduce flows into downstreamreservoirs/lakes during dry years, thus making communities down-stream more water insecure [16].

Second, traditional approaches to quantification of water secu-rity fail to account for the increasing spatial specialization in themodern world; what humans consume and what they produce areincreasingly unrelated. A one-size fits all approach does not workany longer. Most previous definitions of water security were basedon an implicit but flawed assumption that populations everywheredepend on local water resources to satisfy their needs for food,energy and other goods and services. But in the modern world,most humans do not depend on local resources. The very idea thatevery watershed or political unit should have a certain water avail-ability per capita is anachronistic, but still seems to persist in theliterature. Previous approaches were also fundamentally flawedin attributing the entire water footprint of the population to thewatershed or basin in which they resided. In reality, most waterintensive crops such as rice and sugarcane are grown in a handfulof canal command areas.

Production and consumption regions are geographically distinct[19,22]. People living in urban, industrialized watersheds or touristdestinations satisfy most of their material needs from sourcesexternal to the watershed. i.e., by importing food from elsewhere.On the one hand, virtual water transfers offset deficits in locallyavailable water resources [1]; on the other, they allow people indifferent regions to live healthy, productive lives with very differ-ent endowments of water. So how do we define, measure, compareand track water security in places with diverse economies? Differ-ent regions and different groups within those regions have dramat-ically different needs and priorities for water. Moreover, water isonly one factor of production and rarely the primary determinantof economic activity. The latter is driven by a range of factors suchas the comparative advantage of land, labor, energy, governmentpolicy or simple accidents of history [50]. And while the availabil-ity of water may or may not constrain economic activity, the nat-ure of economic activity certainly influences how much water isabstracted and how much is left to the environment.

14 V. Srinivasan et al. /Water Security 1 (2017) 12–20

Third, most indices are snapshots in time; they have no sense ofhistory and offer no predictive insight. Typically, the goal of develop-ing water security indices at a global scale has been to identifywhich regions need intervention but not what those interventionsshould be. The ‘‘so what” question remains the biggest problemwith most global assessments of water security. Once we knowthat a place is water insecure, how do we go about addressingit? The view of water security as a static snapshot of system stateat a particular place and time has been long critiqued by scholars,who argue that it does not capture the complexity of real-worldsystems and the different types of water use [34]. Moreover, it doesnot acknowledge that human action is intrinsic to the water cycle[52]. Social, institutional, engineering and economic infrastructurecannot simply be overlaid on natural processes to evaluate wateravailability for human uses; rather they are a major driver ofchange in water resources systems [23]. The cross-dependenciesor coupling between human and natural systems often leads to‘‘emergent” patterns of form, function, and dynamics, sometimesresulting in surprises that could not have been predicted fromthe knowledge of the behavior of individual elements [23].

The GWP definition [17] may seem inadequate to some politicalscientists who interpret water security differently, emphasizingthe ‘‘securitization” dimension [9]. The definition also misses dis-tributional aspects; one group’s water security may be achievedat the cost of another’s. It is impossible to identify one approachthat would encompass all meanings attributed to water security.We argue, however, that the definition is nonetheless useful anddespite limitations, the dynamic approach proposed in this papercan address the critiques discussed above.

3. Water security as an outcome of a coupled human-watersystem

To address critiques of static indices and account for both theinterconnectedness between human and water systems and thevast number of pathways by which these systems co-evolve, wesuggest framing water security as an outcome of a coupledhuman-water systems. We present this dynamic approach towater security presented in three steps. In Section 3.1, we presenta set of indicators that relate human water use to water availabil-ity. In Section 3.2, we present how these indicators map to differ-ent types of water insecurity. Finally, in Section 3.3, we present aset of causal factors that drive both human water use and wateravailability, so that these indicators can be modeled in the contextof coupled human-water systems.

3.1. Water utilization patterns as outcomes of coupled human-watersystems

We first need to understand how humans use water and howthis relates to what is available. It is surprisingly difficult to initiatea discussion on ‘‘water use” because the different literatures wecite here employ the exact same terms in completely different,even contradictory ways. So first some clarification of terms isnecessary.

In the traditional water resources literature, water use is a gen-eric term that can refer to either ‘‘consumptive” and ‘‘withdrawal”uses [15]. Consumptive water uses refer to uses that remove waterresources from the system i.e., the water is no longer locally avail-able (e.g., irrigation of crops and evaporation from industrial cool-ing are largely consumptive). ‘‘Withdrawals” refer to the waterabstracted from a water body such as a dam or river, but a largefraction may be returned to that system as return flows (e.g.,hydropower plants have high withdrawals, but most of the wateris returned back). Traditionally, water use has been assumed to

refer to blue water only, i.e., water that is abstracted from aquifersand streams [13].

Because water security is concerned with how humans meettheir needs and how much water is left in a watershed for naturalecosystems, we need to broaden the definition of water use toinclude green and grey water use [34]. To do this, we have adoptedterms from the water footprint literature. The ‘‘ConsumptionWater Footprint” (CW) refers to the volume of water embodiedin the goods and services consumed by people living in a region[26]. The goods and services consumed may be made or grownlocally or imported from outside the region. In contrast, the ‘‘Pro-duction Water Footprint” (PW) refers to the volume of water usedto produce goods and services (both material and aesthetic) by peo-ple within the region [26]; the commodities produced may be con-sumed locally or exported out of the watershed. Both terms includeblue, green and grey water use minus any return flows [13].

In our framework, all available physical water is partitionedinto direct and indirect human benefits (Fig. 1). The total physicalwater available in a watershed is precipitation plus any importsfrom outside the watershed. All physical water entering the water-shed must end up somewhere. It may flow out as surface flow (Q).It may be used in agriculture for the production of food, fiber, fod-der or fuel (A). It may also be used for non-agricultural purposessuch as urban lawns or swimming pools or thermal power plantsor industries (U). The sum of A and U is the ProductionWater Foot-print (PW) for the watershed. Additionally, water may be lostthrough ET occurring via natural green spaces (N), which provideindirect cultural, regulating or provisioning services. Because wewant to link water security to water governance, in this paper,the partitioning of uses into A, U or N is based on the how thewater is sourced and managed. Thus, evapotranspiration from anurban garden that is irrigated using piped utility water is ‘‘U”,while evapotranspiration from an irrigated corn crop is ‘‘A”. ETfrom an irrigated urban park is ‘‘U”, while ET from a woodedstretch in a city that’s not actively managed natural is ‘‘N”.

In pre-historic times, all of the available annual renewablewater resource ended up as discharge or evapotranspiration vianatural vegetation; very little was captured by humans. As soci-eties developed and became interconnected, production and con-sumption become increasingly disconnected. Thus, while PW andCW coincide in primitive societies, they tend to diverge in modernsocieties.

Fig. 1-i represents a water utilization pattern in a hypotheticalhunter-gatherer community. In such a landscape, humans do notalter watershed processes much and human water footprints tendto be very small. As societies clear forests and transition to settledagriculture, a small portion of the rainfall endowment gets allo-cated to rain-fed agriculture. Subsistence agriculture communitiesare on average just able to meet their own needs. So their waterfootprint equals the water used to produce goods and services inthe watershed (Fig. 1-ii). While the populations living in theseregions have sufficient food and water in average rainfall years,during drought years, because they are dependent on localresources, consumption would have to drop sharply, leading tofamine.

As societies invest a portion of their wealth in infrastructure,they are able to ‘‘capture” a larger and larger fraction of rainfalland also import water. People increase both their production andconsumption water footprints and become food self-sufficient(Fig. 1-iii). If infrastructure expansion continues, they may eventu-ally begin to generate food surpluses. At this point, PW will beginto exceed CW. If the community continues to invest in infrastruc-ture so that more and more of the rainfall endowment is allocatedto production, then stream flows to the ocean will decline to suchlevels that wetlands and other aquatic ecosystems could besignificantly impacted (Fig. 1-iv). This may also occur through

Fig. 1. Apportionment of available physical water resources across humans and nature. P is precipitation, I refers to physical water entering the watershed from outside eithernaturally or via infrastructure. A refers to water use in agriculture, U refers to water use for non-agriculture uses, Q is flow out of the basin, and N is water used by naturalecosystems. Production Water Footprint (PW), shown as the red box, is the total water abstracted from the watershed for purposes directly beneficial to humans, whileConsumption Water Footprint (CW) use, shown as a grey dotted line, refers to the total water embodied in products consumed by people living in the watershed. (i): Hunter-Gatherers, (ii): Subsistence Agriculture, (iii): Self-sufficiency, (iv): Ecological Destruction, (v): Groundwater Depletion, (vi): A large city with inter-basin imports,(vii)Agricultural Breadbasket, (viii): Tourism dependent economy like a national park. Note: the left-most bar represents the main inputs to the system consisting of precipitationand imports.

V. Srinivasan et al. /Water Security 1 (2017) 12–20 15

groundwater development. Indiscriminate well drilling and theabsence of controls over groundwater extraction (by either licens-ing or pricing) allows farmers to increase water use beyond theannual renewable resource by depleting groundwater (Fig. 1-v).If the CW for the watershed is far greater than the PW, it impliesthat the watershed is a net importer of virtual water. Physicalwater imports, I, allow a watershed to increase the ‘‘rainfall-pie”.I can be comprised of natural water inflows from upstream water-sheds or infrastructural transfers [22].

The three right most bars (Fig. 1-vi to -viii) represent patternsthat may be observed in developed economies with complete spa-tial specialization of economic activity. For instance, Fig. 1-videpicts an urban watershed, where piped water is imported intoa primarily urban watershed.

In Fig. 1, we have implicitly assumed the spatial unit to be awatershed only because it is easy to compute a water balance atthe scale of a watershed. However, there is no inherent reasonwhy the figure could not be applied to political units; it wouldmerely be harder to calculate the physical inflows and outflowsacross system boundaries. Moreover, Fig. 1 is scale independentbecause the axes are ratios. Overall, we would expect smallerwatersheds to be more homogeneous (all urban or all agricultural),while large river basins would have a mix of all water use types.

Boundary issues, however, do challenge our conception of watersecurity, due to the additional challenge of the need to understandboth local and non-local sources of water. To really understandwater security, we would need to define the system boundary asthe spatial domain of all local and non-local water receipts. Forexample, recent research examined local and non-local waterresources dependencies for Flagstaff, Arizona in the United States[33] and the United Kingdom [18]. However, it is not clear howthe quantification of water security would change if a larger orsmaller urban area or watershed were to be considered. In fact, dif-ferent spatial units (i.e., watershed, county, city, country, etc.)

would lead to different understandings of water security, sincenot only do local water resources change with each spatial resolu-tion, but so does reliance on virtual water imports. An importantarea of future research is to evaluate the spatial scaling propertiesof water security. We leave this for future work.

3.2. Linking water utilization patterns to a typology of water insecurity

The different water utilization patterns in Fig. 1 result in differ-ent types of water insecurity. For instance, in Fig. 1-ii, the mainproblem is sufficiency; users do not have access to sufficient quan-tities of (safe, affordable) water to live a healthy, productive life. InFig. 1-iv, the problem is often that over-allocation of resourcesmakes the system vulnerable to drought. Since a large fraction ofthe water is allocated to productive water uses, there will inevita-bly be conflict between human and ecosystem needs duringdrought years. In Fig. 1-v, the problem is long-term unsustainabil-ity, because the productive water uses exceed what is available.Depletion of groundwater means that future generations will notbe as well off as the current one.

Thus a composite view of water insecurity requires us to con-sider all the three components presented in Fig. 1: water resourceutilization (the red box), human demand or consumption (the greydotted line) and human development (the gradient from left toright).

a) The fraction of the available resource is being utilized or thewater resources utilization intensity [14,45]. Water used byhumans to produce food and other goods and services(PW) as a fraction of available water resources (P + I). Againboth PW and P + I are expressed in volumetric terms so thatthe ratio PW/(P + I) has no unit. The ratio represents the frac-tion of the available water resources endowment that isbeing utilized by humans, i.e., it captures the resource con-

16 V. Srinivasan et al. /Water Security 1 (2017) 12–20

straint. As PW/(P + I) creeps closer to 1, the amount of waterleft for nature, both evapotranspiration through natural veg-etation or ecological flows, decreases. It is even possible forPW to exceed P + I in cases of severe groundwater depletion.

b) The capacity of populations to invest in technology to store,distribute and access the water or purchase food and othercommodities on the global market. Places may either investin their own infrastructure or may obtain a higher fraction oftheir agricultural water requirements from external sources,an implicit ‘‘infrastructure sharing” across borders [21]. Weuse GDP/capita as a proxy to capture this ability of societiesto invest in ‘‘trade and technology” denoted as TT.

c) Spatial heterogeneity in production and consumption is cap-tured by the ratio of the production water (PW) and theirconsumption water (CW) footprints. This ratio is inspiredby the internal and external water concept proposed byHoekstra and Mekonnen [19]. However, we suggest a ratio(rather than the difference in [19] to make it scale indepen-dent. Both PW and CW are expressed in volumetric terms sothat the ratio has no unit. We would expect this ratio, PW/CW, to be very different depending on the dominant eco-nomic activities in the region. In an agricultural or water-intensive industrial belt, PW/CW� 1, whereas in an urbanarea with mostly service sector jobs, we would expect PW/CW� 1.

When these three variables are plotted against each other(Fig. 2), we can map out certain spaces (the blue hashed areas)as water insecure; if a watershed has too high a CW but does notproduce enough and has too little money to pay for water infras-tructure or buy food on grain markets, it is likely to be water inse-cure. A community living in a watershed may alleviate physicalwater stress defined as PW/(P + I), by decreasing human abstrac-tion of water or increasing imported water I. As societies becomewealthier (TT increases), they initially increase human waterabstraction to meet their growing needs and even generate foodsurpluses. Over time, they may alter the nature of their economyaltogether by importing more virtual water (a) or either continueto abstract water to the detriment of the environment (b).

3.3. Developing a causal framework to model water insecurity

The idea of a new framework that includes these multiple facetsto understand water security is itself not novel. Indeed, severalauthors have compared them [29] or combined them to developa composite index [45]. The main contribution we offer in thispaper is a dynamic view of water security. Once we understand

Fig. 2. Mapping resource utilization patterns to water insecurity. A given watershed maymarked a and b present two hypothetical trajectories a watershed may follow over time.consumption and production water footprints respectively, and TT is investment in ‘‘tra

what drives societies to utilize water resources in a particularway, we can model and anticipate their occurrence. The goal in thispaper is not yet another index of water security that is based onpublished data, but rather coupled human-water systems modelsthat can generate the indices and trajectories as emergentoutcomes.

But how can we ensure that the models can throw out indica-tors of water insecurity?

Fortunately, previous empirical meta-analyses have found thatpatterns observed in real world case studies could be explainedby just four sets of factors [29,42]:

(i) Culture/ Economy: What societies need water for.(ii) Resource: How much water is available; i.e., the environ-

mental constraint.(iii) Wealth: Whether the society has the capacity to invest in

technology or buy goods (with virtual water embedded).(iv) Governance: Whether there are rules to decide who is

allowed to take out how much water, for what purpose.

For instance, long-term decline in groundwater levels has beenassociated with the presence of a productive aquifer, increasingdemand for human needs, ineffective controls over water use andthe absence of reallocation mechanisms. In developed countries anincreasingly environmentally conscious population is motivated tomeet increasing demand for water in urban areas by decreases inwater use by agriculture (by either land fallowing or a switch toless water-intensive crops). In general, this outcome is observedin regions where there are effective controls over both waterabstraction and environmental protection [42].

Increasingly, coupled human-water systems models are begin-ning to use agent-based models to simulate two-way feedbacksbetween the human and water components of systems. Suchmodels may be able to capture the relationships between gover-nance systems, agent behavior, and resource availability to pre-dict water insecurity. The challenge is that, typically, modelshold agent motivations and the governance or operational rulesconstant over the modeling period. Although a few models havebegun to allow rules to change endogenously [40] these still seemto be operational level changes rather than fundamental constitu-tional changes. In the short to medium term, this is a very reason-able assumption. But what if the socio-political system itself andthe corresponding set of operational and constitutional rulesthemselves change over longer time frames? Models maybe ableto predict historical water insecurity, but not how watershedsbecome less water insecure over time. This requires a more flex-ible modeling approach.

follow one of several paths in the 3-D space defined by the three axes. The red linesThe blue hashed sub-spaces denote water insecure sub-spaces. CW and PW are thede and technology.”

V. Srinivasan et al. /Water Security 1 (2017) 12–20 17

4. Water security as predicted by regime shifts

In recent years, there has been increasing recognition that effec-tive management of water over long periods of time requires anunderstanding of the coevolving dynamics and feedbacks inherentin coupled human-water systems [39,28,46,37]. Place-based‘‘socio-hydrology” studies have begun to study and characterizethe workings of coupled systems over long time frames[20,48,25,41,10,8].

Many case studies from the across the world have identified‘‘regime shifts” arising as a direct result of anthropogenic influences[39,11]. A common theme in these studies is that human responsesare highly nonlinear. People move, lifestyles change, and valuesand norms change. Societies respond to the issues/concerns ofthe day by building infrastructure and trade, constrained by thelegacy effects of their previous decisions.

To understand the concept of regime shift, we need to synthe-size data from many watersheds across the globe. If we plot thetrajectories of these watersheds over time, we should begin tosee some common transitions. Watersheds should follow pre-dictable water resource evolution pathways. They should divergewhen there are differences in macro-economic, governance orinfrastructure decisions.

In order to chart common trajectories, we present a frameworkinspired by the Budyko curve [6], which has revolutionized catch-ment hydrology. It provided a unifying framework for a field thathad previously been fragmented and empirical. However, theframework is only applicable to relatively pristine catchmentsand there has been no comparable framework for human

Fig. 3. Mapping regime shifts in socio-hydrologic systems. The axes represent two ofrepresent the other factors: Technology and Governance.

dominated socio-hydrologic systems. Where the Budyko curveencapsulated the climate (precipitation and radiation) controlson annual water balance of catchments, we present a frameworkthat encapsulates controls on human water security (Fig. 3).

Fig. 3 hypothesizes possible pathways that watersheds may fol-low with regime shifts. It uses the same axes as Fig. 2 but the GDP/-capita (TT) is not displayed for ease of visualization. Instead,increasing wealth is shown with increasingly darker colors. InFig. 3, primitive societies would lie more or less along the PW = CWline; i.e., all production and consumption is local. As societiesbecome wealthier, their economies specialize; some become vir-tual water consumers or importers and others become exporters.If they over-invest in infrastructure then agricultural or industrialproduction could go up, but this would occur at the cost of theenvironment.

Regime shifts occur because of changes in the cultural and polit-ical context of a watershed. In the context of socio-politicalchanges, several factors determining regime shift have been identi-fied, such as community sensitivity, memory of disasters, socialcohesion and technology. For instance, Elshafei et al. [10] character-ize and define ‘‘community sensitivity” to protecting the environ-ment, which fluctuated as the environment or livelihood varied,and prompted community action (through a behavioral responsevariable). van Emmerik et al. [48] estimated ‘‘environmental aware-ness” as a cumulative measure that reflected community sentimentin the Murrumbidgee Basin in Australia as the environmentdegraded and prompted community action to restore the environ-ment. Similarly, a study of the Kissimmee River Basin in Floridaindicated that flood intensity decreased after channelization, which

the causal drivers of water security: Culture/Economy and Resource. The arrows

18 V. Srinivasan et al. /Water Security 1 (2017) 12–20

reduced concerns about flooding [8]. But channelization led to adecrease in wetland storage; as people became concerned aboutecosystem declines, priorities switched back from favoring floodprotection to favoring wetland conservation. Subsequently, man-agement strategies too followed suit. In other words, we need tounderstand what drives societies to shift regimes in the long term.

5. Way forward: Towards a dynamic framework for watersecurity

The perspectives on water security presented in previous sec-tions capture the tension between uniformity and simplicity inchoosing quantitative descriptions of water security on one handand differentiation over space and time and complexity on theother. But in reality they merely capture coupled human-watersystem dynamics at different time-scales (Fig. 4).

Water security as a snapshot at a particular place and point intime has the advantage of being simple and easily scalable, butdoes not really consider feedbacks between human and water sys-tems and does not help inform action. A snapshot in time does nothave any predictive or explanatory power and merely provides arepresentation of the system state over a short period for time(say 1–5 years) over which agencies and governments managethe system.

Water security as an outcome of coupled human-natural sys-tem models accounts for human adaptability to external drivers.Under these circumstances, models can indicate how the coupledhuman-water system might evolve for a given governance, cultureand infrastructure. For example, agent-based models that holdhuman values, norms and beliefs as fixed, coupled with hydrolog-ical models, can be used to make predictions about water securityin the medium term (say 5–25 years), the typical planning horizonfor water agencies.

Over much longer time horizons, the system might undergoregime shifts, which will require changes to model structure andgoverning equations of the coupled human-water system withsocietal changes in governance and culture. A society’s values, normsand beliefs themselves may evolve over time and these in turn mayprompt changes in policy, governance and infrastructure so that,over time, the society might switch to an entirely different trajec-tory. Such regime shifts cannot obviously be predicted in the con-ventional sense that hydrologists do predictions, but anticipatingthese regime shifts can improve predictive insight [43]. These inturn can form the basis of citizen engagement over much longer

Fig. 4. Different perspectives on water security over different timescales.

time frames (say 25–100 years), at which large-scale infrastructureis planned.

6. Implementation of the dynamic framework for watersecurity

The discussion so far has provided the theoretical underpin-nings to defining, classifying, measuring and predicting watersecurity. In this Section, we address implementation of these ideastowards achieving the UN Sustainable Developing Goal (SDG) tar-gets. The UN SDG 6, which deals with water aims to simultane-ously ‘‘achieve universal and equitable access to safe andaffordable drinking water for all”, ‘‘ensure sustainable withdrawalsand supply of freshwater to address water scarcity and substan-tially reduce the number of people suffering from water scarcity”,‘‘implement integrated water resources management at all levels”and ‘‘protect and restore water-related ecosystems, includingmountains, forests, wetlands, rivers, aquifers and lakes” [47].

Unlike the Millennium Development Goals, which focused onaccess to water and sanitation infrastructure, the SDGs are broadin their scope. Operationalizing these goals is likely to be challeng-ing because we have to make trade-offs. Ensuring access forhumans will inevitably have some impact on water-related ecosys-tems, and therefore we need a framework that can evaluate thesetrade-offs. Implementing the SDGs will entail measuring, trackingand funding goals for which we neither have clear metrics noractionable guidelines. An additional source of complexity arisesin that we do not live in a world with a single, omniscientdecision-maker.

Humans are constantly evolving and adapting in response toenvironmental changes and the cumulative action of the adaptiveactions of millions of humans can completely alter the course ofwater resources systems. In its previous incarnation, IntegratedWater Resources Management (IWRM) largely relied onphysically-based models of the water resources system in whichonly parameters such as population and infrastructure wereallowed to change over time. These models were very poorlyequipped to address adaptive actions by people. As a result, seem-ingly obvious solutions had unintended consequences, e.g., thedrip irrigation and virtual water paradoxes in Sivapalan et al.[38]. Moreover, IWRM implementations were stymied by the lackof a clear actionable framework [4,27], in part because the modelsartificially constrained human behavior to what could be modeled.

Any action towards delivering on the SDG 6goalsmust addressthe fundamental weaknesses of previous implementations ofIWRM. In this paper, we suggest that SDG6 goals essentiallyencompass all aspects of water security. We offer a framework todefine, classify, quantify and model water security over differenttime-scales. There are multiple ways in which societies may bewater insecure. Therefore, water security cannot be quantified bya single index. Rather water security should really be seen as a‘‘safe operating subspace” [32] within a three dimensional spacemapping physical resource availability, infrastructure and eco-nomic choices. Coupled human-water system models can then beused to anticipate future watershed trajectories, predict waterinsecurity and inform appropriate action.

Acknowledgements

This paper benefited from discussions with many colleagues inthe socio-hydrology community, reviewer Pieter van der Zaag andone anonymous reviewer. The work on the paper was developedwithin the framework of the Panta Rhei Research Initiative of theInternational Association of Hydrological Sciences (IAHS). Wewould like to acknowledge the US National Science Foundation’s

V. Srinivasan et al. /Water Security 1 (2017) 12–20 19

Socio-Environmental Synthesis Center (SESYNC; NSF award DBI-1052875) for their support of the project ‘‘Toward Socio-hydrologic Synthesis: Modeling the Co-evolutionary Dynamics ofCoupled Human, Water, and Ecological Systems”, which allowedseveral of us to meet and develop these ideas. Srinivasan acknowl-edged funding from IDRC, Canada (Grant No 107086-001) and theMinistry of Earth Sciences, Govt. of India under the Newton BhabhaFund. We are grateful to the SESYNC project participants for dis-cussions that have informed this paper. We thank Günter Blöschlfor critical feedback that helped sharpen our arguments. We are,however, responsible for any errors that remain.

References

[1] J.A. Allan, Virtual water: a strategic resource, Ground Water 36 (1998) 545–546, http://dx.doi.org/10.1111/j.1745-6584.1998.tb02825.x.

[2] K. Bakker, Water security: research challenges and opportunities, Science 337(6097) (2012) 914–915, http://dx.doi.org/10.1126/science.1226337.

[3] M. Barendrecht, A. Viglione, G. Blöschl, A dynamic framework for flood risk,Water Secur. 1 (2017) 12–20.

[4] A.K. Biswas, Integrated water resources management: is it working?, Int J.Water Resour. Dev. 24 (2008) 5–22, http://dx.doi.org/10.1080/07900620701871718.

[5] A. Brown, M.D. Matlock, A Review of Water Scarcity Indices andMethodologies, The Sustainability Consortium. White Paper #106, 2011.

[6] M.I. Budyko, Climate and Life, Academic Press, New York, 1974, p. 508.[7] H.M.L. Chaves, S. Alipaz, An integrated indicator based on basin hydrology,

environment, life, and policy: the watershed sustainability index, WaterResour. Manage 21 (2007) 883–895, http://dx.doi.org/10.1007/s11269-006-9107-2.

[8] X. Chen, D. Wang, F. Tian, M. Sivapalan, From channelization to restoration:sociohydrologic modeling with changing community preferences in theKissimmee River Basin, Florida, Water Resour. Res. 52 (2016) 1227–1244,http://dx.doi.org/10.1002/2015WR018194.

[9] C. Cook, K. Bakker, Water security: debating an emerging paradigm, Glob.Environ. Change. 22 (2012) 94–102, http://dx.doi.org/10.1016/j.gloenvcha.2011.10.011.

[10] Y. Elshafei, J.Z. Coletti, M. Sivapalan, M.R. Hipsey, A model of the socio-hydrologic dynamics in a semiarid catchment: isolating feedbacks in thecoupled human-hydrology system, Water Resour. Res. 51 (2015) 6442–6471,http://dx.doi.org/10.1002/2015WR017048.

[11] Y. Elshafei, M. Tonts, M. Sivapalan, M.R. Hipsey, Sensitivity of emergentsociohydrologic dynamics to internal system properties and externalsociopolitical factors: implications for water management, Water Resour.Res. 52 (2016) 4944–4966, http://dx.doi.org/10.1002/2015WR017944.

[12] M. Falkenmark, J. Lundqvist, C. Widstrand, Macro-scale water scarcity requiresmicro-scale approaches: aspects of vulnerability in semi-arid development,Nat. Resour. Forum 13 (1989) 258–267, http://dx.doi.org/10.1111/j.1477-8947.1989.tb00348.x.

[13] M. Falkenmark, J. Rockström, The new blue and green water paradigm:breaking new ground for water resources planning and management, J. WaterResour. Plan. Manage. 132 (2006) 129–132, http://dx.doi.org/10.1061/(ASCE)0733-9496(2006) 132:3(129).

[14] FAO, Population change-natural resources-environment linkages in the Arabstates region, Rome, 1996.

[15] P.H. Gleick, Water use, Ann. Rev. Env. Resour. 28 (2003) 275–314, http://dx.doi.org/10.1146/annurev.energy.28.040202.122849.

[16] C.J. Glendenning, F.F. Van Ogtrop, A.K. Mishra, R.W. Vervoort, Balancingwatershed and local scale impacts of rain water harvesting in India-A review,Agric. Water Manage. 107 (2012) 1–13, http://dx.doi.org/10.1016/j.agwat.2012.01.011.

[17] Global Water Partnership (GWP), Towards Water Security: A Framework forAction. Mobilising Political Will to Act Stockholm, Sweden, 2000.

[18] A.Y. Hoekstra, M.M. Mekonnen, Imported water risk: the case of the UK,Environ. Res. Lett. 11 (2016) 55002.

[19] A.Y. Hoekstra, M.M. Mekonnen, The water footprint of humanity, Proc. Natl.Acad. Sci. 109 (9) (2011) 3232–3237, http://dx.doi.org/10.1016/B978-0-12-799968-5.00007-5.

[20] J. Kandasamy, D. Sounthararajah, P. Sivabalan, A. Chanan, S. Vigneswaran, M.Sivapalan, Socio-hydrologic drivers of the pendulum swing betweenagricultural development and environmental health: a case study fromMurrumbidgee River basin, Australia, Hydrol. Earth Syst. Sci. 18 (2014)1027–1041, http://dx.doi.org/10.5194/hess-18-1027-2014.

[21] M. Konar, K.K. Caylor, Virtual water trade and development in Africa, Hydrol.Earth Syst. Sci. 17 (2013) 3969–3982, http://dx.doi.org/10.5194/hess-17-3969-2013.

[22] M. Konar, T.P. Evans, M. Levy, C.A. Scott, T.J. Troy, C.J. Vörösmarty, M. Sivapalan,Water resources sustainability in a globalizing world: who uses the water?,Hydrol Process. 30 (2016) 3330–3336, http://dx.doi.org/10.1002/hyp.10843.

[23] P. Kumar, Typology of hydrologic predictability, Water Resour. Res. 47 (2011)W00H05, http://dx.doi.org/10.1029/2010WR009769.

[24] M. Kurian, The water-energy-food nexus Trade-offs, thresholds andtransdisciplinary approaches to sustainable development, Env. Sci. Policy 68(2017) 97–106, http://dx.doi.org/10.1016/j.envsci.2016.11.006.

[25] D. Liu, F. Tian, M. Lin, M. Sivapalan, A conceptual socio-hydrological model ofthe co-evolution of humans and water: case study of the Tarim River basin,western China, Hydrol. Earth Syst. Sci. 19 (2015) 1035–1054.

[26] M.M. Mekonnen, A.Y. Hoekstra, The green, blue and grey water footprint offarm animals and animal products Value of Water Research Report Series No.48, UNESCO-IHE, Delft, The Netherlands, 2010.

[27] B. Mitchell, ‘‘Integrated water resources management: a reassessment” by AsitK. Biswas, Water Int. 29 (2004) 398–399, http://dx.doi.org/10.1080/02508060408691794.

[28] A. Montanari, G. Young, H.H.G. Savenije, D. Hughes, T. Wagener, L.L. Ren, D.Koutsoyiannis, C. Cudennec, E. Toth, S. Grimaldi, G. Blöschl, M. Sivapalan, K.Beven, H. Gupta, M. Hipsey, B. Schaefli, B. Arheimer, E. Boegh, S.J. Schymanski,G. Di Baldassarre, B. Yu, P. Hubert, Y. Huang, A. Schumann, D.A. Post, V.Srinivasan, C. Harman, S. Thompson, M. Rogger, A. Viglione, H. McMillan, G.Characklis, Z. Pang, V. Belyaev, ‘‘Panta Rhei—Everything Flows”: change inhydrology and society—The IAHS Scientific Decade 2013–2022, Hydrol. Sci. J.58 (2013) 1256–1275, http://dx.doi.org/10.1080/02626667.2013.809088.

[29] J.C. Padowski, S.M. Gorelick, B.H. Thompson, S. Rozelle, S. Fendorf, Assessmentof human–natural system characteristics influencing global freshwater supplyvulnerability, Environ. Res. Lett. 10 (2015) 104014, http://dx.doi.org/10.1088/1748-9326/10/10/104014.

[30] C. Perry, Efficient irrigation; inefficient communication; flawedrecommendations, Irrig. Drain. 56 (2007) 367–378.

[31] P. Raskin, P. Gleick, P. Kirshen, G. Pontius, K. Strzepek, ComprehensiveAssessment of the Freshwater Resources of the World. Water Futures:Assessment of Long-range Patterns and Problems Stockholm, Sweden, 1997.

[32] J. Rockström et al., A safe operating space for humanity, Nature 461 (2009)472–475, http://dx.doi.org/10.1109/TIP.2011.2165549.

[33] R.R. Rushforth, B.L. Ruddell, The vulnerability and resilience of a city’s waterfootprint: the case of Flagstaff, Arizona, USA, Water Resour. Res. 52 (2016)2698–2714, http://dx.doi.org/10.1002/2015WR018006.

[34] H.H.G. Savenije, Water scarcity indicators; the deception of the numbers, Phys.Chem. Earth, Part B: Hydrol. Oceans Atmos. 25 (2000) 199–204.

[35] C.A. Scott, S. Vicuña, I. Blanco-Gutiérrez, F. Meza, C. Varela-Ortega, Irrigationefficiency and water-policy implications for river basin resilience, Hydrol.Earth Syst. Sci. 18 (2014) 1339–1348, http://dx.doi.org/10.5194/hess-18-1339-2014.

[36] D. Seckler, R. Barker, U. Amarasinghe, Water scarcity in the twenty-firstcentury, Int. J. Water Resour. Dev. 15 (1999) 29–42, http://dx.doi.org/10.1080/07900629948916.

[37] M. Sivapalan, G. Blöschl, Time scale interactions and the coevolution ofhumans and water, Water Resour. Res. 51 (9) (2015) 6988–7022, http://dx.doi.org/10.1002/2015WR017896.

[38] M. Sivapalan, M. Konar, V. Srinivasan, A. Chhatre, A. Wutich, C.A. Scott, J.L.Wescoat, I. Rodríguez-Iturbe, Socio-hydrology: use-inspired watersustainability science for the Anthropocene, Earth’s Future 2 (2014) 225–230, http://dx.doi.org/10.1002/2013EF000164.

[39] M. Sivapalan, H.H.G. Savenije, G. Blöschl, Socio-hydrology: a new science ofpeople and water, Hydrol. Process. 26 (2012) 1270–1276, http://dx.doi.org/10.1002/hyp.8426.

[40] A. Smajgl, L.R. Izquierdo, M. Huigen, Modeling endogenous rule changes in aninstitutional context: the adico sequence, Adv. Complex Syst. 11 (2008) 199–215.

[41] V. Srinivasan, Reimagining the past – use of counterfactual trajectories insocio-hydrological modelling: the case of Chennai, India, Hydrol. Earth Syst.Sci. 19 (2015) 785–801, http://dx.doi.org/10.5194/hess-19-785-2015.

[42] V. Srinivasan, E.F. Lambin, S.M. Gorelick, B.H. Thompson, S. Rozelle, The natureand causes of the global water crisis: syndromes from a meta-analysis ofcoupled human-water studies, Water Resour. Res. 48 (2012) W10516, http://dx.doi.org/10.1029/2011WR011087.

[43] V. Srinivasan, M. Sanderson, M. Garcia, M. Konar, G. Blöschl, M. Sivapalan,Prediction in a socio-hydrological world, Hydrol. Sci. J. (2016), http://dx.doi.org/10.1080/02626667.2016.1253844, in press.

[44] V. Srinivasan, K.C. Seto, R. Emerson, S.M. Gorelick, The impact of urbanizationon water vulnerability: a coupled human-environment system approach forChennai, India, Glob. Environ. Change 23 (1) (2013) 229–239, http://dx.doi.org/10.1016/j.gloenvcha.2012.10.002.

[45] C. Sullivan, Calculating a water poverty index, World Dev. 30 (2002) 1195–1210, http://dx.doi.org/10.1016/S0305-750X(02)00035-9.

[46] S.E. Thompson, M. Sivapalan, C.J. Harman, V. Srinivasan, M.R. Hipsey, P. Reed,A. Montanari, G. Blöschl, Developing predictive insight into changing watersystems: use-inspired hydrologic science for the anthropocene, Hydrol. EarthSyst. Sci. 17 (2013) 5013–5039, http://dx.doi.org/10.5194/hess-17-5013-2013.

[47] UN Water, Water Security and the Global Water Agenda: A UN-WaterAnalytical Brief Hamilton, Ontario, Canada, 2013.

[48] T.H.M. van Emmerik, Z. Li, M. Sivapalan, S. Pande, J. Kandasamy, H.H.G.Savenije, A. Chanan, S. Vigneswaran, Socio-hydrologic modeling to understandand mediate the competition for water between agriculture development andenvironmental health: murrumbidgee River basin, Australia, Hydrol. EarthSyst. Sci. 18 (2014) 4239–4259, http://dx.doi.org/10.5194/hess-18-4239-2014.

[49] G.E. van Halsema, L. Vincent, Efficiency and productivity terms for watermanagement: a matter of contextual relativism versus general absolutism,Agric. Water Manage. 108 (2012) 9–15.

20 V. Srinivasan et al. /Water Security 1 (2017) 12–20

[50] S. Verma, D.A. Kampman, P. van der Zaag, A.Y. Hoekstra, Going against theflow: a critical analysis of inter-state virtual water trade in the context ofIndia’s National River Linking Program. Phys, Chem. Earth, Parts A/B/C 34(2009) 261–269.

[51] C.J. Vörösmarty, P.B. McIntyre, M.O. Gessner, D. Dudgeon, A. Prusevich, P.Green, S. Glidden, S.E. Bunn, C.A. Sullivan, C.R. Liermann, P.M. Davies, Global

threats to human water security and river biodiversity, Nature 467 (2010)555–561, http://dx.doi.org/10.1038/nature09549.

[52] T. Wagener, M. Sivapalan, P.A. Troch, B.L. McGlynn, C.J. Harman, H.V. Gupta, P.Kumar, P.S.C. Rao, N.B. Basu, J.S. Wilson, The future of hydrology: an evolvingscience for a changing world, Water Resour. Res. 46 (5) (2010) W05301, http://dx.doi.org/10.1029/2009WR008906.