The challenge of unprecedented floods and droughts in risk management
0 likes35 views
This document analyzes a dataset of 45 pairs of flood or drought events that occurred in the same area an average of 16 years apart. It finds that while risk management has generally reduced the impacts of floods and droughts, it faces difficulties managing impacts from unprecedented extreme events. In two cases where the second event was much more hazardous, improved risk governance and investment led to lower impacts. However, the challenges of managing unprecedented extremes are concerning given projections of more severe hydrological events due to climate change.
The challenge of unprecedented floods and droughts in risk management
- 1. 80 | Nature | Vol 608 | 4 August 2022 Article Thechallengeofunprecedentedfloodsand droughtsinriskmanagement Heidi Kreibich1 ✉, Anne F. Van Loon2 , Kai Schröter1,3 , Philip J. Ward2 , Maurizio Mazzoleni2 , Nivedita Sairam1 , Guta Wakbulcho Abeshu4 , Svetlana Agafonova5 , Amir AghaKouchak6 , Hafzullah Aksoy7 , Camila Alvarez-Garreton8,9 , Blanca Aznar10 , Laila Balkhi11 , Marlies H. Barendrecht2 , Sylvain Biancamaria12 , Liduin Bos-Burgering13 , Chris Bradley14 , Yus Budiyono15 , Wouter Buytaert16 , Lucinda Capewell14 , Hayley Carlson11 , Yonca Cavus17,18,19 , Anaïs Couasnon2 , Gemma Coxon20,21 , Ioannis Daliakopoulos22 , Marleen C. de Ruiter2 , Claire Delus23 , Mathilde Erfurt19 , Giuseppe Esposito24 , Didier François23 , Frédéric Frappart25 , Jim Freer20,21,26 , Natalia Frolova5 , Animesh K. Gain27,28 , Manolis Grillakis29 , Jordi Oriol Grima10 , Diego A. Guzmán30 , Laurie S. Huning6,31 , Monica Ionita32,33,34 , Maxim Kharlamov5,35 , Dao Nguyen Khoi36 , Natalie Kieboom37 , Maria Kireeva5 , Aristeidis Koutroulis38 , Waldo Lavado-Casimiro39 , Hong-Yi Li4 , María Carmen LLasat40,41 , David Macdonald42 , Johanna Mård43,44 , Hannah Mathew-Richards37 , Andrew McKenzie42 , Alfonso Mejia45 , Eduardo Mario Mendiondo46 , Marjolein Mens47 , Shifteh Mobini48,49 , Guilherme Samprogna Mohor50 , Viorica Nagavciuc32,34 , Thanh Ngo-Duc51 , Thi Thao Nguyen Huynh52 , Pham Thi Thao Nhi36 , Olga Petrucci24 , Hong Quan Nguyen52,53 , Pere Quintana-Seguí54 , Saman Razavi11,55,56 , Elena Ridolfi57 , Jannik Riegel58 , Md Shibly Sadik59 , Elisa Savelli43,44 , Alexey Sazonov5,35 , Sanjib Sharma60 , Johanna Sörensen49 , Felipe Augusto Arguello Souza46 , Kerstin Stahl19 , Max Steinhausen1 , Michael Stoelzle19 , Wiwiana Szalińska61 , Qiuhong Tang62 , Fuqiang Tian63 , Tamara Tokarczyk61 , Carolina Tovar64 , Thi Van Thu Tran52 , Marjolein H. J. Van Huijgevoort65 , Michelle T. H. van Vliet66 , Sergiy Vorogushyn1 , Thorsten Wagener21,50,67 , Yueling Wang62 , Doris E. Wendt67 , Elliot Wickham68 , Long Yang69 , Mauricio Zambrano-Bigiarini8,9 , Günter Blöschl70 & Giuliano Di Baldassarre43,44,71 Riskmanagementhasreducedvulnerabilitytofloodsanddroughtsglobally1,2 ,yet theirimpactsarestillincreasing3 .Animprovedunderstandingofthecausesof changingimpactsisthereforeneeded,buthasbeenhamperedbyalackofempirical data4,5 .Onthebasisofaglobaldatasetof45 pairsofeventsthatoccurredwithinthe samearea,weshowthatriskmanagementgenerallyreducestheimpactsoffloodsand droughtsbutfacesdifficultiesinreducingtheimpactsofunprecedentedeventsofa magnitudenotpreviouslyexperienced.Ifthesecondeventwasmuchmorehazardous thanthefirst,itsimpactwasalmostalwayshigher.Thisisbecausemanagementwas notdesignedtodealwithsuchextremeevents:forexample,theyexceededthedesign levelsofleveesandreservoirs.Intwosuccessstories,theimpactofthesecond,more hazardous,eventwaslower,asaresultofimprovedriskmanagementgovernanceand highinvestmentinintegratedmanagement.Theobserveddifficultyofmanaging unprecedentedeventsisalarming,giventhatmoreextremehydrologicaleventsare projectedowingtoclimatechange3 . Observeddecreasingtrendsinthevulnerabilitytofloodsanddroughts, owingtoeffectiveriskmanagement,areencouraging1 .Globally,human andeconomicvulnerabilitydroppedbyapproximately6.5-and5-fold, respectively, between the periods 1980–1989 and 2007–2016 (ref. 2 ). However,theimpactsoffloodsanddroughtsarestillsevereandincreas- ing in many parts of the world6 . Climate change will probably lead to afurtherincreaseintheirimpactsowingtoprojectedincreasesinthe frequencyandseverityoffloodsanddroughts3 .Theeconomicdamage offloodsisprojectedtodoubleglobally7 andthatofdroughtstotriple in Europe8 , for a mean temperature increase of 2 °C. The purpose of risk management is to reduce the impact of events through modification of the hazard, exposure and/or vulnerability: accordingtoUnitedNations(UN)terminology9 ,disasterriskmanage- mentistheapplicationofdisasterriskreductionpoliciesandstrategies to prevent new disaster risk, reduce existing disaster risk and manage residual risk, contributing to the strengthening of resilience against, andreductionof,disasterlosses.Hazardisaprocess,phenomenonor humanactivitythatmaycauselossoflife,injuryorotherhealthimpacts, property damage, social and economic disruption or environmen- tal degradation; exposure is the situation of people, infrastructure, https://doi.org/10.1038/s41586-022-04917-5 Received: 19 August 2021 Accepted: 30 May 2022 Published online: 3 August 2022 Open access Check for updates A list of affiliations appears at the end of the paper.
- 2. Nature | Vol 608 | 4 August 2022 | 81 housing,productioncapacitiesandothertangiblehumanassetslocated inhazard-proneareas;andvulnerabilityistheconditionsdeterminedby physical,social,economicandenvironmentalfactorsorprocesses10–13 thatincreasethesusceptibilityofanindividual,acommunity,assetsor systems to the impacts of hazards. To be effective, risk management needs to be based on a sound understanding of these controlling risk drivers14,15 .Paststudieshaveidentifiedincreasingexposureasaprimary driver of increasing impacts3,4 , and vulnerability reduction has been identified as key for reduction of impacts16,17 . However, ascertaining thecombinedeffectofthedriversandtheoveralleffectivenessofrisk managementhasbeenhamperedbyalackofempiricaldata4,5 . Hereweanalyseanewdatasetof45 pairsoffloodordroughtevents that occurred in the same area on average 16 years apart (hereinafter referredtoaspairedevents).Thedatacomprise26 floodand19 drought pairedeventsacrossdifferentsocioeconomicandhydroclimaticcon- texts from all continents (Fig. 1a). We analyse floods and droughts together,becauseofthesimilarityofsomeofthemanagementmeth- ods (for example, warning systems, water reservoir infrastructure), the potential for trade-offs in risk reduction between floods and droughts and therefore value for the management communities to learn from each other18 . The impact, quantified by direct (fatali- ties, monetary damage), indirect (for example, disruption of traffic or tourism) and intangible impacts (for example, impact on human health or cultural heritage), is considered to be controlled by three drivers:hazard,exposureandvulnerability3 .Thesedriversarequantified usingalargerangeofdifferentindices—forexample,thestandardized precipitationindex,thenumberofhousesintheaffectedareaandrisk awareness, respectively (Supplementary Table 1). These three driv- ers are considered to be exacerbated by management shortcomings. Hazardmaybeexacerbatedbyproblemswithwatermanagementinfra- structuresuchasleveesorreservoirs19 .Exposureandvulnerabilitymay be worsened by suboptimal implementation of non-structural meas- uressuchasrisk-awareregionalplanning20 orearlywarning21 ,respec- tively. We analyse management shortcomings and their effect on the threedriversexplicitly,asthisisthepointatwhichimprovementscan start—forexample,bytheintroductionofbetterstrategiesandpolicies. 12 40 16 26 25 1 3 14 39 28 18 4 35 20 34 13 37 6 19 33 2 36 31 42 41 44 10 21 22 11 23 9 15 27 45 8 38 24 7 17 32 43 5 30 29 Large decrease (–2) Small decrease (–1) Change in impact Drought Flood No change (0) Small increase (+1) Large increase (+2) a b Impact Hazard Exposure Vulnerability Management shortcomings 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 19 21 20 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 Fig. 1 | Location of flood and drought paired events coloured according to changes in impact and their indicators of change. a, Location of flood and drought paired events (n = 45). Numbers are paired-event IDs. b, Indicators of change, sorted by impact change. Impact is considered to be controlled by hazard, exposure and vulnerability, which are exacerbated by risk management shortcomings. Maps of the paired events coloured according to drivers and management shortcomings are shown in Extended Data Fig. 1.
- 3. 82 | Nature | Vol 608 | 4 August 2022 Article Dataavailabilityunderstandablyvariesamongthepairedevents,and this can introduce inconsistency and subjectivity. The analyses are therefore based on indicators of change, to account for differences between paired events in respect of measured variables, data quality anduncertainty.Theseindicatorsofchangerepresentthedifferences betweenthefirstevent(baseline)andthesecond,categorizedaslarge decreases/increases(−2/+2),smalldecreases/increases(−1/+1)andno change(0)(SupplementaryTable 2).Tominimizethesubjectivityand uncertainty of indicator assignment, a quality assurance protocol is implemented and indicators of change with sub-indicators are used. Themajorityofpairedeventsshowdecreasesinmanagementshort- comings (71% of paired events; Fig. 1b), which reflects that societies tendtolearnfromextremeevents22 .Mostcasesalsoshowadecreasein vulnerability(80%ofpairedevents)associetiestypicallyreducetheir vulnerabilityafterthefirsteventofapair21 .Thefivepairedeventswith alargedecreaseinimpact(darkblue,topleftinFig. 1b)areassociated with decreases or no change of all three drivers. Driversofchangesinimpact Changes in flood impacts are significantly and positively correlated withchangesinhazard(r = 0.64,P ≤ 0.01),exposure(r = 0.55,P ≤ 0.01) and vulnerability (r = 0.60, P ≤ 0.01) (Fig. 2a), which is in line with risk theory3 .Althoughapreviousanalysisofeightcasestudies21 identified vulnerability as a key to reduction of flood impacts, this new, more comprehensive,datasetsuggeststhatchangesinhazard,exposureand vulnerability are equally important, given that they correlate equally strongly with changes in flood impact. Changes in drought impacts aresignificantlycorrelatedwithchangesinhazardandexposure,but notwithchangesinvulnerability(Fig. 2c).Thissuggeststhatchanges invulnerabilityhavebeenlesssuccessfulinreducingdroughtimpact than flood impact, which is also consistent with those event pairs for which only vulnerability changed (Extended Data Table 1). However, quantificationofthecontributionofindividualdriversisdifficultwith this empirical approach because there are only a limited number of casesinwhichonlyonedriverchanged.Therearethreecasesinwhich only vulnerability changed between events, two cases in which only hazardchangedandnocaseinwhichonlyexposurechanged(Extended DataTable 1).Additionally,pairedeventswithoutachangeinhazard(0) are analysed in more detail to better understand the role of exposure and vulnerability (Extended Data Fig. 2). In all these paired events, areductioninimpactwasassociatedwithareductioninvulnerability, highlightingtheimportanceofvulnerability.Infiveoftheseeightcases withadecreaseinimpacttherewasalsoadecreaseinexposure,whereas P > 0.05 P ≤ 0.05 P > 0.01 P ≤ 0.01 Drought a b c d Impact Hazard Exposure Vulnerability Mgmt shortc Impact Hazard Exposure Vulnerability Mgmt shortc Hazard Exposure Vulnerability Mgmt shortc Flood Decrease in impact Increase in impact Frequency Frequency Correlation coefficient One case Ten cases Impact 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Hazard Exposure Vulnerability Mgmt shortc Impact Hazard Exposure Vulnerability Mgmt shortc Hazard Exposure Vulnerability Mgmt shortc * Frequency ** ** Indicator change Indicator change +2 +1 0 –1 –2 +2 +1 0 –1 –2 +2 +1 0 –1 –2 +2 +1 0 –1 –2 Five cases Fig.2|Correlationmatrixandhistogramsofindicatorsofchange. a, c,Correlationmatrixofindicatorsofchangeforflood(a)anddrought(c) pairedevents.ColoursofsquaresindicateSpearman’srankcorrelation coefficientsandtheirsize,theP value.b, d,Histogramsofindicatorsofchange offlood(b) anddrought(d)stratifiedbydecrease(n = 15andn = 5pairedevents forfloodanddrought,respectively)andincrease(n = 5andn = 8pairedevents, respectively)inimpact.TheasteriskdenotesthesuccessstoriesofBox 1; doubleasterisksdenotepairsforwhichthesecondeventwasmuchmore hazardousthanthefirst(thatis,'unprecedented').Mgmtshortc,management shortcomings.
- 4. Nature | Vol 608 | 4 August 2022 | 83 inonecase(floodsinJakarta,Indonesiain2002and2007(ID 18))there was a large increase in exposure. In the paired event of droughts in California,UnitedStates(1987–1992and2011–2016,ID 36)anincrease in exposure and a reduction in vulnerability increased impact, which points to the more important role of exposure in comparison with vulnerability in this drought case (Extended Data Fig. 2). Generallythechangesindriversarenotsignificantlycorrelatedwith each other, with the exception of hazard and exposure in the case of floods (r = 0.55, P ≤ 0.01) (Fig. 2a). This finding may be explained by theinfluenceofhazardonthesizeoftheinundationarea,andthuson thenumbersofpeopleandassetsaffected,whichrepresentexposure. The sensitivity analysis suggests that the correlation pattern is robust,asvisualizedbythecoloursinExtendedDataFig. 3.Thepattern of P values is also robust for flood cases, although these become less significant for drought because of the smaller sample size (Extended Data Fig. 3). Wesplitthepairedeventsintogroupsofdecreasingandincreasing impacttoevaluatetheirdriversseparately(Fig. 2b,d).Overall,thepat- tern is similar for floods and droughts. Most flood and drought pairs withdecreasingimpactshoweitheradecreaseinhazard(tenpairs,50%) ornochange(eightpairs,40%).Exceptionsaretwofloodpairsthatare success stories of decreased impact despite an increase in hazard, as detailed in Box 1. The change in exposure of the pairs with decreased impacts (Fig. 2b,d) ranges from a large decrease to a large increase, whereasvulnerabilityalwaysdecreased.Allcaseswithalargedecrease invulnerability(−2)areassociatedwithadecreaseinimpacts.Overall, the pattern suggests that a decrease in impacts is mainly caused by a combinationoflowerhazardandvulnerability,despiteanincreasein exposure in 25% of cases. Theroleofhazardandvulnerabilityinimpactreductioncanbeexem- plifiedbythepairofriverinefloodsinJakarta,Indonesia(ID 4inFig. 1). The 2007 event had a flood return period of 50 years, whereas it was 30 years for the 2013 event23 (that is, the hazard of the second event wassmaller).Vulnerabilityhadalsodecreasedasaresultofimproved preparednessresultingfromafloodriskmappinginitiativeandcapac- itybuildingprogrammesimplementedafterthefirstflood,toimprove citizens'emergencyresponse,aswellasbyanimprovementinofficial emergency management by establishment of the National Disaster ManagementAgencyin2008.Additionally,exposurewassubstantially reduced. Whilst the first flood caused 79 fatalities and direct damage of €1.3 billion, the second event caused 38 fatalities and €0.76 billion of direct damage. AnotherexampleisapairofCentralEuropeandroughts(ID 9).Dur- ingthe2003event,theminimum3-monthStandardizedPrecipitation EvapotranspirationIndexwas−1.62whereasin2015itwas−1.18—that is,thehazardofthesecondeventwassmaller24 .Thevulnerabilitywas alsolowerinthesecondevent,becausethefirsteventhadraisedpublic awarenessandtriggeredanimprovementininstitutionalplanning.For instance, the European Commission technical guidance on drought managementplans25 wasimplemented.Manyreservoirswerekeptfilled untilthebeginningofsummer2015,whichalleviatedwatershortages for various sectors and, in some cities (for example, Bratislava and Bucharest), water was supplied from tanks26 . Additionally, water use and abstraction restrictions were implemented for non-priority uses includingirrigation26 .Theimpactwasreducedfrom€17.1billionto€2.2 billion,despiteanincreaseinexposurebecauseofthelargerdrought extent affecting almost all of Europe in 2013. Most flood and drought pairs with an increase in impact also show alargerhazard(11cases,85%;Fig. 2b,d).Forsixofthesepairedevents (46%),thesecondeventwasmuchmorehazardousthanthefirst(haz- ard indicator-of-change +2), whereas this was never the case for the pairswithdecreasingimpact.Ofthosepairswithanincreaseinimpact, 12 (92%) show an increase in exposure and nine (69%) show a small decreaseinvulnerability(vulnerabilityindicator-of-change−1).Overall, the pattern suggests that the increase in impact is mainly caused by a Box 1 Success stories of decreased impact despite increased hazard The dataset includes two cases in which a lower impact was achieved despite a larger hazard of the second event, making these interesting success stories (Fig. 3). Both cases are flood paired events, but of different types (that is, pluvial and riverine floods (Table 1)). These cases have in common that institutional changes and improved flood risk management governance were introduced and high investments in integrated management were undertaken, which led to an effective implementation of structural and non-structural measures, such as improved early warning and emergency response to complement structural measures such as levees (Table 1). Table 1 | Characteristics and commonalities in flood management of the two success stories. Pluvial floods in Barcelona, Spain (ID 12) Riverine floods in Danube catchment in Germany and Austria (ID 15) Event characteristics 1995 2018 2002 2013 Hazard (hazard indicator-of-change +1) Duration, 4 h; average event precipitation, 38 mm Duration, 21 h; average event precipitation, 45 mm 7,700 m³ s−1 peak discharge at gauge Achleiten 10,100 m³ s−1 peak discharge at gauge Achleiten Impacts (impact indicator-of-change −1) €33.6 milliona €3.5 million €4 billiona €2.32 billion Commonalitiesinmanagementchanges:potentialfactorsofsuccess Institutional changes, improved governance Reorganization of early warning and emergency response after 1995, with improved collaboration between municipality, Catalonia and State Agency of Meteorology Flood information service (HORA) for Austria went online in 2006; reorganization of flood warning and emergency response units with improved collaboration across federal states and transnationally High investments in structural and non-structural measures About €136 milliona invested in structural measures alone, following the Integrated Sewerage Plan of Barcelona Around €3.6 billiona invested in flood risk management between events on structural and non-structural measures, including new legislation and building codes in Germany and Austria Strongly improved early warning and emergency response New radar and lightning network plus operative mesoscale meteorological models in Catalonia, real-time control system based on rain gauge network and water level monitoring in Barcelona Technical improvements in weather forecasting in Germany, much higher penetration rate of flood warnings and more effective flood response actions among citizens a Calculated as costs at the time of the second event.
- 5. 84 | Nature | Vol 608 | 4 August 2022 Article combinationofhigherhazardandexposure,whichisnotcompensated by a small decrease in vulnerability. Theroleofhazardandexposureinincreasingimpactisillustratedby apairofpluvialfloodsinCorigliano-RossanoCity,Calabria,Italy(ID 40). This2015eventwasmuchmorehazardous(+2)thanthatin2000,with precipitationreturnperiodsofmorethan100and10–20 years,respec- tively27 .Also,the2000eventoccurredduringtheoff-seasonfortourism inSeptemberwhereastheexposurewasmuchlargerin2015,because theeventoccurredinAugustwhenmanytouristswerepresent.Inter- ruption of the peak holiday season caused severe indirect economic damage.Anotherexampleisapairofdroughts(ID 33)affectingNorth Carolina, United States. Between 2007 and 2009, about 65% of the state was affected by what was classified as an exceptional drought, with a composite drought indicator of the US Drought Monitor of 27 months28 , whereas between 2000 and 2003 only about 30% of the statewasaffectedbyanexceptionaldroughtof24 months28 .Thecrop lossesin2007–2009wereabout€535million,whereastheywere€497 million in 2000–2003, even though vulnerability had been reduced duetodroughtearlywarningandmanagementbytheNorthCarolina Drought Management Council, established in 2003. Effectsofchangesinmanagementondrivers The correlations shown in Fig. 2a,c also shed light on how manage- ment affects hazard, exposure and vulnerability and thus, indirectly, impact.Forfloodpairedevents,changesinmanagementshortcomings are significantly positively correlated with changes in vulnerability (r = 0.56,P ≤ 0.01),andbotharesignificantlypositivelycorrelatedwith changes in impact (Fig. 2a). For drought, however, these correlations arenotsignificant(Fig. 2c).Thus,achievingdecreasesinvulnerability, and consequently in impact, by improving risk management (that is, reducing management shortcomings) seems to be more difficult for droughts than for floods. This difficulty may be related to spillover effects—that is, drought measures designed to reduce impacts in one sector can increase impacts in another. For example, irrigation to alleviate drought in agriculture may increase drought impacts on drinking water supply and ecology29 . The paired floods in the Piura region, Peru (ID 13) illustrate how effective management can reduce vulnerability, and consequently impact. At the Piura river, maximum flows of 3,367 and 2,755 m3 s−1 were recorded during the 1998 and 2017 events, respectively (that is, hazard showed a small decrease (−1)). Around 2000, the national hydrometeorologicalservicestartedtoissuemedium-rangeweather forecasts that allowed preparations months before the 2017 event. In2011,theNationalInstituteofCivilDefenceandtheNationalCentre for the Estimation, Prevention, and Reduction of Disaster Risk were foundedwhich,togetherwithnewlyestablishedshort-rangeriverflow forecasts,allowedmoreefficientemergencymanagementofthemore recent event. Additionally, non-governmental organizations such as Practical Action had implemented disaster risk-reduction activities, includingevacuationexercisesandawarenesscampaigns30 .Allofthese improvements in management decreased vulnerability. The impact of the second event was smaller, with 366 fatalities in 1998 compared with 159 in 2017, despite an increase in exposure due to urbanization and population increase. Whenthehazardofthesecondeventwaslargerthanthatofthefirst (+1, +2), in 11 out of 18 cases (61%) the impact of the second event was also larger, irrespective of small decreases in vulnerability in eight of these cases (light blue dots/triangles in Fig. 3). There are only two pairedeventsinourdatasetforwhichadecreaseinimpactwasachieved despite the second event being more hazardous (highlighted by the green circle in Fig. 3). These cases are considered success stories and arefurtherdiscussedinBox 1.Forthetwopairedevents(ID 21and30) for which the only driver that changed was hazard (+1), the impacts did not change (0) (Extended Data Table 1). Water retention capacity of 189,881,000 m³ and good irrigation infrastructure with sprinkling machineswereapparentlysufficienttocounteracttheslightincreasein hazardforthedroughtpairedeventinPolandin2006and2015(ID 21). The improved flood alleviation scheme implemented between the pairedfloodevents(2016and2018),protectedpropertiesinBirming- ham,UnitedKingdom(ID 30).Thereare,however,sevencasesforwhich the second event was much more hazardous (+2) than the first (high- lighted by the purple ellipse in Fig. 3)—that is, events of a magnitude that locals had probably not previously experienced. We term these events, subjectively, as unprecedented; almost all had an increased impact despite improvements in management. One unprecedented pluvial flood is the 2014 event in the city of Malmö,Sweden(ID45).Thiseventwasmuchmorehazardousthanthat experienced a few years before, with precipitation return periods on averageof135and24 years,respectively,for6 hduration31 .Thelargest 6 hprecipitationmeasuredatoneofninestationsduringthe2014event corresponded to a return period of 300 years. The combined sewage system present in the more densely populated areas of the city was overwhelmed,leadingtoextensivebasementfloodingin2014(ref. 31 ). Thedirectmonetarydamagewasabout€66millionasopposedto€6 million in the first event. An unprecedented drought occurred in the CapeTownmetropolitanareaofSouthAfrica,in2015–2018(ID 44).The drought was much longer (4 years) than that experienced previously in2003–2004(2 years).AlthoughtheBergRiverDamhadbeenadded to the city’s water supply system in 2009, and local authorities had developedvariousstrategiesformanagingwaterdemands(forexam- ple, water restrictions, tariff increases, communication campaign), Large decrease Small decrease Change in vulnerability Drought Flood No change Small increase Large increase Change in impact +2 +1 0 –1 –2 Change in hazard –2 –1 0 +1 +2 Fig.3|Relationshipbetweenchangeinhazardandchangeinimpacts. Categoriesare:lowerhazardandlowerimpact,tencases;higherhazardand higherimpact,11 cases;lowerhazardandhigherimpact,onecase;higher hazardandlowerimpact,twocases.Circlesandtrianglesindicatedroughtand floodpairedevents,respectively;theircoloursindicatechangeinvulnerability. Greencirclehighlightssuccessstories(n = 2)ofreducedimpact(−1)despitea smallincreaseinhazard(+1).Purpleellipseindicatespairedevents(n = 7) withlargeincreaseinhazard(+2)—thatis,eventsthatweresubjectively unprecedentedandprobablynotpreviouslyexperiencedbylocalresidents.
- 6. Nature | Vol 608 | 4 August 2022 | 85 the second event caused a much higher direct impact of about €180 million32 because the water reserves were reduced to virtually zero. Eventhoughitisknownthatvulnerabilityreductionplaysakeyrole inreducingrisk,ourpaired-eventcasesrevealthatwhenthehazardof thesecondeventwashigherthanthefirst,areductioninvulnerability alonewasoftennotsufficienttoreducetheimpactofthesecondevent to less than that of the first. Our analysis of drivers of impact change revealstheimportanceofreducinghazard,exposureandvulnerability to achieve an effective impact reduction (Fig. 2). Although previous studieshaveattributedahighprioritytovulnerabilityreduction17,21 ,the importanceofconsideringallthreedriversidentifiedheremayreflect thesometimeslimitedefficiencyofmanagementdecisions,resultingin unintendedconsequences.Forexample,leveeconstructionaimingat reducinghazardsmayincreaseexposurethroughencouragingsettle- mentsinfloodplains33,34 .Similarly,constructionofreservoirstoabate droughts may enhance exposure through encouraging agricultural development and thus increase water demand35,36 . Eventsthataremuchmorehazardousthanprecedingevents(termed unprecedented here) seem to be difficult to manage; in almost all the cases considered they led to increased impact (Fig. 3). This finding mayberelatedtotwofactors.First,largeinfrastructuresuchaslevees andwaterreservoirsplayanimportantroleinriskmanagement.These structuresusuallyhaveanupperdesignlimituptowhichtheyareeffec- tive but, once a threshold is exceeded, they become ineffective. For example, the unprecedented pluvial flood in 2014 in Malmö, Sweden (ID 45) exceeded the capacity of the sewer system31 and the unprec- edented drought in Cape Town (ID 44) exceeded the storage water capacity37 . This means that infrastructure is effective in preventing damage during events of a previously experienced magnitude, but often fails for unprecedented events. Non-structural measures, such as risk-aware land-use planning, precautionary measures and early warning, can help mitigate the consequences of water infrastruc- ture failure in such situations21 , but a residual risk will always remain. Second,riskmanagementisusuallyimplementedafterlargefloodsand droughts,whereasproactivestrategiesarerare.Partofthereasonfor thisbehaviourisacognitivebiasassociatedwiththerarityandunique- nessofextremes,andthenatureofhumanriskperception,whichmakes peopleattachalargesubjectiveprobabilitytothoseeventstheyhave personally experienced38 . Ontheotherhand,twocasestudieswereidentifiedinwhichimpact wasreduceddespiteanincreaseinhazard(Box 1).Ananalysisofthese casestudiesidentifiesthreesuccessfactors:(1)effectivegovernance ofriskandemergencymanagement,includingtransnationalcollabo- ration such as in the Danube case; (2) high investments in structural and non-structural measures; and (3) improved early warning and real-time control systems such as in the Barcelona case. We believe thereispotentialformoreuniversalapplicationofthesesuccessfac- torstocounteractthecurrenttrendofincreasingimpactsassociated withclimatechange3 .Thesefactorsmayalsobeeffectiveintheman- agement of unprecedented events, provided they are implemented proactively. Onlinecontent Anymethods,additionalreferences,NatureResearchreportingsum- maries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author con- tributions and competing interests; and statements of data and code availabilityareavailableathttps://doi.org/10.1038/s41586-022-04917-5. 1. Jongman, B. et al. Declining vulnerability to river floods and the global benefits of adaptation. Proc. Natl Acad. Sci. USA 112, E2271–E2280 (2015). 2. Formetta, G. & Feyen, L. Empirical evidence of declining global vulnerability to climate-related hazards. Glob. Environ. Change 57, 101920 (2019). 3. IPCC. Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation (eds Field, C.B. et al.) (Cambridge Univ. Press, 2012). 4. Bouwer, L. M. Have disaster losses increased due to anthropogenic climate change? Bull. Am. Meteorol. Soc. 92, 39–46 (2011). 5. Ward, P. J. et al. Natural hazard risk assessments at the global scale. Nat. Hazards Earth Syst. Sci. 20, 1069–1096 (2020). 6. Economic Losses, Poverty & Disasters (1998–2017) (UNISDR (United Nations Office for Disaster Risk Reduction) and CRED (Centre for Research on the Epidemiology of Disasters), 2018); https://www.cred.be/unisdr-and-cred-report-economic-losses-poverty-disasters- 1998-2017 7. Dottori, F. et al. Increased human and economic losses from river flooding with anthropogenic warming. Nat. Clim. Change 8, 781–786 (2018). 8. Cammalleri, C. et al. Global Warming and Drought Impacts in the EU (Publications Office of the European Union, 2020); https://doi.org/10.2760/597045 9. Terminology on Disaster Risk Reduction (UNDRR (United Nations Office for Disaster Risk Reduction), 2017); www.undrr.org/terminology 10. Cutter, S. L., Boruff, B. J. & Shirley, W. L. Social vulnerability to environmental hazards. Soc. Sci. Q. 84, 242–261 (2003). 11. Turner, B. L. et al. A framework for vulnerability analysis in sustainability science. Proc. Natl Acad. Sci. USA 100, 8074–8079 (2003). 12. Eakin, H. & Luers, A. L. Assessing the vulnerability of social-environmental systems. Annu. Rev. Environ. Resour. 31, 365–394 (2006). 13. Eriksen, S. et al. Adaptation interventions and their effect on vulnerability in developing countries: help, hindrance or irrelevance? World Dev. Rev. 141, 105383 (2020). 14. Kreibich, H. et al. Costing natural hazards. Nat. Clim. Change 4, 303–306 (2014). 15. De Ruiter, M. C. et al. Why we can no longer ignore consecutive disasters. Earths Future 8, e2019EF001425 (2020). 16. Di Baldassarre, G. A. et al. Perspectives on socio-hydrology: capturing feedbacks between physical and social processes. Water Resour. Res. 51, 4770–4781 (2015). 17. Mechler, R. & Bouwer, L. M. Understanding trends and projections of disaster losses and climate change: is vulnerability the missing link? Clim. Change 133, 23–35 (2015). 18. Ward, P. J. et al. The need to integrate flood and drought disaster risk reduction strategies. Water Secur. 11, 100070 (2020). 19. Raikes, J. et al. Pre-disaster planning and preparedness for floods and droughts: a systematic review. Int. J. Disaster Risk Reduct. 38, 101207 (2019). 20. Johnson, K. A. et al. A benefit–cost analysis of floodplain land acquisition for US flood damage reduction. Nat. Sustain. 3, 56–62 (2019). 21. Kreibich, H. et al. Adaptation to flood risk: results of international paired flood event studies. Earths Future 5, 953–965 (2017). 22. Birkland, T. A. Focusing events, mobilization, and agenda setting. J. Public Policy 18, 53–74 (1998). 23. Budiyono, Y. et al. River flood risk in Jakarta under scenarios of future change. Nat. Hazards Earth Syst. Sci. 16, 757–774 (2016). 24. Ionita, M. et al. The European 2015 drought from a climatological perspective. Hydrol. Earth Syst. Sci. 21, 1397–1419 (2017). 25. Drought Management Plan Report (European Commission, 2007); https://ec.europa.eu/ environment/water/quantity/pdf/dmp_report.pdf 26. Van Lanen, H. A. J. et al. Hydrology needed to manage droughts: the 2015 European case. Hydrol. Process. 30, 3097–3104 (2016). 27. Petrucci,O.et al.Civilprotectionanddamaginghydrogeologicalevents:comparativeanalysis of the 2000 and 2015 events in Calabria (southern Italy). Adv. Geosci. 44, 101–113 (2017). 28. NDMC (National Drought Mitigation Center) U.S. Drought Monitor https://droughtmonitor. unl.edu (2020). 29. Garrick, D. E. et al. Managing the cascading risks of droughts: institutional adaptation in transboundary river basins. Earths Future 6, 809–827 (2018). 30. French, A. & Mechler, R. Managing El Niño Risks Under Uncertainty in Peru (International Institute for Applied Systems Analysis, 2017); http://pure.iiasa.ac.at/id/eprint/14849/1/ French_Mechler_2017_El%20Ni%C3%B1o_Risk_Peru_Report.pdf 31. Sörensen, J. & Mobini, S. Pluvial, urban flood mechanisms and characteristics— assessment based on insurance claims. J. Hydrol. 555, 51–67 (2017). 32. Muller, M. Cape Town’s drought: don’t blame climate change. Nature 559, 174–176 (2018). 33. White, G. F. Human Adjustment to Floods (Univ. of Chicago Press, 1945). 34. Wenger, C. Better use and management of levees: reducing flood risk in a changing climate. Environ. Rev. 23, 240–255 (2015). 35. Kallis, G. Coevolution in water resource development: the vicious cycle of water supply and demand in Athens, Greece. Ecol. Econ. 69, 796–809 (2010). 36. Di Baldassarre, G. et al. Water shortages worsened by reservoir effects. Nat. Sustain. 1, 617–622 (2018). 37. Savelli, E. et al. Don’t blame the rain: social power and the 2015–2017 drought in Cape Town. J. Hydrol. 594, 125953 (2021). 38. Merz, B. et al. Charting unknown waters—on the role of surprise in flood risk assessment and management. Water Resour.Res. 51, 6399–6416 (2015). Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. © The Author(s) 2022
- 7. 86 | Nature | Vol 608 | 4 August 2022 Article 1 GFZ German Research Centre for Geosciences, Section Hydrology, Potsdam, Germany. 2 Institute for Environmental Studies, Vrije Universiteit Amsterdam, Amsterdam, the Netherlands. 3 Leichtweiss Institute for Hydraulic Engineering and Water Resources, Division of Hydrology and River basin management, Technische Universität Braunschweig, Braunschweig, Germany. 4 Department of Civil and Environmental Engineering, University of Houston, Houston, TX, USA. 5 Lomonosov Moscow State University, Moscow, Russia. 6 University of California, Irvine, CA, USA. 7 Department of Civil Engineering, Istanbul Technical University, Istanbul, Turkey. 8 Center for Climate and Resilience Research, Santiago, Chile. 9 Department of Civil Engineering, Universidad de La Frontera, Temuco, Chile. 10 Operations Department, Barcelona Cicle de l’Aigua S.A, Barcelona, Spain. 11 Global Institute for Water Security, University of Saskatchewan, Saskatoon, Saskatchewan, Canada. 12 LEGOS, Université de Toulouse, CNES, CNRS, IRD, UPS, Toulouse, France. 13 Department of Groundwater Management, Deltares, Delft, the Netherlands. 14 School of Geography, Earth and Environmental Sciences, University of Birmingham, Birmingham, UK. 15 Agency for the Assessment and Application of Technology, Jakarta, Indonesia. 16 Department of Civil and Environmental Engineering, Imperial College London, London, UK. 17 Department of Civil Engineering, Beykent University, Istanbul, Turkey. 18 Graduate School, Istanbul Technical University, Istanbul, Turkey. 19 Faculty of Environment and Natural Resources, University of Freiburg, Freiburg, Germany. 20 Geographical Sciences, University of Bristol, Bristol, UK. 21 Cabot Institute, University of Bristol, Bristol, UK. 22 Department of Agriculture, Hellenic Mediterranean University, Iraklio, Greece. 23 Université de Lorraine, LOTERR, Metz, France. 24 CNR-IRPI, Research Institute for Geo-Hydrological Protection, Cosenza, Italy. 25 INRAE, Bordeaux Sciences Agro, UMR ISPA, Villenave dʼOrnon, France. 26 University of Saskatchewan, Centre for Hydrology, Canmore, Alberta, Canada. 27 Environmental Policy and Planning Group, Department of Urban Studies and Planning, Massachusetts Institute of Technology, Cambridge, MA, USA. 28 Department of Economics, Ca’ Foscari University of Venice, Venice, Italy. 29 Lab of Geophysical-Remote Sensing & Archaeo-environment, Institute for Mediterranean Studies, Foundation for Research and Technology Hellas, Rethymno, Greece. 30 Pontificia Bolivariana University, Faculty of Civil Engineering, Bucaramanga, Colombia. 31 California State University, Long Beach, CA, USA. 32 Alfred Wegener Institute Helmholtz Center for Polar and Marine Research, Palaeoclimate Dynamics Group, Bremerhaven, Germany. 33 Emil Racovita Institute of Speleology, Romanian Academy, Cluj-Napoca, Romania. 34 Forest Biometrics Laboratory, Faculty of Forestry, Ștefan cel Mare University, Suceava, Romania. 35 Water Problem Institute Russian Academy of Science, Moscow, Russia. 36 Faculty of Environment, University of Science, Ho Chi Minh City, Vietnam. 37 Environment Agency, Bristol, UK. 38 School of Chemical and Environmental Engineering, Technical University of Crete, Chania, Greece. 39 Servicio Nacional de Meteorología e Hidrología del Perú, Lima, Peru. 40 Department of Applied Physics, University of Barcelona, Barcelona, Spain. 41 Water Research Institute, University of Barcelona, Barcelona, Spain. 42 British Geological Survey, Wallingford, UK. 43 Centre of Natural Hazards and Disaster Science, Uppsala, Sweden. 44 Department of Earth Sciences, Uppsala University, Uppsala, Sweden. 45 Civil and Environmental Engineering, The Pennsylvania State University, State College, PA, USA. 46 Escola de Engenharia de Sao Carlos, University of São Paulo, São Paulo, Brasil. 47 Department of Water Resources & Delta Management, Deltares, Delft, the Netherlands. 48 Trelleborg municipality, Trelleborg, Sweden. 49 Department of Water Resources Engineering, Lund University, Lund, Sweden. 50 University of Potsdam, Institute of Environmental Science and Geography, Potsdam, Germany. 51 University of Science and Technology of Hanoi, Vietnam Academy of Science and Technology, Hanoi, Vietnam. 52 Institute for Environment and Resources, Vietnam National University Ho Chi Minh City, Ho Chi Minh City, Vietnam. 53 Institute for Circular Economy Development, Vietnam National University Ho Chi Minh City, Ho Chi Minh City, Vietnam. 54 Observatori de l’Ebre, Ramon Llull University – CSIC, Roquetes, Spain. 55 School of Environment and Sustainability, University of Saskatchewan, Saskatoon, Saskatchewan, Canada. 56 Department of Civil, Geological and Environmental Engineering, University of Saskatchewan, Saskatoon, Saskatchewan, Canada. 57 Dipartimento di Ingegneria Civile, Edile e Ambientale, Sapienza Università di Roma, Rome, Italy. 58 University of Applied Sciences, Magdeburg, Germany. 59 Center for Environmental and Geographic Information Services, Dhaka, Bangladesh. 60 Earth and Environmental Systems Institute, The Pennsylvania State University, State College, PA, USA. 61 Institute of Meteorology and Water Management National Research Institute, Warsaw, Poland. 62 Key Laboratory of Water Cycle and Related Land Surface Processes, Institute of Geographical Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing, China. 63 Department of Hydraulic Engineering, Tsinghua University, Beijing, China. 64 Royal Botanical Gardens Kew, London, UK. 65 KWR Water Research Institute, Nieuwegein, the Netherlands. 66 Department of Physical Geography, Utrecht University, Utrecht, the Netherlands. 67 Civil Engineering, University of Bristol, Bristol, UK. 68 School of Natural Resources, University of Nebraska-Lincoln, Lincoln, NE, USA. 69 School of Geography and Ocean Science, Nanjing University, Nanjing, China. 70 Institute of Hydraulic Engineering and Water Resources Management, Technische Universität Wien, Vienna, Austria. 71 Department of Integrated Water Systems and Governance, IHE Delft, Delft, the Netherlands. ✉e-mail: Heidi.Kreibich@ gfz-potsdam.de
- 8. Methods The concept of paired events aims at comparing two events of the same hazard type that occurred in the same area21 to learn from the differencesandsimilarities.Thisconceptisanalogoustopairedcatch- ment studies, which compare two neighbouring catchments with different vegetation in terms of their water yield39 . Our study follows the theoretical risk framework that considers impact as a result of threeriskcomponentsordrivers3 :hazard,exposureandvulnerability (ExtendedDataFig. 4).Hazardreflectstheintensityofanevent,such as a flooded area or drought deficit—for example, measured by the standardized precipitation index. Exposure reflects the number of peopleandassetsintheareaaffectedbytheevent.Consequently,the change in exposure between events is influenced by changes in the populationdensityandtheassetsintheaffectedarea(socioeconomic developments), as well as by changes in the size of the affected area (changeofhazard).Vulnerabilityisacomplexconcept,withanexten- sive literature from different disciplines on how to define, measure and quantify it13,40–42 . For instance, Weichselgartner43 lists more than 20 definitions of vulnerability, and frameworks differ quite substan- tially—for example, in terms of integration of exposure into vulner- ability11 orseparatingthem3 .Reviewsandattemptstoconvergeonthe variousvulnerabilityconceptsstressthatvulnerabilityisdynamicand that assessments should be conducted for defined human–environ- ment systems at particular places12,44,45 . Every vulnerability analysis requires an approach adapted to its specific objectives and scales46 . The paired event approach allows detailed context and place-based vulnerability assessments that are presented in the paired event reports, as well as comparisons across paired events based on the indicators-of-change. The selection of sub-indicators for the char- acterization of vulnerability is undertaken with a particular focus on temporal changes at the same place. All three drivers—hazard, exposure and vulnerability—can be reduced by risk-management measures.Hazardcanbereducedbystructuralmeasuressuchaslevees orreservoirs19 ,exposurebyrisk-awareregionalplanning20 andvulner- ability by non-structural measures, such as early warning21 . Our comparative analysis is based on a novel dataset of 45 paired events from around the world, of which 26 event pairs are floods and 19 are droughts. The events occurred between 1947 and 2019, and the average period between the two events of a pair is 16 years. The number of paired events is sufficiently large to cover a broad range of hydroclimatic and socioeconomic settings around the world and allows differentiated, context-specific assessments on the basis of detailed in situ observations. Flood events include riverine, pluvial, groundwater and coastal floods47–50 . Drought events include mete- orological,soilmoistureandhydrological(streamflow,groundwater) droughts51 .Therationaleforanalysingfloodsanddroughtstogether isbasedontheirpositionatthetwoextremesofthesamehydrological cycle, the similarity of some management strategies (for example, warningsystems,waterreservoirinfrastructure),potentialtrade-offs in the operation of the same infrastructure52 and more general inter- actions between these two risks (for example, water supply to illegal settlements that may spur development and therefore flood risk). There may therefore be value in management communities learning from each other18 . The dataset comprises: (1) detailed review-style reports about the events and key processes between the events, such as changes in risk management (open access data; Data Availability statement); (2) a key data table that contains the data (qualitative and quantita- tive) characterizing the indicators for the paired events, extracted from individual reports (open access data); and (3) an overview table providing indicators-of-change between the first and second events (Supplementary Table 3). To minimize the elements of subjectivity and uncertainty in the analysis, we (1) used indicators-of-change as opposedtoindicatorsofabsolutevalues,(2)calculatedindicatorsfrom a set of sub-indicators (Supplementary Table 1) and (3) implemented aqualityassuranceprotocol.Commonly,morethanonevariablewas assessed per sub-indicator (for example, flood discharges at more than one stream gauge, or extreme rainfall at several meteorological stations). A combination or selection of the variables was used based onhydrologicalreasoningonthemostrelevantpieceofinformation. Special attention was paid to this step during the quality assurance process,drawingonthein-depthexpertiseoneventsofoneormoreof ourco-authors.Theassignmentofvaluesfortheindicators-of-change, includingqualityassurance,wasinspiredbytheDelphiMethod53 thatis builtonstructureddiscussionandconsensusbuildingamongexperts. Theprocesswasdrivenbyacoregroup(H.K.,A.F.V.L.,K.Schröter,P.J.W. andG.D.B.)andwasundertakeninthefollowingsteps:(1)onthebasis of the detailed report, a core group member suggested values for all indicators-of-changeforapairedevent;(2)asecondmemberofthecore group reviewed these suggestions; in case of doubt, both core group membersrecheckedthepairedeventreportandprovidedajointsug- gestion; (3) all suggestions for the indicators-of-change for all paired eventswerediscussedinthecoregrouptoimproveconsistencyacross pairedevents;(4)thesuggestedvaluesoftheindicators-of-changewere reviewedbytheauthorsofthepaired-eventreport;andfinally(5),the complete table of indicators-of-change (Supplementary Table 3) was reviewedbyallauthorstoensureconsistencybetweenpairedevents. Compound events were given special consideration, and the best possible attempt was made to isolate the direct effects of floods and droughts from those of concurrent phenomena on hazard, exposure and impact, based on expert knowledge of the events of one or more oftheco-authors.Forinstance,inthecourseofthisiterativeprocessit becameclearthatfatalitiesduringdroughteventswerenotcausedby alackofwater,butbytheconcurrentheatwave.Itwasthusdecidedto omitthesub-indicator‘fatalities’indroughtimpactcharacterization. The potential biases introduced by compound events were further reducedbytheuseoftherelativeindicators-of-changebetweensimilar event types with similar importance of concurrent phenomena. The indicator-of-change of impact is composed of the following sub-indicators:numberoffatalities(forfloodsonly),directeconomic impact,indirectimpactandintangibleimpact(SupplementaryTable 1). Floodhazardiscomposedofthesub-indicatorsprecipitation/weather severity, severity of flood, antecedent conditions (for pluvial and riv- erinefloodsonly),aswellasthefollowingforcoastalfloodsonly:tidal levelandstormsurge.Droughthazardiscomposedofthedurationand severity of drought. Exposure is composed of the two sub-indicators people/area/assets exposed and exposure hotspots. Vulnerability is composed of the four sub-indicators lack of awareness and precau- tion,lackofpreparedness,imperfectofficialemergency/crisismanage- ment and imperfect coping capacity. Indicators-of-change, including sub-indicators,weredesignedsuchthatconsistentlypositivecorrela- tions with impact changes are expected (Supplementary Table 1). For instance,adecreasein'lackofawareness'leadstoadecreaseinvulner- abilityandisthusexpectedtobepositivelycorrelatedwithadecrease inimpacts.Managementshortcomingsarecharacterizedbyproblems withwatermanagementinfrastructureandnon-structuralriskmanage- ment shortcomings, which means that non-structural measures were notoptimallyimplemented.Thesesub-indicatorswereaggregatedinto indicators-of-change for impact, hazard, exposure, vulnerability and managementshortcomings,toenableaconsistentcomparisonbetween floodanddroughtpairedevents.Thissetofindicatorsisintendedtobe ascomplementaryaspossible,butoverlapsarehardtoavoidbecause of interactions between physical and socioeconomic processes that controlfloodanddroughtrisk.Althoughthemanagementshortcom- ing indicator is primarily related to the planned functioning of risk management measures, and hazard, exposure and vulnerability pri- marilyreflecttheconcreteeffectsofmeasuresduringspecificevents, thereissomeoverlapbetweenthemanagementshortcomingindicator and all three drivers. Supplementary Table 1 provides definitions and
- 9. Article examplesofdescriptionormeasurementofsub-indicatorsforfloodand droughtpairedevents. Thechangesareindicatedby−2/2forlargedecreaseorincrease,−1/1 forsmalldecreaseorincreaseand0fornochange.Inthecaseofquantita- tivecomparisons(forexample,precipitationintensitiesandmonetary damage), a change of less than around 50% is usually treated as a small changeandaboveapproximately50%asalargechange,butalwayscon- sideringthespecificmeasureandpairedevents.SupplementaryTable 2 providesrepresentativeexamplesfromfloodanddroughtpairedevents showinghowdifferencesinquantitativevariablesandqualitativeinfor- mationbetweenthetwoeventsofapaircorrespondtothevaluesofthe sub-indicators,rangingfromlargedecrease(−2)tolargeincrease(+2). Weassumethataneventisunprecedentedinasubjectiveway—thatis,it hasprobablynotbeenexperiencedbefore—ifthesecondeventofapair ismuchmorehazardousthanthefirst(hazardindicator-of-change+2). Spearman’srankcorrelationcoefficientsarecalculatedforimpact, drivers and management shortcomings, separated for flood and drought paired events. Despite the measures taken to minimize the subjectivityanduncertaintyofindicatorassignment,therewillalways be an element of subjectivity. To address this, we carried out a Monte Carlo analysis (1,000 iterations) to test the sensitivity of the results whenrandomlyselecting80%offloodanddroughtpairedevents.For eachsubsamplecorrelation,coefficientsandP valueswerecalculated to obtain a total of 1,000 correlation and 1,000 P value matrices. The 25th and 75th quantiles of the correlation coefficients and P values were calculated separately (Extended Data Fig. 3). Dataavailability Thedatasetcontainingtheindividualpairedeventreports,thekeydata tableandSupplementaryTables 1–3areopenlyavailableviaGFZData Services(https://doi.org/10.5880/GFZ.4.4.2022.002). Sourcedataare provided with this paper. 39. Brown, A. E. et al. A review of paired catchment studies for determining changes in water yield resulting from alterations in vegetation. J. Hydrol. 310, 28–61 (2005). 40. Cutter, S. L. & Finch, C. Temporal and spatial changes in social vulnerability to natural hazards. Proc. Natl Acad. Sci. USA 105, 2301–2306 (2008). 41. Hinkel, J. ‘‘Indicators of vulnerability and adaptive capacity’’: towards a clarification of the science–policy interface. Glob. Environ. Change 21, 198–208 (2011). 42. Tate, E. Social vulnerability indices: a comparative assessment using uncertainty and sensitivity analysis. Nat. Hazards 63, 325–347 (2012). 43. Weichselgartner, J. Disaster mitigation: the concept of vulnerability revisited. Disaster Prev. Manage. 10, 85–94 (2001). 44. Adger, W. N. Vulnerability. Glob. Environ. Change 16, 268–281 (2006). 45. Birkmann, J. Framing vulnerability, risk and societal responses: the MOVE framework. Nat. Hazards 67, 193–211 (2013). 46. Thywissen, K. Components of Risk—a Comparative Glossary (UNU-EHS, 2006); http://collections.unu.edu/view/UNU:1869 47. Tarasova, L. et al. Causative classification of river flood events. Wiley Interdiscip. Rev. Water 6, e1353 (2019). 48. Rosenzweig, B. R. et al. Pluvial flood risk and opportunities for resilience. Wiley Interdiscip. Rev. Water 5, e1302 (2018). 49. Ascott, M. J. et al. Improved understanding of spatio‐temporal controls on regional scale groundwater flooding using hydrograph analysis and impulse response functions. Hydrol. Proc. 31, 4586–4599 (2017). 50. Danard, M., Munro, A. & Murty, T. Storm surge hazard in Canada. Nat. Hazards 28, 407–431 (2003). 51. Tallaksen, L. & Lanen, H. A. J. V. Hydrological Drought. Processes and Estimation Methods for Streamflow and Groundwater (Elsevier, 2004). 52. Van den Honert, R. C. & McAneney, J. The 2011 Brisbane floods: causes, impacts and implications. Water 3, 1149–1173 (2011). 53. Okoli, C. & Pawlowski, S. D. The Delphi method as a research tool: an example, design considerations and applications. Inform. Manage. 42, 15–29 (2004). Acknowledgements The presented work was developed by the Panta Rhei Working Groups 'Changes in flood risk' and 'Drought in the Anthropocene' within the framework of the Panta Rhei Research Initiative of the International Association of Hydrological Sciences. We thank the Barcelona Cicle de l’Aigua S.A., Barcelona City Council, Environment Agency (United Kingdom), Länsförsäkringar Skåne, Steering Centre for Urban Flood Control Programme in HCMC (Vietnam), VA SYD and the West Berkshire Council (United Kingdom) for data. The work was partly undertaken under the framework of the following projects: Alexander v on Humboldt Foundation Professorship endowed by the German Federal Ministry of Education and Research (BMBF); British Geological Survey’s Groundwater Resources Topic (core science funding); C3-RiskMed (no. PID2020-113638RB-C22), financed by the Ministry of Science and Innovation of Spain; Centre for Climate and Resilience Research (no. ANID/ FONDAP/15110009); CNES, through the TOSCA GRANT SWHYM; DECIDER (BMBF, no. 01LZ1703G); Deltares research programme on water resources; Dutch Research Council VIDI grant (no. 016.161.324); FLOOD (no. BMBF 01LP1903E), as part of the ClimXtreme Research Network. Funding was provided by the Dutch Ministry of Economic Affairs and Climate; Global Water Futures programme of University of Saskatchewan; GlobalHydroPressure (Water JPI); HUMID project (no. CGL2017-85687-R, AEI/FEDER, UE); HydroSocialExtremes (ERC Consolidator Grant no. 771678); MYRIAD-EU (European Union’s Horizon 2020 research and innovation programme under grant agreement no. 101003276); PerfectSTORM (no. ERC- 2020-StG 948601); Project EFA210/16 PIRAGUA, co-founded by ERDF through the POCTEFA 2014–2020 programme of the European Union; Research project nos. ANID/FSEQ210001 and ANID/NSFC190018, funded by the National Research and Development Agency of Chile; SECurITY (Marie Skłodowska-Curie grant agreement no. 787419); SPATE (FWF project I 4776-N, DFG research group FOR 2416); the UK Natural Environment Research Council-funded project Land Management in Lowland Catchments for Integrated Flood Risk Reduction (LANDWISE, grant no. NE/R004668/1); UK NERC grant no. NE/S013210/1 (RAHU) (W.B.); Vietnam National Foundation for Science and Technology Development under grant no. 105.06-2019.20.; and Vietnam National University–HCMC under grant no. C2018-48-01. D.M. and A. McKenzie publish with the permission of the Director, British Geological Survey. The views expressed in this paper are those of the authors and not the organizations for which they work. Author contributions H.K. initiated the study and led the work. H.K., A.F.V.L., K. Schröter, P.J.W. and G.D.B. coordinated data collection, designed the study and undertook analyses. All co-authors contributed data and provided conclusions and a synthesis of their case study (the authors of each paired event report were responsible for their case study). M. Mazzoleni additionally designed the figures, and he and N.S. contributed to the analyses. H.K., G.D.B., P.J.W., A.F.V.L., K. Schröter and G.D.B. wrote the manuscript with valuable contributions from all co-authors. Funding Open access funding provided by Helmholtz-Zentrum Potsdam Deutsches GeoForschungsZentrum - GFZ. Competing interests The authors declare no competing interests. Additional information Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41586-022-04917-5. Correspondence and requests for materials should be addressed to Heidi Kreibich. Peer review information Nature thanks Elizabeth Tellman, Oliver Wing and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Reprints and permissions information is available at http://www.nature.com/reprints.
- 12. ExtendedDataFig.3|Resultsofthesensitivityanalyses.a–dCorrelation matrixofindicators-of-changefor25thand75thquantilesofcorrelation coefficientsandp-values,respectively(a, c)and75thand25thquantilesof correlationcoefficientsandp-values,respectively(b, d)separateforfloodand droughtpairedevents.Quantilesofcorrelationcoefficientsand p-valueswere calculatedseparately;coloursofsquaresindicateSpearman’srankcorrelation coefficients;sizesofsquaresindicatesp-values.Fig. 2a,cisaddedtotheright toeasecomparison.
- 14. Extended Data Table 1 | Overview of the indicators-of-change of paired events where only one of the three drivers has changed