global environmental changeluedeke/allerlei/bukarest2018/flood_risk.pdf · 7 april 2015 received in...

Global Environmental Change 35 (2015) 199–212

Ensemble flood risk assessment in Europe under high end climatescenarios

Lorenzo Alfieri*, Luc Feyen, Francesco Dottori, Alessandra BianchiEuropean Commission—Joint Research Centre, Ispra, Italy

A R T I C L E I N F O

Article history:Received 7 April 2015Received in revised form 19 August 2015Accepted 6 September 2015Available online 24 September 2015

Keywords:EURO-CORDEX climate projectionsRCP 8.5Shared socioeconomic pathways (SSP)Flood riskPotential flood damage

A B S T R A C T

At the current rate of global warming, the target of limiting it within 2 degrees by the end of the centuryseems more and more unrealistic. Policymakers, businesses and organizations leading internationalnegotiations urge the scientific community to provide realistic and accurate assessments of the possibleconsequences of so called “high end” climate scenarios.This study illustrates a novel procedure to assess the future flood risk in Europe under high levels of

warming. It combines ensemble projections of extreme streamflow for the current century based onEURO-CORDEX RCP 8.5 climate scenarios with recent advances in European flood hazard mapping.Further novelties include a threshold-based evaluation of extreme event magnitude and frequency, analternative method to removing bias in climate projections, the latest pan-European exposure maps, andan improved flood vulnerability estimation.Estimates of population affected and direct flood damages indicate that by the end of the century the

socio-economic impact of river floods in Europe is projected to increase by an average 220% due toclimate change only. When coherent socio-economic development pathways are included in theassessment, central estimates of population annually affected by floods range between 500,000 and640,000 in 2050, and between 540,000 and 950,000 in 2080, as compared to 216,000 in the currentclimate. A larger range is foreseen in the annual flood damage, currently of 5.3 Bs, which is projected torise at 20–40 Bs in 2050 and 30–100 Bs in 2080, depending on the future economic growth.ã 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND

license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Contents lists available at ScienceDirect

Global Environmental Change

journa l home page : www.e l sev ier .com/ loca te /g loenv cha

1. Introduction

Flood risk is the combination of the probability of a flood eventand of the potential adverse consequences for human health, theenvironment, cultural heritage and economic activity associatedwith a flood event (EU Floods Directive, European Commission,2007). Key component of flood risk assessments is the accurateestimation of the flood hazard (i.e. magnitude and frequency offloods) and of the potential impact on human activities. The latteris usually identified as the product of exposure, that is, “people,property, systems, or other elements present in hazard zones thatare thereby subject to potential losses”, and of vulnerability, that is,“the characteristics and circumstances of a community, system orasset that make it susceptible to the damaging effects of a hazard”(UNISDR, 2009). All three components of flood risk, namely hazard,exposure and vulnerability, are subject to changes in time due to

* Corresponding author at: European Commission—Joint Research Centre, TP 122,Via E. Fermi 2749, 21027 Ispra, VA, Italy.

E-mail address: [email protected] (L. Alfieri).

http://dx.doi.org/10.1016/j.gloenvcha.2015.09.0040959-3780/ã 2015 The Authors. Published by Elsevier Ltd. This is an open access article un

socio-economic development and the possible influence of achanging climate. This makes the assessment of present and futureflood risk a particularly challenging task.

Literature works on future flood risk assessment commonlyestimate the flood hazard component from climatic projections ofatmospheric variables, which are then used to estimate the futurestreamflow extremes through suitable hydrological models andextreme value statistical analysis (e.g. Dankers and Feyen, 2009;Rojas et al., 2012; Tramblay et al., 2014; Ward et al., 2014).Appraising the impact of floods on population and assets requiresrelatively detailed information on topography, asset distribution,population density and potential damage functions. As a conse-quence, most flood risk assessment studies are performed at thescale of regions, countries or river basins, where local informationis more easily accessible (e.g. Bubeck et al., 2011; Falter et al., 2015;Foudi et al., 2015; Hall et al., 2003; Kok and Grossmann, 2009;Linde et al., 2011). Global and continental scale applications areless numerous in the literature (Arnell and Gosling, 2014; Dankerset al., 2014; Feyen et al., 2012; Hirabayashi et al., 2013; Jongmanet al., 2014; Ward et al., 2013; Winsemius et al., 2012). The limitedavailability of consistent large scale datasets and the increased

der the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

http://crossmark.crossref.org/dialog/?doi=10.1016/j.gloenvcha.2015.09.004&domain=pdfmailto:[email protected]://dx.doi.org/10.1016/j.gloenvcha.2015.09.004http://dx.doi.org/10.1016/j.gloenvcha.2015.09.004http://www.sciencedirect.com/science/journal/09593780www.elsevier.com/locate/gloenvcha

Table 1EURO-CORDEX climate scenarios used in this study (adapted from Alfieri et al.(2015)).

Institute GCM RCM Driving ens member

1 KNMI EC-EARTH RACMO22E r1i1p12 SMHI HadGEM2-ES RCA4 r1i1p13 SMHI EC-EARTH RCA4 r12i1p14 MPI-CSC MPI-ESM-LR REMO2009 r1i1p15 CLMcom MPI-ESM-LR CCLM4-8-17 r1i1p16 SMHI MPI-ESM-LR RCA4 r1i1p17 CLMcom EC-EARTH CCLM4-8-17 r12i1p1

200 L. Alfieri et al. / Global Environmental Change 35 (2015) 199–212

computing cost, particularly in ensemble simulation approaches,play in favour of coarser resolution modelling, though this impliesinevitable simplifications in methods and results accuracy.

Lugeri et al. (2010) performed a pan-European flood riskassessment on limited computing resources, by using hazard mapsderived from a topographic index which only depends on themorphology of the terrain and do not need any hydrological orhydraulic simulation. In a following study, Rojas et al. (2013) usedan ensemble of 12 bias corrected climate projections into adistributed hydrological model, to derive the future statisticaldistribution of discharge extremes. These were translated intoflooded area at 100 m grid resolution using a planar approximationof water levels and then used for flood risk assessment. Despite thestep forward brought by the hydrological modelling, this approachdoes not constrain flood volumes and can lead to largeoverestimation of the flooded area in low-lying regions. Advancesin mapping the flood hazard can be achieved with inundationmodels coupled with suitable high resolution topography (Schu-mann et al., 2014; Yan et al., 2015). Recent works have shown thathigh resolution flood hazard mapping is already feasible atcontinental (Alfieri et al., 2014) and global scale (Fluet-Chouinardet al., 2015; Sampson et al., 2015), paving the way to anunprecedented level of detail in large scale flood impact assess-ments.

This work makes use of the pan-European flood hazardmapping procedure by Alfieri et al. (2014), which is for the firsttime fully integrated into a high resolution flood risk assessment atcontinental scale. This is combined with projections of the futureflood hazard (Alfieri et al., 2015) driven by an ensemble of the latestclimate scenarios adopted by the Intergovernmental Panel onClimate Change (IPCC). Climate projections from 1970 to 2100 arerun through a distributed hydrological model and resultingstreamflow is analysed statistically to estimate future changesin the flood hazard in Europe. We produced an updated ensembleevaluation of the future impact of floods in Europe by combiningthe occurrence and magnitude of future discharge peaks withexposure maps for the current climate and information on flooddefences. Future estimates of expected economic damage andpopulation affected are produced, considering first only climaticdrivers and then including the effect of possible socio-economicdevelopment pathways coherent with the considered climatescenarios. Three main methodological novelties are introduced, toovercome shortcomings pointed out in previous large scale floodrisk assessments:

- Flood hazard maps are derived by a 2D hydraulic model, ratherthan through simplified approaches, and then integratedconsistently in the flood risk assessment.

- The frequency of extreme peak discharges is assessed through apeak over threshold selection, which addresses the limitation ofblock-maxima analysis on annual floods in the presence ofchanges in the frequency of extreme events.

- An alternative to bias correcting climate projections is proposed,which is performed in the impact assessment and therefore doesnot modify atmospheric variables nor the energy balance.

In addition, the risk assessment hereby described is among thefirst examples of its kind making use of the following up-to-dateinput datasets:

- EURO-CORDEX regional climate scenarios over Europe based onRepresentative Concentration Pathways (RCP).

- Socio-economic growth based on the recently developed SharedSocioeconomic Pathways (SSP).

- Updated geographic information on the European land coverand population density.

- A coherent dataset of flood vulnerability information.

The following section describes the meteorological datasets andstatic maps of exposure and vulnerability to floods used in themodelling approach. Section 3 gives details on the modellingapproach, while results are shown in Section 4. Final Sections 5 and6 include discussions and conclusions of the presented work.

2. Data

2.1. Meteorological data

2.1.1. Observed dataThe EFAS Meteo dataset (Ntegeka et al., 2013) is used in this

work to calibrate the hydrological model and to derive the currentflood hazard maps in Europe. EFAS-Meteo is a pan-European5 � 5 km2 resolution gridded daily dataset of a number ofmeteorological variables, with temporal availability from 1990 on-wards. It includes precipitation, surface temperature (mean,minimum and maximum), wind speed, vapour pressure, radiationand evapotranspiration (potential evapotranspiration, bare soiland open water evapotranspiration). The dataset was created aspart of the development of the European Flood Awareness System(EFAS, Thielen et al., 2009) and is continuously updated in nearreal-time. The source data used to generate the EFAS-Meteodataset consists mostly of point observations which are collectedand stored by two databases managed by the JRC, namely the EU-FLOOD-GIS and the MARS database (Rijks et al., 1998). On average,observations from more than 6000 stations for precipitation andmore than 3000 stations for temperature are spatially interpolatedto generate daily meteorological gridded values to use as input inLisflood. Daily maps of evapotranspiration to use in Lisflood areestimated for the same period through the Penman-Monteithformula, by including an average of 2000 stations providing vapourpressure, incoming solar radiation and wind speed.

2.1.2. Climate projectionsSeven climate projections (see Table 1) with high concentration

scenario (i.e. RCP 8.5) through the XXI century were taken from theEURO-CORDEX database (Jacob et al., 2014). The climatic scenarioswere produced by downscaling three GCM with four RCM on a gridresolution of 0.11� (i.e. �12.5 km in Europe). All three driving GCMare rated in the top 25%, according to a performance evaluation ofCMIP5 models carried out by Perez et al. (2014), in their ability toreproduce spatial patterns and climate variability over the north-east Atlantic region. Indeed, this region is dominated by the NorthAtlantic Oscillation, a prominent climate fluctuation pattern of theNorthern Hemisphere (Hurrell et al., 2003). Atmospheric variablesused in this work are average, minimum and maximum surface airtemperature, total precipitation, surface air pressure, 2-metrespecific humidity, 10-metre wind speed and surface downwellingshortwave radiation. Variables are provided as daily maps for

L. Alfieri et al. / Global Environmental Change 35 (2015) 199–212 201

historical runs from 1970 to 2005 and climate scenarios from2006 to 2100.

2.2. Static datasets

Static datasets are used in this work to model flood exposureand vulnerability information.

Flood exposure information is given by the 100 m resolutionmap of European population density by Batista e Silva et al. (2013)and by the refined version of the CORINE Land Cover proposed byBatista e Silva et al. (2012), which has 100 m resolution andminimum mapping unit of 1 ha.

Vulnerability to floods is included in the form of damagefunctions and through a flood protection map. Country specificdepth-damage (DD) functions from Huizinga (2007) were used tolink flood depth with the corresponding direct economic damage.A piece-wise linear function from 0 to 6 m flood depth is definedfor each of the 45 land use classes included in the refined CORINELand Cover. As previously done by Rojas et al. (2013), national DDfunctions were rescaled at the NUTS2 level (see http://ec.europa.eu/eurostat/statistics-explained/index.php/Glossary:NUTS)according to the ratios between the NUTS2 and the country GDPper capita. Spatial information on the flood protection level inEurope is obtained from the 5 km resolution map produced byJongman et al. (2014). In their study, flood protections weremodelled for most European basins using a three-step procedure

Fig. 1. Schematic view of the flood risk mapping approach under future climate scenariointerpretation of the references to colour in this figure legend, the reader is referred to

and a validation of the estimated values was carried out for13 locations in 10 different countries. For those countries wherethe protection level was missing while the hazard and exposureinformation was available for the risk assessment (i.e. parts ofNorway, Croatia and Republic of Macedonia), a constant value of100 years return period was used as protection level as proposedby Rojas et al. (2013).

3. Methods

3.1. Numerical models

Numerical methods used in the following can be classified into(1) hydro-meteorology models, (2) statistical methods of extremevalues and of ensemble forecasting and (3) a damage model for thesocio-economic impact assessment. The three hydro-meteorologymodels are briefly described below, while we refer to the publishedliterature for methodological details and relevant past examples.

Lisvap (Burek et al., 2013b) is a pre-processor that calculatespotential evapo-transpiration from gridded meteorological data.Daily evapo-transpiration maps based on the historical and futureclimate scenarios for each ensemble’s model were estimated usingthe Penman–Monteith formulation. Output maps were then usedas input to the hydrological model Lisflood.

Lisflood (Burek et al., 2013a; van der Knijff et al., 2010) is adistributed physically based rainfall-runoff model combined with

. The dashed red box shows the novelty contribution introduced by this work. (For the web version of this article.)

http://ec.europa.eu/eurostat/statistics-explained/index.php/Glossary:NUTShttp://ec.europa.eu/eurostat/statistics-explained/index.php/Glossary:NUTS


a routing module for river channels. It is capable of reproducingmost hydrological processes taking place in the different climatesof the Earth, and it has been successfully applied in a number ofcase studies and operational systems at spatial scales ranging fromsmall flash-flood prone catchments to large river basins (Alfieriet al., 2011; Thielen et al., 2009; Thiemig et al., 2013; Wanders et al.,2014). For this work Lisflood was run on the European domain at5 km spatial resolution and daily time step. More details on themodel setup, the parameter calibration and the hydrologicalsimulations driven by the climate projections of Table 1 are givenby Alfieri et al. (2015).

2D hydraulic simulations are performed with Lisflood-FP (Batesand De Roo, 2000; Neal et al., 2011), using flood hydrographs withstatistical features derived by Lisflood hydrological simulations. Inthe configuration used for this work, Lisflood-FP was set up at100 m resolution using the SRTM Digital Elevation Model (Jarviset al., 2008) and roughness coefficients linked to the highresolution land use map by Batista e Silva et al. (2012).

3.2. Modelling approach

The workflow is shown in Fig. 1, where it is broken down intothree main components. In a first step we used the approachdescribed by Alfieri et al. (2014) with the EFAS Meteo dataset (seeSection 2.1.1) as input to produce 100 m resolution maps of floodextent and flood depth in Europe for the observed climate andreturn periods TF = {10, 20, 50, 100, 200, 500} years. Output floodhazard maps produced herein are likely to improve the previousversion by Alfieri et al. (2014) thanks to an enhanced hydrologicalmodel calibration and an improved meteorological forcing whichis currently based on a larger number of weather stations and amore consistent spatial interpolation algorithm for meteorologicaldata. Differently from previous projections of future Europeanflood risk assessment, the current flood hazard is here modelledusing observed meteorological parameters, rather than the outputof climatic scenarios (e.g. as in Feyen et al., 2012; Jongman et al.,2014; Rojas et al., 2013).

Fig. 2. Example of areas of influence at 100 m resolution (left) and corresponding river nelinked to the grid points with same colour on the right. (For interpretation of the referearticle.)

In parallel, we used flood hazard estimates under high-endclimate scenarios from the work by Alfieri et al. (2015) in the formof magnitude and recurrence of projected discharge peaks.Extreme value statistical distributions were fitted on the simulateddischarge annual maxima of the control period (1976–2005), toderive analytical relations between extreme discharges Q andprobability of occurrence (and consequently of their return periodT), in each point of the European river network at 5 km gridresolution. The 2-years discharge peak is commonly consideredrepresentative of river bank-full conditions (e.g. Carpenter et al.,1999). Hence, historical (1976–2005) and future (2006–2100)simulated peaks larger than the 2-years value were transformedinto return periods by inverting the relations between dischargeand return period. In this step, a Gumbel extreme valuedistribution was assumed for annual maximum discharges, asdescribed by Alfieri et al. (2015).

The socio-economic impact assessment (bottom part in Fig.1) isthe main focus of this article. First, flood hazard maps werecombined with depth-damage functions and with a populationdensity map as described in Section 2.2 to derive potential damage(PD) and potential population affected (PPA) by floods for thereturn periods TF, assuming no flood protection. Note that, in thistask, damage is estimated as a function of the flood depth, whilepopulation is considered affected for any positive flood depth (e.g.Ward et al., 2013). Key step to improving the computing efficiencyis the aggregation of 100 m resolution maps of PD and PPA to 5 kmresolution through the definition of Areas of Influence (AoI). This iscarried out by linking each peak discharge in the river network at5 km resolution to a flood hydrograph with constant return periodand in turn to the neighbouring area at 100 m resolution where theflood depth is highest as compared to forcing the hydraulic modelwith flood hydrographs taken from any other grid point in the 5 kmriver network. This creates a univocal link between 100 m and 5 kmresolution data, by preserving the total PD and PPA. A graphicalexample of AoI is given in Fig. 2 for the region of Paris, in France, fora 100-years flood return period. Colours in Fig. 2 show the linkbetween 100 m AoI and corresponding 5 km grid points, while 45�

twork at 5 km grid resolution (right) for TF = 100 years. Coloured areas on the left arences to colour in this figure legend, the reader is referred to the web version of this


hatching is displayed on cells which hydrograph do not contributeto the highest flood depth in any point in the 100 m map.

For each of the seven climate scenarios, the estimated returnperiod T of discharge peaks was linked to the correspondingdamage and population affected, by interpolating linearly betweenthe six maps previously calculated for TF = {10, 20, 50, 100, 200,500}. 500 years PD and PPA was assigned to simulated dischargepeaks larger than the 500 years event, to limit the increasinguncertainty range affecting analytical distributions estimated onconsiderably shorter time windows (i.e. 30 years in this case). Oneshould note that, besides the linking between event magnitudeand impact, this step can be seen as an alternative approach to biascorrecting climate projections. Recent works have pointed outsome limitations induced by bias correcting the output of GCM orRCM (Ehret et al., 2012; Huang et al., 2014; Muerth et al., 2013;Themeßl et al., 2012). These can be summarized in (1) violating theenergy balance and breaking the link between atmosphericvariables; (2) general improvement of the average conditionsbut not necessarily of the extreme values; (3) the quality of the biascorrection strongly depends on that of the underlying observationdataset, and often implies the upscaling to a coarser grid size withthe detriment of losing small scale variability in climate informa-tion that is crucial for flood hazard estimation. The alternativeapproach proposed in this work consists of: (1) using the rawoutput of climate projections; (2) estimating the probability ofoccurrence, and consequently the return period, of simulatedextreme discharge peaks of the future by relating them toanalytical extreme value distributions fitted on the currentsimulated climate (i.e. the baseline scenario of each correspondingclimate projection); (3) matching the return period of simulatedpeaks with the corresponding estimated impact (i.e. potential

Fig. 3. Potential population affected (left) and damage (right) for the 100-years return below 10,000 persons affected and 250 Ms of damage are not shown.

damage and population affected) for the same probability ofoccurrence, taken from the observed datasets through piece-wiselinear interpolation between the six return periods.

Additional vulnerability information is included in the floodrisk assessment by setting the damage and population affected tozero whenever the discharge peak has a return period smaller thanthe flood protection level for the corresponding river section. Giventhe inherent extremely low probability of occurrence of floodevents, impact estimates are aggregated in space and time toincrease their robustness, to produce country-wide and Europe-wide estimates of expected annual damage (EAD) and expectedannual population affected (EAPA) over 30-years time slicesincluding a baseline scenario (1976–2005) and three futurescenarios with RCP 8.5 (2006–2035, 2036–2065, 2066–2095). Inthe remainder, the four time slices are referred to as 1990, 2020,2050 and 2080, named after their median year.

Alternative flood risk scenarios are evaluated by including twoshared socioeconomic pathways (SSP, O’Neill et al., 2014) in themodel chain. van Vuuren and Carter (2014) suggest that the RCP8.5 are compatible with socio-economic development driven bymitigation challenges (SSP5) or both mitigation and adaptationchallenges (SSP3). Gross domestic product (GDP) and populationprojections from the Organisation for Economic Co-operation andDevelopment (OECD) for SSP5 and SSP3 were acquired in the formof 5-years multipliers and applied to the exposure layers (i.e.population density and damage functions) to include socio-economic features in the future population affected and expecteddamage estimation. As SSP are provided in the form of countryaggregated information, this step assumes that the growth in GDPand population can be considered homogeneously distributedwithin each country.

period flood (assuming failure of flood protections), aggregated for the AoI. Values


4. Results and discussion

4.1. Potential and actual flood impact

Potential flood impact maps (PPA and PD) aggregated on a 5 kmgrid through the AoI are produced for the six return periods TF. Anexample of PPA and PD for the 100-years return period is shown inFig. 3. Maps refer to population estimates of 2006 and to GDPPurchasing Power Standards of 2007. The risk assessment shown inthis work is performed in the area with light grey background inFig. 3, which includes the EU-28 countries (except Malta andCyprus), Norway and the Republic of Macedonia. Fig. 3 showshotspots of risk to damage and population in Europe, in case offailure of flood defences. AoI with less than 10,000 persons affectedand 250Ms of damage are not shown in the figure. Highestpotential impact is mainly located in densely urbanized areas alonglarge rivers and in the flood plains, such as the cities of Paris andLondon, the Netherlands, the lower Po plain and various citiesalong the Danube River. The impact information was aggregated atcountry level and shown in Table 2, both as absolute values ofpotential population affected and damage (PPA and PD) and asratios to the country population and GDP, respectively (PPAR andPDR). Similarly, the potential impact for the same return period isincluded in Table 2, which accounts for current flood protectionstandards by setting to zero the PPA and PD in all river sectionswith flood protection equal or larger than 100 years. Relativefigures for the countries in Table 2 indicate average values of 1.2%and 4.8%, respectively, for population affected and of 1.8% and 5.5%,respectively, for economic damage, with relatively large differ-ences among countries. Striking examples are those of Hungaryand the Netherlands where impact levels of a 100-years flood areparticularly low (i.e.

Fig. 4. Country aggregated expected annual population affected (ensemble mean) in the baseline scenario (a) and relative changes in time slice 2020 (b), 2050 (c), and 2080(d). Impact of climate change only.


Agency (EEA, 2010), reporting annual figures between 4.3 and 8 Bsof losses and 262,000 people affected by flood events in Europe.

Future projections of EAD and EAPA in the current century showa vast majority of positive changes and an increasing trend for mostcountries. Positive changes in 2080 are estimated in excess of+500% in Italy, Hungary, Austria and Slovakia for EAPA and inBelgium, Austria, Slovenia and Slovakia for EAD, with maximumvalues roughly at +750% for Austria. Overall mean projections ofannual population affected by floods under climate change in theconsidered domain are estimated at 493 K (+128%), 538 K (149%)and 701 K (+224%) for the time slices 2020, 2050 and 2080,respectively. Similarly, the expected annual damage is projected torise to 11 Bs (+108%), 13.1 Bs (+148%) and 17.3 Bs (+227%) throughthe three time slices. Negative changes are found overall in four

countries. In the time slice 2080 they are projected to take placeonly in Finland (�8% in EAPA and �15% in EAD) and in Lithuania(�30% in EAPA and �28% in EAD), due to a reduction in snowmelt-driven floods (see Alfieri et al., 2015).

Spatially aggregated mean values of EAD and EAPA per year areshown in Fig. 6 (dark grey and red lines), together with theensemble spread (grey and pink shades) given by the seven modelrealizations. Relative changes from the baseline average values canbe read in the y-axis on the right. For comparison, the figure showsthe effect of including socio-economic development in theanalysis, using SSP3 and SSP5 (only ensemble mean, togetherwith 10-years moving average). Projections of population affectedunder SSP3 (SSP5) lie below (above) the estimate accounting onlyfor climate change, with relatively small differences by the end of

Fig. 5. Country aggregated expected annual damage (ensemble mean) in the baseline scenario (a) and relative changes in time slice 2020 (b), 2050 (c), and 2080 (d). Impact ofclimate change only.


the century. 10-years moving averages peak around 2080, whenthe differences in population affected versus the baselinesimulation are projected at +230%, +159% and +366% for thescenarios of “No SSP”, SSP3 and SSP5, respectively. On the otherhand, changes in damage appear considerably larger than those inpopulation affected. 10-years moving averages centred in 2093 areprojected at +240%, +465% and +2430% for “No SSP”, SSP3 andSSP5 versus the baseline, thus indicating larger uncertainty levelsin the economic growth.

Future risk projections for each time slice are then broken downat country level and shown in Figs. 7 and 8 for the three socio-economic pathways. These figures highlight countries that

contributes the most to the overall change in flood risk throughthe current century, together with the associated uncertainty ofthe ensemble (coloured bar) and their mean value (vertical dash).The two figures clearly show an increasing spread of the modelensemble in time, though with some exceptions, notably in the twocentral time slices (i.e. 2020 and 2050).

When only the climate forcing is considered (i.e. “No SSP”scenario), countries with mean EAD larger than 1 Bs; by the end ofthe century are Italy (4.6 Bs), France (2.1 Bs), UK (2.0 Bs) andGermany (1.8 Bs), though Poland is projected to reach 1.2 Bs bythe time slice 2020, later decreasing at about 1 Bs by the end of thecentury. Whereas SSP are included, EAD is projected to rise further

Fig. 6. Simulated damage and population affected per year and relative changefrom the baseline scenario (Europe-wide aggregated figures). Future scenariosinclude no SSP (with ensemble spread in pink), SSP3, and SSP5, together with their10-years moving average. (For interpretation of the references to colour in thisfigure legend, the reader is referred to the web version of this article.)


in all countries, reaching an average five-fold increase for theSSP5 in the time slice 2080 as compared to the “No SSP” scenario,which becomes seven-fold for UK and France.

Similar considerations can be drawn for trends in populationaffected. The average EAPA per country by 2080, consideringclimate change only, is the highest in Germany (112 K), Italy (94 K),France (93 K), UK (71 K), Poland (58 K) and Croatia (54 K), the latterat a staggering 1.2% of its current country population. Figures areon average 30% higher in the SSP5, while the general decrease inthe overall population modelled in the SSP3 partly compensates forthe climate related increase of flood risk. As a result, in theSSP3 scenario, 14 countries out of the 28 considered show a rise ofEAPA in the time slice 2020 followed by a roughly constant ordecreasing trend in the subsequent future.

To conclude, Fig. 9 displays relative annual figures of populationaffected and damage in the considered countries, obtained bydividing numbers in Figs. 7 and 8 by the country population andGDP, respectively. One can note that the assumption of spatiallyhomogeneous change of exposure in the SSPs, as stated inSection 3.2, makes these graphs valid for all three configurationsof future socio-economic developments. Data in Fig. 9 are keyindicators of the relative impact of current and future floods on theeconomy of countries. Values of relative affected population areclearly dominated by Croatia (HR), which already high baselinemean value is projected to reach 13m in 2050, together with aworst-case scenario above 20m in the entire second half of thecentury. All other countries are foreseen to remain within the 5muntil 2100, with largest central estimates in time slice 2080 in the

Republic of Macedonia (4.8m), Slovenia (2.9m), Austria (2.7m),Romania (2.2m) and Belgium (2.1m). Similarly, projections ofrelative annual damage show the largest central estimates in2080 mostly in Balkan countries, with Croatia (5.2m), Republic ofMacedonia (4.9m), Romania (3.6m), Bulgaria (3.4m) and Hungary(3.1m) above the threshold of 3m of the respective country GDP.Fig. 9 also stresses the importance of the ensemble approach in thistype of studies, pointing out regions with large uncertainty in theclimatic projections, which can be significant even in the baselinescenario (e.g. as in Macedonia, Croatia, Latvia).

5. Discussion

The modelling framework proposed and applied for the floodrisk assessment proved to be capable of reproducing coherentvalues of population affected and economic damage for thebaseline scenario, if compared to past estimates at European (ABI,2005; EEA, 2010) and country/regional scales (e.g. see Linde et al.,2011; Lugeri et al., 2010; Penning-Rowsell, 2015), especially in lightof the considerable uncertainty affecting available estimates offlood impact (Penning-Rowsell, 2015). Such result is even morerelevant if one considers that no calibration was performed on thesocio-economic impact model nor on the climatic variables.Historical and future flood peaks coming from climate scenariosare assigned realistic impact values by matching the quantiles(hence return periods) of their statistical distribution with those ofthe observed dataset. This step relies on the assumption that thedistribution of discharge peaks in the observed dataset (1990–2013) can be considered representative of the baseline time span(1976–2005). Note that the proposed approach is focused on theflood risk due to riverine floods in river basins with upstream arealarger than 500 km2 (see Alfieri et al., 2014). Hence, the impact dueto flash floods, surface water flooding and coastal floods is notaccounted for.

Results indicates that overall relative changes in the futuresocio-economic impact in Europe follow the correspondingchanges in the flood hazard as found by Alfieri et al. (2015). Inthe latter, the authors found that changes in the frequency of floodpeaks are likely to impact the future flood hazard more than thecorresponding changes in magnitude. In particular, the frequencyof flood peaks with high return period is projected to risesignificantly in most of Europe, even in regions where the overallfrequency of severe (i.e. larger than the bankfull conditions)discharge peaks is projected to decrease. The authors showed thatthe increase in frequency is not imputable to significant changes inthe different climate forcing used as input (i.e. EURO-CORDEX, ascompared to previous assessments based on SRES scenarios), butto the different approach to estimating the frequency of occurrenceof discharge peaks and to some limitations in the block maximaapproach in estimating extreme discharge values larger than therange of data used for the distribution fitting. As a result, we arguethat flood risk assessments based on block maxima analysis tend tounderestimate the impact of future floods in presence of increasingfrequency of extreme events. This finding explains the consider-able differences between future risk projections obtained in thiswork (e.g. +224% in EAPA and +227% in EAD by 2080) and those ofthe previous assessment by Rojas et al. (2013), whose figures offuture flood risk in Europe in 2080 are projected to increase by 90%in EAPA and by 120% in EAD, as compared to the 1981–2010 baseline.

In addition, differences in the assumed flood protectionstandards were shown to change dramatically the results ofimpact assessments (Rojas et al., 2013; Ward et al., 2013). The useof a sound map of flood protection levels as that derived byJongman et al. (2014) is a key component which improves thepresented evaluation of the flood risk at national level, as

Fig. 7. Country aggregated EAPA for the four time slices (mean value and ensemble spread) for the three cases of no SSP (left), SSP3 (center), and SSP5 (right). (Forinterpretation of the references to colour in the text, the reader is referred to the web version of this article.)


compared to linking the protection standards to the country GDP(Feyen et al., 2012) or to the assumption of a constant protectionlevel for the entire simulation domain (Rojas et al., 2013). Withregard to the modelling of flood depth and extent, no directcomparison can be done with previous assessments, whichextrapolated the flood hazard maps using future climate scenariosas input and therefore cannot be benchmarked. However, theproposed approach has the strength of using only one set of floodhazard maps, which are based on the observed climate and canthus be validated against flood records and national/regionalhazard maps (Alfieri et al., 2014). It follows that any improvementin mapping the flood hazard in the current climate can beintegrated in this risk assessment approach and improve theestimates of the current and future flood risk.

6. Conclusions

This work describes a Europe-wide socio-economic flood riskassessment based on an ensemble of the latest regional climatescenarios adopted by the IPCC and suitable socio-economicpathways. Simulated flood risk scenarios for the current centuryare representative of high level greenhouse gas concentration inthe atmosphere (RCP 8.5), which corresponds to the exceedance of+4 �C warming before year 2100, as compared to pre-industrialaverages (Alfieri et al., 2015).

Compared to earlier continental flood risk assessments(Dankers and Feyen, 2009; Jongman et al., 2014; Rojas et al.,2013), this work addresses several limitations by incorporating anumber of methodologies and state-of-the-art datasets that have

Fig. 8. Country aggregated EAD for the four time slices (mean value and ensemble spread) for the three cases of no SSP (left), SSP3 (center), and SSP5 (right). (Forinterpretation of the references to colour in the text, the reader is referred to the web version of this article.)


recently become available. Noteworthy novelties are summarizedas follows:

- Flood depth and extent are estimated through a recentlydeveloped procedure to map the flood hazard at pan-Europeanscale (Alfieri et al., 2014). This is a major advance compared tothe commonly used planar approximation of water levels, as thenew approach is based on a mass-conservative 2D hydraulicmodelling and on coherent flood hydrographs derived for a set ofdifferent probability levels.

- The evaluation is based on consistent combinations of the newRCP and SSP scenarios adopted by the IPCC for the nextgeneration of climate impact assessments.

- The future flood hazard is incorporated in the flood riskassessment through the selection of simulated peaks overthreshold, rather than through analytical curves fitted on annualstreamflow maxima. This enables a more consistent evaluationof the frequency of future extreme events, rather than just theirmagnitude (see Alfieri et al., 2015).

- We propose an alternative approach to bias correct climatesimulations, which consists of a statistical matching betweenthe frequency of occurrence of future flood peaks and thecorresponding impact (i.e. damage and population affected) forthe same return period, taken from the observed climate.

- New high resolution maps of population density and land useincreased the spatial accuracy of the impact assessment.

Fig. 9. Country aggregated annual relative population affected (left) and relative damage (right) for the four time slices (mean value and ensemble spread).



- Finally, the assumption of GDP-based or constant floodprotection levels in Europe has been replaced by the morerealistic flood defence map by Jongman et al. (2014).

Ensemble mean estimates of 5.3 Bs of damage and 216,000 peo-ple affected by river floods every year in Europe are well within therange of the observed values found in the literature, pointing outthe suitability of the proposed modelling approach for futureimpact assessments. By forcing the model with high end climaticprojections, the socio-economic impact of river floods in Europe isprojected to increase by an average of 220% by the end of thecentury, due to climate change only. The coupling of climatescenarios (RCP 8.5) with coherent projections of socio-economicgrowth (SSP3 and SSP5) led to an overall evaluation of the futureflood risk and the related uncertainty. Central estimates ofpopulation annually affected in 2050 are within 500,000 and640,000 and within 540,000 and 950,000 in 2080. Largervariability is foreseen in the future economic growth andconsequently in the expected damage of flooding, with centralestimates at 20–40 Bs in 2050 and 30–100 Bs per year in 2080.

High-end climate scenarios are hereby shown to be linked witha significantly larger impact of future river floods on the Europeaneconomy and society. Extreme flood events are catastrophic andunpredictable until days or hours before they take place.International coordination is fundamental to prepare and put intoaction effective mitigation and adaptation plans, together withraised awareness and resilience to better cope with naturalcatastrophes and streamline the recovery process.

Acknowledgement

The research leading to these results has received funding fromthe European Union Seventh Framework Programme FP7/2007-2013 under grant agreement no. 603864 (HELIX: “High-EndcLimate Impacts and eXtremes”; www.helixclimate.eu).

References

ABI, 2005. Financial Risks of Climate Change—Summary Report. Association ofBritish Insurers.

Alfieri, L., Smith, P.J., Thielen, J., Beven, K.J., 2011. A staggered approach to flash floodforecasting—case study in the Cevennes region. Adv. Geosci. 29, 13–20. doi:http://dx.doi.org/10.5194/adgeo-29-13-2011.

Alfieri, L., Salamon, P., Bianchi, A., Neal, J., Bates, P., Feyen, L., 2014. Advances in pan-European flood hazard mapping. Hydrol. Process. 28, 4067–4077. doi:http://dx.doi.org/10.1002/hyp.9947.

Alfieri, L., Burek, P., Feyen, L., Forzieri, G., 2015. Global warming increases thefrequency of river floods in Europe. Hydrol. Earth Syst. Sci. Discuss. 12, 1119–1152. doi:http://dx.doi.org/10.5194/hessd-12-1119-2015.

Arnell, N.W., Gosling, S.N., 2014. The impacts of climate change on river flood risk atthe global scale. Clim. Change 1–15. doi:http://dx.doi.org/10.1007/s10584-014-1084-5.

Bates, P., De Roo, A.P., 2000. A simple raster-based model for flood inundationsimulation. J. Hydrol. 236, 54–77. doi:http://dx.doi.org/10.1016/s0022-1694(00)00278-X.

Batista e Silva, F., Lavalle, C., Koomen, E., 2012. A procedure to obtain a refinedEuropean land use/cover map. J. Land Use Sci. 1–29. doi:http://dx.doi.org/10.1080/1747423X.2012.667450.

Batista e Silva, F., Gallego, J., Lavalle, C., 2013. A high-resolution population grid mapfor Europe. J. Maps 9, 16–28. doi:http://dx.doi.org/10.1080/17445647.2013.764830.

Bubeck, P., De Moel, H., Bouwer, L.M., Aerts, J.C.J.H., 2011. How reliable areprojections of future flood damage? Nat. Hazards Earth Syst. Sci. 11, 3293–3306.doi:http://dx.doi.org/10.5194/nhess-11-3293-2011.

Burek, P., Knijff van der, J., Roo de, A., 2013a. LISFLOOD, Distributed Water Balanceand Flood Simulation Model Revised User Manual 2013. Publications Office,Luxembourg.

Burek, P., van der Knijff, J., Ntegeka, V., 2013. LISVAP, Evaporation Pre-Processor forthe LISFLOOD Water Balance and Flood Simulation Model—Revised UserManual (No. doi: 10.2788/26160), EUR 26167 EN. Joint Research Centre—Institute for Environment and Sustainability. EUR 26167 EN. 36 pp.

Carpenter, T.M., Sperfslage, J.A., Georgakakos, K.P., Sweeney, T., Fread, D.L., 1999.National threshold runoff estimation utilizing GIS in support of operationalflash flood warning systems. J. Hydrol. 224, 21–44.

Dankers, R., Feyen, L., 2009. Flood hazard in Europe in an ensemble of regionalclimate scenarios. J. Geophys. Res. Atmos. 114 doi:http://dx.doi.org/10.1029/2008jd011523.

Dankers, R., Arnell, N.W., Clark, D.B., Falloon, P.D., Fekete, B.M., Gosling, S.N., Heinke,J., Kim, H., Masaki, Y., Satoh, Y., Stacke, T., Wada, Y., Wisser, D., 2014. First look atchanges in flood hazard in the Inter-Sectoral Impact Model IntercomparisonProject ensemble. Proc. Natl. Acad. Sci. U. S. A. 111, 3257–3261. doi:http://dx.doi.org/10.1073/pnas.1302078110.

EEA, 2010. Mapping the Impacts of Natural Hazards and Technological Accidents InEurope—An Overview of the Last Decade. European Environment Agency (ISBN978-92-9213-168-5 No. EEA Technical report No. 13/2010).

Ehret, U., Zehe, E., Wulfmeyer, V., Warrach-Sagi, K., Liebert, J., 2012. HESS opinionsshould we apply bias correction to global and regional climate model data?Hydrol. Earth Syst. Sci. 16, 3391–3404. doi:http://dx.doi.org/10.5194/hess-16-3391-2012.

European Commission, 2007. Directive 2007/60/EC of the European Parliament andof the Council of 23 October 2007 on the assessment and management of floodrisks.

Evans, E.P., Johnson, P., Green, C., Varsa, E., 2000. Risk assessment and programmeprioritisation: the Hungary flood study. MAFF Flood and Coastal DefenceConference .

Falter, D., Schröter, K., Dung, N.V., Vorogushyn, S., Kreibich, H., Hundecha, Y., Apel, H.,Merz, B., 2015. Spatially coherent flood risk assessment based on long-termcontinuous simulation with a coupled model chain. J. Hydrol. 524,182–193. doi:http://dx.doi.org/10.1016/j.jhydrol.2015.02.021.

Feyen, L., Dankers, R., Bódis, K., Salamon, P., Barredo, J.I., 2012. Fluvial flood risk inEurope in present and future climates. Clim. Change 1–16.

Fluet-Chouinard, E., Lehner, B., Rebelo, L.-M., Papa, F., Hamilton, S.K., 2015.Development of a global inundation map at high spatial resolution fromtopographic downscaling of coarse-scale remote sensing data. Remote Sens.Environ. 158, 348–361. doi:http://dx.doi.org/10.1016/j.rse.2014.10.015.

Foudi, S., Osés-Eraso, N., Tamayo, I., 2015. Integrated spatial flood risk assessment:the case of Zaragoza. Land Use Policy 42, 278–292. doi:http://dx.doi.org/10.1016/j.landusepol.2014.08.002.

Hall, J.W., Evans, E.P., Penning-Rowsell, E.C., Sayers, P.B., Thorne, C.R., Saul, A.J., 2003.Quantified scenarios analysis of drivers and impacts of changing flood risk inEngland and Wales: 2030–2100. Environ. Hazards 5, 51–65. doi:http://dx.doi.org/10.1016/j.hazards.2004.04.002.

Hirabayashi, Y., Mahendran, R., Koirala, S., Konoshima, L., Yamazaki, D., Watanabe, S.,Kim, H., Kanae, S., 2013. Global flood risk under climate change. Nat. Clim.Change 3, 816–821. doi:http://dx.doi.org/10.1038/nclimate1911.

Huang, S., Krysanova, V., Hattermann, F.F., 2014. Does bias correction increasereliability of flood projections under climate change? A case study of large riversin Germany. Int. J. Climatol. doi:http://dx.doi.org/10.1002/joc.3945.

Huizinga, H.J., 2007. Flood Damage Functions for EU Member States. HKV Lijn inwater, Lelystad, the Netherlands.

Hurrell, J.W., Kushnir, Y., Ottersen, G., Visbeck, M., 2003. An overview of the NorthAtlantic oscillation. Geophys. Monogr. Am. Geophys. Union 134, 1–36.

Jacob, D., Petersen, J., Eggert, B., Alias, A., Christensen, O.B., Bouwer, L.M., Braun, A.,Colette, A., Déqué, M., Georgievski, G., Georgopoulou, E., Gobiet, A., Menut, L.,Nikulin, G., Haensler, A., Hempelmann, N., Jones, C., Keuler, K., Kovats, S., Kröner,N., Kotlarski, S., Kriegsmann, A., Martin, E., van Meijgaard, E., Moseley, C., Pfeifer,S., Preuschmann, S., Radermacher, C., Radtke, K., Rechid, D., Rounsevell, M.,Samuelsson, P., Somot, S., Soussana, J.-F., Teichmann, C., Valentini, R., Vautard,R., Weber, B., Yiou, P., 2014. EURO-CORDEX: new high-resolution climate changeprojections for European impact research. Reg. Environ. Change 14, 563–578.doi:http://dx.doi.org/10.1007/s10113-013-0499-2.

Jarvis, A., Reuter, H.I., Nelson, A., Guevara, E., 2008. Hole-filled Seamless SRTM dataV4. International Centre for Tropical Agriculture (CIAT).

Jongman, B., Hochrainer-Stigler, S., Feyen, L., Aerts, J.C.J.H., Mechler, R., Botzen, W.J.W., Bouwer, L.M., Pflug, G., Rojas, R., Ward, P.J., 2014. Increasing stress ondisaster-risk finance due to large floods. Nat. Clim. Change 4, 264–268. doi:http://dx.doi.org/10.1038/nclimate2124.

de Kok, J.-L., Grossmann, M., 2009. Large-scale assessment of flood risk and theeffects of mitigation measures along the Elbe River. Nat. Hazards 52, 143–166.doi:http://dx.doi.org/10.1007/s11069-009-9363-6.

Te Linde, A.H., Bubeck, P., Dekkers, J.E.C., de Moel, H., Aerts, J.C.J.H., 2011. Future floodrisk estimates along the river Rhine. Nat. Hazards Earth Syst. Sci. 11, 459–473.doi:http://dx.doi.org/10.5194/nhess-11-459-2011.

Lugeri, N., Kundzewicz, Z.W., Genovese, E., Hochrainer, S., Radziejewski, M., 2010.River flood risk and adaptation in Europe—assessment of the present status.Mitig. Adapt. Strateg. Glob. Change 15, 621–639. doi:http://dx.doi.org/10.1007/s11027-009-9211-8.

Muerth, M.J., Gauvin St-Denis, B., Ricard, S., Velázquez, J.A., Schmid, J., Minville, M.,Caya, D., Chaumont, D., Ludwig, R., Turcotte, R., 2013. On the need for biascorrection in regional climate scenarios to assess climate change impacts onriver runoff. Hydrol. Earth Syst. Sci. 17, 1189–1204. doi:http://dx.doi.org/10.5194/hess-17-1189-2013.

Neal, J., Villanueva, I., Wright, N., Willis, T., Fewtrell, T., Bates, P., 2011. How muchphysical complexity is needed to model flood inundation? Hydrol. Process. doi:http://dx.doi.org/10.1002/hyp.8339.

Ntegeka, V., Salamon, P., Gomes, G., Sint, H., Lorini, V., Thielen, J., Zambrano-Bigiarini, M., 2013. EFAS-Meteo: A European Daily High-resolution GriddedMeteorological Data Set for 1990–2011.

O’Neill, B.C., Kriegler, E., Riahi, K., Ebi, K.L., Hallegatte, S., Carter, T.R., Mathur, R., vanVuuren, D.P., 2014. A new scenario framework for climate change research: the

http://www.helixclimate.euhttp://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0005http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0005http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0010http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0010http://dx.doi.org/10.5194/adgeo-29-13-2011http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0015http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0015http://dx.doi.org/10.1002/hyp.9947http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0020http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0020http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0020http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0025http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0025http://dx.doi.org/10.1007/s10584-014-1084-5http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0030http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0030http://dx.doi.org/10.1016/s0022-1694(00)00278-Xhttp://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0035http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0035http://dx.doi.org/10.1080/1747423X.2012.667450http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0040http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0040http://dx.doi.org/10.1080/17445647.2013.764830http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0045http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0045http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0045http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0050http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0050http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0050http://10.2788/26160http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0060http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0060http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0060http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0065http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0065http://dx.doi.org/10.1029/2008jd011523http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0070http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0070http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0070http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0070http://dx.doi.org/10.1073/pnas.1302078110http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0080http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0080http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0080http://dx.doi.org/10.5194/hess-16-3391-2012http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0090http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0090http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0090http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0095http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0095http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0095http://dx.doi.org/10.1016/j.jhydrol.2015.02.021http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0100http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0100http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0105http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0105http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0105http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0105http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0110http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0110http://dx.doi.org/10.1016/j.landusepol.2014.08.002http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0115http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0115http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0115http://dx.doi.org/10.1016/j.hazards.2004.04.002http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0120http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0120http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0120http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0125http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0125http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0125http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0130http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0130http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0135http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0135http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0140http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0140http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0140http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0140http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0140http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0140http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0140http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0140http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0140http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0145http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0145http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0150http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0150http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0150http://dx.doi.org/10.1038/nclimate2124http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0155http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0155http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0155http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0160http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0160http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0160http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0165http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0165http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0165http://dx.doi.org/10.1007/s11027-009-9211-8http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0170http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0170http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0170http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0170http://dx.doi.org/10.5194/hess-17-1189-2013http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0175http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0175http://dx.doi.org/10.1002/hyp.8339http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0185http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0185


concept of shared socioeconomic pathways. Clim. Change 122, 387–400. doi:http://dx.doi.org/10.1007/s10584-013-0905-2.

Penning-Rowsell, E.C., 2015. A realistic assessment of fluvial and coastal flood risk inEngland and Wales. Trans. Inst. Br. Geogr. 40, 44–61. doi:http://dx.doi.org/10.1111/tran.12053.

Perez, J., Menendez, M., Mendez, F.J., Losada, I.J., 2014. Evaluating the performanceof CMIP3 and CMIP5 global climate models over the north-east Atlantic region.Clim. Dyn. 1–18. doi:http://dx.doi.org/10.1007/s00382-014-2078-8.

Rijks, D., Terres, J.M., Vossen, P., 1998. Agrometeorological Applications for RegionalCrop Monitoring and Production Assessment (No. Tech. Rep. EUR 17735 EN). EC-Joint Research Centre, Ispra, Italy.

Rojas, R., Feyen, L., Bianchi, A., Dosio, A., 2012. Assessment of future flood hazard inEurope using a large ensemble of bias-corrected regional climate simulations. J.Geophys. Res. D Atmos. 117 doi:http://dx.doi.org/10.1029/2012JD017461.

Rojas, R., Feyen, L., Watkiss, P., 2013. Climate change and river floods in the EuropeanUnion: socio-economic consequences and the costs and benefits of adaptation.Glob. Environ. Change 23, 1737–1751. doi:http://dx.doi.org/10.1016/j.gloenvcha.2013.08.006.

Sampson, C., Smith, A., Bates, P., Neal, J., Alfieri, L., Freer, J., 2015. A high-resolutionglobal flood hazard model. Water Resour. Res. doi:http://dx.doi.org/10.1002/2015WR016954 Accepted Author Manuscript.

Schumann, G.J.-P., Bates, P.D., Neal, J.C., Andreadis, K.M., 2014. Technology: fightfloods on a global scale. Nature 507, 169. doi:http://dx.doi.org/10.1038/507169e.

Themeßl, M.J., Gobiet, A., Heinrich, G., 2012. Empirical-statistical downscaling anderror correction of regional climate models and its impact on the climate changesignal. Clim. Change 112, 449–468. doi:http://dx.doi.org/10.1007/s10584-011-0224-4.

Thielen, J., Bartholmes, J., Ramos, M.-H., De Roo, A., 2009. The European flood alertsystem—part 1: concept and development. Hydrol. Earth Syst. Sci. 13, 125–140.

Thiemig,V., Rojas, R.,Zambrano-Bigiarini, M.,DeRoo,A.,2013. Hydrological evaluationof satellite-based rainfall estimates over the Volta and Baro-Akobo Basin. J. Hydrol.499, 324–338. doi:http://dx.doi.org/10.1016/j.jhydrol.2013.07.012.

Tramblay, Y., Amoussou, E., Dorigo, W., Mahe’, G., 2014. Flood risk under futureclimate in data sparse regions: linking extreme value models and floodgenerating processes. J. Hydrol. 519, 549–558. doi:http://dx.doi.org/10.1016/j.jhydrol.2014.07.052.

UNISDR, 2009. UNISDR Terminology on Disaster Risk Reduction.van der Knijff, J.M., Younis, J., de Roo, A.P.J., 2010. LISFLOOD: a GIS-based distributed

model for river basin scale water balance and flood simulation. Int. J. Geogr. Inf.Sci. 24, 189–212.

van Vuuren, D.P., Carter, T.R., 2014. Climate and socio-economic scenarios forclimate change research and assessment: reconciling the new with the old.Clim. Change 122, 415–429. doi:http://dx.doi.org/10.1007/s10584-013-0974-2.

Wanders, N., Bierkens, M.F.P., de Jong, S.M., de Roo, A., Karssenberg, D., 2014. Thebenefits of using remotely sensed soil moisture in parameter identification oflarge-scale hydrological models. Water Resour. Res. 50, 6874–6891. doi:http://dx.doi.org/10.1002/2013wr014639.

Ward, P.J., Jongman, B., Weiland, F.S., Bouwman, A., van Beek, R., Bierkens, M.F.P.,Ligtvoet, W., Winsemius, H.C., 2013. Assessing flood risk at the global scale:model setup, results, and sensitivity. Environ. Res. Lett. 8, 044019. doi:http://dx.doi.org/10.1088/1748-9326/8/4/044019.

Ward, P.J., van Pelt, S.C., de Keizer, O., Aerts, J.C.J.H., Beersma, J.J., van den Hurk, B.J.J.M., Te Linde, A.H., 2014. Including climate change projections in probabilisticflood risk assessment. J. Flood Risk Manag. 7, 141–151. doi:http://dx.doi.org/10.1111/jfr3.12029.

Winsemius, H.C., Van Beek, L.P.H., Jongman, B., Ward, P.J., Bouwman, A., 2012. Aframework for global river flood risk assessments. Hydrol. Earth Syst. Sci.Discuss. 9, 9611–9659. doi:http://dx.doi.org/10.5194/hessd-9-9611-2012.

Yan, K., Di Baldassarre, G., Solomatine, D.P., Schumann, G.J.-P., 2015. A review of low-cost space-borne data for flood modelling: topography, flood extent and waterlevel. Hydrol. Process. doi:http://dx.doi.org/10.1002/hyp.10449.

http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0185http://dx.doi.org/10.1007/s10584-013-0905-2http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0190http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0190http://dx.doi.org/10.1111/tran.12053http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0195http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0195http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0195http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0200http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0200http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0200http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0205http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0205http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0205http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0210http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0210http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0210http://dx.doi.org/10.1016/j.gloenvcha.2013.08.006http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0215http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0215http://dx.doi.org/[183_TD$DIFF][179_TD$DIFF]10.1002/2015WR016954http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0220http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0220http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0225http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0225http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0225http://dx.doi.org/10.1007/s10584-011-0224-4http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0230http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0230http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0235http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0235http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0235http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0240http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0240http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0240http://dx.doi.org/10.1016/j.jhydrol.2014.07.052http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0250http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0250http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0250http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0255http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0255http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0255http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0260http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0260http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0260http://dx.doi.org/10.1002/2013wr014639http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0265http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0265http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0265http://dx.doi.org/10.1088/1748-9326/8/4/044019http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0270http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0270http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0270http://dx.doi.org/10.1111/jfr3.12029http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0275http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0275http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0275http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0280http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0280http://refhub.elsevier.com/S0959-3780(15)30040-6/sbref0280

Ensemble flood risk assessment in Europe under high end climate scenarios1 Introduction2 Data2.1 Meteorological data2.1.1 Observed data2.1.2 Climate projections

2.2 Static datasets

3 Methods3.1 Numerical models3.2 Modelling approach

4 Results and discussion4.1 Potential and actual flood impact4.2 Flood risk

5 Discussion6 ConclusionsAcknowledgementReferences

global environmental changeluedeke/allerlei/bukarest2018/flood_risk.pdf · 7 april 2015 received in...

Documents