Increasing flood risk under climate change: a pan-European assessment of the benefits of four adaptation strategies
- First Online:
- Cite this article as:
- Alfieri, L., Feyen, L. & Di Baldassarre, G. Climatic Change (2016) 136: 507. doi:10.1007/s10584-016-1641-1
- 2.6k Downloads
Future flood risk in Europe is likely to increase due to a combination of climatic and socio-economic drivers. Effective adaptation strategies need to be implemented to limit the impact of river flooding on population and assets. This research builds upon a recently developed flood risk assessment framework at European scale to explore the benefits of adaptation against extreme floods. The effect of implementing four different adaptation measures is simulated in the modeling framework. Measures include the rise of flood protections, reduction of the peak flows through water retention, reduction of vulnerability and relocation to safer areas. Their sensitivity is assessed in several configurations under a high-end global warming scenario over the time range 1976–2100. Results suggest that the future increase in expected damage and population affected by river floods can be compensated through different configurations of adaptation measures. The adaptation efforts should favor measures targeted at reducing the impacts of floods, rather than trying to avoid them. Conversely, adaptation plans only based on rising flood protections have the effect of reducing the frequency of small floods and exposing the society to less-frequent but catastrophic floods and potentially long recovery processes.
Recent research provides a considerable body of evidence on the effect of anthropogenic greenhouse gas emissions on the Earth’s climate (Stocker et al. 2013). Despite the inevitable uncertainty affecting climatic projections, an increasing number of scientific studies suggest that global warming will exceed 2 °C and range up to 6 °C by the end of the century (Betts et al. 2011; Friedlingstein et al. 2014), following Representative Concentration Pathways (RCP) with radiative forcing up to 8.5 W/m2. In Europe, such a scenario is likely to be linked to a sharp increase in flood risk (Feyen et al. 2012; Alfieri et al. 2015b), making adaptation plans a vital component of current and future disaster risk reduction strategies (Adger et al. 2005; Brandimarte et al. 2009). Flood risk reduction is tackled through structural and non-structural measures involving flood zoning, land-use planning and private precautionary measures, with notable differences in the approach from country to country, even within Europe (Kreibich et al. 2015).
While the number of coordinated flood reduction plans is steadily growing, particularly at community level (e.g., Stahre 2008; Reinhardt et al. 2011), most flood risk prevention actions performed in the past decades focused on corrective rather than preventive measures. After a flood had hit, a recurrent case of flood management was to reinforce and rise flood protections up to a level that would safely confine the peak flow of the river in case a similar event occurred again in the future (see e.g., Fenn et al. 2014). Yet, more and more research studies based on past events acknowledge dykes heightening as measures of last resort or even examples of maladaptation (Hallegatte 2009; Zurich 2014; Wenger 2015), as they give a misleading impression of complete safety which is at odds with the catastrophic consequences in case of failure during flood events (e.g., Di Baldassarre et al. 2015). The last two decades have seen a progressive policy shift towards programs to give “room for rivers” (Rohde et al. 2006; Opperman et al. 2009), aimed to increase the storage space of rivers by restoring floodplains and thus reducing the flood depth by spreading floodwaters over wider areas. Other adaptation options such as relocation to safer areas or flood proofing of buildings require deeper commitment of homeowners and have thus found limited applications in practice (McLeman and Smit 2006; Bichard and Kazmierczak 2012). Yet, insurance programs and disaster financing schemes have large potential in steering the flood risk management in the private and public sectors (Keskitalo et al. 2014; Jongman et al. 2014).
Quantifying the benefits of adaptation measures is crucial for planning nation-wide coordinated actions for flood risk reduction in view of future socio-economic dynamics and the potential intensification of the hydrological cycle and of its extremes (Alfieri et al. 2015a). Current and past research has described extensively the damage reduction potential of a wide range of adaptation options (ABI 2003; Arnbjerg-Nielsen and Fleischer 2009; Kreibich et al. 2011; Woodward et al. 2011). However, only few studies have quantified the benefits of adaptation strategies through simulation approaches, especially in view of future climate change. Among these, Poussin et al. (2012) used a modelling framework to investigate the benefits of spatial zoning, wet and dry flood-proofing on the future flood risk in the Meuse River. At European level, Rojas et al. (2013) and Jongman et al. (2014) used an ensemble of regional climate projections to assess the sensitivity of increased flood protection standards and of risk transfer financing on riverine flood risk throughout the XXI century.
This work relies on the new flood risk assessment framework proposed by Alfieri et al. (2015b) to illustrate the benefits of adaptation in reducing expected damages and population affected by river floods in Europe under 4 °C global warming by the end of the century. Adaptation is here intended as of the Intergovernmental Panel on Climate Change (IPCC) terminology (IPCC 2001), hence measures aimed at reducing the sources or enhance the sinks of greenhouse gases, i.e., classified as “mitigation” measures, are not considered in this work. The risk assessment framework comprises hydrological modelling, threshold-based evaluation of extreme event magnitude and frequency, fully integrated 2D flood hazard mapping, updated exposure maps, country-specific depth-damage functions and improved vulnerability information to estimate current and future flood risk. Within this framework, we consider four different adaptation options and evaluate their effectiveness in risk reduction. Each adaptation option is therefore simulated in 8 to 12 different configurations to assess the sensitivity of its implementation on the resulting flood risk. Risk reduction estimates are obtained by aggregating the results of seven ensemble simulations in space, over 28 European countries, and in time, through three 30-year time slices (TS), to strengthen the robustness of the analysis.
2 Data and methods
Continuous daily streamflow simulations from 1976 to 2100 (Alfieri et al. 2015a), forcing a distributed hydrological model (Lisflood, van der Knijff et al. 2010) with an ensemble of seven EURO-CORDEX (Jacob et al. 2014) RCP 8.5 downscaled regional climate scenarios over Europe. The General Circulation Models (GCM) driving the regional models chosen are rated in the top 25 %, according to a performance evaluation of CMIP5 models carried out by Perez et al. (2014), in their ability to reproduce spatial patterns and climate variability over the north-east Atlantic region, that is the most influential on the European weather patterns.
Estimation of potential population affected (PA) and expected damage (ED) of river floods in Europe in the current climate, through a combination of hydrological and high-resolution (100 m) hydraulic modeling. Output flood extent and depths are coupled with an impact model based on population density and depth-damage relations to estimate PA and ED for selected return periods.
Each flow peak of the future streamflow scenario exceeding the local flood protection levels is assigned an impact (PA and ED) through linear interpolation among the return periods estimated for the current climate.
The contribution of the future socio-economic development foreseen at country level is added through a set of 5-year multipliers provided by the Organisation for Economic Co-operation and Development (OECD). Data and related description can be found at the link https://tntcat.iiasa.ac.at/SspDb/dsd.
This work takes future socio-economic developments based on SSP5 (O’Neill et al. 2014), consistent with high mitigation challenges due to high-end warming and willingness to take adaptation measures against climate impacts. SSP5 assumes a world with rapid economic growth as opposed to relatively small changes in population (van Vuuren and Carter 2014).
It is worth noting that the average impact estimates for the baseline period are quantitatively in agreement with reported figures at European level (see Alfieri et al. 2015b), thus supporting the suitability of the impact model and the underlying datasets for future climate projections.
2.1 Adaptation measures for flood risk reduction
Four types of adaptation measures were considered and implemented to different extents, to assess their sensitivity to the corresponding risk reduction. Each adaptation option targets the reduction of flood risk by acting on one of the three components of the risk formula, namely hazard, exposure and vulnerability. In the figures and the related discussions, multiplicative and reduction rates associated to each adaptation option defined below are referred to as “sensitivity factors”.
2.1.1 Increase of flood protection levels
It aims at reducing the vulnerability of people and assets to extreme streamflow conditions. It requires limited space as it normally consists of elevating the river banks, through permanent or temporary barriers, to increase the maximum streamflow that the watercourse can fully contain and convey downstream without causing damage. This keeps the flood storage to minimum levels hence the magnitude of the flood peak can remain unchanged for long river reaches. As a consequence, its implementation (and maintenance) need be homogeneous within each river basin as local weaknesses would represent preferential triggering points for flooding. In the simulation framework, the return period of current flood protections in Europe, expressed in years, was increased by a set of 12 constant rates ranging between 5 % and 2500 %, where the upper bound was derived by the findings of the post-event adaptation scenario, described in Sect. 3.1.
2.1.2 Reduction of the peak flows
This adaptation option aims at reducing the flood hazard through a reduction and a delaying of peak flows during extreme events. Peak reduction is achieved by setting up areas within or aside the river network that can be flooded in a controlled manner when the river stage reaches critical levels. In addition, peak flows are reduced by reservoirs, sustainable urban drainage systems (SUDS, e.g., Pasche et al. 2008), retarding basins, infiltration basins, and through targeted land management plans such as afforestation and river renaturation (Reinhardt et al. 2011). In this study, we run the impact model with a set of 11 different reduction factors between 5 % and 95 % applied to the return period (i.e., the average recurrence interval) of simulated discharge peaks.
2.1.3 Reduction of vulnerability
It includes all adaptation options which can be modelled through a progressive reduction of the vulnerability, including the implementation of early warning systems, dry and wet flood proofing, and floating buildings, among others (see Strangfeld and Stopp 2014; Kreibich et al. 2015; Pappenberger et al. 2015). In the impact model, the adaptive measure is implemented through a multiplicative factor, ranging between 0 and 1, applied to the damage curves and to the population density layer. One should note that this measure does not reduce the frequency of flooding events but rather the consequences of the flooding, hence the reduction in population affected is to be seen as a reduction of the degree of disruption to the population and their activities.
It reduces the exposure of people and assets at risk of flooding by moving them to areas with negligible risk (King et al. 2014). In this adaptation option, some assumptions are taken to identify areas where the relocation would occur. Past events showed that flood relocation is primarily driven by economic evaluations and mostly occurs after catastrophic events which makes the reconstruction costs of the same magnitude of buying a new property (Kick et al. 2011; López-Carr and Marter-Kenyon 2015). The relocation mask was defined as the set of areas with 3 or more meters of flood depth following an event with return period of 20 years, assuming no flood protections in place. By definition, in these areas, flooding has a 50 % probability to occur in a 13.5 year period, so it is likely to be experienced by permanent residents once or more in their lifetime. In addition, on European average, 3 m correspond to roughly 75 % of the maximum potential flood damage for several land use classes including residential, commerce and industrial, among others (Huizinga 2007). One can note that the criteria defining the relocation mask are independent of the local exposure and vulnerability, thus suggesting the following two considerations: 1) this measure leads to higher benefits in countries with considerable developments along rivers and potentially large flood depths, while little risk reduction can be achieved in wide flood plains where the flooding rarely reaches high depths. 2) The relocation mask is evaluated independently of the local flood protection standards, as a failure in the protections would likely induce a very large impact and a difficult recovery process. In the impact assessment, we tested 8 different relocation ratios between 5 % and 100 %, to be applied as multiplicative factors to people and assets located within the area defined by the relocation mask. These modified exposure layers are then used within the risk assessment framework to estimate the impact of future flood peaks and their corresponding inundation depths.
3.1 Post-event adaptation
3.2 Sensitivity analysis of adaptation strategies
The results and applicability of the proposed adaptation measures should be considered in light of their inherent assumptions and limitations. Sensitivity factors approaching 100 % reduction of the peak flow and of the vulnerability (Fig. 2 to Fig. 4) are unrealistic with technologies currently available. Simulations in the upper range of sensitivity are shown for completeness of the analysis as well as to show the effect of the climate uncertainty at different sensitivity levels. In real world applications, peak flow reduction rates rarely exceed 50 % (Pasche et al. 2008; Reinhardt et al. 2011) and tend to decrease with the event magnitude and with the catchment area. With regard to vulnerability reduction, early warning systems are known to yield profitable cost-benefit ratios (Pappenberger et al. 2015), though with relatively low risk reduction ratios (Meyer et al. 2012). On the other hand, structural measures for vulnerability reduction lead to higher risk reduction rates, at the expense of more considerable investments.
An additional risk component is due to the probability of failure of the flood protections for event magnitudes lower than the design standards, as often occurs in flood events (Apel et al. 2006; Serre et al. 2008; Zurich 2014).
Heightening river dykes reduces the probability of overflowing thus minimizing the floodplain storage and increasing the magnitude of peak flows downstream.
Rising flood protections and the consequent reduction in the frequency of flooding events favors the loss of flood memory, leading to increasing exposure in flood-prone areas (Di Baldassarre et al. 2015). This dynamic, usually referred to as “levee effect”, is characterized by potentially long flood-free periods followed by catastrophic events and large flood losses.
The latter point is supported by the results in Fig. 5 showing how, for fixed risk reduction values, rising flood protections lead to larger socio-economic impacts than in the case of relocation and vulnerability reduction. Past research has shown that the difficulty and the time of recovery of population and ecosystems increase more than linearly with the relative impact of events (Romme et al. 1998; Me-Bar and Valdez 2004), leading the system to a complete collapse in case of extreme disasters. In this regard, the European Union Solidarity Fund (EUSF) was set up to support EU member states significantly affected by disasters, to help and speed up the recovery process. However, Jongman et al. (2014) suggested that the expansion of the EUSF budget to compensate for future large scale floods is infeasible with the projected increasing trend in flood losses for the current century. In addition, such compensation mechanism might be a disincentive for governments to undertake active risk reduction efforts. On the other hand, empirical evidence suggests that recurrent flooding is usually associated with decreasing vulnerability (e.g., Wind et al. 1999; Kreibich and Thieken 2009; Jongman et al. 2015), due to the enhanced resilience and coping capacity acquired by the society during previous events (so-called “adaptation effect”).
A final comment is devoted to the uncertainty of climate projections and their impact on adaptation. The benefits of methods relying on reducing the exceedance of flood thresholds (i.e., rising flood protections, reducing peak flow) heavily depend on the future climate scenario. In some cases, the magnitude of future climate extremes is within a relatively wide range around that of local flood protections, so that the consequent ensemble range of estimated risk reduction can be large. Striking examples are those of the UK, TS 2050 (Fig. 3), with up to 50 % of uncertainty in the risk reduction, and other examples in the Supplement, such as for Belgium (TS 2020), Denmark (TS 2050), Estonia (TS 2020), Hungary (TS 2050), Luxemburg (TS 2080) and Netherlands (TS 2080). Uncertainty in risk reduction consistently decreases in the case of relocation and disappear altogether in vulnerability reduction, as these measures rely on reducing the consequences of a flooding event, rather than trying to avoid it. In addition, despite our effort to characterize and possibly minimize the climatic uncertainty, one should be aware of other sources of uncertainty (e.g., in the hydrological and hydraulic modeling, in the space-time discretization, in the impact model, among others) which affect complex modeling framework such as the one presented in this work.
In a 4 °C global warming scenario, the socio-economic impact of river floods in Europe is likely to triple before the end of the century (Alfieri et al. 2015b). In this work, we implemented a relatively simple model applicable to large areas, which can be used to assess the sensitivity of linear changes in different components of the risk formula, to the overall flood risk under selected climatic scenarios. In comparison to previous similar works, the key feature of this research is the use of a high-resolution modeling framework combined with a large number of simulated scenarios, resulting from the product of four adaptation options, an average of ten rates of implementation of each option, seven regional climate models spanning 125 years of climatic data, and two flood impact indicators. We showed how four different classes of adaptation options can reduce the future flood risk to compensate for the impact of climate change. Research findings suggest that current relative flood impact levels can be retained or even decreased in the future decades, provided that coordinated and effective adaptation plans are promptly prepared and put into action.
Under the projected increase in frequency and magnitude of river floods, traditional approaches based only on rising indefinitely local flood protections are not sustainable in the long term. The combined effect of these two dynamics is likely to exacerbate the “levee effect” by reducing the frequency of moderate events and exposing the society to few catastrophic floods, followed by potentially long and painful post-event recovery. We recommend future adaptation strategies to be based on a combination of different measures working in synergy and optimized at the level of river basins, rather than through independent actions over selected river reaches. In agreement with previous research (Zurich 2014; Di Baldassarre et al. 2015), we have showed that adaptation efforts should give priority to measures targeted at reducing the consequences of hazardous events, rather than trying to avoid their occurrence. In particular, relocation and vulnerability reduction measures should be further developed, due to their two key features of 1) reducing the impacts of all floods without reducing their frequency, thus strengthening the resilience of societies and ultimately the “adaptation effect”; and 2) reducing the effects of uncertainty in future climate on the consequent risk reduction due to adaptation measures. Further adaptation measures to reduce the peak flow should make use of natural retention capacity upstream, while rising flood protections should be seen as last resort, to compensate for the residual risk in areas where other options cannot be implemented. In the latter case, best practice in the realization of new structures include 1) the need for gradual and non-catastrophic failure in case of overload, and 2) building in redundancy, so that a single failure in the system would not compromise the overall flood risk protection capacity.
The research leading to these results has received funding from the European Union Seventh Framework Programme FP7/2007-2013 under grant agreement no 603864 (HELIX: “High-End cLimate Impacts and eXtremes”; www.helixclimate.eu).
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided 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.