1

NepalFbFFeasibilityStudyHighlights

● TherearemultiplehazardsinNepalwhichcouldbenefitfromthedevelopmentofFbFsystems.● ThehazardprioritizednowisriverinefloodduetothehighimpactsoffloodsintheTerairegion,its

alignmentwiththeNepalRedCrossSocietystrategicplan,andtheabilitytoforecastnegativeimpactswithsufficientleadtime.

● Itmayalsobepossibletoincludeatriggeringmodelforflashfloodswhichwouldprovideaveryshorttimeperiod(3-6hours)tocompleteno-costactions.

● Therecommendedtriggeringmodelforriverinefloodwouldbeamulti-steptriggerusingmultipleforecastswhicharesuitableatdifferentleadtimes.

● ThereisastronglogictoactingearlyinanticipationoffloodsintheTerairegion,primarilybecauseroadaccesstocommunitiesisimmediatelycompromisedduringfloodevents.NepalRedCrossSocietiesbranchstaffandvolunteershaveexpressedfrustrationwithhavingto‘sitontheirhands’whileaffectedpopulationsareinaccessible.

● InpreviousyearsofextremefloodingtheNepalRedCrossSocietyhasneededtoemployexpensivestrategiestoprovidesupporttodesperatepeople,suchastheuseofhelicopterstoairdropfoodtopersonsstrandedontheirroofs.Assuch,itisexpectedthatawelldevelopedandimplementedFbFprogramforanextremeeventwouldhaveafavourablecost-benefitratioasopposedtoactiononlyinaresponsecapacity.

● ThereisastrongFbFcommunityofpracticeinNepalwithpositiveintegrationintogovernmenttechnicalservices,andopportunitiestoworkcloselytogetherunderthenewARRCCprogram.

● TheDanish-fundedpilotprogramforthe2018monsoonseasoninonedistrictoftheTerai(Bardiya)didnottriggerbuttheprocessofsettinguptheprogramhascreatedmultiplelessonslearnedandhasestablishedastrongunderstandingoftheFbFconceptwithintheNepalRedCrossSocietyDisasterManagementteamatheadquartersandthroughoutthebranchinBardiya.

● ItisrecommendedthattheprocesstowriteandvalidateanEAPfortheDREFbyFbAbeginassoonaspossibletobetrigger-readyforthe2020monsoonseason.

● ItisrecommendedthatNRCStakesadvantageoftheworkcompletedonthe2018pilottobetrigger-readyforasmall-scaleFbFprograminthe2019monsoonseason.Ifsuchasystemwastotrigger,itwouldreducefloodimpactsandprovideavaluablelearningopportunitytofeedintothedevelopmentofanEAPtoaccessfundsfromtheDREF.

● TherearemultipleinvestmentsofPartnerNationalSocietiesandtheIFRCinNepalthatcanbebuiltofffortheestablishmentofanFbFsystem,includingtheprovisionofdatamanagersatthedistrictemergencyoperationscentres,andthePERprocess.

● Thereisahighlevelofheterogeneitywithinthefloodproneareaintermsofbothvulnerabilityandpreparedness.DevelopingasystemthatallowsNRCStoselectthecommunitiesexpectedtoexperiencethemostextremeriverfloodingconditions,andselectingactionssuitableforcommunitieswithdifferinglevelsofpreparednessandfamiliaritywithNRCS,willbeessentialforthesystem’ssuccess.

● Theabilitytotriggereffectivelyrestsonstrongintegrationwithgovernmentsystematalllevels.● AdditionalhighpotentialopportunitiestoexploreinfutureincludetheintegrationoftheFbF

systemwithsocialprotectionsystems,andtheinclusionofotherhazards-mostlynotablycoldandheatwaves.

2

TableofcontentsHighlights 1

Introduction 3Aboutthisreport 3Background 3FbFactivitiesinNepaltodate 4

Initialforecastanalysisperhazard 5Institutionsinterviewedasbackgroundfortriggerassessment 5SituationAnalysisofAvailableForecastsforFbF 5

HazardAnalysis 6Relevantinformationonnon-prioritizedhazards: 7

Flashflood 7Drought 8Landslides 8Epidemics 8Coldandheatwaves 9

FloodForecastAnalysis 9Background:experienceofthe2017floods 9Currentlyavailablefloodforecasts 10SuitabilityofavailableforecastsforFbF 11

FloodRiskAnalysis 15VulnerabilityandPreparedness 15Exposure 16Dangerlevels 17PossibilitytodevelopanImpact-basedforecast(IBF) 17Jurisdictionfordisseminatingtheforecast/earlywarning 19NextstepsforintegratingtheavailableforecastsanddataintoanFbFsystem 19

ActionsAnalysis 20Buildingoffofthe2018FbFpilotandpre-existingpreparednessefforts 21Feedbackonactionsandplannedactionimplementationfromthefield 22Actionsshort-listing 23

Scaleofintervention 23ReadinessandImmediatepre-disasteractions 24

CapacityAssessments 25NationalSocietyCapacity 25

3

General 25Finance 25

TechnicalServiceCapacity 26CapacityoftheFbFCommunityofPractice 26

Recommendationsonareasforcapacitybuildingforstakeholders 26

NextSteps 27PlansforMonsoonseason2019 27

Rationale 27Fundingpracticalities 27Designrecommendations 27Additionaldata/informationneededtosetthetriggerfor2019 28Nextstepstobetrigger-readyforthe2019Monsoonseason 28

ProposedongoingsupportfromRCCC 29

AdditionalOpportunities 29Opportunitiestoaddressadditionalhazardsinfuture 29OpportunitiestoexploresocialprotectioninNepalalignedwithFbF 29

Conclusion 30Annex 30

Annex1:Documentsreviewedpriortothefieldstudy 30

IntroductionAboutthisreportThis report outlines the findings of the feasibility study, which aimed to identify highpotential applications of the FbF concept to reduce the impacts of hydro-meteorologicaldisasterevents.ThestudywasconductedjointlybyMeghanBailey(RCCC),AhmadulHassan(RCCC)andManishDhungel(DanishRedCross)withfundingfromDanishRedCrossandin-country support from theNepalRedCrossSociety (NRCS). In-country interviewsand fieldvisitstoflood-proneareaswereconductedFeb11-21,2019.Thisstudybuildsofftheworkthat has beendonepreviously to set up the FbFpilot project in BardiyaDistrict, and theworkofother stakeholders furthering theFbFagenda inNepal–mostnotably theWorldFood Programme (WFP) and Practical Action Consulting (PAC), in collaboration with theNepalDepartmentforHydrologyandMeteorology(DHM).ThereportwasfinalizedinJune2019.Itcanbesharedwidely.BackgroundForecast-basedFinancing(FbF)(alsocalledForecast-basedAction/FbA)isamechanismthatuses climate and weather forecasts totrigger timely humanitarian action, before a

4

hazardhitsthe exposedpopulation. These actions are automaticallyfunded when the triggering forecast is released. This mechanism has been progressivelydeveloped since 2008. It began with several bilaterally funded pilot projects, whichestablishedanevidence-base forhowtodesignand implementFbFsystems.The lessons,best practices and recommendations from these projects led to the establishment of acentral pool of funding as part of the DREF, whereby national societies can develop FbFEarly Action Protocols (EAPs) to receive funding to implement activities in advance of ananticipated disaster. The development of EAPs of a sufficiently high standard to pass thevalidationprocessoftheFbAbyDREFmechanismoffersagoalpostinthedevelopmentofFbFsystems.ThisfeasibilitystudyisanearlysteptowardsdevelopinganFbAbyDREFEAP,whichwilllikelyincludeadditionaltrialingoftriggersandactionsintheinterim.FbFactivitiesinNepaltodateTherehasbeenconsiderableworkundertakeninNepalalready.ThereisanestablishedFbFworking group including the World Food Programme, Practical Action Consulting, IFRC(workinggrouphosts),theNepalRedCrossSocietyandDanishRedCross.Thereisaviewtoexpandthisworkinggrouptoalso includetheDepartmentofHydrologyandMeteorology(DHM). Although they are currently not part of the working group, all members of theworkinggrouphaveestablished relationshipswithDHM.TheWorldFoodProgrammehasbeenintheprocessofsettingupanFbFsystemsince2015.Theyhaveformedcloseworkingrelationships with the government hydrological and meteorological technical services toinformtriggersandhavecompletedasubstantialmappingexerciseofavailablevulnerabilitydata. Their triggers were established jointly through a consultancy with Practical ActionConsulting. Importantly, as World Food Programme do not have ongoing operationalpresenceontheground,theirpost-triggeractionsareplannedtobe implementedbyRedCross branch staff and volunteers. They have established MoUs at branch level. Theprogram has yet to be triggered. As will be highlighted later in the report, effectivecoordinationwithWorldFoodProgrammewillbeessentialasboththeRedCrossandWFPpost-triggeractivitieswillbecompletedbyRedCrosschapterlevelstaffandvolunteers.ThroughtheDanishRedCross’sInnovation/HumGrant,apilotFbFprogramwasestablishedin Bardiya district for the 2018 monsoon season. As 2018 was not an extreme year forflooding, the programme rightly did not trigger. The establishment of the pilot offeredmultiple lessons learnedwhich are thoroughly detailed in Danish Red Cross reports. TheNepal Red Cross chapterwas heavily involved in the establishment of the pilot and nowhave a high level of understanding of the FbF concept, and operational constraints. Thebranch has an informationmanagement specialist seconded to the governmental districtlevelemergencyoperationscentre.Thisindividualisheavilyinvolvedinmonitoringthedatacoming in fromobservational stations (river level and rain gauges)whichare relevant forflooding throughout thedistrict.As such, the staffandvolunteershaveanaboveaverageawarenessoffloodmonitoringandanticipatingfloodingconditionsintheirjurisdiction.Theset-up of an Red Cross information management specialist seconded to the districtemergencyoperationscentre,andfacilitatingongoinginformationsharingaboutfloodriskbothtotheRedCrosschapter,andtothedistrictchiefoperatingofficerwhoisresponsibleforissuingwarnings,willbeanimportantassettoanyFbFsystem.

5

InitialforecastanalysisperhazardCentraltothefeasibilityassessmentofanFbFsystemistheestablishmentofwhichdisasterevents, if any, canbe anticipatedwith sufficient reliability inorder to triggerpre-disasteraction.Intheinitialdaysofthisstudy,thefollowinginstitutionswereinterviewedtorelatedtotriggerfeasibility.Institutionsinterviewedasbackgroundfortriggerassessment

Hazard Government NGO/PNS/UN Scientific/Knowledgecenter

Flashflood

Meteorological ForecastingDivision of Department ofHydrology and Meteorologyunder the ministry of Energy,WaterResourcesandIrrigation

PracticalAction ICMOD

RiverFlood Hydrological Division andForecastingSection,DHM

PracticalAction

WFPICMOD

Landslide Meteorological ForecastingDivision,DHM

PracticalAction

Drought Meteorological ForecastingDivision,DHM ICMOD

Heat/coldwave Meteorological ForecastingDivision,DHM

Basedon these interviews the following situation analysis of currently available forecastswhichcouldformthetriggerinwholeorinpartforanFbFsystemwascompleted.SituationAnalysisofAvailableForecastsforFbFHazard Forecast LeadTime SourceFlashflood Real time precipitation

monitoringandgaugetomeasurecorrelationusedforforecast.

Upstream gauge monitoring andrelation with the downstreamgaugeisabletoforecastthefloodinthedownstreamriverbasin.

This tool is still at research level.BothPracticalActionandICIMODare working to improve the leadtimebyaddingarainfallforecast.

3-6hours Practical Action ICIMODMeteorologicalForecasting Division,DHM

RiverFlood DHMfloodhavebeenforecasting 1-3days Hydrological Division and

6

in the major river systems inNepal from 2017 and calibratedin 2018. The MIKE 11hydrodynamic flood model useforKoshi,WestRaptiandBagmatiriver basins and HEC HMS floodmodel are simulated in Karnali,BabaiandNarayaniriverbasin.

ForecastingSection,DHM

Landslide Currently no forecasting isavailable.Research with international andlocal partners to develop acomputational framework basedonhydrology, geology and socio-economicfactorsfundedbyDFID.

Unknown PracticalActionTribhuvanuniversity

Drought GlobalSeasonrainfallforecast Months MeteorologicalForecasting Division,DHM

Ministry of AgricultureDevelopment

FAO

ICMODHeat/coldwave Nationally, a temperature

forecast exists but not aheat/cold index. A heat/coldindex would be required toanticipate the impacts of heatand cold waves as temperaturealone is insufficient. The ClimateCentre could support in thedevelopment of a heat index inthenextphaseofClimateCentresupportforNepal.

3-7days MeteorologicalForecasting Division,DHM

HazardAnalysisSixhazardtypeswereconsideredbasedonthehistoricalimpactsofthehazard,theabilitytoforecastit,theprioritiesoftheNepalRedCrossaccordingtotheirstrategicplan,whichitistechnicallyfeasibletodevelopandimpact-basedforecasttotriggeraction,andwhethermeaningful action could be taken in the lead time afforded by an early warning. Note,althoughearthquakesareamajorhazardinNepal,itisnotpossibletoforecastthemintimeandspace,astheyarebasedonseismicactivity.Thefollowingmatrixhighlightsthehazardsconsidered(flood,drought,landslide,epidemic,coldwave,andheatwave).

7

Canforecastthehazard?

Cancurrentlyforecasttheimpact?

PrioritizedinNRCSstrategicplan?

TechnicallyfeasibletomakeanIBF?

Meaningfulearlyactiontotake?

Flood(riverine)

Yes(atcertaintimescales)

no yes yes yes

Flood(flash)

Yes(atveryshortleadtimes)

no yes yes yes

Drought Yes(unsureofskill)

unsure no unsure Unsure(notwithinRedCrossstrategicadvantageasprimaryimpactsconcernfoodsecurity)

Landslide Indevelopment no yes unsure yes

Epidemic Somefactorstechnicallyfeasible

no yes Somediseasetypes yes

ColdWave

Yes,butnationalforecastsarequalitative

no no yes yes

HeatWave

Yes,butnationalforecastsarequalitativeonly

no no yes yes

Through this preliminary assessment, riverine flood emerged as the priority hazard forwhichtoassessfeasibilitythoroughly.However,informationonotherhazardswascollectedatalowerintensitytosupportdecisionmakingastowhetherFbFshouldbedevelopedforadditionalhazards.Relevantinformationonnon-prioritizedhazards:FlashfloodTheimpactsofflashfloodsaresevere,frequentlyresultinginlossoflifeintheTerai,andthehazardhasbeenprioritizedbyNRCS.Flashfloodwasnotprioritizedforthedevelopmentofa full FbF system because of the very short lead time for FbF action (3-6 hours). This isprimarily because the ability to release financing on this timescale to ground level isunlikely. However, certain no-cost actions, such as passing evacuation orders, would befeasible on this time scale. If systemswere developedwhich could act rapidly enough totakeadvantageof3-6hour leadtimes,suchactionscouldmeaningfullyreduceflashfloodimpacts,includinglossoflife.Theseno-costactionscouldbeincludedinabroaderfloodingEAPforareaspronetoflashflooding.Inpracticalterms,thiswouldmeandevelopingafloodEAP that includes both flash and riverine components, with no-cost actions outlined forflashfloodandamoretypicalsetofcostedactionsforriverinefloods.TheClimateCentreiscurrently involved in a flash flooding research program with NASA, which could provide

8

technicalguidanceinthedevelopmentofflashfloodtriggers.However,the current feasibility assessmentdidnot explore indetailwhat the capacityof any localactorsare tobeoperatingon timescalesof thismagnitude. It is likely that sucha systemwould need to includemultiple simulations towork out the idealmethod of post-triggeroperation.Ifthereisaninteresttopursuethisoption,technicalsupportinthisareacanbeprovidedbytheclimatecentreinthenextphase.DroughtDroughtwasnotprioritizedprimarilybecause incomparison tootherhazards inNepal, itdoesnotproduceimpactsrequiringhumanitarianintervention.DroughtwasnotidentifiedasapriorityinthestrategicplanofNRCS.AlthoughWASHimpactsassociatedwithdroughthave been documented, drought impacts in Nepal tend to affect the food security andlivelihood sectors, which is not the strategic advantage of NRCS. Apart from a seasonalforecastforrainfall(3-monthleadtime),thereisnospecificdroughtearlywarningproductproducedbytheDepartmentofHydrologyandMeteorologyonwhichanFbFtriggercouldbebased.LandslidesLandslides are of considerable interest for an FbF program. Landslides have beenparticularly harmful in the months and years following the 2015 earthquake. The quakedestabilized the surface of many mountainous areas, making it more susceptive tolandslides.ThereiscurrentlyaDFIDfundedresearchprogramunderSHEARexaminingtheability to anticipate landslides in Nepal. A team from international and local researchinstitutesarenowworking todevelop the computational frameworkbasedonhydrology,geology and socio-economic factors. The Nepal-based institutions are the NGO PracticalActionandTribhuvanUniversity. Thisprogram includes theplanneddevelopmentof low-costcensorsbyImperialCollegeLondon,whichcouldcontributetoalandslideearlywarningsystem.Theresultsoftheresearchprogramonlandslidescouldproduceatooltoanticipatelandslides, which could in turn be operationalized for an FbF system. The research iscurrentlyunderway.Resultsareexpected,includinganindicationofbothleadtimeandskillofanywarningtheyareabletoproduce, inearly2020.Atthattime,theRCCCwillreviewtheresultsandwillprovideguidanceastowhetherFbFtoaddresstheimpactsoflandslidesis feasible. NRCS would be well placed to engage in this space. NRCS has extensiveexperience with landslide response and some experience with landslide early warningsystems, including small-scale community based approaches involving setting of alarmnoises for people to evacuate (with a lead-time of a couple ofminutes). The hazard is apriorityareaforNRCS.EpidemicsMultiple epidemics including dengue and cholera, which have links relationships tohydrometeorology are common to Nepal. The case load often increases in specific areasfollowingfloodingeventsorheavyrainfallrequiringhumanitarianinterventionfromNRCS.Atpresentthereisnoearlywarningsystemforthesepublichealthissues.Shouldasystembedevelopedthatcouldanticipateanuptickincaseloadsasaresultofweatherconditionsorotherdisasterimpacts,anFbFsystemcouldtriggerearlyactiontoreducetheimpactsofepidemics.Appropriateactionswouldmost likelycomefromtheWASHsector,which isahigh capacity area of the NRSC. As such, NRSC would be well-placed to implement an

9

epidemic FbF system should awarning systembedeveloped. It is notpossibleatthistimefortheClimateCentretohelpproducesuchanearlywarningsystemfrom scratch, as this would be a major scientific undertaking requiring considerableresources.ColdandheatwavesCurrently only qualitative forecasts are available. These forecasts do not provide aprobabilityofanevent,butratherthemeteorologistslookatthemodelresultsandprovideaqualitative indicationofwhether aheatwaveor coldwave is likely. Inorder touse thisforecast,wewouldneedtoevaluatewhether,inthepast,thesequalitativeindicationshaveindeed lead toextremeeventsornot.Anotheroption is toworkwith themeteorologicalproviderstoensurethattheyprovideawarningwhenacertainprobabilityofaheatwaveorcold wave is forecasted by thesemodels, which would then give us a confidence in theprobabilityofactinginvainandthefrequencyofthetrigger.ItispossibletocompletethistypeofanalysiswiththeviewtodevelopanFbFsysteminfutureifthereisadesireamongstakeholders to do so. The Climate Centre is able to provide the technical support todevelop a heat index for a cold and heat wave trigger. Currently there are majorinvestments beingmade in India to be able to anticipate and understand the impacts ofheat and coldwaves. There is an opportunity to build off these investments as the Teraiarea would have very similar conditions. There are multiple health and WASH relatedactions,with a strong evidence base,which could be triggered in an FbF system for coldand/or heatwave. Coldwaveswithin the Terai occur at the frequency and severity thatwould be suitable for the FbA by DREFmechanism. As temperature extremes affect theelderly and very young disproportionately, this hazard could be a high potential area toexplore integration within the national social protection system (see section on socialprotection for more details). Globally there is increased interest in early action on heatwaves,astheyarestronglycorrelatedwithmortality.Thehazardtendstobedeprioritizedbecausethedeathsarenotclassifiedspecificallyashavingbeentheresultofaheatwave-rather they are classified as dehydration, high blood pressure, etc. As such they can bethoughtofasa‘silentemergency’.Thereisstrongevidencethatthefrequencyandseverityofheatwaveswillincreaseasaresultofglobalclimatechange.

FloodForecastAnalysisBackground:experienceofthe2017floodsDuring the 2017 flood, 17 peoplewere reported to have died, and 337,000 peoplewereaffected,with91,000ofthosepeopletemporarilydisplaced.Atthetimeofthe2017floodstherewasnotaneffectiveforecastingsystem.BasedonthisexperiencetheGovernmentofNepaltookinitiativetostrengthentheDHMfloodforecastingsystemtoreducelossoflifeand property through the creation of the flood forecasting divisionwith three sub-units.Previoustothe2017floodingevent,floodforecastingwastheremitofasingleunit.AssuchcapacityasgreatlyimprovedwithinDHManditremainsandinstitutionalpriority.Likewise,inthe2018strategicplanoftheNepalRedCrossSociety,floodremainsatoppriority.

10

CurrentlyavailablefloodforecastsDHMstartedfloodforecastingwitha leadtimeof3days inthemajorsixriverssysteminNepal.Themodelwascalibratedin2018.InNepal,thesixmajorrivers(Karnali,Babai,WestRapti,Narayani,BagmatiandKoshi)basinsaremodelledasshowninFigure1.

Figure1:LocationmapsofsixriverbasinsinNepal

Therearetwotypesofforecastingmodelswhichareusedforthesesixmajorriverbasins.TheMIKE 11 hydrodynamic floodmodel has been adopted for the three river basins i.e.Koshi,WestRaptiandBagmaandtheotherthreeriverbasinsadoptedHECHMS.The initial floodmodelforthe3-dayforecastassessment(reportedbyDHM)showsabout40cmerrorbetweentheriverlevelspredictedandobservedintheupstream,and20cminthedownstream.Theerrorintheforecastismostlyinthesmalltributariesintheupsteambasinandmuchlessinthedownstreamrelativelylargerrivers.Assuch,theFbFsystemwillfocusontriggeringbasedonthebehavioursofthelargerrivers.DHMisworkingtoimprovetheforecastinthesmallerrivers,hencecurrentlytheseforecastsarenotdisseminated.Forthepurposesofunderstandingnegative flood impacts,a20cmerrorconsiderablyreducestheutilityof the3-day forecast.Onthecontrary, theoneday leadtimeforecast iswithintheacceptable rangeaccording to theDHM. In theminor riversbasinsonlyoneday leadtime flood forecastwas disseminated in year 2018, as the confidence is low for 2-3 daysleadtime.Theforecastmessageispreparedusingavailablemodels-MIKE11/HEC-HMS,GLOFASandthemodel developed by RIMES. TheDHMdisseminateswarningswith respect to dangerlevels. Danger levels are pegged on flood return periods (i.e. a 1 in 5 year flood) whichcorrelate with specific known impacts (i.e. flow overtop a riverbank). If the forecastindicatesconditionswillsurpassaspecificdangerlevel,awarningisissued.

11

TheDHMuseacolour systemto indicate risk level.Thecolouryellowrepresents thewater reaching the levelassociatedwitha1 in2-5year returnperiod (thebank full level). The colour red (also referred to as the ‘danger level’) represents floodwaters associated with a 1 in 5-8 year return period. At this level, rural roads typicallybecome submerged in affected areas, cutting off access to communities, with only themajorhighways remaining sufficientlyelevated.Highwaysbecomesubmerged (alsocalledthehighwaytoplevel)infloodwatersassociatedwitha1in25-yearreturnperiod.The1in2-5 year (rural roads submerged) would be a sufficiently extreme anticipated event totriggertheFbAbyDREF,ifanEAPweretobedeveloped.SuitabilityofavailableforecastsforFbFCentral to theassessmentofFbF feasibility isanassessmentofhowreliable theavailableforecasts are. This provides an indication of how frequently a trigger, based on a givenforecast,mighthaveafalsealarm(type1error)ormissanupcomingevent(type2error).Theseassessmentsareoftenlookingatforecast‘skill’,whichisrepresentedbyaskillscore.Suchassessmentscanbesensitiveas it involvescriticallyexaminingtheproductproducedbyanotherorganisation.However,itisnotmeanttobeacritique,butratheranassessmentofwhethertheforecastavailableissuitablefortheveryspecificpurposesofanFbFsystem,involving thedisbursementof large amount of humanitarian funds. FortunatelyDHMhasbeenveryopenandcollaborativewithmultiplestakeholderswithintheFbFcommunityofpractice,includingthefeasibilitystudyteam.Themodelsimulationresults(producedbyDHM)offorecastsfromtheyear2008to2013witha lead timeof1-3daysare shown inFigure2.Thereare120water level stationsofwhich90 stations are telemetric. 200 stationsmeasuredischarge, outofwhich160wereusedforthemodel.Figure2showsthatthemodelwasnotabletosimulatethepeaksata3-day leadtime.Resultshows improvementfora2-day leadtimebutthereliability isstillinsufficientforanFbFtriggerastherewouldbetoomanymissedeventsandfalsealarmstobe able to confidentlydistribute funds.However, the forecast for a lead timeof onedaysufficientlymatches theobserveddischarge rate indicating the forecast isable to identifymostfloodingevents(withreliabilityvaryingslightlybetweenmajorrivers).

12

(a)LeadtimeDay3(sourceDHM)

(b)LeadtimeDay2(sourceDHM)

13

c)LeadtimeDay1(sourceDHM)

Figure2:ModelSimulationresult.(a)Leadtime3days,(b)leadtime2daysand(c)lead

time1day.(sourceDHM)Figure 2 shows that the forecast with one day lead time is better matched with theobserveddischargedata.AbiascorrectionhasbeenappliedtotheabovefiguresbyDHMusingastatisticalmethodtoaddthedistributionoferrortothe initialvaluetoproduceacorrelationbetweentheforecastandtheobservation.Afterthebiascorrection,theresultsof the forecast output are improved for the one-day lead time forecast. Due to lack ofaccesstotheforecastandobserveddata,theskillanalysiswascarriedusingapresentationon the forecast by DHM. Assessment of the skill of forecasts was carried out using thedirectionofwaterflow(waterwillrise)andfoundthatthemodelforecastedthatthewaterwouldrise9times.Ofthose9times,thewaterroseasexpected5timesanddidnotrise4times,asshowninthefollowingtable.Hence,FalseAlarmRatio(FAR)offloodforecastforonedayleadtimeisabout0.44(4/9).TheskillwillimproveasthegovernmentconsiderstheimprovementofEWSapriorityandputsdirectresourcestowardsit.

Observe

ForecastItem Rise FallRise 5 4Fall 3 6

RIMES’s assessment found that the Nash Sutcliffe coefficient (a statistical method forassessing thegoodnessof fitwithinhydrologicalmodels)varies river to riverand foundavariationbetween0.53and0.7.Inaddition,theyfoundverygoodagreementatDebganjonNarayaniriver.ThismeansthattheforecastismostreliablefortheNarayaniRiver.

In addition to the forecasts produced by DHM, there are other globalmodelswhich cancontribute to the trigger. The GLOFAS (Global Flood Awareness System)model has been

14

usedinotherFbFprograms.ThereliabilityofGLOFASvariesfromplaceto place and as such a specific assessmentwas needed for the Terai region.DHMbeganusingGLOFASin2015.Theyhavecalibratedthemodelwith25observationstationdatasets.From2018,GLOFAShasbeenproducingaflooddepthmapwithwith7daysleadtime.DHMhasbeenusingthisproductbutdoesnotreceivetherawordigitaldata,andassuchcouldnotprovidedataforthismodeltothestudyteam.Duringthe interviewwith ICIMOD,thetechnical specialistsalso felt confident touse the1day lead time,andnotat longer leadtimesgiventhecurrentforecastlimitations.Giventhatthenationalforecastsbeyond1-dayleadtimecurrentlyisnotsufficientlyreliableforanFbFsystem,andgiventhata1-dayleadtimelimitswhatactionscanbedone,itisrecommendedthattheGLOFASmodelbeusedasareadinesstriggertoinitiatereadinessactionsata7dayleadtime,which(assumingastopmechanismisnotinitiated)wouldbebefollowedbya1dayleadtimeactiontriggerusingtheDHMforecast.

DHMmadeananalysisonriversfortheirnationalforecast.ThefollowingsampleoftheanalysisshowstheforecastoftheriversanditsNashSutcliffecoefficientwhichisshowninthechartbelow(note,aNashSutcliffeabove0.7showsgoodskill):

Station NashSutcliffeKumalGaun(KaliG) 0.57Kotagaun(KaliG) 0.58Sisaghat(Madi) 0.48

Bimalnagar(Marsyan) 0.42Kalikhola(Trishuli) 0.52Devghat(Narayani) 0.76Rajaiya(EastRapti) 0.1

Based on this analysis, FbF should be feasiblewith 1-3 days of lead time in the differentlarge rivers ofNepal. The FbFproject canprovide further analysis to estimate the risk offalsealarms,whichwillhelpselect the lead-timeof the triggerand integrateotherglobalmodelsasneeded.

As was presented in the primary results of the feasibility study in-country, therecommendedtriggersystemwouldlooklikethis:

15

Note:adigitalversionofthiswillbepreparedfortheEAP.Timeiswrittenasitpertainstotheanticipateddisasterevent.Zero(0)isthefloodevent,whichwebelievetobethetimethatruralroadsceasetobeaccessible.Timebeforetheeventrangesfrom-7daysto-1day.Thisshowsfromlefttoright:normalmonsoonpreparednessactivitiesbefore-7days.TheGLOFASmodel isusedasa readiness triggerat -7days to initiate readinessactions tobeimplementedbetween-7daysand-1day.Duringthistimeperiodastopmechanismcouldbeinitiatedatanytimeiftheforecastsbeingmonitoredindicatethattherisklevelhasgonebelowthethreshold,inwhichcaseactivitieswouldstop.Ifnostopmechanismisinitiated,readinessactivitiescontinueuntiltheactivationtrigger.TheactivationtriggerisinitiatedbytheDHM1-dayfloodforecastat -1day/ -24hours.After this triggerthere is24hourstoimplement the high-cost pre-disaster actions before rural roads are no longer accessible.Followingthisperiod,ifneeded,normalresponseandrecoveryactivitieswouldbeinitiated.

FloodRiskAnalysisVulnerabilityandPreparednessDuringfieldvisits,communitiessharedstoriesofveryhighwaterlevels,goinguptothetopsof tall trees and just below the level of roofs. In some cases, ropes were tied betweenhousesandtreestoallowindividualstopullthemselvesthroughthewatermanyfeetaboveground-level from treetop to roofmake it tohigher ground. Inprevious floods theNepalRed Cross has made emergency distributions of foods to families stranded on roofs viahelicopterdrop. It is clear that floods inNepal represent significant risk tohuman life, inadditiontolarge-scaledamagetoproperty,andlivelihoods.Further,theactivitiesthathavebeen taken in the past during extreme flooding events to access destitute, strandedfamilies,have insome instances (e.g. the2014 floods) includedemergencydistributionoffoodandnon-food itemsdroppedviahelicopters to families strandedon rooftops. Theseactions are extremely high cost. As such, it is likely that an FbF system (implemented atscale)inanticipationofanextremefloodingeventofthismagnitudewouldhaveafavorablecost-benefitratio.Thereisahighlevelofheterogeneitybetweencommunities intermsofexposuretofloodwaters, their preparednessmeasures, theexistenceof elevatedevacuation shelters, theirknowledge and use of existing early warnings, and their familiarity and exposure to the

16

NRCS. The study team observed neighbouring villages with entirelydifferentlevelsofcommunityorganisation.Onevillageclearlyhadalongtermrelationshipwith NRSC, was in the process of building an elevated cement evacuation shelter, hadindividuals trained in search and rescue,waswell awareof floodwarning textmessages,and has a community organisation with pooled funds into which households madecontributions with a communally held bank account that could be accessed duringemergency times, including during flood events. During the focus group discussion theymentionedmultipleactionsatbothhouseholdandcommunity level that theyarealreadytaking in the time period directly before a flood and while their roads and home areinundated, includingmoving the elderly, disabled, pregnant and young children to higherground.During focusgroupdiscussions communitymembersdescribed roadaccessbeingwipedoutduringfloodeventsandthustheysentasmalldelegationorable-bodiedpersonstowadethroughthewatertobuypreparedfoodsandwatertobringbacktothebroadercommunity.Theseitemsinthisinstancewereboughtoncreditandpaidforlaterwhentheycouldaccesstheircommunalbankaccount.In contrast, the neighbouring community appeared to have a very low level of selforganisation, includingnocommunal loansandsavingsorcontingencyfunds,noexposureto search and rescue training, little to no awareness of existing flood forecasting earlywarnings,andlittleconnectiontoNRCSoranyotherexternalorganisation.Admittedlythestudy teamwas only able to visit a small number of communities and so an analysis ofvulnerability and exposure must be primarily informed by secondary sources and theexperientialknowledgeofNRCSandotherstakeholders.However,thecontrastbetweenthetwoneighbouringcommunitieswasstrikinganddemonstratednotonlytheheterogeneityof preparedness and organisation at community level but also highlighted two keychallenges;Firstly, thechallengetodesignactionsthatwouldbesuitableforcommunitieswithvariablepreparednessandself-organisation,andsecondly, the importanceofvillage-level targeting post-trigger so as not to exclude communities that are poorly networkedwithexternalorganisationsandwith low levelsof self-organisation if their vulnerability iscomparativelyhigh.ExposureExposuretofloodwaterswithintheTeraivariesenormously.Themostflood-pronedistrictsandsub-districtsarewellunderstoodbyallparties.However,howexposurevariesatvillageand household level is a significant gap. There appeared to be a belief held amongstakeholders thatproximity to theriverbankwas theprimarycharacteristicofexposure -thatthehousesandvillagesclosesttotheriver’sedgewerenecessarilythemostatriskofflood impacts.However,otherfactorsare important indicatorsofexposure. Inadditiontoproximity to rivers, elevation (which is often visuallymasked in the plains environment),positionvis-a-vistheupstream-downstream,andthebuildingofmomentumoffloodwatersastheymoveacrossthelandscapewillbenecessarytoconsider.Further,therearelikelytobe changes in vulnerability in future with the establishment of new large-scale watermanagementinfrastructureprojects.Therearemultiplelargeprojectsinprogressorattheplanning stages throughout the Terai to better manage water. This will have a positiveeffect on reducing flooding in some areas and a negative effect on others. It will beimportant for NRCS to keep abreast of these developments and review all availabledocumentationtounderstandwhichcommunitiesmayhaveanincreasedrisk-especiallyif

17

thosecommunitieshavenottypicallybeenflood-proneandassuchmayhavefewinternalcopingcapacities.An impact-based FbF system is meant to be able to provide guidance on where to takeaction following a trigger. Given there is a such a high level of variability in bothpreparednessandexposureintheflood-pronedistricts,effectivecommunity-leveltargetingwill be essential. A communitymapping exercise using geo-location such as Open StreetMap may be beneficial to understand the differing vulnerability. This would be anopportunitytomarkwhichcommunities(A)haveafunctioningcommunityorganisation,(B)have a suitable evacuation shelter sufficient for their population, (C) whether there areelevatedareas/homeswhichcouldprovidetemporaryshelter, (D)haveanelevatedwatersource,(E)theirelevation,and(F)whethertheirfloodriskislikelytochangeinthenext1-5yearsasaresultoflargewaterinfrastructureprojects.DangerlevelsTheworkthathasbeendonebyDHMonflooddangerlevelsthusfarhasfocusedonroadsubmersion.Although there is likely tobe somevariation from road to road, broadly thefirst danger level ofDHM is set at the flood levelwhereby it is expected thatmain ruralroadswillbesubmerged.Thiscorrespondswitha1in2-5yearreturnperiod,whichisalsothe minimum threshold for an FbA by DREF EAP. The submersion of roads is a logicalthreshold for FbF actions as it limits access to communities in need. Thenational societystaffandvolunteersatchapterlevelinBardiyaexpressedfrustrationabout‘sittingontheirhands’ waiting for water levels to recede before being able to support communities.Triggeringandcompletingactionsinadvanceofwhenruralroadsbecomeimpassablehasaclearlogic.Givenalloftheabove,itisrecommendedtousethesamedangerlevelasthatusedbyDHMforriverfloodswithareturnperiodof1 in2-5years. Inpracticethiswouldmeantriggeringinanticipationofeventsofthismagnitudeorgreater.PossibilitytodevelopanImpact-basedforecast(IBF)IBFisquicklybecomingthegoldstandardforsevereweatherwarnings,althoughithasbeenimplemented in very few low-income country contexts to date. IBF aims to go beyondforecastedwhat theweatherwillbe,and instead forecastswhat theweatherwilldo (theimpacts).Work has been initiated globally by the United KingdomMeteorological Office(UKMO)andendorsedbytheWorldMeteorologicalOrganization(WMO).InitialdiscussionshavealreadytakenplacebetweenUKMOandNepalstakeholdersincludingDHMandWFPabout the establishment of an IBF system in Nepal to produce robust severe weatherwarnings for multiple hazards, supported by the Asia Regional Resilience to a ChangingClimate (ARRCC) programme.Development of an IBF systemby the respectivemandatedauthorities is an opportunity to enable FbF systems. The warnings produced by an IBFsystemcouldbeusedastheFbFtrigger.ThetypeofIBFwhichcouldbeestablishedinNepalis nicely explained in this video focusing on the system in South Africa(https://www.youtube.com/watch?v=uwpswXH2X9Y).Atpresent, thereare informalpractices to integrate lotsofdata sourcesandexperientialknowledge already happening. For example,multiple district-level stakeholders explained

18

lookingattheoutputsoftherivergauge(whicharepresented inreal-time on the government website: http://hydrology.gov.np/#/river_watch?_k=n4pc2t) andlookingat the rain forecastupstream,andqualitativelymakingadecisionaboutexpectedflooding in the area based on their experiential knowledge of what happens when boththoseindicatorsareindicatinghighrisk,atwhatpointduringthemonsoonseason,andforwhichcommunities(basedonatacitunderstandingofelevationandlocationasitpertainsto localdykes,embankmentsandotherstructuresaffected thebehaviorof floodwaters).Basedon this, thepolice embeddedwithin thedistrict emergencyoperations centres aretheninformedoftheelevatedriskoffloodingsothattheycantakeaction(notpre-defined).In thisway, an informal IBF is already takingplace in someareasof theTerai.Anationallevel IBF would make the kind of calculations already happening mentally about whichindicators produce impacts and where, and integrate this kind of thinking in a morestandardized, empirically-grounded way, at higher scales, and for multiple hazards. Thesystem would allow for a clear integration of vulnerability and exposure indicators tohighlight where the most severe expected impacts will occur, which areas/elements areexposedtothehazardsandwhoorwhat infrastructure ismost likelytobeaffectedgiventheirvulnerabilityconditions.NegotiationsarecurrentlyunderwaytoestablishsuchasysteminNepal.Afullsystemcouldbeoperationalby2022.Inthemeantime,forecastingtheimpactoffloodsinNepalcanstillbe done with sufficient vulnerability and exposure data, building off the informal workanticipating impacts that is already happening at district-level. To do this, FbF projectstypically tryandproducean impactcurvewhichcharts thedamageproducedbyahazardwith the return period (extremity) of that hazard to more reliably set the threshold foraction.AnanalysistoproduceanimpactcurvewillbecompletedbytheClimateCentrefortheFbAbyDREFEAP in thecomingyear.There is insufficient timeandaccess todata (atpresent) to complete such an analysis for the 2019 monsoon season, and as such it isrecommendedthatthethresholdbesetatthebankfulllevel,whichisthecurrentthresholdusedbyDHMforfloodalerts.Producinganappropriateimpactcurvetoproduceanimpact-based FbF trigger for the coming years rests on close collaborationwithWFP.WFP havealready completed extensive work on vulnerability mapping through their VulnerabilityAnalysisMapping(VAM)tool,andassuchitdonenotmakesensetotryanddoaseparatepiece of work on vulnerability. It will be important to collaborate closely with WFP toestablish joint triggersand jointly integrate theirvulnerabilityandexposurework into themulti-step flood forecasting system designed for the FbF program. This will also involveaccessing or creating location data with the view to create a bespokemap of the Terai.InitialdiscussionswithWFPindicatetheyarehappytoengage,butoutliningclearrolesandresponsibilitiesstillneedstobedone.The Asia Regional Resilience to a Changing Climate (ARRCC) programme offers a uniqueopportunity to develop an FbF system that is co-produced and co-implemented bymanynational actors inNepalwith support fromexternal partners, including RCCC andUKMO.The trigger level for the 2019monsoon season will be set for one river basin using theexisting danger level (bank full level, 1 in 2-5 year return period) set by the nationalgovernment for issuingwarnings,as this isasmallpilotsystemdesignedprimarily togainknowledgeonactionsimplementationandthereisinsufficienttimetodoadditionaltriggeranalysis. However, in the long term, amore sustainable, integrated approach is possible

19

under the ARRCC program. Specifically, this program will include IBFlearning labs, through which key actors will be able to come together and co-designcomponentsoftheIBFsystem,whichideallywillalsobetheFbFtriggeringmodelinthelongterm.Jurisdictionfordisseminatingtheforecast/earlywarning

The jurisdiction for disseminating warnings at district level rest with the Chief DistrictOfficer, who in turn is in close coordination and communication with high levels ofgovernment, including DHM headquarters. The FbF concept was explained to the ChiefDistrictOfficer inBardiyawheretheFbF2018pilottookplaceandwherethe2019trial ismeant to takeplace.The feedback fromtheChiefDistrictOfficerwas thatanFbFsystemmustmaintain the integrityof theDHMforecast. Inotherwords, theDHMforecastmustremainintactandnotaltered.Further,ChiefDistrictOfficerindicatedthataddingalayerofvulnerabilityandexposuredatainadditiontothehazardforecastwouldbeacceptable.Thisis fine because we do not need to make any alterations to the 1-day forecast, which issufficiently reliable for an FbF system. However, it will be important tomake sure in allcommunications that the DHM 1-day forecast is the basis of the activation trigger. It ispossible that presentations of the recommended triggering system, once the GLOFASreadiness trigger and theexposure and vulnerabilitydata layers are added,willmask theunderlying forecastused. Itwillbe important that is remainsvisiblyanapplieduseof theDHMoutput.Lastly,therecommendedtriggeringsystemoutlinedonpage14hasnotbeenpresentedtoDHMeitheratheadquartersoratdistrictlevel.Itwillbeimportanttopresenttherecommendedtriggeringsystem(-7dayGLOFASmodel,-1dayDHMtriggersetc)togettheir feedback and approval so as not to overstep jurisdictions as it pertains to thedisseminationofwarnings/forecasts.

NextstepsforintegratingtheavailableforecastsanddataintoanFbFsystem

Note:theserecommendationsareforthedevelopmentofanFbFbyDREFEAPatnationalscale,andmanyofthesestepsarenotnecessary/feasibleforthecomingmonsoonseason.ThestudyteamrecommendsthefollowingstepstointegratetheavailableforecastsintoanoperationalFbFsystem.ItmaybepossibleforsomeofthesestepstobecompletedaspartoftheARRCCprogram.ThiswillneedtobeplannedoutwithARRCCstakeholdersatalaterdate.

Step 1: Identify a climatologist or hydrologist in the government who can be the pointpersonforFbF,andtraintheprofessionalontheimplementationproceduresforthetrigger,perhapsincludingastudyvisittoanothercountryFbFprojectoracollaboratinguniversity.

Step2:MoUwithDHMtohaveaccesstotheforecastdigitaldataandadvisorysupport.TheRCCC hydrologist (Dr. Hassan) will collect this observation and forecast data from DHM.Withsupport fromtechnicalexpert fromDHMand localuniversity, thepartieswill jointlycomputetheskilloftheforecast(vis-a-vistheimpactlimit).

20

Step 3: Through consultation with community-level stakeholders andtheNS,confirmwhethera5-yearreturnperiod(floodingofroads)istheappropriateimpactlevel atwhich the communitywill needhelpat the timeofdisaster.UsinghistoricaldatafromtheDHM,confirmtheriverlevelsthataccordwiththisreturnperiod.

Step4:CollectandorganizethevulnerabilitydatafromWFPsuchthatitisinaformatwhichcanbeintegratedwithahazardforecast,withsupportfromRCCC.

Step 5: Develop the relation between hazard intensity against the impact from historicdamage/impact data for different exposure levels. This should be developed in closecooperation and coordination between the department of disastermanagement and theNepalRedCrossNationalSociety,withtechnicalsupportfromtheClimateCentre.

Step6:Developthemodelfortriggeringusingtheforecast,impactcurveandvulnerabilitydata.Thismodelshouldshowthepotentiallyinundatedroads,houses,andinfrastructureatthetriggerfloodlevelandhigher,sothatthenationalsocietycanseeclearlywheretotakeactionwhenthetriggerisreached.

Step7:Completenationalsocietycapacitybuildingatallthestagesoftheworkmentionedintheabovestepsoftriggerdevelopment.Note:capacitybuildingforotheraspectsoftheFbFsystembeyondthetriggerareoutlineinthesectiononcapacitybuilding.

ActionsAnalysisOneofthemajorlessonslearnedcomingfromtheFbFprojectsthathavebeencompletedso far is the importanceof selecting thebestpossibleactions.At times in theearlier FbFprojects, too little focus was placed on action selection as compared to theoperationalization of the triggering mechanism. Ultimately, regardless of the skill of atriggering mechanism, if the actions that the trigger initiates are not very helpful forreducingthespecificdisasterimpactsthataremeanttobeaddressed,theFbFsystemwillnothelpbeneficiariestoreducehazardimpacts.TheselectionofactionsforanyFbFsystemisanimportantonethatmustbalanceanumberof considerations, including the effectiveness of a given action to reduce specific disasterimpacts, the clear benefit in doing the action before the disaster rather than after thedisaster,theabilityforthenationalsocietyandotherpartnerstorealisticallycompletetheaction after the warning but before the disaster hits (thus working very quickly), thesuitabilityof theaction for the local context, theexperience/skillsof locallyengagedstaffand volunteers to complete the specific action, the scalability of the action(s), and thecomplementarityofactions(ifmorethanoneisbeingimplemented).Thereisnoperfectsetofactionsforanygivencontextorcapacitylevel.Whatisneededisaconsiderationofalltheabovefactors,thedevelopmentofacleartheoryofchangeforeachaction selected, and a willingness to re-evaluate with each trigger in a transparent andcriticallyreflexivemanner,sothatovertimethesysteminitiatesthebestactionspossible.

21

Moreextensiveexplorationofpotentialactionstoshortlistforthe2020monsoonandonwardswillbeneeded,especiallyifflashfloodingisgoingtobeaddedtotheEAPonriverine/fluvialfloods,asthiswouldrequireanassessmentofactionsthatcouldbecompletedwithinafewhours.Forthe2019floods,itisrecommendedthatapaireddownversionoftheactionsidentifiedforthe2018pilotbeused.Buildingoffofthe2018FbFpilotandpre-existingpreparednesseffortsThe2018pilotprojectaffordedavaluableopportunitytoexplorepotentialFbFactionsandalsogaugeresponsesbythepilotcommunityandNRCSchaptertothoseactions.Thesetofactionsforthepilotwascomprehensive,withmultiplearrangements inplacesuchaspre-purchaseagreements.TherelevantNRCSchapterofficewaswell-awareofwhattheywouldhavedoneshouldtheyhaveatrigger.Equallytheselectedfocuscommunitywasalsowellawareofwhatwasexpectedandbroadlyreadytotakeaction.Thelessonslearnedcomingoutofthepilotareverywelldocumentedandthecritical reflectionsofthe implementingstaffcontainedwithinthereportswerehelpfulandinformativetothisstudy.Titlesofthosereportscanbefoundintheannexofthisreport.As is the case with many small-scale pilot projects with a pre-defined focal community,thereareaspectsofthepost-triggeractionsplannedforthepilotthatwouldbedifficulttotaketoscaleinalargeFbFsystem.Mostnotablythoseare:

1. In the pilot there were a large number of different items requiring different pre-purchasingagreementsintheadjacentmarkettowntothefloodpronearea.Scalingthisapproachwouldmeancreatingtheseagreementsforvendors inmarkettownsover large areas of the flood prone regions of the Terai. Thiswould incur a largeoverhead cost,whichmay in turn need to be re-done each season dependent onvendor availability. It is recommended that the number of pre-purchasingagreementsbepaireddowntoanumberthatismoremanageabletotaketoscale.This would involve reducing the items included for distribution and also makingthoseagreementsinhubsthatcanservicelargerareas(ideallyoneperdistrict).

2. The2018pilotselectedacommunityasthefocalcommunitythathasahighlevelofpreparedness.Althoughitwasnotpossibletodoacomprehensiveassessmentastohowtypicaltheirlevelofpreparednesswasduringthestudymission,observationsofothercommunities thatwerevisited inneighbouringareasandconversationswithlocal Red Cross staff and volunteers suggest that the pilot community’s level ofpreparednessandinternalorganisationaremuchhigherthanaverage.GiventhatanFbFsystemoperatingatscaleneedstobeabletoworkinanycommunitywheretheanticipatedimpactishighest,actionsneedtobesuitableforcommunitieswithlowerlevels of preparedness. In principle, for FbF systems in all countries, the actionsshould be able to be completed within communities that have no pre-existingtrainingorpre-existingrelationshipwiththeRedCross.

3. The 2018 pilot included significant pre-trigger contact with the focal community,includinganagreementthathouseholdswouldsupplylargebamboostosupporttheset up of evacuation shelters. Actions requiring extensive pre-trigger contact aredifficulttoscalebecause itwouldrequirethat levelofcontact(andanypre-triggeragreements) tobedone invirtuallyeverycommunitywithin thebasis that is floodprone.This isbecause inanFbFsystemoperatingatscale,communitiestoreceivesupportarenotpre-selected,butratherselectedatthetimeofthetriggerwhenthe

22

areas at highest anticipated risk become known (based on theforecast + exposure data + vulnerability data). For the context of this flood FbFprogram in the Terai,whichwillmost likely use a 7 day readiness trigger, contactwith the communities that receive support would begin at this 7 day readinesstrigger.Anythingbeforethe7dayreadinesstriggerwouldneedtohappenineveryfloodpronecommunityinthebasin,whichisunrealisticandwouldincursignificantcost.

TherearemultiplepreparednessinvestmentsthathavebeenmadeintheTeraiwhichformastrongbasisforanFbFsystem.CareshouldbemadetoensureanFbFsystemcapitalizesonalloftheseassets.Thelistofinvestmentsthathavebeenmadeinthisareaisextensiveanditwasbeyondthescopeofthisstudytocaptureallofthem.However,thosethatthestudyteambecomeawareofthatarehighlyrelevantforFbFinclude:

1. ThecoordinationoftheDistrictEmergencyOperationsCentre(DEOC),whichworksin close collaboration with radio stations and other media outlets to disseminatewarnings. An FbF system could include pre-disaster messages which aredisseminatedinthesamefashion.

2. The establishment of Information Management Officers seconded from the RedCross to DEOC to support disaster monitoring. In addition to fostering very closecollaboration between the government and the Red Cross at district level, theseofficers increase the capacity of the DEOC to monitor forecasts and hydro-meteorologicalobservationsandtoanticipatedisasterimpacts.

FeedbackonactionsandplannedactionimplementationfromthefieldDuringthefieldvisit therewereafewfeedbackpointsmadebyresidentsoftheFbFpilotcommunityandchapterstaff/volunteersthatworthreiteratinghere:

1. Itwas communicated that there is insufficient provision for volunteers. The studyteamtookthistomeaninsufficientbudgetforvolunteerperdiemsspecifically.Thiswasexpressedasageneralpoint,butalsopertainingtotheFbFpilot.Assuch,itwillbe important to do a full costing, in advance of the pilot season, of how manyvolunteersareneededoverhowmanydaystocompleteallnecessaryactionsoftheFbF trigger, as well as any pre-season actions, and post-trigger impact evaluationactivities.

2. ItwascommunicatedthatthereisalackofclarityofwhatneedstobedonebyRedCrosschapterstaffandvolunteerspost-triggerforWFP. Itwasexpressedthattheywanttobereadytodowhatisexpectedofthemandnottodisappointanyparties.Further, therewillalsoneedtobeclarityonwhat isexpectedofchapterstaffandvolunteersforactivitiesofboththeWFPandRedCrossFbFprogrammes,especiallyiftheybothtriggerinthesameareasatthesametime(whichislikelygiventhereisaviewtoharmonisetriggers).Thiswillrequirestrongcoordinationandpre-planningforlotsofdifferentpost-triggerscenarios.

3. Aconcernabouttheuseofunconditionalcashwasexpressed.Specificallyaconcernaboutmisuseof cash (spendingon inappropriate things) andalso that cashwouldencourageamindsetofdependenceanddiscourageself-reliance.Ifcashisselectedas a method to support households, these concerns will need to be addressed.Sharing the literature base on cash, or perhaps studies on cash use for FbF (forexampleBangladeshorMongolia) ina format that isaccessiblecouldassist in this

23

effort. New cash-focused programs beginning inNepal such asthe new ECHO-funded programme with cash components will also help raiseawarenessabouttheusesofunconditionalcash.

4. Thechapterstaffandvolunteers inBardiyaexpressedapreferenceforFbFactionsthatfocusonfood,potablewaterandtheprovisionoftoilets.

5. It was communicated by members of the FbF pilot community that they wereuncomfortable with the agreement for them to provide bamboo for the shelters.Although they understood the rationale, they worried that in the stress ofevacuatingthemselvesandtheirbelongingsthattheywouldbetoooverwhelmedtodotheharvesting.

Actionsshort-listingAccordingtointerviews,theprimarybarriertosupportingfloodaffectedpersonsisaccess.Typically, the Red Cross chapter is aware that households and individuals are in need ofsupportbutareinoneofthefollowingscenarios:

A. stranded inevacuationareas (officialorunofficial)with little tonoaccess to food,safewater,latrines,medicineand/orshelter

B. stranded on their rooftops or in trees, but thewater levels not rising or recedingwithlittletonoaccesstofood,safewater,latrines,medicineand/orshelter

C. strandedontheirrooftopsor intreeswithwater levelsrising,requiring immediatesearchandrescuesupport

Additional analysis is needed to understand how correlated these scenarios are to thereturnperiodsoffloods.However,givenanFbFsystemwouldtriggerfor1in5yearfloods,ormoreextremefloodingevents,allofthesescenariosarepossiblefollowinganFbFtrigger.InthescenarioC,whichisthemostextreme,thisleveloffloodingislikelytobefollowedbya large-scale response effort. The selection of actionsmust bear inmind these differentscenarios.It ispossiblethatsomeactionswouldhelpbeneficiariesinscenarioAandBbutbe entirely useless in scenario C. The process of selected actions for the FbF systemwillrequiredcarefulconsiderationofwhatactionswouldhavethehighestpotentialtoreducethelikelihoodofbeneficiariesfindingthemselvesinanyofthesescenariospost-flooding.ScaleofinterventionWhilemostFbFinterventionshavedoneinterventionsatthehouseholdlevel(e.g.providingcash or NFIs to households directly), it is not a given that all FbF interventions need tohappenathouseholdlevel.Itispossiblethatanactioncouldhappenataninstitutionallevel(e.g.providingequipmenttovolunteers),atacommunitylevel(e.g.providingacommunalwater purification unit, or providing communal supplies to an evacuation site) or at abroaderlevelinvolvingthewide-scaledisseminationofmessaging.Duringthefieldvisits,itbecameclearthattherearemanyvillage-levelgroups,suchasvillagedevelopmentgroups,loansandsavingsgroups,etc.Itmaybepossibletoprovidesupportsatthegrouplevel,suchas cash transfers to established community groups with bank accounts, to increase thecoverage of the FbF intervention and build off the investments that have beenmade inestablishing andmaintaining these community structures. Action at the community levelinvolvesworkingoutwhichareasarecoveredbyafunctioninggroupwhichcouldbegivensupport. If a mapping of functioning community groups does not already exist, then a

24

mappingexercisemightneed tobeundertaken to targetat this level.Decisionsaboutthescaleofinterventionisultimatelyuptothelocalstakeholders.ReadinessandImmediatepre-disasteractionsIf the suggested triggeringmodel (-7 day readiness trigger and -1 day pre-disaster actiontrigger)isadopted,thentwosetsofcomplementaryactionsareneeded.Readinessactionsare meant to be lower cost, no regret actions which can be taken to get ready for asubsequentpre-disaster triggerwhen there is insufficient skill tobeable tobeginhigher-costactionswithmoreleadtime.Giventhereisahighlevelofuncertaintyatthisleadtime,they should all be actions that the national society is comfortable taking in a false alarmscenario. Incontrast,pre-disasteractionstaken-24hoursbeforethedisastercanbedonewithahigherlevelofcertaintythattheforecastiscorrect.Theseactionsmustbeabletobereasonably completed in the time frame afforded by the forecast, which in this instancewouldbe24hours.Note,thatthe-24hourperiodpre-disastercouldcommenceatanytimeofday, including inthemiddleof thenight,onapublicholidayetc,andsoplansmustbemade that are suitable to fully take advantageof the24hours, regardlessof the timeoftriggering.The final selectionof actionsneeds tobedoneby local stakeholders. This study canonlymakerecommendationsofwhichactionstoconsiderbasedoninterviewsduringthestudy,althoughsomeactionsarelikelytobeessentialforthefunctioningofthesystem(markedbelow).Actionsrecommendedtobedonefollowingthereadinesstrigger(initiatedat-7dayleadtimeandcompletedbefore-1day)areasfollows:

1. Activation of the CDMC (including the police, NDEOC, Red Cross and others)[essential]

2. Selection of villages/households to support, based on where the highest risk isindicated[essential]

3. Refresher training for plans of evacuation, search and rescue practices (and othertypesoftrainingsifrelevant)

4. Volunteervisitationtoareashighlightedbythetriggermodelashavingthehighestlevelofrisk.Duringthistimethey:

a. Communicatetheelevatedrisk,uncertaintyof the forecast,plans (-24houractivities)ifalaterforecastconfirmstherisklevel

b. Identification of areas of high elevation based on the communitymappingexerciseorcommunityperception/expertjudgement

5. Volunteervisitationtosafeareastoinspectitmeetsminimumcriteriaforasafearea(criteriatobedeveloped,butmustincludeaminimumelevationsoasnottoalsobesubmergedinanextremeevent)

6. Volunteers position basic supplies, based on needs, to safe areas [supplies to bedetermined, but should be very low cost at this stage]. This could include theinstallation of a temporary toilet (pending a review as to whether these can beprocuredandinstalledwithintheleadtime).

Recommended immediate pre-disaster actions (initiated at -24 hour lead time to becompletedbeforeroadsbecomeinaccessible)areasfollows:

25

1. Dissemination of warning via radio, loudspeaker, in person byvolunteers, and/or SMSwith clear guidance on the steps affected persons shouldtaketoreducetheirrisk

2. Distributionstohouseholds(itemstobedetermined,butcouldincludefood,tarps,WASHitems,cash).

3. Distributionofsearchandrescueessentialitemstovolunteersintheareaidentifiedashavingthehighestrisk.

CapacityAssessmentsNationalSocietyCapacityGeneralNRCScurrentlyundergoingseveralchanges.Thesocietyiswrappinguplargescaleresponseand recoveryactivities following the2015earthquake,whichmay result ina reductionofcorestaffatheadquarterslevel.ThenumberofPNSdelegatesisalsoplannedtobereducedwithseveralPNSsplanningtocoordinatewithshareddelegates.Thelong-standingheadofdisastermanagementhasrecentlyretiredwithanewstaffpersoncomingfromthehealthsectortotakehisplace.Althoughitisexpectedthatthistransitionwillbesmooth,someofthe institutional knowledge may be lost. In particular, the now retired head of disastermanagementwaswell-versedintheFbFconcept,thepilotwhichwasundertaken,andthelessonslearnedcomingoutofit.Goingforward,thestaffpersonsworkingonFbFwillneedtomakesurethenewseniorlevelofthedisastermanagementdepartmentiswellengagedintheprocess.GivenaPreparednessforEffectiveResponse(PER)processhasrecentlybeeninitiated,thisdocumentation was used to assess general NS capacity, rather than doing a light-touchcapacity assessment during the two-weekmission toNepal. A reviewof theDRCE reportindicates many strengths relevant to FbF including good coordination between partners,timely deployment of volunteers, appropriate reporting, strong external communication,strongcommunityengagement, andappropriateuseandmanagementof stocksandpre-positioned supplies. Weaknesses relevant to FbF include ill-defined scales at which torespond, lack of some procedures for internal responsemanagement, lack of safety andsecurity policies to keep volunteers and staff safe, and little evidence of the use of earlywarnings forhazards toactproactively.Theseassessmentsweremore thorough than thelimited amount of capacity assessment which could be done as part of this study. Theassessment resultsdocumented in theDRCEare largely commensuratewithobservationsmadeduringthemission.FinanceDuringinterviewswiththefinancedepartment,theywereveryopentotheFbFconceptbutexpressed some concerns about how quickly funds could move from headquarters todistrictchapterlevelthroughtheDREFmechanism.Forthe2019monsoonseasonthiswillnot be a problem as the fundswill bemade available bilaterally through theDanish RedCross, which currently keeps sufficient budget in-country. For the FbA by DREF (to beconsideredfor lateryears),both IFRCandFbF-relatedstaffatHQwillneedtoengagethe

26

finance team in thedecisionsaroundhowthe fundswillbemanaged.There aremany options acceptable to IFRC in this regard, including using funds held in-countrywhicharelaterreplenished.TechnicalServiceCapacityIt was not possible to do ameaningful assessment of DHM’s capacity, apart from a skillanalysisoffloodforecasts(discussedonpage9).However,thesetupatdistrict level (theissuanceofwarnings,datamanagement, integrationforRedCrosschaptertogovernmentemergencyoperations)appearshighly conducive toanFbF system. ICIMOD (InternationalCentre for IntegratedMountainDevelopment) is also headquartered inNepal,which is avaluableasset.ICIMODandDHMhaveastrongcollaboration.CapacityoftheFbFCommunityofPracticeThis project benefits from an established FbF community of practice. There is an FbFworking group which meets regularly and multiple partners who have been involved indevelopingFbFsystems,withstaffwhohaveattendedFbFdialogueplatforms.ItisastrongfoundationfortheNRCStobeginlargerscaleforecastbasedwork.Coordinationwillneedtobeprioritisedformultipleaspectsoftheprogram-mostnotablythetriggeringmodalities(forecast,dangerlevel,integrationingovernmentsystems)andtheimplementationofpost-triggeractions(locationsforeachinstitutiontosupportandtheNRCSground-levelstaffandvolunteers to implement the actions). Additional MOUs may need to be developed toachievetheseaims.Recommendationsonareasforcapacitybuildingforstakeholders

ThestakeholdersofinterestinthiscasewouldbeprimarilyDHM(includingcounterpartsatlower levels of government), local governments, NRCS and relevant partner nationalsocieties, relevant disastermanagement authorities (at all levels), and othermembers oftheFbFworkinggroupincludingWFPandPracticalAction.

1. AnypartyusingtheoutputoftheFbFtriggermusthavethecompetenciestoa. understand and interpret the output (which will be a map with layers of

hazard, exposure and vulnerability information) accurately with anunderstandingoftheuncertaintyinherentwithinit.

b. usetheoutputto identifycommunities totarget forearlyaction.Fromthisbasis, apply additional targeting criteria (if any), to identify specifichouseholdswhichwillbereceivingforecastbasedsupport(forexample,non-fooditems).

c. havethesoftsystemsinplacetobeabletoworkveryquicklyduringtheleadtimeaffordedbythewarning.Thismayrequiredry-runsandotherexercisesto practiceworking at the speed required (including at headquarters level,suchastheadministrationoffinanceetc.).

d. evaluate the impact of those actions in a rigorous way. The RCCC M&Especialist can support on some aspects of this as needed. Evaluating the

27

impact will be necessary for continually improving theEAPandforfurtheringthelearningofthelargerFbFcommunityofpractise.

2. Parties involved in the original design of the FbF triggering systemmust build thecapacity to evaluate the trigger post-disaster, update the trigger as necessary forsubsequentEAPswithminimalexternalsupport,andeventuallydeveloptriggersforadditionalhazards.

3. TheNepalRedCrossSociety,IFRC,andsupportingpartnernationalsocietiesshouldworktogethertobuildthecorecompetenciesthatsupporttheFbFsystem,includingdata preparedness, GIS skills, strong financial management, and capabilitiesassociatedwithspecificactionsincludingcashreadiness,WASH,health,andshelter-specific skills as well as building an understanding of the FbF concept and triggermodel.

4. The earlywarning systems (EWS) inNepal are highly variable district to district intermsoftheirfunctioning,reliability,usewithintheircatchmentareasandsoforth.TheweakEWSsneedtobestrengthenedsuchthattheyprovidetheearlywarningservicethatthepublicisentitledto.Thiswillrequireadditionalinvestment.

5. Access to safe evacuation shelters with sufficient sanitation provision and thecapacity to shelter sufficient numbers of people, remains a major challenge. Forextreme flooding events, there is no replacement for necessary shelterinfrastructure. The FbF working group must continue to advocate for emergencyshelter provision and make clear that FbF action will not replace the need forfamiliestoevacuatetosafeareasinextremeevents.

NextStepsPlansforMonsoonseason2019RationaleGiventheinvestmentthatwasmadeinthe2018FbFpilot,andtheenthusiasmaboutFbFatboth chapter and headquarters level, it is recommended that the national societyimplement an additional pilot for the coming season. Working in Bardiya District isrecommendedas theRedCrosschapter isalreadyvery familiarwithFbF.Therearemorethan125floodpronecommunitiesinBardiyaDistrict.FundingpracticalitiesThe Danish Red Cross is able to bilaterally fund the post-trigger costs, if the programmetriggers in the 2019 monsoon season. It was understood by the study team during themissionthatthereissufficientfundsheldincountryforanactivationofthisscalewhichcanlaterbereplenished.Assuch,theabilitytomakeimmediatetransfersfromoutsideNepalisnotneeded.ThepracticalitiesofgettingfundingtoBardiyaDistrictimmediately,iftriggered,needstobereviewedbytheDanishRedCrossincountry.

DesignrecommendationsAsthiswillbeapracticeforhowanFbFsystemwouldfunctionatscale,itshouldnotuseapre-defined community.Rather, it shouldbe ready to takeaction inanyof the125 floodpronevillagesinBardiya,wherethetriggermodelindicatesthehighestrisk.Thenumberofvillagesthatcouldcoveredthisseasonstillneedstobedetermined.Itisrecommendedthat

28

this trial run covers approximately 500households to gain experienceworkingatscalebasedonaforecast.Ifthereisatrigger,itislikelythattheareahighlightedas having the highest needwill be larger than the program is able to support. Targetingcriteriawillneedtobedeveloped.Ifthereareareashighlightedathighestriskthatincludesvillages that theRedCrosshasnotpreviouslyworkedwith, it is recommendedthat thosecommunities are included to gain practice working in areas without an establishedrelationshipwith the chapter. The actions recommended for this season are the sameasthose outlined on page 23. A smaller set of actions may need to be prioritised for thisseason.Additionaldata/informationneededtosetthetriggerfor20191. TheexactlocationsoftheforecastingpointsfromDHMfortheselectedbasin(affectingBardiyaDistrict),aswellastheobservationdataforthoselocations. Ideally,thiswouldbe10yearsofobservationdataforthesestations,but5yearswouldbesufficient.Itispossiblethatthisdatamayneedtobepurchased.2.Requestthe2017floodexposuremapfromDHM.3. Request a topographicmap of Bardiya to understand the elevation of communities ascomparedtoroads,toidentifywhichcommunitiesarebelowtheroadlevel.4. Request to use WFPs VAM. This can be combined with the 2017 exposure map andtopographymapfortheselectedbasintoguidevillageleveltargetingpost-trigger.Nextstepstobetrigger-readyforthe2019MonsoonseasonThefollowingstepsneedtobecompleted.TheRCCCcansupportremotelyonallsteps,asrequested.

1. Setthetriggera. Presenttheoutlinedtriggermodeltorelevantstakeholderstogetfeedback

andbuy-in (NRCS,PNS, IFRC,DHM,WFP,PAC). This should ideallybedonethroughtheFbFworkinggroup.

b. Establish MoU with DHM (if not possible for this monsoon season, thendefinitelyforthefollowingseasons),includingtheagreementtoprovidereal-time notification for when the FbF trigger has been reached to relevantindividualswithintheRedCrossandotherpartners.

c. RequestadditionaldatafromDHMandWFP(outlinedabove)d. RequestsupportfromWFPforvulnerabilitymapping(ideallygetpermission

tousetheirVAMoutput).e. Makeafinaldecisiononthetriggermodelbasedonfeedback.

2. Definethescaleofintervention(e.g.thenumberofbeneficiaries)3. Selecttheactionsfollowingbothtriggersanddevelopatheoryofchangeforthose

actions4. Makeapost-triggercoordinationplanwithWFPiftheyintendtobeoperationalfor

thecomingmonsoonseason.5. Make an outline of how the systemwill be evaluated (the trigger, actions) if the

systemtriggers.6. ArrangefinancinginternallywithinDanishRedCrossforpre-triggerset-upactivities,

post-triggercostsintheeventualitythatthesystemtriggers.7. MakeanewcontractanddeliverableswithRCCCforongoingsupport.

29

ProposedongoingsupportfromRCCCBasedonthestudyteam’sunderstandingofwhatthemainsupportneedsare,theareasofongoingsupportareasfollows:

1. Remote support to be trigger-ready for smaller scale bilaterally funded trigger for2019season

2. RemotesupporttodeveloptheEAPforFbAbyDREFthroughthedesignphaseandallstepsofthevalidationprocess.Thiscouldincludein-countrysupport,ifneeded.

3. Support for post-trigger impact evaluation to contribute to the ongoing learningwithintheFbFcommunityofpractice

4. Supporttheco-developmentofanImpact-BasedForecastingsystemwithmandatedauthoritiesinNepalandtheUKmeteorologicaloffice.

AdditionalOpportunitiesOpportunitiestoaddressadditionalhazardsinfutureThestudyteamfoundthattheFbFprogrammeshouldstartwithfloods,butcouldincludeno-costactionsforflashfloods.Iftheleadtimeforflashfloodsimproves,costedactionsforflashfloodscouldbeintroduced.Theleadtimewouldneedtoincreaseto24hoursormore.Itiscurrently3-6hours.ThenationalsocietyshouldmaintainclosecontactwithPracticalActiontobeawareoftheresearchoutcomesonthelandslideforecastingsystem.Ifthissystembecomesoperational,itwouldbepossibletodevelopanFbFlandslideprogramme.FbFforheatwavesandcoldwavescouldbeviableincertainpartsofthecountry.Ifthereisinterestamongnational stakeholders, thisEAPcouldbedevelopedquicklyafter the floodEAPhasbeenfinished.OpportunitiestoexploresocialprotectioninNepalalignedwithFbFThere is growing interest in the field of adaptive social protection, with an increasingappreciation that social protection is a valuable modality to support vulnerable peopleaffectedbynaturalhazards.TheintegrationofFbFintosocialprotectionsystemscouldbeavaluableopportunitytoreachvulnerablehazard-affectedpeople.Also,giventheexperienceofDRCandNRCSinsupportingpersonswithdisabilities,DRCandNRCSarewell-placedtosupport in this area.During the fieldmission, a representative ofUNICEF (which is a keypartner in the implementation of social protection in Nepal) expressed an interest inintegrating the triggering system developed into their internal protocol, but morediscussionswould need to take place. If there is interest among national stakeholders toexplore furtherhow the FbF systemcouldbe integrated into social protection systems inNepal,thisisanareawheretheRCCCcouldprovideadditionalsupport.Inparticular,therewouldbea strong logic to integrating an FbF system for cold andheatwaveswith socialprotection systems that target the elderly, infants, chronically ill, and persons withdisabilitiesinbothruralandurbansettings.

30

ConclusionThere is high feasibility for FbF inNepal.Nepal uniquely benefits from a pre-existing FbFcommunityofpractice.Riverine floodhasbeenhighlightedas the firsthazard toaddress,but other hazards including flash floods and heat and cold wave have potential. Whilecapacityneedstobebuiltinsomeareas,theNRCSiswell-positionedtoimplementanFbFprogram. If local partners are committed, it will be possible to develop an EAP of asufficientlyhighstandardsoastopassthevalidationprocess.AnFbFsystemoperatingatscale in Nepal has the potential to substantially reduce disaster impacts before they areexperienced.AnnexAnnex1:Documentsreviewedpriortothefieldstudy

● ForecastBasedFinancePilotProjectBardiya–LessonsLearned2018● ForecastBasedFinanceDiscussionPaper(unofficial)2018● FloodHazardMaps● GISBasedFloodHazardMappingandVulnerabilityAssessmentofPeopleDueto

ClimateChange:ACaseStudyfroKanakaiWatershed,EastNepal● NepalFloodSnapshot● StandardOperatingProcedures,ForecastBasedFloodPreparedness,Bardiya● StandardOperatingProcedures,ForecastBasedFloodPreparedness,Banke● StandardOperatingProcedures,ForecastBasedFloodPreparedness,Dang● DepartmentofMeteorologyandHydrologyFloodBulletins● WFPPre-InterventionSurvey● ForecastbasedFinancingandEmergencyPreparednessforClimateRisks(FbF)● DanishRedCrossForecastbasedFinanceConceptNote● DanishRedCrossForecastbasedFinanceOnePager● DepartmentofHydrology&MeteorologyHydrologicalNetworkofNepal(Sept.

2015)● DepartmentofHydrology&MeteorologyStudyofMajorWeatherEvents● DepartmentofHydrology&MeteorologyListofHydrologyStations● NRCSVCAs(limitednumbertranslatedintoEnglish)