A history of futures: A review of scenario use in water policy studies in the Netherlands

(1)

A

history

of

futures:

A

review

of

scenario

use

in

water

policy

studies

in

the

Netherlands

M. Haasnoot

a,b,c,

*

,

H. Middelkoop

b

a_Deltares,_P.O._Box_177,₂₆₀₀_MH_Delft,_The_Netherlands

b_Utrecht_University,_Department_of_Physical_Geography,_P.O._Box_80115,₃₅₀₈_TC_Utrecht,_The_Netherlands

c_University_of_Twente,_Department_of_Water_Engineering_and_Management,_P.O._Box_217,₇₅₀₀_AE_Enschede,_The_Netherlands

1. Introduction

Theworld’s river deltas areincreasingly vulnerabledue to

pressures fromclimate change, relative sea level rise and

populationgrowth (Syvitski et al., 2009;Vo¨ro¨smarty, 2009).

Therefore,denselypopulateddeltassuchastheNetherlands

requirewell-designedwatermanagementforfloodprotection

andforcopingwithvaryingwaterdemandsandavailability.

Watermanagementdecisionsshouldbringsolutionsthat

willsustainforseveraldecades,implyingthattheyshouldbe

adequate even in case of changes in pressures. However,

uncertainties about the future make decisionmaking less

straightforward. Therefore, policymakers increasingly use

robustnessasindicatorindecisionmaking.Arobuststrategy

performsrelativelywellacrosswiderangeofpossiblefutures

(Lempertetal.,2006)andotheruncertainties.Water

manage-mentfacesuncertaintiesarisingfrom(1)naturaluncertainties

such as trends and extreme weather events; (2) social

uncertaintiesdueto shiftsinhuman response and values;

(3) technological uncertainties through modelling future

statesandimpact(e.g.Haasnootetal.,2011).

Scenarioanalysisisamethodfordealingwith

uncertain-ties, and aims to assess possible impacts and to design

policies (e.g. Carter et al., 2007). Scenarios are coherent

descriptionsofalternativehypotheticalfuturesthat reflect

differentperspectivesonpast,presentandfuture

develop-ments, whichcanserveasabasis foraction(VanNotten,

2005). Since itsfirst useinmilitary planning inthe 1950s

(Bradfieldetal.,2005;Brown,1968;KahnandWiener,1967),

scenarioanalysishasbeenappliedinavarietyofareas,such

as business development (Bradfield et al., 2005; Van der

Heijden,1996;Wack,1985),environmentalplanning(Alcamo, 2009, 2001; Peterson et al., 2003) and climate change

mitigation andadaptation (Hulme andDessai, 2008; IPCC,

2000;Rosentrater,2010;Wigleyetal.,1980).Scenarioshave

alsobeenusedforrobustdecisionmakingincaseofcomplex

problems with deepuncertainty,such as long-termwater

management underchangingconditions(e.g.Lempertand

Schlesinger, 2000;Dewaretal.,1993;LempertandBankes,

a

r

t

i

c

l

e

i

n

f

o

Publishedonline5April2012

Keywords: Scenarios Watermanagement Climatechange Uncertainty TheNetherlands

a

b

s

t

r

a

c

t

The futureof humanlife inthe world’sriverdeltas depends onthe successof water

management.Todealwithuncertaintiesaboutthefuture,policymakersintheNetherlands

haveusedscenariostodevelopwatermanagementstrategiesforthecoastalzoneofthe

Rhine–Meuse delta. In this paper we reflect on six decades of scenario use in the

Netherlands, and provide recommendations forfuture studies. Based on twocriteria,

‘Decisionrobustness’and‘Learningsuccess’,weconcludethat(1)thepossibilitiesforrobust

decisionmakingincreasedthroughaparadigmshiftfrompredictingtoexploringfutures,

butthescenariomethodisnotyetfullyexploitedfordecisionmakingunderuncertainty;and

(2)thescenariosenabledlearningaboutpossibleimpactsofdevelopmentsandeffectiveness

ofpolicyoptions.Newscenarioapproachesareemergingtodealwiththedeepuncertainties

watermanagersarecurrentlyfacing.

# ₂₀₁₂_Elsevier_Ltd._All_rights_reserved.

*Correspondingauthorat:Deltares,P.O.Box177,2600MHDelft,TheNetherlands.Tel.:þ318823358175.

E-mailaddress:marjolijn.haasnoot@deltares.nl(M.Haasnoot).

Available

online

at

www.sciencedirect.com

journalhomepage:www.elsevier.com/locate/envsci

1462-9011/$–seefrontmatter# ₂₀₁₂_Elsevier_Ltd._All_rights_reserved.

(2)

2003;Lempertetal.,2006;Groves,2006;Kwakkeletal.,2010

orMiddelkoop etal.,2004; VanAsselt andRotmans, 2002; Dessai and Hulme, 2007 for examples related to water management).

Toenablelifeinalow-lyingdelta,theDutchhavehada

longhistoryofcontrollingandmaintainingthewatersystem.

IntheNetherlands,scenarioshavebeenusedsincethe1950s

to prepare water management for the future. After six

decadesofexperience,wereflectonscenariouseinwater

management in the Netherlands, and identify possible

improvementsforfuturestudies. Thisevaluationprovides

moreinsightinpolicymakingonwatermanagementinriver

deltasunderuncertaintytosupportthecurrentdevelopment

of the next generation scenarios for climate adaptation

studies.

This paper provides a review of scenario use in water

management studies on the Rhine–Meuse delta in the

Netherlands,and evaluatesthe lessonsthatcanbederived

from this experience. We seek to answer the following

questions:Whatwastheevolvementofscenariouseinwater

management?Didthescenariosprovideprospectforrobust

decisionmaking?Didthescenariosenablelearningfor

policy-makersand/orscientists?Aftergivingahistoricalperspective,

weevaluatethescenariousebasedontwocriteria:‘Decision

robustness’ and ‘Learningsuccess’.We endthe paperwith

conclusionsandrecommendationsforfuturewater

manage-mentstudies.

2. Approach

for

evaluating

the

scenario

use

Forourchronologyonscenariouseinwatermanagementin

the Netherlands we reviewed all national water policy

documents,thekeyresearchstudiesonclimateandwater,

andrelatedclimatescenariostudies.Inaddition,weusedour

own experience, based on participation in several water

policystudiessincethe1990s,andtheexperienceofseveral

colleagues,whowereinvolvedinearlierwaterpolicystudies

orclimatescenariostudies.Wepresentthestudiesfromthe

Netherlands against the (inter)national context (see Fig.1

Fig.1–HistoricalperspectiveondevelopmentsinnationalwaterpolicydocumentsintheNetherlands,keyresearchstudies

onclimateandwater,climatescenariostudiesandthecontextinwhichthesestudiesweremade.PWM=NationalPolicy

(3)

for overview and supplementary information for more characteristics).

ForouranalysisweadoptedtwocriteriausedbyHulmeand

Dessai(2008b)inaframeworkforclimatescenarioevaluation,

whichwefurtherrefertoasthe‘Decisionrobustness’andthe

‘Learningsuccess’.

The‘Decisionrobustness’criterioncanbeaddressedwith

the following question: ‘do the scenarios contain a sufficient

representation of relevant knowable uncertainties to offer the

prospectthatdecisionstakenwithsupportofthescenarioswillbe

robust?’ Robustness is an important criterion for good

decisions under uncertainty (Rosenhead et al., 1972; Metz

et al.,2001),especially bypolicymakersfacing deep

uncer-tainty(Lempertet al., 2006; Grovesand Lempert, 2007). By

including uncertainties in decisionmaking it is possible to

identifystrategiesthatperformrelativelywellundervarious

differentpossiblefutures(robuststrategies),ortomakea

well-thought-outdecisiononwhetherornottoadaptastrategyin

viewofa specificuncertainty. Assessingthe robustness of

decisionsisrelevant,because decisionsinvolvelarge

high-costinvestments,andcanhavelargeimplicationsforsociety.

Therefore, water management decisions should be

cost-effectiveforseveraldecades,evenifthefutureturnsoutto

bedifferentfromwhatwasanticipated.

Intuitively,onemightconsiderthefollowingquestionasa

criterion for evaluating the ‘Decision robustness’ (in

retro-spect):‘wasthedecisiontakena‘good’decision?’However,there

aresomefundamentalproblemsinansweringthisquestion.

Firstly,majorwatermanagementdecisionshaveoftenalong

implementationtime,orinvolvestrategieswitha

consider-ablelife-time(e.g.tensofyears).Yet,formanystudiesthetime

passedhasbeentooshorttodecidewhetherdecisionshave

turnedouttobesuccessful.Secondly,andmoreimportant,we

canonlyevaluatedecisionsagainstthesinglepastwehad,

whichisonlyonerealisationofallpossiblefuturesthatcould

haveevolvedafterthedecisionwastaken.Forexample,dueto

inherentclimatevariabilityandthestochasticnatureofthe

occurrenceofextremes,prolongedperiodscanpasswithout

extremeevents,eveninthecaseofclimatechange.Ifitwas

decided that anticipatory strategies were not needed, this

decisionwouldhavebeenevaluatedas‘good’,asaresultofthe

fortuitousabsenceofextremeevents.Inother–equallylikely–

realisations of the future, in which some extreme events

occurred,thisdecisionwouldhavebeenjudgedas‘bad’.So,

judgingadecisionagainstasinglepastdoesnotprovide a

soundindicationofitsrobustnessorpotentialsuccess;such

evaluation requires confronting the result to a range of

realisationsofthefuture.Inourpaper,therefore,wefocus

on whether the decisionprocess –based on the scenarios

considered–providedprospectsforrobustdecisions.

Indicatorsforthe‘Decisionrobustness’ criterionshould,

therefore, reflect whether relevant uncertainties are

suffi-ciently represented. Relevant uncertainties have significant

anddistinguishedimpactontheoutcomes,andconsequently

thedecisionmaking(cf.IPCC,2001). Forwatermanagement

thisinvolvesuncertaintiesinbothwaterdemandand

availabil-ity.Thismeansthatscenariosshouldincludeuncertaintiesin

climate,sealevel andriverdischarges,thatallaffectwater

availability,aswellasuncertainties insocio-economicand

social developments (e.g. land use and the accepted flood

damage),thatdeterminesocietalrequirementsandthusthe

waterdemand.Adifferentkindofrelevantuncertaintyarises

frominteractionsbetweenthewatersystem,societyandwater

management.Forexample,floodsanddroughtsmayraisethe

needforadditional ornewmeasures,ormoreprofoundly,it

may influence societal perspective (e.g. how we evaluate

systemandourexpectationsofthefuture),andmaytriggera

waterpolicyresponsewhichmaythenaffectthewatersystem.

Theresultingwatermanagementresponsewillthenaffectthe

watersystemanditsfutureresponsetoextremes.Uncertainty

inthepolicyresponsefurtheraddstothetotaluncertaintyon

the water system in the future. In retrospective, water

management in the Netherlands has indeed strongly been

drivenbybothfloods(e.g.in1993and1995)anddroughtevents

(e.g. the summer of 1976), and socio-economic trends (e.g.

increasingvaluationofnatureandculturalheritage).Forrobust

decisionmaking scenarios should, therefore, consider the

dynamic interactions among climate, society and water

managementastheseevolveinthecourseoftimeandinfluence

theperformanceofpolicyoptions.

To determine whether uncertainties were sufficiently

representedforrobustdecisionmaking,weanalysedtherange

anddiversityoftheconsideredscenariosusingthefollowing

indicators:thenumberofscenarios,thevarietyintherangeof

outcomesencompassed,thevarietyinalternatives,andthe

temporal and dynamic nature of the scenarios. Using the

rangeofascenarioasindicatorfor‘Decisionrobustness’does

notmeanthatdecisionmakingshouldbebasedonlyonthe

extremesnorthatabroaderrangeinitselfisbetter.Instead,

several alternative scenarios should be considered that

encompass a relevant and plausible range of futures.

Alternativescenariosgobeyondthefrequentlyused‘business

asusual’scenariosderivedbyextrapolationofongoingtrends,

andcomprisechangesindevelopmentsinthecourseoftime.

Regardingthetemporalnatureofascenarios,scenarioscanbe

‘snapshots’describingamomentinthefuture,or‘transient’

scenariosdescribingtheevolvementtoacertainpointinthe

future(VanNotten,2005).Thedynamicnatureofascenario

referstowhetherascenarioisessentiallybasedonagradual

extrapolationoftrends, orwhether itencompassesevents,

discontinuities, or even surprises which change gradual

developmentsabruptly(VanNotten,2005).Whatisconsidered

‘plausible’or‘relevant’issubjecttodifferentinterpretations,

and depends on one’s expectations about the future and

understandingofthesystem.Awayofdealingwiththistype

of uncertainty – often referred to as perspective-based

uncertainty–isincludingsuchdifferentperspectivesinthe

scenarios(cf.Middelkoopetal.,2004;VanAsseltetal.,9582).

The‘Learningsuccess’criterionreferstothequestion:did

the scenarios enable learning for policymakers and scientists?

Answeringthisquestionisrelevanttoindicatethevalueof

scenarioanalysis,andtoimprovefuturescenariouseinwater

managementstudies.Althoughtherearemanydefinitionsof

learning, mosttheorists agreethat learningisa changein

knowledgeorbehaviourasaresultofexperience(e.g.Kolb,

1984;Driscoll,1994).Althoughwecouldnotprovide

quantita-tive measures, we determined indications of the learning

effect from reflection and underpinnings indicated in the

reports. We give some examples: (1) A policy report that

(4)

astartingpointoftheirstudy(‘Scenariostudiesshowthatclimate

changewillhaveanimpactonthehydrologicalwatersystem.’).(2)A

policy document mentioning a contextual development or

eventasareasontoadaptapolicyorascenario(‘Eventxraised

awarenessthatanewscenario/approachisneeded.’).(3)Aresearch

studystatingthatpreviousresultsshowed‘X’,but‘Y’isunclear,

andwillbestudied.Therefore,weanalysedtheevolutionofthe

scenariocontentanduse,thestudy’ssubject,andthe

science-policyinteraction,and usethisinformation incombination

withourexperienceandtheexperienceofourcolleagues,to

estimatethe‘Learningsuccess’.

3. Historical

perspective

on

scenario

use

in

water

management

studies

3.1. Theemergenceofconcepts

Theemergenceofconceptofanthropogenicglobalwarming

hasbeencharacterisedbydifferentmilestones(e.g.Peterson

etal.,2008;Weart,2010).Mid-19thcentury,Tyndallsuggested

that atmospheric changes could explain ice ages (Tyndall,

1861).Arrheniuswasthefirsttoquantifythecontributionof

CO2tothegreenhouseeffect(Arrhenius,1896).Inthe1950s,

progressinunderstandingofclimatecyclesresultedinthe

Milankovitchtheory,explainingcyclesatglacial-interglacial

time scales (Milankovitch, 1930). After 1950, tools became

available for measuring greenhouse gases. Keeling (1960)

showed a faster CO2 increase than Arrhenius’ estimate.

Togetherwithavailabledataontheglobaltemperaturethis

ledto theideathat increasingCO2could resultin marked

climate change (Revelleet al., 1965). In the 1970s, climate

modelsweredevelopedandusedtostudythecombinedeffect

ofcoolingthroughaerosolsandwarmingthroughCO2.After

warmingtrends,reportedinthe1940s,amultidecadecooling

was observed (Mitchell, 1963). Although scientific articles

described both potential future warming and cooling, the

media (e.g. Gwynne, 1975) mainly covered a future cooler

world(Petersonetal.,2008).Inthemid-1970s,thediscussion

inthemediabecamedichotomous:theclimatecouldbecome

warmerorcooler(Mathews,1976).

The scenario concept originates from the 1950s and is

ascribedtoHermanKahnatthattimeworkingattheRAND

Corporation(VanAsseltet al.,2010).Hedemonstratedwith

scenarios thatUS militaryplanning was basedon ‘wishful

thinking’insteadof‘reasonableexpectations’(Bradfieldetal.,

2005). In the 1970s, scenarios were used to explore the

sustainabilityofnaturalresources.‘Thelimitstogrowth’of

theClubofRomeisawell-knownexample(Meadowsetal.,

1972). Using scenarios and the World3computermodel the

study showed that a long-term perspective can identify

problems in current policies (Van Asselt et al., 2010). In

businessdevelopment,ShellOilisconsideredthefirsttouse

scenarioplanning(VanderHeijden,1996;Wack,1985).

3.2. Towardsfirstscenariosinwatermanagement(1953– 1988)

Afteramillenniumofadaptationinresponseto(flood)events,

theDutchshiftedtoanticipatorywatermanagementinthe

courseofthetwentiethcentury.The1916stormfloodalong

theZuiderzeeinitiatedtheimplementationofexistingplans

forthe Afsluitdijk, alargedefencestructureseparatingthe

Zuiderzeefromthesea.The1953stormsurge,whichkilled

1835andaffected750,000people,triggeredaparadigmshift.

Policymakers learned that the deterministic approach was

inadequate.Fromtheperspectivethat‘thisshouldneverhappen

again’,theystatedthattheprobabilityofoccurrenceofsuchan

eventshouldbeverysmall.Accordingly,ana-prioriaccepted

exceedance probabilityand correspondingwaterlevelwere

determined,resultingindesignconditionsfortheDeltaWorks

(DeltaCommittee,1960),thelargedefensestructuresinthe

southwestdelta.Thiswasthefirstuseoffutureconditions.A

relativesealevelrisebasedonextrapolationofmeasurements

wasincludedinthedesignofthedefensestructures,because

ofitslifetime(100–200years)(Rijkswaterstaat,2008).However,

apotentiallyacceleratedsealevelriseduetoclimatechange

wasnotconsidered.Thisprobabilisticapproachwasadopted

forallprimaryflooddefences.

AlongwiththeDeltaWorkstheDutchgovernmentdecided

fordevelopinganationalpolicyonwatermanagement,andto

document thisinaNationalPolicyMemorandumon Water

Management (PWM). As safetywas ensured withthe Delta

WorksandtheAfsluitdijk,the1stPWMfocusedonfreshwater

supply(Rijkswaterstaat,1968).Althoughclimatechangeand

sealevelrisewerementioned,assessmentsconsideredonlyan

increaseinwaterdemand.Uncertaintiesaboutfuture

devel-opmentswereacknowledged,butnobandwidthwasgiven.The

documentstatedthat‘theinfluenceofthesedevelopments(climate

change andupstreamwateruse)onthe totalwateravailabilityis

consideredtobesmall.Itishoweverimportanttokeepmonitoringthese

developments.’(Rijkswaterstaat,1968,p.137).

In the 1980s, scenarios became mainstream in futures

research(Mossetal.,2010).Also,intheNetherlandsscenario

analysis emerged. This was probably supported by the

cooperationwiththeRANDCorporationforthePAWN-study

(Policy Analysis for the Water management of the

Netherlands) (RAND Corporation, 1983; Rijkswaterstaat,

1985)thatprovidedthescientificsupport forthe2ndPWM

(Rijkswaterstaat,1984).

Inthe2ndPWM,thegovernmentstatedthatrevisionofthe

1stPWMwasneededdueto:‘societaldevelopments,changesin

insight and stakeholders of the water system. For example, the

prognosesforthefuturewaterdemandsforagricultureanddrinking

andindustrywaterneedtoberevisedandtheimportanceofsectors

like industry, shipping and nature has been acknowledged’

(Rijkswaterstaat, 1984, p. 7). The 2nd PWM emphasised

improving watermanagementfromacost-benefit

perspec-tive.Thiswasaparadigmshift;insteadofensuringwaterfor

all users,policywas nowonlyimplementedifthe benefits

were larger than the costs. Trends in water use were

considered for agriculture, drinking and industry water in

thepolicyanalysis.ThePAWN-studymentionsthat‘atplaces

wheretheuncertaintyintheresultshasanimpactontheconclusions,

either asensitivityanalysis isexecuted ordifferent scenariosare

described.’(Rijkswaterstaat,1985,p.138).Thestudyconcluded

that even in case of the ‘maximum trend scenario’ for

irrigation, wherein manyfarmerswould use sprinklers, no

large interventions were needed. These conclusions were

(5)

3.3. Climatechangescenariosandimpactanalysisonthe watersystem(1988–1998)

By the end of the 1980s, experimentswith Global Climate

Models (GCMs) indicated that the signal of anthropogenic

warmingwouldsoonemergefromnaturalvariability(Hansen,

1988;Mossetal.,2010). TheInternationalPanelonClimate

Change(IPCC)publishedits firstassessment includingfour

scenariosin1990(IPCC,1990).Thescenario‘businessasusual’

(BaU) assumedno or few policiesto limit greenhouse gas

emission andwas presented with alower, best and upper

estimate. The other three ‘accelerated policy’ scenarios

described future climates after emission reduction. In the

secondassessmentreport,theBaUscenariowaselaboratedin

theIS92scenarios(IPCC,1995).Dutchresearchersdeveloped

the globalmodel IMAGE forimpact assessment andpolicy

development regarding greenhouse gases (Rotmans, 1990;

Alcamoetal.,1999).

In this period, the first studies on climate and water

appearedintheNetherlands.Inacoastaldefensestudythree

sealevelrisescenarioswereconsidered,namely:the‘policy’

scenarioincludingsea level afterglobal implementationof

climatechangemitigationpolicies;the‘anticipatory’scenario

describing the best guess; and the ‘unfavourable’ scenario

describingthebestguessplusstandarddeviation(DeRonde

andVogel,1988). Basedonthesescenarios,thesubsequent

ISOS (ImpactofSealevel riseOn Society) studyquantified

impacts, and identifiedpolicy options (Rijkswaterstaatand

DelftHydraulics,1988).Thestudyfocusedonsafetyagainst

flooding,usingscenarioson sealevel rise,river discharges,

windand tidal conditions.The ISOSstudy wasthe first to

includechangesinriverdischarges inthescenarios.

Socio-economic developments were excluded because of their

uncertainty.

Nowthatsafetyandwatersupplyweremanagedwell,the

government shifted its focus to water quality because:

‘pollution,togetherwithoverexploitationofwaterandanunbalanced

spatialplanninghaveresulted inanunsustainablewatersystem’

(Rijkswaterstaat, 1988, p. 5). Accordingly, the 3rd PWM,

entitled‘Waterfornowandthefuture’,focusedonecological

andchemicalwaterqualityprovidedthatsafetywas

guaran-teed. TheBrundtland report (Brundtland, 1987), which put

sustainabilityhigh onthe internationalpoliticaland public

agenda, clearly inspired this quality focus. Policymakers

definedfuturetargetsbasedonpastconditions,andidentified

policyoptionstoreachthesetargetconditionsunderdifferent

scenarios.Thescenarios includedextrapolationsofongoing

waterdemandtrendsandtheintendedresultof

environmen-tal policy defined by the Ministry of the Environment.

Although this ministrypublishedthree estimates,onlythe

centralestimatewasconsidered.

While research studies extended their scope by using

integratedscenarios, policymakerswere focusingonsafety

issues.Triggeredbythe1993and1995floodeventsandthe

increasedattentiontoclimatechangeandsealevelrise,the

Dutch government installed the committee Tielrooy to

analysewhethercurrentwatermanagementwassufficiently

preparedforfutureclimatechange andsea level rise.This

committeeadoptedthreeoftheKNMI1999scenarios,which

weresimilartotheKNMI1997scenarios,butignoredthe‘dry’

scenario, because this scenario contained complementary

signalscomparedtotheotherscenarios(wetterandwarmer,

drier andwarmer, drierand colder).Socio-economic

devel-opmentswereonlyconsideredinaqualitativesense.Inthe

finalreport,guidingprinciplestoprepareforclimatechange

were explicitlyputforward:‘anticipateinstead ofreact, create

moreroomforwater,anddonotonlydischarge,butalsostorewater’

(CW21,2000).Asanalternativeforconfiningwaterinnarrow

zonesbetweendikes,creatingmoreroomforwaterwasan

upcomingparadigminrivermanagement,aimingat

decreas-ingwaterlevelsintimesofpeakdischarges,andenhancing

nature’s qualityat thesame time(DienstLandelijkGebied,

1999;Silvaetal.,2000).Regardingcoastalzonemanagement,

thegovernmentdecidedin2000todoubletheamountofsand

forbeachnourishmentinresponsetonewinsightson

long-term morphological developments (Rijkswaterstaat and

IMAU,2000).

In2003,severalgovernmentalorganisationsagreedina

so-called National Water Agreement (NWA) to define and

implement strategies for coping with climate change and

sealevelriseby2015,andtoexplorethenecessarystrategies

for2050(MinisterievanVerkeerenWaterstaat,2003).Water

boardsshouldadopttheguidingprinciplesofthecommittee

Tielrooy,and‘atleastusetheircentralestimatescenariofor2050

withanoutlookto2100todevelopmeasures’.

Until this period, policymakersneglected ‘drought’ asa

possible effectof climatechange.In 2002,the government

studiedthebalancebetweenfreshwaterdemandandsupply

(RIZA,2005).Thedrysummerof2003wasawelcomesurprise

forgettingthesubjectonthepoliticalagenda.KNMIupdated

the 1999 scenarios and re-introduced a ‘dry’ scenario ina

revisedversionbasedonRCMresults(Beersma,2001).Forthe

analysisalsolandusechangeswereincludedaswell.

3.4. Newclimatescenariosandadaptationpolicyin legislation(2006topresent)

Based on extended and improved information of amongst

others the IPCC’s fourth assessment (IPCC, 2007), KNMI

developed new climate scenarios; KNMI’06 scenarios (Van

denHurketal.,2007;Katsmanetal.,2008).Asuncertaintydue

toemissionscenarioswassmallerthantheuncertaintydueto

climate models, temperature was used as discriminating

factor.Asecondrelevantfactorwasthe circulationregime.

This resulted two scenarios with a moderate temperature

increase (þ18C) and two withstrong temperature increase

(þ28C),whichwerefurtherdistinguishedbyastrongorweak

changeofatmosphericcirculationoverEurope.Forsealevel

riseabandwidthwasgiventocoverthelargevarietyinthesea

levelrisespredictedbydifferentclimatemodelsfordifferent

globalwarmingscenarios.ThefourKNMI’06scenarioswerea

problem for the water managers as this precludes the

selection of a central estimate, as was prescribed in the

NWA of2003,and theadequacyofdesignedpolicyoptions

neededtobereconsidered.TheNWAwasupdatedin2008,and

prescribedfordifferentwaterrelatedproblemstheuseofonly

one of the KNMI’06 scenarios (Ministerie van Verkeer en

Waterstaat, 2008). In 2009, KNMI reflectedon the KNMI’06

report based on new scientific understanding and recent

(6)

KNMIdidnotseethe needfordefiningnewscenarios,the

scenarioswiththemoderatetemperaturechangeswerenow

consideredlessplausiblethanthosewiththelargerchanges.

Consequently,againtheguidelinesintheNWA(Ministerievan

VerkeerenWaterstaat,2008)wasoutdated.Forexample,for

studiesondroughttheNWAprescribedtousethe‘moderate

dry’scenario,whileaccordingtotheupdateofKNMIforthis

kindofsituationsthe‘strongerdry’scenariowouldbemore

plausible.

In 2007, the government established the second Delta

committeeforidentifyingactionstopreventfuturedisasters

(Kabatet al.,2009;DeltaCommittee,2008),astheexpected

futureclimatechangeandsealevelrise‘cannolongerbeignored’

(DeltaCommittee,2008,p.5).NexttotheKNMI’06scenarios,

thecommitteeconsideredahigh-endscenarioexistingofa

plausibleupperlimitofsealevelrisesin2100and2200fora

robustnesstestofpoliciesandinvestments(Katsmanetal.,

2011; Vellinga et al., 2008). The high-end scenario learnt

policymakersthattheNetherlandscanovercomesealevelrise

and climate change, but that the water system has to be

adapted.TheadviceresultedinaDeltaActandispresently

beingelaboratedonintheso-calledDeltaProgramme.

Climatechangeandsealevelrisewerenowonthepolitical

and public agenda.In the 5th PWM(Rijkswaterstaat,2009)

climatechangeandsea levelriseplayedanimportantrole.

The report had a separate chapter about dealing with

uncertaintiesonclimatechange.ThefourKNMI’06scenarios

were described indetail, while socio-economic trendsand

futuretargetsweredescribedqualitatively.Againascenario

wasprescribedforstrategydevelopment, meaningthatthe

system should be prepared for coping with the situation

describedinaspecificscenario.Thereportstated,that‘Forthe

choiceofascenariothesocietalriskisimportant.Forsafetyissuesthe

riskislarger,thanfordrainageandwaterloggingissues.Incaseof

lowflexibilityandhighsocietalrisk,thereisapreferencefortheupper

limitsofclimatechange.’(Rijkswaterstaat,2009,p.28).Thereport

mentionsthedifficultiesofincludingnewscientific

informa-tion:‘Theavailabilityofrepeatedlynewscenariosresultsintherisk

thatdecisionmakingwillbepostponedduetotheuncertainties...On

theonehanditisstrivedtousemostrecentinsightswhileontheother

hand stable assumptions are needed for decisionmaking and

implementation. New insightscannot result in new assumptions

and evaluations.’ (Rijkswaterstaat, 2009, p. 27). The report

identifiedpolicyoptionstoreachthedescribedtargets,and

presented a planning scheme with research and decision

milestones.

AtEuropeanlevel,the FloodDirective(2007/60/EC)came

intoforcein2007.Thisdirectiveaimsatmappingandreducing

floodriskand,asoneofthemeasures,mappingflood-prone

areascategorisedtolow,medium(likelyreturnperiod100

years),andhighprobability.TheFloodDirectivereferstothese

categories as scenarios. The 5th PWM states that it will

incorporatethisDirectiveintheDutchlegislationinthenext

planningperiod.

3.5. Dealingwithuncertaintiesaboutthefuture:new approaches(2006topresent)

After2000,the awareness raised thatuncertainty overthe

futurewillremainandcannotbeeliminated(cf.VanAsselt,

2000).More research does notautomaticallyreduce

uncer-tainty but may even increase it. Taleb (2007) emphasized

futureuncertaintywiththeintroductionofthe‘BlackSwans’

concept. These are unforeseen occurrences (unknown

unknowns)withalowprobabilityofoccurrencebuthaving

a large impact. Although from a differentfield, the recent

‘economiccrisis’raisedawarenessthat(unexpected)events

influence ourworldview.Newapproachesfordealingwith

uncertainties emerged (e.g. Carter et al., 2007; Dessai and

Hulme, 2004; Russill and Nyssa, 2009). Gladwell (2000)

introduced the ‘tipping points’ concept to describe the

catchinessofbehaviourandideas.MoserandDilling(2007)

used tipping points to conceptualise social change, and

definedit as‘momentsintime whereanormally stable oronly

gradually changing phenomena suddenly takes a radical turn.’

(MoserandDilling,2007,p.492).

IntheNetherlands,discussionsonscenarioupdatesledto

a new approach, using the systems vulnerability to define

Adaptation Tipping Points (ATP) indicating whether, and

underwhatconditions,currentwatermanagementstrategies

willcontinuetobeeffectiveunderdifferentclimatechanges

(Kwadijketal.,2010).Incaseofnewscenarios,onlythetiming

ofan ATPneedstobeupdated.Eventsandsurpriseswere

recognised as triggers for adaptation, societal change and

learning:notonlythefutureendpoint,butalsothepathwayto

this point is important. Therefore, a method to explore

AdaptationPathwayswasdeveloped.Byexploringpathways

withtransientscenarios,andincludingthedynamic

interac-tionbetweenthewatersystemandsociety,policymakerscan

identify robust and flexible pathways or identify lock-ins

(Haasnootetal.,2011,inpress;Offermansetal.,2011).

Also,atapolicylevelnewconceptsemerged.Recently,both

theScientificCouncilforGovernmentPolicyandtheAdvisory

CouncilfortheMinistryofTransportandWaterManagement

advisedtoconsideruncertaintyexplicitly(VanAsseltetal.,

2010;RaadvoorVerkeerenWaterstaat,2009).Thelatterstates

that ‘weshould notonly be preparedforexpected butuncertain

futureclimates,butalsoforunknownuncertainties,so-calledBlack

Swans.’Accordingly,policydevelopmentshouldincorporate

proactiveadaptationbyusingscenariosforcharacterisationof

uncertainties,andindicatorstomonitorthenecessityofpolicy

revision.Thecouncilalsostatesthat‘policybasedonanextreme

scenarioisliabletoproveundulyexpensiveorunnecessary’(p.53).

ThisstatementisincontrastwiththesecondDelta

Commit-tee.Thescientificcouncilrequestedattentionfornormative

foresightsincludingavarietyofvaluesandperspectives(Van

Asseltetal.,2010).

ThechairoftheDeltaProgrammementionedthat:‘Oneof

the biggestchallenges is dealing with uncertainties in the future

climate,butalsoinpopulation,economyandsociety.Thisrequiresa

newwayofplanning,whichwecalladaptivedeltaplanning.Itseeks

tomaximiseflexibility;keepingoptionsopenandavoiding‘lock-in’

(Kuijken,2011).Thesewerestartingpointsforanewapproach

forscenariodesign(Bruggemanetal.,2011).Byanalysingwhat

makes policiesforsafetyandwatersupplyvulnerable, four

climateandlandusescenarioswithsmallandlargeimpact

wereestablished.

Originatingfromthe1990s,butbecomingpracticeinthe

pastyears,istheparadigmshiftoccurringintheNetherlands

(7)

engineer-ing structures to a more ‘soft’ approach using natural

dynamicsofthesystemitself(cf.Inman,2010).Thechanging

approach involves restoration of wetlands, beaches and

naturalfloodplains,andisreferredtoas‘ecological

engineer-ing’,‘buildingwithnature’or‘greenadaptation’(e.g.

Aarnin-khof et al., 2010; Waterman, 2008; Van Koningsveld and Mulder,2004). Theseapproachesarenovelways ofdealing

with uncertainty: instead of fighting unpredictable future

events,adaptingtowhatishappening(Inman,2010).

4. Key

findings

4.1. Didthescenariosenablerobustdecision-making?

The central issue related to this question is whether the

scenariossufficientlyrepresentedrelevantknowable

uncertain-tiesforenablingrobustdecisionmakingonwaterpolicies.We

observedthatscenariosinpolicyanalysisshiftedfrom

describ-ingfuturewaterdemandtowateravailabilityafterthe3rdPWM.

Forthe1stPWMpolicymakersexpectednorelevantchangesin

wateravailability.Researchstudiesfocusedmainlyonwater

availabilityscenariosintermsofclimatechange,sealevelrise

andriverdischarges.Thus,few studiesincludedall relevant

knowableuncertaintiesforlong-termwatermanagement.

Whethertherelevantuncertaintiesweresufficiently

repre-sented can be assessed from the number, value range,

temporalanddynamicnatureandtheamountofalternatives.

Overthepastdecades,thenumberofscenarioshasincreased

from one to multiple scenarios, thereby increasing the

representeduncertaintyrange.Allresearchstudiesincluded

severalscenarios;first onlyclimatescenarios,later studies

alsoincludedsocio-economicdevelopments.Thefirstpolicy

documents considered a single scenarioonly, while policy

studiesinthepast15yearsusedthreetofourscenarios.Still,

the guidelines forclimate adaptation followingfrom these

policydocumentsrecommendedusingonlyonescenariofor

the design of water policies (Ministerie van Verkeer en

Waterstaat,2003,2008).Hence,althoughpolicymakers

recog-nised uncertainty about the future with several scenarios,

they persisted focusing on a ‘best estimate’ of the future

climateintermsofabestprediction,untilKNMI(deliberately)

presentedfourscenariosin2006(VandenHurketal.,2007).

Thereafter,policymakersselectedoneofthesefourscenarios

as ‘best scenario’ for strategy development for a specific

problemsuchassafetyorwatersupply.Thus,inpractisethe

rangeoftheuncertaintieswasnotfullyconsidered.

Although an increasingnumber of scenarios was

intro-duced, most scenarios remained to be extrapolations of

trends.Thisisreflectedbythescenarionames.Thefirstfour

policydocumentsmerelyused‘business-as-usual’scenarios

called‘trend’,‘autonomous developments’and ‘prognoses’.

Fewpolicystudiesincludeda‘maximumtrend’,‘worsecase’

scenario. Only a few background studies tried to include

alternatives,suchasthe‘discontinuity’scenarioforthe4th

PWM.Incontrast,researchstudiesexploredmorealternatives

byconsideringseveralscenariossuchas‘worsecase’,‘lower/

central/upper’estimates,‘dry’and‘cooling’scenarios.

Thedynamicandtemporalnatureofthescenarioswere

limitedtodefiningafewprojectionhorizons,inmostcasesthe

years2050and2100.Scenariosdescribedfortheseyearswere

projections of climate and external context,resulting in a

snapshotofthefuturesituationbeyondcontrolofthewater

managers.Likewise,socio-economicdriversofwaterdemand

were consideredasindependent ‘policy driven’or

‘autono-mousdevelopments’, whichweregradual extrapolationsof

trendsintothefuture.Adaptationoptionswerethen

formu-latedandevaluatedagainstexternalconditionsatonefuture

point.Scenarioanalysisforwatermanagementwas,thus,a

one-way pressure-impact analysis without response from

societyorwatermanagement,unlikeglobalmodels,suchas

IMAGE(Rotmans,1990).Asaresult,thewaterpolicystudies

haveignoredthedynamicpathintothefuturewithnatural

(year-to-year)variability,extremeevents,thepotentiallylarge

role of societal response to climate events and water

managementresponsetoclimate-associatedeventsor

chang-ingsocio-economicperspectives.Itisonlyinrecentscientific

studiesthatthisinteractionisrecognized,andthatscenarios

arebecomingcompletedwiththesenewrelevantdimensions

oftime-series,dynamicinteractionandsurprises(Haasnoot

etal.,inpress).

The range of the values used in the scenarios is an

additional indicator for the sufficient representation of

uncertainty (see Figs. 2 and 3 for climate scenarios and

supplementary information for socio-economic

develop-ments).The1stand2ndPWMusedonevaluebasedontrends

for water demand,but extended the range due toclimate

variabilitybyanalysingyearswithdifferentnetprecipitation

and discharge. Three studies translated socio-economic

developments into land use maps. The projection year of

thesescenariosextendedfrom2015to2050to2100resultingin

an increase of the considered acreage change and the

bandwidth for urban and nature, but not for agriculture.

Regarding the climate scenarios, the bandwidth of the

emission and global temperature changes in the IPCC

scenarios hasbecomelarger.Previousclimatescenariosfor

theNetherlandshadsimilarrangesfortheglobaltemperature

astheIPCCscenarios,butrecentscenariosdifferfromtheIPCC

assessments.Thebandwidthforglobaltemperatureriseused

intheNetherlands(Fig.2)isremarkablysmallerthantheIPCC

scenariosatthattime.ThisiscausedbythefactthattheKNMI

scenariosrepresentapproximately80%ofthetotalrangeof

the output of the climate models, while IPCC scenarios

presented the complete range. However, it is uncertain

whether water managers and the general public in the

Netherlands areaware ofthisdifference,and onlysee the

smaller uncertaintyrange. Overtheyears,KNMI’s scenario

valuesforsummerprecipitationhavechangedconsiderably,

incontrasttothewintervalues.Theintroductionofthe‘dry’

scenariosreflectstheawarenessoflargeruncertaintyabout

futuresummerclimate,asnotonlythemagnitude,butalso

directionofthechangedifferedinthescenarios.

ThedifferenceinprojectionsofsealevelrisebetweenIPCC

and theDutchscenarios isstriking(Fig.2).Whilethe IPCC

scenariosshowatrendtonarrowerrangesandsmallervalues

forsealevelrise,theKNMIkeptthesamerangeandthevalues

werelargerthantheIPCC.Thesedifferencescanmainlybe

explained from the different uncertainties included in the

scenarios (e.g. the uncertainty in the contribution of ice

(8)

sheetswasnotincludedinthesealevelscenariovalues,but

only described in the report. These uncertainties were,

however,includedinthenationalKNMIscenarios,together

with recent (scenario and field) studies which were not

available at the time of the AR4 (Katsman et al., 2011). In

addition, regional differences due to variation in ocean

temperature, distribution of melt water over the oceans,

and – in some studies – tectonic subsidence contribute to

differencesbetweenthescenariostudies.Forexample,inthe

1990s studiesvalueswerederivedfromtheIPCCestimates,

supplementedwiththenaturaltrendandsubsidenceofthe

Netherlands (VanAsseltet al., 9582). TheDeltaCommittee

includedatectonicsubsidenceof10cm/year(Vellingaetal.,

2008),whilethestudiesinthe1990sincludedasubsidenceof

5cm/year. Thehigh-end sealevel rise exploredby the2nd

DeltaCommitteewasdiscussedthoroughlyamong

research-ers and policymakers. The valueswere larger than in the

KNMI’06 scenarios,because the DeltaCommittee aimed at

definingan‘upperplausible’limitofsealevelrisebyincluding

awider rangeofuncertainties andmechanismsunderlying

sealevelrisefortheNetherlands.Remarkably,thisupperlevel

isnotthatmuchhigherthantheupperendsoftheuncertainty

rangesputforwardin1990inthenationalstudies.

4.2. Didthescenariosenablelearning?

Generally,scenarioanalysisinwaterpolicystudiesenabled

fourdifferentlessons:(1)insightinimpactsofclimatechange

and socio-economic developments, as a result of several

national,butalsoglobalstudies(e.g.IPCCreports,ISOSand

NRPstudies);(2)theneedandeffectivenessofpolicies,such

the2ndPWMortheATPstudy;(3)theneedforadaptationof

targetsand/orpoliciesasaresultofcomparingscenarioswith

monitoringresults(e.g.2ndand3rdPWM);and(4)awareness

aboutpossibleimpactsofclimateandsocio-economic

devel-opments.Forexample,thesecond DeltaCommitteewidely

communicated its results through readable reports and

YouTube videos accessible for the general public. This

receivedalotofmediaattention,and raisedtheawareness

Fig.2–ValuesforglobalandlocalsealevelrisefortheNetherlands(left)andglobaltemperaturechange(right)in2100for

nationalandglobalclimatescenarios(referenceyear1990).FAR,SAR,TARandAR4referrespectivelytothe1st,2nd,3rd

and4thIPCCreport,NRPisNationalResearchProgramme,CT21=CommitteeTielrooy,DC=secondDeltaCommittee,and

PWM=NationalPolicyMemorandumonWaterManagement.ScenariosfortheNetherlandsareingrey.IntheDCstudy,

theglobaltemperaturerangeincludedforthesealevelrisewaslarger(dashedline)thanfortheclimateparameterssuchas

precipitation(solidline).IntheAR4reportsealevelrisevalueswerepresentedforthescenarios(solidline),andadditional

uncertainsealevelrisewasdescribedinthereport(dashedline).

Fig.3–Valuesforprecipitationchange(w=winter;

s=summer)in2100fordifferentnationalclimate

scenarios.PWM=NationalPolicyMemorandumonWater

Management,NRPisNationalResearchProgramme,

CT21=CommitteeTielrooy,andDC=secondDelta

(9)

oftheimportancefordevelopingwatermanagementstrategies

to prepare for the future. Furthermore, their ‘worst case’

scenariodeliberatelyprovokedlotsofdiscussionamongwater

managersintheNetherlands,whichenhancedtheexchangeof

ideas,andthusinvolvedalargedegreeoflearningaccordingto

thechairofthecommittee(Veerman,2010).Floodanddrought

eventscorrespondingwiththescenarios,butalsothepublic

debateaboutissues(e.g.climatechange,creditcrisis)

acceler-atedtheinfluenceofstudyresultsinpolicyimplementation.

Both scenario analysis in water management and the

science-policy interaction have clearly evolved inthe past

twenty years. In retrospective we can distinguish five

evolutionsthatreflectthelearningprocessofscientistsand

policymakers:

1. Fromfloodprotectiontointegratedwatermanagement:Thisshift

wassupportedbylessonsontheeffectivenessofpoliciesin

scenarioanalysis.Afterthemajorfloodingof1953,water

managementfocusedonfloodprotection.However,inthe

courseoftime,andwiththestep-wisecompletionofthe

Delta works, attention was givento other water-related

problems. In the PWMs the focus changed from water

supplyforeconomicpurposes,viaacost–benefitanalysis

for maintaining water availability to water quality and

nature, and eventually introducing the concept of

‘inte-gratedwatermanagement’,whichthe5thPWMextended

with spatial planning issues. Also, the scientific studies

showalearningprocessthroughanevolutioninthestudied

subjects. The first research studies focused on safety

againstcoastalflooding,whichwaslaterextendedtolarge

rivers and regional watersystems and finallyto impact

assessmentsofwaterservices.

2. Towards integrated scenarios: This shift was initiated by

awarenessthatbothwateravailabilityandwaterdemandare

relevantforwaterpolicymaking,aswellastheglobaland

Europeanshifttointegratedstudies.Also,scenariostudies

showed therelevance ofintegratedstudies for

decision-making.Althoughcomingfromadifferentstartingpoint,

bothscientificandpolicystudiesmovedtowardsintegrated

scenarios.Scientificstudiesfirstusedclimatescenarios.By

theendofthe1990s, socio-economicdevelopmentswere

consideredincreasinglyrelevant.Afteronlyevaluatingland

usechangetrendsand‘autonomous’socio-economic

devel-opments,integratedscenarioscomprisingbothclimateand

socio-economiccomponentsweredefinedtoexplore

differ-entwatermanagementstyles.Thescenariocontentinthe

PWMschangedincorrespondencewiththepurposeofthe

PWMsfromwaterdemandtrendstoclimatescenarios,while

at present integrated scenarios are considered. Still, the

integratedscenariosarenotyetfullyemployedforimpact

assessmentorpolicydevelopment.Furthermore,the

influ-enceofsocietalperspectives(e.g.onpolicytargets)remains

tobefullyincorporatedinpolicymaking.

3. Frompredicting toexploring the future:Whilepolicymakers

experienced that the future turned out differently than

envisioned,andsomeeventsoccurredascompletesurprise,

evidencegrewthatwecannotpredictthefuture.Initially,

prognosesonlyappliedtopossiblechangesinwaterdemand.

Estimatesoffuturefloodmagnitudes–asrequiredforthe

probabilistic flood protection approach – were based on

autonomous developments or expert judgement. These

‘predictandact’studiesslowlyshiftedtoan‘exploreand

anticipate’approachforwhichseveralscenarioswereused.

Still,theinitialuseof‘bestguess’or‘centralestimate’climate

scenariosreflectsthedesireofpredictingfutureconditions,

althoughnowassociatedwithbandsofuncertainty.Withthe

IPCC-SRESandKNMI’06scenarios,therecognitionthatthe

futureisuncertainandthatthereisno‘mostlikely’future,

hasincreasinglysettledinwatermanagement.Accordingly,

researchandpolicystudiesnotonlyaimedatimprovingthe

understanding of future developments such as climate

changeandreducinguncertainties,butalsoondeveloping

methodsfordealingwithuncertaintiesaboutthefuture.This

observed shift correspondswith observationsoffuturists

(VanAsseltetal.,2010;Slaughter,2002;VantKlooster,2008).

Both approaches, also referred to as forecasting and

foresight, are still used next to each other (Van Asselt

et al., 2010). Also, in water management the predictive

approachisstillusedwhenitcomestoshorttermactions

suchasfloodforecastinganddeterminingthe(long-term)

designdischarge.Forshorttermdroughtmanagementboth

forecastsandscenarios(foresights)areused.Someanalysts

propose to use probabilistic scenarios, but we have not

observedthesescenariosinthestudiesreviewed,butthis

could be initiated by the EU FloodDirective’s approach,

which prescribes touse scenarioswith floods with low,

mediumandhighprobability.

4. Interactionscience,policyandevents:Mostuncertaintiesabout

the futurewerefirst investigated byscientists,andlater

incorporated in policy, especially if events seemed to

support the trendsindicated byscenarios. For example,

the 3rd and 4th PWMdocumentsmentioned potentially

relevant impacts of climate based IPCC results and

scientific research in the preceding decades. In recent

years,the turn-overratefromscientificstudies towater

management has speeded-up. Scientific studies involve

stakeholders and while novel approaches in scenario

analysis emerge briefly after being introduced in the

scientificworldinwatermanagementapproachesaswell.

5. Fromfightingwatertoaccommodatingandadaptingtowater:

Sincethe1960,awarenessraiseaboutpotentialeffectsof

climatechangeasaresultofscenariostudies,and flood

events.Thisawarenesstriggeredashiftfromfocusingon

‘hard’ defensive infrastructures for flood protection to

‘softer’ measures for integrated water management, by

usingnaturalprocessesandaccommodatingwater(e.g.4th

PWM).Thus,insteadofstaticinfrastructureswithalonglife

time,easilyadaptablepoliciestochanging,unpredictable

boundaryconditionswerechosen.

5. Conclusions

and

recommendations

Thisreviewdescribestheuseofscenariosinwater

manage-ment studiesintheNetherlands overthepast60years.To

identify what we have learnt from this experience, we

analysedwhetherthescenariosenabledrobust

decisionmak-ingandlearning.

The opportunities for robust decisionmaking resulting

(10)

especiallyinpolicymaking.Althoughthenumberof

scenar-ios increased, for the strategy development often one

scenario was appointed for design conditions. Rarely, all

relevant uncertainties were included. Especially in the

policy documents uncertainties in water demand or

avail-ability were considered, while none included social

(per-spective-based) uncertainty. The number of alternative

futures increased, but scenarios mainly remained based

onextrapolationoftrends.Almostallscenariosusedwere

snapshotsat2 or3 timehorizons, therebyignoring

path-waystowardstheendpoint,anddisregardingthepossibility

that events may drastically change such pathways. All

scenariosweresurprisefree.The‘decisionrobustness’can

thusbeimproved.

Differences in value range between different scenario

studies can often be explained by reading details and

communicating with the developers, which indicates that

communicationonassumptionsisimportantforappropriate

scenariouse.

Thescenariosenabledlearningaboutpossibleimpactsof

developments,the need andeffectivenessofpolicies, and

theneedforadaptationofpolicies.Inaddition,thescenarios

raised awareness about potential future problems. The

historicalperspectiveshowsaclearscience-policy

interac-tion.Forexample,firstusedinresearchstudies,thepolicy

documents took climatechange and sea level rise up,as

important developments to consider in strategy

develop-ment; sometimes with a little help of a flood or drought

event.We observedseveralparadigm shifts reflecting the

learning process of scientists and policymakers: (a) from

flood control to integrated water management, (b) from

predictingtoexploringthefuturewithintegratedscenarios,

and(c)fromfightingwatertoaccommodatingandadapting

towater.

Dealingwithuncertaintiesappearstobeastruggle,given

theparadoxbetweenthedesire toexplorepotentialfutures

usingseveraldifferentscenarios,andthepreferenceofwater

managerstodesignpoliciesbasedonasinglescenariothatis

notfrequently updated.However, watermanagers needto

facethatthefutureisinherentlyuncertain,andscenariosare

alwayslikelytobeupdatedbynewscenariosastheyresult

from a process of design and construction at a specific

moment and location (Hulme and Dessai, 2008b). These

uncertaintiesshouldnotbeusedasaconstrainttodevelop

adaptationmeasuresforwatermanagement(cf.Dessaietal.,

2009;HulmeandDessai,2008b).

We provide five recommendations for improving water

policydevelopmentunderuncertainty:

1. For sustainable decisionmaking water managers should

considerseveralscenariostoexploretherelevantrangeof

theuncertainties,andnotselectingthemostlikelyfutureor

prescribinga‘design’scenario.

2. Newapproaches areavailable, which can together with

scenarioanalysissupportthedevelopmentofsustainable

measures.Severalmethodsinvolvemanycomputational

experimentstoanalysetheeffectsofuncertainparameters

(e.g.‘ExploratoryModeling’Bankes(1993))toseekforrobust

decisions(Lempertetal.,2006;LempertandBankes,2003)

oroptimal solutions(‘InfoGap’theory Ben-Haim(2001)).

Walker et al. (2001) describe a planning process with

differenttypesofactions(e.g.‘mitigatingactions’,‘hedging

actions’)andsignpoststomonitorifadaptationisneeded.

Also,adaptation tippingpoints(Kwadijketal., 2010)and

exploring adaptation pathways with transient scenarios

(Haasnootetal.,2011)canbeofassistance.

3. Scenario developers should clearly communicate the

assumptions, purpose and limitations of scenarios, and

the conditions under which the scenarios were made

(processandtimelimits).

4. Tailoredscenariosareneededtoensurerelevantscenarios

andappropriateuse.Todeveloptailoredscenarioswater

managers should assess the system’s vulnerability and

communicatethistoscenariodevelopers.

5. To improve scenarios and their use, evaluation of past

scenariosremainsuseful.Forthispurpose,evaluationon

‘Decision robustness’ and ‘Learning success’ deserve

further elaboration in terms of more explicit criteria

concerninge.g.comparisonwithstudy’sobjectives,

stake-holder involvement, pathway analysis, more precise

addressingofthelearningeffect(wholearned whatand

how?).

6. Insteadofrespondingtofloodanddroughtevents,

policy-makers could identify triggers (Walker et al., 2001) and

adaptation pathways(Haasnootetal.,2011).Thetriggers

givesignals whenit istimetomake adecisionand the

adaptationpathwaysallowforidentifyingrobustoptions

andlock-ins.

Summarizing,exploringthefuturewithseveralscenarios,

analysing the vulnerability and good communication with

scenariodevelopersmayhelpwatermanagerstodealwith

uncertainties,andmakesustainabledecisions.

Acknowledgements

WethankMarjoleinvanAsselt(MaastrichtUniversity,WRR)

andEelcovanBeek(Deltares)fortheirusefulsuggestionsona

previousversion ofthis manuscript,PietHoekstra(Utrecht

University)forprovidinginformationonthedevelopmentsin

coastal water management, and Jules Beersma, Caroline

Katsman (KNMI) and Gu¨nther Ko¨nnen(formerly KNMI)for

feedbackontheclimatescenariostudies.

Appendix

ASupplementary

Data

Supplementarydataassociatedwiththisarticlecanbefound,

intheonlineversion,atdoi:10.1016/j.envsci.2012.03.002.

r

e

f

e

r

e

n

c

e

s

Aarninkhof,S.G.J.,VanDalfsen,J.A.,Mulder,J.P.M.,Rijks,D.,

2010.Sustainabledevelopmentofnourishedshorelines

innovationinprojectdesignandrealisation.In:PIANCMMX

Congress, Liverpool,UK.

Alcamo,J.,2001.Scenariosastoolsforinternational

environmentalassessment.TechnicalReport.European

(11)

Alcamo,J.,2009.EnvironmentalFutures.ThePracticeof

EnvironmentalScenarioAnalysis.Developmentsin

IntegratedEnvironmentalAssessment.Elsevier,

Amsterdam,TheNetherlands.

Alcamo,J.,Leemans,R.,Kreileman,E.,1999.GlobalChange

Scenariosofthe21stCentury,ResultsfromtheIMAGE2.1

Model.Pergamon/ElsevierScience,Oxford.

Arrhenius,S.,1896.Ontheinfluenceofcarbonicacidintheair

uponthetemperatureoftheground.London,Edinburgh,

andDublinPhilosophicalMagazineandJournalofScience

(5thSeries)41,237–275.

Bankes,S.C.,1993.Exploratorymodelingforpolicyanalysis.

OperationsResearch41,435–449.

Beersma,J.,2001.RainfallgeneratorfortheRhineBasin.

Descriptionof1000-yearsimulations.TechnicalReport.

KNMI.

Ben-Haim,Y.,2001.Info-gapvalueofinformationinmodel

updating.MechanicalSystemsandSignalProcessing15,

457–474.

Bradfield,R.,Wright,G.,Burta,G.,Carnish,G.,VanDerHeijden,

K.,2005.Theoriginsandevolutionofscenariotechniquesin

longrangebusinessplanning.Futures37,95–812.

Brown,S.,1968.Scenariosinsystemsanalysis.In:Quade,E.,

Boucher,W.(Eds.),SystemsAnalysisandPolicyPlanning:

ApplicationsinDefence.AmericanElsevierPublishingCo.,

NewYork.

Bruggeman,W.,Hommes,S.,Haasnoot,M.,TeLinde,A.,vander

Brugge,R.,2011.Deltascenarios:scenariosforrobustness

analysisofstrategiesforfreshwatersupplyandwatersafety

(Deltascenario’s:Scenario’svoorrobuustheidanalysevan

maatregelenvoorzoetwatervoorzieningenwaterveiligheid).

TechnicalReport.Deltares(inDutch).

Brundtland,G.H.,1987.OurCommonFuture.NumberISBN

0-19-282080-XinReportoftheWorldCommissionon

EnvironmentandDevelopment.OxfordUniversityPress,

Oxford.

Carter,T.,Jones,R.,Lu,X.,Bhadwal,S.,Conde,C.,Mearns,L.,

ONeill,B.,Rounsevell,M.,Zurek,M.,2007.Newassessment

methodsandthecharacterisationoffutureconditions.In:

ClimateChange2007:Impacts,Adaptationand

Vulnerability.ContributionofWorkingGroupIItotheFourth

AssessmentReportoftheIntergovernmentalPanelon

ClimateChange.CambridgeUniversityPress,Cambridge.

CW21,2000.Dealingdifferentlywithwater,watermanagement

forthe21stcentury(AndersOmgaanmetWater.

Waterbeleidvoorde21eeeuw).MinistryofTransport,Public

WorksandWaterManagement/UnionofWaterboards,The

Hague,TheNetherlands(inDutch).

DeRonde,J.G.,Vogel,J.A.,1988.Coastaldefenceafter1990.

Technicalreport6:sealevelrise(Kustverdedigingna1990.

Technischrapport6.Zeespiegelstijging).TechnicalReport.

RijkswaterstaatMinisterievanVerkeerenWaterstaat(in

Dutch).

DeltaCommittee,1960.FinalreportoftheDeltaCommittee

(RapportDeltacommissie.Eindverslageninterimadviezen).

TechnicalReport(inDutch).

DeltaCommittee,2008.Workingtogetherwithwater:aliving

landbuildsforitsfuture.FindingsoftheDutchDelta

Committee2008.MinisterievanVerkeerenWaterstaat,The

Hague,TheNetherlands.Availablefromhttp://

www.deltacommissie.com/en/advies.

Dessai,S.,Hulme,M.,2004.Doesclimateadaptationpolicyneed

probabilities? ClimatePolicy4,107–128.

Dessai,S.,Hulme,M.,2007.Assessingtherobustnessof

adaptationdecisionstoclimatechangeuncertainties:acase

studyonwaterresourcesmanagementintheeastof

England.GlobalEnvironmentalChange17,59–72.

Dessai,S.,Hulme,M.,Lempert,R.J.,Pielke,R.J.,2009.Climate

prediction:alimittoadaptation.In:Adger,W.,Lorenzoni,

I.,OBrien,K.(Eds.),AdaptingtoClimateChange:

Thresholds,Values,Governance.CambridgeUniversity

Press,Cambridge, pp.64–78.

Dewar,J.A.,Builder,C.H.,Hix,W.M.,Levin,M.H.,1993.

Assumption-basedplanning:aplanningtoolforvery

uncertaintimes.TechnicalReportMR-114-A.RAND,Santa

Monica.

DienstLandelijkGebied,1999.WorkingTogetherwithNaturein

theRiverRegions.

Driscoll,M.,1994.PsychologyofLearningforInstruction.Allyn

&Bacon,Massachusetts.

Gladwell,M.,2000.TheTippingPoint.HowLittleThingsCan

MakeaBigDifference.LittleBrown,London.

Groves,D.G.,2006.Newmethodsforidentifyingrobust

long-termwaterresourcesmanagementstrategiesforCalifornia.

Ph.D.thesis.PardeeRANDGraduateSchool.

Groves,D.G.,Lempert,R.J.,2007.Anewanalyticmethodfor

findingpolicy-relevantscenarios.GlobalEnvironmental

Change17,76–85.

Gwynne,P.,1975.Thecoolingworld.Newsweek(April).

Haasnoot,M.,Middelkoop,H.,Offermans,A.,vanBeek,E.,Van

Deursen,W.P.A.,inpress.Exploringpathwaysfor

sustainablewatermanagementinriverdeltasinachanging

environment.ClimaticChange,

doi:10.1007/s10584-012-0444-2.

Haasnoot,M.,Middelkoop,H.,VanBeek,E.,VanDeursen,

W.P.A.,2011.Amethodtodevelopsustainablewater

managementstrategiesforanuncertainfuture.Sustainable

Development19,369–381.

Hansen,J.,1988.GlobalclimatechangesasforecastbyGoddard

instituteforspacestudiesthree-dimensionalmodel.Journal

ofGeophysicalResearch93,9341–9364.

Hulme,M.,Dessai,S.,2008.Negotiatingfutureclimatesfor

publicpolicy:acriticalassessmentofthedevelopmentof

climatescenariosfortheUK.EnvironmentalScience&

Policy11,54–70.

Hulme,M.,Dessai,S.,2008b.Predicting,deciding,learning:can

oneevaluatethesuccessofnationalclimatescenarios?

EnvironmentalResearchLetters3,7.

Inman,M.,2010.Workingwithwater.NatureReportsClimate

Change39–41.

IPCC,1990.ClimateChange1990:FirstAssessmentReport.

CambridgeUniversityPress,UK.

IPCC,1995.AdaptationsandMitigationofClimateChange:

Scientific-TechnicalAnalyses.ContributionofWorking

GroupIItotheSecondAssessmentoftheIntergovernmental

PanelonClimateChange.CambridgeUniversityPress,

Cambridge,UK.

IPCC,2000.Specialreportonemissionsscenarios.

IntergovernmentalPanelonClimateChange.

IPCC,2001.Climatechange2001:synthesisreport.IPCCThird

AssessmentReport.CambridgeUniversityPress,Cambridge,

UK.

IPCC,2007.ClimateChange2007:SynthesisReport.Cambridge

UniversityPress,UK.

Kabat,P.,Fresco,L.O.,Stive,M.J.F.,Veerman,C.P.,vanAlphen,

J.S.L.J.,Parmet,B.W.A.H.,Hazeleger,W.,Katsman,C.A.,2009.

Dutchcoastsintransition.NatureGeosciences2,450–452.

Kahn,H.,Wiener,A.J.,1967.TheYear2000:AFrameworkfor

SpeculationontheNextThirty-threeYears.Macmillan,New

York.

Katsman,C.,Hazeleger,W.,Drijfhout,S.,vanOldenborgh,G.,

Burgers,G.,2008.Climatescenariosofsealevelriseforthe

northeastAtlanticocean:astudyincludingtheeffectsof

oceandynamicsandgravitychangesinducedbyicemelt.

ClimaticChange91,351–37410.1007/s10584-008-9442-9.

Katsman,C.,Sterl,A.,Beersma,J.,vandenBrink,H.,Church,J.,

Hazeleger,W.,Kopp,R.,Kroon,D.,Kwadijk,J.,Lammersen,

(12)

H.,Vaughan,D.,Vellinga,P.,Vermeersen,L.,vandeWal,R.,

Weisse,R.,2011.Exploringhigh-endscenariosforlocalsea

levelrisetodevelopfloodprotectionstrategiesfora

low-lyingdeltatheNetherlandsasanexample.ClimaticChange

109,617–645.

Keeling,C.D.,1960.Theconcentrationandisotopicabundance

ofcarbondioxideintheatmosphere.Tellus12,200–203.

KleinTank,A.M.G.,Lenderink,G.J.,2009.ClimateChangeinThe

Netherlands.SupplementstotheKNMI’06Scenarios.KNMI,

DeBilt,TheNetherlands.

Kolb,D.A.,1984.ExperientialLearning:ExperienceastheSource

ofLearningandDevelopment.PrenticeHall,Englewood

Cliffs,NJ.

Kuijken,W.,2011.http://www.deltacommissaris.nl/nieuws/

toespraken/thedeltaprogrammeinthenetherlandsthedelta worksofthefuture.aspx(accessedon18.01.11).

Kwadijk,J.C.J.,Haasnoot,M.,Mulder,J.,Hoogvliet,M.,Jeuken,

A.,Krogt,R.,Oostrom,N.,Schelfhout,H.,Velzen,E.,

Waveren,H.,Wit,M.,2010.AdaptingtoSea-levelRiseinThe

Netherlands.WileyInterdisciplinaryReviews:Climate

Change.

Kwakkel,J.H.,Walker,W.E.,Marchau,V.A.W.J.,2010.Adaptive

airportstrategicplanning.EuropeanJournalofTransport

andInfrastructureResearch10,249–273.

Lempert,R.J.P.S.,Bankes,S.,2003.Shapingthenextone

hundredyears:newmethodsforquantitative,longterm

policyanalysis.TechnicalReportMR-1626-RPC.RAND,Santa

Monica.

Lempert,R.J.,Groves,D.G.,Popper,S.W.,Bankes,S.C.,2006.A

general,analyticmethodforgeneratingrobuststrategies

andnarrativescenarios.ManagementScience52,514–528.

Lempert,R.J.,Schlesinger,M.E.,2000.Robuststrategiesfor

abatingclimatechange.ClimaticChange45,387–401.

Mathews,S.W.,1976.What’shappeningtoourclimate?

NationalGeographicMagazine150,576–615.

Meadows,D.H.,Meadows,D.L.,Randers,J.,Behrens,W.W.,1972.

TheLimitstoGrowth.AReportfortheClubofRome’sProject

onthePredicamentforMankind.UniverseBooks,NewYork,

NY,USA.

Metz,B.,Davidson,O.,Swart,R.,Pan,J.,2001.Climatechange

2001:mitigation.ContributionofWorkingGroupIIItothe

ThirdAssessment[TAR],ReportoftheIntergovernmental

PanelonClimateChange(IPCC).CambridgeUniversityPress,

Cambridge,UK.

Middelkoop,H.,VanAsselt,M.B.A.,Van’tKlooster,S.A.,Van

Deursen,W.P.A.,Kwadijk,J.C.J.,Buiteveld,H.,2004.

PerspectivesonfloodmanagementintheRhineand

Meuserivers.RiverResearchandApplications20,

327–342.

Milankovitch,M.,1930.MathematischeKlimalehreund

AstronomischeTheoriederKlimaschwankungen.Vol.1.

PartAofHandbuchderKlimatologie.GebruderBorntrager.

MinisterievanVerkeerenWaterstaat,2003.Nationalwater

agreement(NationaalBestuursakkoordWater).Technical

Report(inDutch).

MinisterievanVerkeerenWaterstaat,2008.Updatenational

wateragreement(NationaalBestuursakkoordWater

Actueel).TechnicalReport(inDutch).

Mitchell,J.M.J.,1963.Ontheworld-widepatternofsecular

temperaturechange.In:ChangesofClimate:Proceedingsof

theRomeSymposiumOrganizedbyUNESCOandtheWorld

MeteorologicalOrganization,Rome, pp.161–181.

Moser,S.,Dilling,L.,2007.Towardthesocialtippingpoint:

creatingaclimateforchange.In:CreatingaClimatefor

Change:CommunicatingClimateChange&Facilitating

SocialChange.CambridgeUniversityPress,NewYork, pp.

491–516.

Moss,R.H.,Edmonds,R.A.,Hibbard,K.A.,Manning,M.R.,Rose,

S.K.,vanVuuren,D.,Carter,T.R.,Emori,S.,Kainuma,M.,

Kram,T.,Meehl,G.A.,Mitchell,J.F.B.,Nakicenovic,N.,Riahi,

K.,Smith,S.J.,Stouffer,R.J.,Thomson,A.M.,Weyant,J.P.,

Wilbanks,T.J.,2010.Thenextgenerationofscenariosfor

climatechangeresearchandassessment.Nature463,

747–756.

Offermans,A.,Haasnoot,M.,Valkering,P.,2011.Amethodto

exploresocialresponseforsustainablewatermanagement

strategiesunderchangingconditions.Sustainable

Development19,312–324.

Peterson,G.D.,BeardJr.,T.,Beisner,B.,Bennett,E.,Carpenter,

S.,Cumming,G.,Dent,C.,Havlicek,T.,2003.Assessing

futureecosystemservices:acasestudyofthenorthern

HighlandsLakedistrict,Wisconsin.ConservationEcology

7,1.

Peterson,T.,Connolley,W.,Fleck,J.,2008.Themythofthe1970

globalcoolingscientificconsensus.BulletinoftheAmerican

MeteorologicalSociety89,1325–1337.

RaadvoorVerkeerenWaterstaat,2009.Whiteswans,Black

swans.Adviceonpro-activeadaptationtoclimatechange

(Wittezwanen,zwartezwanen.Adviesoverproactieve

adaptatieaanklimaatverandering).TechnicalReport(in

Dutch).

RANDCorporation,1983.Policyanalysisofwatermanagement

fortheNetherlands,summaryreport.TechnicalReport.

RandCorporation.

Revelle,R.,Broecker,W.,Craig,H.,Kneeling,C.,Smagorinsky,J.,

1965.Restoringthequalityofourenvironment:reportofthe

environmentalpollutionpanel.TechnicalReport.President’s

ScienceAdvisoryCommittee,TheWhiteHouse.

Rijkswaterstaat,1968.NationalPolicyMemorandumonWater

Management(DewaterhuishoudingvanNederland).1st

PWM.TechnicalReport(inDutch).

Rijkswaterstaat,1984.NationalPolicyMemorandumonWater

Management(DewaterhuishoudingvanNederland).2nd

PWM.TechnicalReport.Rijkswaterstaat(inDutch).

Rijkswaterstaat,1985.Policyanalysisforthewater

managementintheNetherlands(Beleidsanalysevoorde

waterhuishoudinginNederland/WitteNota).Technical

Report(inDutch).

Rijkswaterstaat,1988.Thirdnationalpolicymemorandumon

watermanagement.Waterfornowandthefuture(Derde

Notawaterhuishouding.Watervoornuenlater).3rdPWM.

TechnicalReport.Rijkswaterstaat(inDutch).

Rijkswaterstaat,2008.Notaforthe2ndDeltaCommittee.Life

timeofstormsurges(Kennisnotitiet.b.v.Deltacommissie.

Vraag4aDelevensduurvanstormvloedkeringen).Technical

Report(inDutch).

Rijkswaterstaat,2009.Nationalpolicymemorandumonwater

management2009–2015(NationaalWaterplan2009–2015).

5thPWM.TechnicalReport(inDutch).

Rijkswaterstaat,DelftHydraulics,1988.Impactofsealevelrise

onsociety,acasestudyfortheNetherlands.Technical

Report.Rijkswaterstaat,DelftHydraulics.

Rijkswaterstaat,IMAU,2000.Theothersideoftheclimate

(Dekeerzijdevanhetklimaat).TechnicalReport(in

Dutch).

RIZA,2005.Severnessandextentofwatershortagesinthe

Netherlands.FinalreportDroughtstudy(Aard,ernsten

omvangvanwatertekorteninNederland.Eindrapport

droogtestudie).TechnicalReport(inDutch).

Rosenhead,J.,Elton,M.,Gupta,S.K.,1972.Robustnessand

optimalityascriteriaforstrategicdecisions.Operational

ResearchQuarterly(1970–1977)23,413–431.

Rosentrater,L.D.,2010.Representingandusingscenariosfor

respondingtoclimatechange.WileyInterdisciplinary

Reviews:ClimateChange1,253–259.

Rotmans,J.,1990.IMAGE:anintegratedmodeltoassessthe

greenhouseeffect.PhDthesis.KluwerAcademicPublishers,