Integrated Environmental Assessment and Management — Volume 5, Number 1—pp.

17–26
Ó 2009 SETAC 17

Flood Risk Assessment in European River Basins—Concept,

Methods, and Challenges Exemplified at the Mulde River
Volker Meyer,* Dagmar Haase, and Sebastian Scheuer`
UFZ–Helmholtz Centre for Environmental Research, Department of Economics, Permoserstraße 15, 04318 Leipzig, Germany
`Martin-Luther-University Halle-Wittenberg, Institute for Geosciences, Department Thematic Cartography and Remote Sensing,
06099 Halle, Germany

Review
(Received 20 August 2008; Accepted 2 October 2008)

EDITOR’S NOTE:
This is 1 of 12 papers prepared by participants attending the workshop ‘‘Risk Assessment in European River Basins—State of the Art and
Future Challenges’’ held in Liepzig, Germany on 12–14 November 2007. The meeting was organized within the framework of the European
Commission’s Coordination Action RISKBASE program. The objective of RISKBASE is to review and synthesize the outcome of European
Commission FP4–FP6 projects, and other major initiatives, related to integrated risk assessment–based management of the water/

Special Series
sediment/soil environment at the river basin scale.

ABSTRACT
Flood risk assessment is an essential part of flood risk management, a concept that is becoming more and more popular
in European flood policy and is part of the new European Union flood directive. This paper gives a brief introduction into
the general concept and methods of flood risk assessment. Furthermore, 3 problems in the practical application of flood
risk assessment, particularly on the river basin scale, are discussed: First, uncertainties in flood risk assessment; second, the
inclusion of social and environmental flood risk factors; and third, the consideration of the spatial dimension of flood risk.
In the 2nd part of the paper a multicriteria risk mapping approach is introduced that is intended to address these 3
problems.

Keywords: Risk assessment Flood Multicriteria analysis

INTRODUCTION: FLOOD RISK ASSESSMENT AND out where these risks are unacceptable high, risk reduction
MANAGEMENT aims at finding measures to decrease flood risk. Thereby, the
In recent years a shift in flood policy in Europe from the old whole spectrum of flood risk reduction measures is consid-
concept of ‘‘flood protection’’ or ‘‘flood defense’’ to the new ered. That means measures aimed at reducing the flood
paradigm of ‘‘flood risk management’’ can be recognized hazard, like dikes or retention measures, are taken into
(Schanze 2006; Johnson et al. 2007). To simplify matters, account, but also measures that focus on the reduction of
flood protection aims at preventing flood hazards up to a vulnerability, like land use restrictions in the flood plain,
certain magnitude by providing a certain protection level warning systems, or insurance.
(e.g., against floods up to an exceedence probability once in This concept of flood risk management is the basis for the
100 y). Such protection levels are mostly established by new European Union ‘‘directive on the assessment and
means of structural measures such as dikes, reservoirs, flood management of flood risk’’ (Directive 2007/60/EC). This
polders, or river channel improvements directive prescribes risk assessment and mapping as well as
In contrast, flood risk management does not only considers the development of flood risk management plans, aimed at
the hazard but also the possible consequences. Therefore, reducing adverse consequences (see especially Articles 4, 6,
flood risk management measures not only aim to control flood and 7 of the directive). Furthermore, the directive deter-
waters but also consider possibilities to reduce the vulner- mines that these actions should be carried out at a river basin
ability for flooding; for example, by reducing the number of scale.
elements at risk and/or their susceptibility (Messner and The term risk is understood in the context of flood risk
Meyer 2006). Furthermore, flood risk management attempts management as the probability of negative consequences
to adjust flood protection to the risk situation by concentrat- (FLOODsite Consortium 2005; Schanze 2006). By consid-
ing risk reduction measures to areas with high expected ering the negative consequences or damage of floods for the
damage in order to spend public funds in an economically whole range of probabilities, flood risk can be expressed as
efficient way (Messner and Meyer 2006). the expected annual average damage due to flooding,
Flood risk management can be broadly divided in 2 steps whereas damage covers economic, social, as well as environ-
(Schanze 2006): Flood risk assessment and flood risk mental consequences. This understanding goes back to the
reduction. While the objective of risk assessment is to provide definition of risk introduced by Knight (1921; e.g., Köck
information on current or future flood risks in order to find 2001; Hansjürgens 2004). Especially in social science this
definition of risk is often criticized because it suggests that
risk is something measurable, objective, and naturally given
* To whom correspondence may be addressed: [email protected]
(e.g., Banse and Bechmann 1998), but the measurability of
Published on the Web 10/10/2008.
risk is often restricted, for example, because of high
18 Integr Environ Assess Manag 5, 2009—V Meyer et al.

Figure 1. Damage–probability curve.

uncertainties. Furthermore, sociologists like Renn (1998) Apart from these general components, damage evaluation
argue that risk is not naturally given but is always associated approaches differ considerably in detail. Regarding their
with human decisions or actions. We nevertheless apply this spatial scale and accuracy level the existing methods can be
risk formula in the following because we believe that even broadly differentiated into macro-, meso-, and micro-scale
an uncertain estimation of a risk measure can be a valuable approaches (Messner et al. 2007). Macro-scale approaches
basis of information for new human decisions. often rely on land use information with a low spatial
For the practical application of flood risk assessment, flood resolution and/or low typological differentiation in order to
damage has to be estimated for flood events of different reduce the effort of analysis and hence permit considering
probability in order to construct a damage-probability curve large river basins as a whole (see IKSR 2001; Sayers et al.
(see Figure 1). The risk (or the annual average damage) is 2002). Micro-scale approaches (for example, Penning-Row-
shown by the area or the integral under the curve. An sell et al. 2003) on the other hand try to achieve more
approximation of this area can be expressed using the accurate results by applying very detailed, object-oriented
following equations (DVWK 1985; Eqns.1 and 2): land use data, as well as value and susceptibility information.
However, this requires more effort, which often restricts
X
k
D̄ ¼ D½i 3 DPi ð1Þ these approaches to small areas. Meso-scale approaches (for
i¼1 example, Klaus and Schmidtke 1990; Kok et al. 2004) fit in
somewhere between macro- and micro-scale approaches with
DðPi  1Þ þ DðPi Þ regard to both accuracy and effort. Hence, they are often
D½i ¼ ð2Þ applied in small- or medium-sized river basins, coastal
2
stretches, or dike ring areas. In our example in the second
where D̄ is the annual average damage, D[i] the mean damage
part of this paper, we therefore apply a meso-scale approach
of 2 known points of the curve, and DPi ¼ jPi  Pi1j the
for the Mulde River basin.
probability of the interval between those points.
As this last example shows the new approach of flood risk
This shows that the ex-ante evaluation of flood damages is
management still has to face problems with regard to its
an essential part of risk assessment. A large variety of
practical application, especially on a river basin scale. Three of
approaches currently exists for the estimation of flood
these practical problems will be described in the following
damage. Usually they deploy the following kind of input data
section. After that, a multicriteria risk assessment approach
in order to estimate flood damage (Messner et al. 2007):
will be briefly introduced which considers these 3 problems.
 Inundation characteristics, such as data especially on the
estimated area and inundation depth of a flood event, CURRENT PROBLEMS IN RISK ASSESSMENT ON A
calculated by hydrodynamic models; RIVER BASIN SCALE
 Information on number and type of the exposed elements
at risk (people, properties, biotopes, etc.), usually Uncertainties in risk assessment: A trade-off between
gathered from land use data sources; accuracy and effort
 Information concerning the value of these elements at risk Even though great improvements within methods of flood
(either in monetary or nonmonetary terms); damage evaluation and risk assessment have been made
 Information on the susceptibility of these elements at during the last decades, the uncertainties in the results are still
risk, usually expressed by depth–damage relationships. high; for example, due to data problems in all parts of risk
Flood Risk Assessment in European River Basins—Integr Environ Assess Manag 5, 2009 19

analysis (Handmer 2003; Downton and Pielke 2005; Nacht- measured on its own scale. Generally speaking, the key to
nebel 2007; Apel et al. 2008). These uncertainties cannot be the aggregation of the different risks is a weighting of the
reduced completely. Hence, an objective flood risk may exist different criteria. This requires an involvement of decision
but will never be exactly quantified. makers and stakeholders in the assessment process.
However, as stated above, flood risk assessment approaches
differ concerning the degree of accuracy they are able to The spatial dimension of risks and risk reducing effects is
achieve. Thus, it is a question of which degree of uncertainty often not considered
in flood risk assessment one is willing to accept with respect The spatial distribution of risks as well as the effects of flood
to the objective of the study. The choice of an appropriate risk mitigation measures is rarely taken into account. Thus, the
assessment approach is hence always a trade-off between evaluation and selection of appropriate reduction measures are
accuracy and effort. In order to support decisions on concrete mostly based on their overall net benefit. Therefore, it is often
risk reduction measures for a specific site, detailed micro-scale not considered which areas benefit most from a measure and
approaches should be applied. On the other hand these which areas do not. This may lead to spatial disparities of flood
micro-approaches are mostly not applicable on a river basin risk that are not desirable or even acceptable.
scale due to their high effort and enormous data require- Furthermore, little attention is given at present to the
ments. For flood risk assessment on a river basin scale, effects flood risk reduction measures have to areas down-
therefore, mainly macro- or meso-scale approaches are stream in the river basin. So, for example, for the evaluation
applied, which could lead to considerable uncertainties in of dikes often only local risk reducing effects are considered as
the results, especially with regard to the spatial accuracy of benefits while negative effects on the flood peak in down-
the results, because land use data sources or damage functions stream areas are neglected.
with a high level of aggregation are applied.
In any case, pseudo-accuracy should be avoided by MULTICRITERIA RISK MAPPING APPROACH
documenting and, if possible, quantifying the uncertainties In the following discussion, an approach is briefly intro-
within the risk assessment results. This can be either duced that is intended to address the 3 problems mentioned
documented by a range (i.e., a lower and upper and maybe above. This multicriteria risk mapping approach has been
a mean value), by a standard deviation figure, or by means of a developed in the context of the FLOODsite-project and is
probability distribution. described in greater detail in Meyer et al. (2007) and Meyer et
Thereby, the request for transparent documentation of the al. (2008). By combining GIS-based risk mapping with
uncertainties of risk assessment can be satisfied (Köck 2001). multicriteria analysis (Malczewski 1999) our approach is able
However, it should be noted that risk assessment should not to show the spatial distribution of risks and to consider not
be the only decision-determining criteria but only one source only economic but also social and environmental risk criteria.
of information within the decision-making process. At least for the economic risk criterion a very basic approach
is applied in order to document the uncertainties associated
Social and environmental flood risks are often neglected
with flood risk assessment. The approach is tested at the
The current practice of flood risk assessment as described at Vereinigte Mulde River in the Free State of Saxony, Germany.
the beginning of this paper still focuses on economic damages, This river stretch is approximately 60 km long (i.e., the
especially damage to buildings and their inventories. In approach should be capable of dealing with flood risk
contrast, social and environmental effects of flooding—for assessment on a basin scale). However, for better visual-
instance, loss of life, stress, or destruction of biotopes—are ization, the maps in the following are restricted to a small 4 3
often not considered or considered in a very limited manner. 4 km section around the city of Eilenburg, located in the
This is partly because they are not, or at least not easily, northern part of the area.
measurable in monetary terms and hence not comparable with As risk criteria the evaluation criteria described in Table 1
economic damages. In consequence, flood risk assessment is are applied. With this set of criteria we consider all 3
often incomplete and hence biased, because an important dimensions of sustainability (the social, economic, and
intangible part of the overall risk is neglected. This should be environmental dimensions). However, in order to keep the
modified by the new European Union flood directive (Direc- approach applicable this set of criteria is quite small and
tive 2007/60/EC), which demands in Article 6 a risk assess- simple. For a more comprehensive assessment, criteria can be
ment and mapping of social, economic, and environmental added or the given criteria might be further improved.
flood risk and also directly refers to the Water Framework The general procedure of the multicriteria risk mapping
Directive (Directive 2000/60/EC). However, an aggregation of approach is shown in Figure 2. The basis for all damage
these different risk dimensions is not necessarily required. evaluations is inundation depth data for flood events with a
Nevertheless, approaches exist that make it possible to different exceedence probability of discharge (1:10, 1:25,
include the ‘‘intangible’’ effects in an overall risk assessment. 1:50, 1:100, 1:200, 1:500) calculated by Wenk (Schanze et al.
The traditional approach of environmental economics would 2008) using HEC-RAS, a quasi 2D hydrodynamic model
be to monetize such public goods by methods like contingent (www.hec.usace.army.mil/software/hec-ras/). This is com-
valuation or hedonic pricing (Hanley and Spash 1993). bined with information on the spatial distribution of elements
However, such valuations of life and environmental goods at risk for the different criteria (Step 1) in order to calculate
are still often criticized and not only due to ethical reasons absolute damage or affected units, respectively, for these
(see e.g., Hansjürgens 2004 for a discussion of such different flood events (Step 3). Based on these different
criticisms). damage figures and their associated probabilities the expected
Another way of including intangibles in an overall assess- annual average damage or flood impact is calculated by means
ment is multicriteria analysis (see e.g., Keeney and Raiffa of Equation 1 (Step 5). The units of measurement for each of
1993; Malczewski 1999). Here, each criterion can be the criteria are shown in Table 1. Each of the different criteria
20 Integr Environ Assess Manag 5, 2009—V Meyer et al.

Table 1. Evaluation criteria for flood risk for all 3 dimensions of sustainability

Flood risk Damage unit

dimension Criteria Subcriteria [. . ./year] Description Method/data
Economic Annual average E Damages on Meso-scale damage
damage assets (buildings, evaluation approach,
inventories etc.) based on official
statistics, land use data,
and depth–damage
functions
Environmental Aggregated Erosion Affected: yes/no Areas with Data from flood
environmental risk potential erosion in 2002
potential of
pollutants
Accumulation Affected: yes/no Areas with Data from flood
potential accumulation in 2002
potential of
pollutants
Inundation of Affected: yes/no Areas of biotopes Identification of
oligotrophic vulnerable to vulnerable biotopes
biotopes inundation based on biotope-type
data
Social Annual average Number of people Number of people Meso-scale approach,
affected population affected affected at their based on official
home statistics and land
use data
Probability of Affected: yes/no Vulnerable Point data of
vulnerable community locations vulnerable community
community locations like hospitals, schools, locations based on
of being affected elderly peoples’ homes, own surveys
etc.

risk maps is then standardized to values between 0 and 1 here by applying 3 different sets of depth–damage curves.
(Step 6), weighted (Step 7), and finally, aggregated to a Three different sets of damage functions were applied, taken
multicriteria risk map (Step 8). All these calculations are from the KRIM project at the German North-Sea coast (Mai
carried out for a grid with a spatial resolution of 10 m. The et al. 2007), a damage evaluation study from the River Rhine
procedure for the different criteria will be described in more (IKSR 2001), and from the Dutch standard method (Kok et
detail below. al. 2004). Damage evaluation is carried out for every
For the economic risk criterion we focus on direct damages, combination of asset value map and damage function set,
like damage to buildings and inventories, because this is the resulting in 6 different damage estimations for each flood
most important economic damage category (Messner et al. event (Step 3).
2007). For the assessment of direct damages a meso-scale Based on these different damage values a mean damage can
damage evaluation approach is applied (see Meyer 2005). The be calculated for each grid cell as well as a minimum and
general procedure is the following: The total value of assets at maximum damage value (Step 4). The different damage
risk (Step 1) and the spatial distribution are estimated based on estimations for the inundation events considered are then
data from official statistics combined with land use data. For used to calculate the annual average damage according to the
different economic sectors the net value of fixed assets, which Equation 1 (Step 5). This is conducted for the mean as well as
includes residential and nonresidential buildings and fixed for the minimum and maximum damage estimations so that
inventories as well as the value of stocks and household goods, the final output is a mean, minimum, and maximum annual
are calculated based on official statistics. The different value average damage per grid cell, accordingly. The spatial
distribution of the mean annual average damage for the city
categories are then assigned to corresponding land use
of Eilenburg is shown in Figure 3.
categories, taken from ATKIS-DLM, the German official
In order to assess the environmental risk of an inundation
topographic data source. In order to consider methodological
in the Mulde floodplains, 3 subcriteria have been selected
uncertainties 2 different spatial modeling keys are applied here.
(Table 1):
Relative depth–damage curves are then used to calculate
the damaged share of these values, depending on inundation 1. The erosion potential of areas (i.e., where pollutants
depth (Step 2). Methodological uncertainties are considered might be mobilized during a flood).
Flood Risk Assessment in European River Basins—Integr Environ Assess Manag 5, 2009 21

Figure 2. Stepwise procedure of the multicriteria risk mapping approach.

2. The accumulation potential of areas (i.e., where new The area size is an appropriate proxy for all 3 criteria
polluted sites might be created by a flood). The representing the potential risk existing. The data recorded for
identification of both erosion and accumulation areas is the delineation of environmental risk zones stem from a study
based on data from the Mulde flood in 2002. on the remote-sensing and GIS-based identification of the
3. Inundation of oligotrophic biotopes (i.e., biotopes which areas affected by the Mulde flood in 2002 (Haase et al. 2003).
might be negatively affected by a longer inundation). After intersecting the maps for these subcriteria with
These areas are identified by biotope data for the area. inundation data (Step 1) a simple Boolean yes/no damage
Since we were aiming to develop a simple assessment tool function is applied for each criterion and grid cell, depending
we focused on a limited number of subcriteria for the on whether the area is affected or not. Because the 3 criteria
environmental risk. For regional stakeholders in the Mulde are different in terms of their environmental impact but may
basin the 3 criteria listed were the most crucial showing what occur simultaneously during one unique flood event, we
material re-allocation occurs in the riparian zone during the suggest calculating a sum of the values given for each
flood (subcriteria 1 and 2) and where more long-term affected subcriterion to estimate a first environmental impact poten-
sites that might be irreversibly damaged (subcriterion 3) are. tial of a flood. The list is not complete and has to be amplified
22 Integr Environ Assess Manag 5, 2009—V Meyer et al.

Figure 3. Annual average damage (AAD) (City of Eilenburg): Mean estimation.

Figure 4. Environmental risk: Standardized values (0–1).

Flood Risk Assessment in European River Basins—Integr Environ Assess Manag 5, 2009 23

Figure 5. Annual affected population and vulnerable community locations at risk and their probability of being flooded.

according to both other case study requirements and general of the annual average affected population can be calculated
improvements. (Step 5; see Figure 5).
Analogous to the calculation of economic damage, damage As ‘‘vulnerable community locations’’ the locations of
maps for environmental consequences can be produced for hospitals, schools, and homes of the elderly and children are
each flooding event (Step 3). Each raster cell can hereby identified (Step 1). For reasons of simplicity we assume that
achieve ‘‘damage values’’ between 0 and 3. Based on these each of the community locations has the same vulnerability,
different damage maps an environmental risk map is no matter, for example, the size of a hospital or school. For a
calculated by using Equation 1 (Step 5). This risk value can more detailed study it would of course be desirable to include
be interpreted as annual average environmental flood impact, more variables that would express the vulnerability of each
expressed in the point scale described above. In Figure 4 these community location. Intersecting the vulnerable community
values are already standardized in values from 0 to 1 (Step 6). locations map with the inundation maps leads to the
determination of affected community locations per inunda-
As social risk criteria 2 very simple indicators are used: The
tion scenario (Step 3). By applying Equation 1 an approx-
annual average affected population and the probability of
imate estimation of the probability of being affected can be
vulnerable community locations like hospitals, schools, and
calculated for each vulnerable community location (Step 5;
elderly people’s homes being flooded (see Table 1). The
Figure 5).
annual average affected population is chosen as a criterion For the aggregation of the different risk maps (Step 8) a
because it is a simple but comprehensive indicator for the simple additive weighting approach is used (see e.g.,
flood risk of the local population in general. The social hot Malczewski 1999). The general model for this is
spot criterion is chosen in order to emphasize a bit more on X
spatial concentration of people with a high susceptibility, like Vi ¼ wj vij ð3Þ
children, the elderly, or chronically ill people. j

For the determination of the affected population, first of all Where Vi is the overall value or utility of the grid cell i, vij is
the spatial distribution of the population is calculated by a the standardized value of grid cell i regarding criterion j (Step
meso-scale approach (Step 1; Meyer 2005). By intersecting 6), and wj is the standardized weight for criterion j (Step 7).
this population density map with the inundation data the This means a weighted standardized average of the single
number of affected people can be estimated for each event criterion values for each area is calculated.
(Step 3). With ‘‘affected people’’ we mean the number of For a real site-specific assessment this would of course
inhabitants whose home or at least the first floor of their require the involvement of decision makers in order to
dwelling is flooded, regardless of how many of these determine the weights given to each criterion. In our
inhabitants are at home at that time and on which floor of example such a participation of decision makers did not
the building they live. According to Equation 1, the number take place (yet) so arbitrary set of weights are chosen. Figure
24 Integr Environ Assess Manag 5, 2009—V Meyer et al.

Figure 6. Standardized multicriteria risk: Large weight on economic and population criteria (0.4 each).

Figure 7. Comparison of risk value distribution for different sets of weights: Large weight (0.4) on the economic and population criterion (Mean_pop4), on the
population (Pop4), on the economic (Econ4), on the environmental (Ecol4), and on the vulnerable community location criterion (Hot4).

6 shows, for example, the results if a relatively high weight of example, we compare the risk value distribution of the area
0.4 is given to the economic and the population criterion, as shown in Figure 6 for different weight sets. In addition to the
these 2 criteria are often considered as the most important weight set displayed in Figure 6 (Mean_Pop4) it also includes
protection goods. 4 different weights set where a large weight of 0.4 is given to
However, the approach also allows showing the sensitivity each one of the criteria (Pop4, Econ4, Ecol4, and Hot4). All
of the results against different sets of weights. In Figure 7, for weight sets show a very high number of square meters with
Flood Risk Assessment in European River Basins—Integr Environ Assess Manag 5, 2009 25

very low risk values (0.001–0.1). The ‘‘Mean_pop4’’ weight risk assessment is of course only a first step. Further research
set, which assigns large weight to 2 criteria, has a further is still needed on how to deal with such uncertain information
subdominant peak in the value area 0.3 to 0.4 while the other in the decision-making process in flood risk management and
weight sets show subpeaks in lower risk values (0.1–0.3). how to determine an appropriate level of accuracy for the
In practice, such a sensitivity analysis of weight sets can be different decision problems.
used to visualize the changes in the risk maps if weight sets of Second, our approach showed how social, environmental,
different stakeholder groups are used. In real world decision and economic risk criteria can be aggregated, but it also
situations often several different stakeholder groups with unveiled how important the involvement of decision makers
different preference structures are involved in the decision- in the evaluation process is, as the selection of criteria and the
making process. Representatives from a nature conservation determination of weights given to them substantially deter-
organization, for example, will likely give a high weight to mine the results of the assessment process.
environmental criteria, while representatives from a chamber In our example we did not involve stakeholders yet, so the
of commerce will tend to give high weight on economic exemplary results shown are still arbitrary. However, liter-
criteria. Our approach can be used to compare the results of ature provides some examples and procedures for how to
different weight sets in order to find compromise solutions. integrate stakeholders in multicriteria analysis (e.g., Proctor
But the sensitivity with regard to uncertainties in the criteria and Drechsler 2006). Consequently, the next step would be
values can also be shown by conducting the same calculation to interlink such participation approaches with our multi-
with the minimum and maximum criteria value map. criteria assessment approach.
However, in our example this is only possible for the Third, a number of problems also arise from the mapping
economic criterion. of flood risk. The accuracy of spatial flood risk assessment
depends on the resolution of the underlying grid. However,
SUMMARY AND CONCLUSIONS the use of high resolution grids is limited by at least 2 factors:
The new paradigm of flood risk management is obviously Computational implementation (i.e., system memory avail-
getting more and more popular in flood policy in Europe, and ability and addressing) and the level of detail and resolution of
it is the conceptual background for the new European Union the corresponding input data like inundation and land use
flood directive. The assessment of flood risks is an essential data. This also means that the mapping of flood risk is always
part of this approach. This paper briefly described the general associated with inaccuracies or uncertainties and the right
concept of assessing flood risk as well as different approaches level of spatial resolution is maybe not easy to find.
for flood damage evaluation. Furthermore, if spatial disparities are identified by risk
We furthermore outlined 3 problems in the current mapping it is still an open question how to deal with them.
practical application of flood risk assessment: Our approach is able to identify high risk areas but it does not
yet answer the question of how much they should be reduced
1. How to deal with uncertainties in the results of flood risk and how risk reduction efforts can be distributed ‘‘fairly’’ in
assessment; river basins (cf. Johnson et al. 2007).
2. How to include social and environmental flood risks; Acknowledgment—We would like to thank 2 anonymous
3. How to properly consider the spatial distribution of reviewers for their fruitful comments. Furthermore, we
flood risks. would like to thank Gerald Wenk and Michael Rode for
We introduced a multicriteria risk mapping approach providing the inundation data. The work described in this
which tries to consider these 3 problems. First, at least for publication was supported by the European Community’s
the economic criterion, methodological uncertainties in the Sixth Framework Programme through the grant to the budget
results are documented by mean, maximum, and minimum of the Integrated Project FLOODsite, Contract GOCE-CT-
risk values. Second, our approach includes an exemplary set of 2004-505420. This paper reflects the authors’ views and not
economic as well as social and environmental flood risk those of the European Community. Neither the European
criteria. These different criteria can be aggregated by means of Community nor any member of the FLOODsite Consortium
a multicriteria decision rule being a simple additive weighting is liable for any use of the information in this paper.
approach. Third, the spatial distribution of flood risks as well
as risk reducing effects can be displayed by a GIS-based risk
REFERENCES
mapping approach (i.e., all risk assessment calculations are
Apel H, Merz B, Thieken AH. 2008. Quantification of uncertainties in flood risk
carried out for a grid with a resolution of 10 m). assessments. International Journal of River Basin Management 6:149–162.
However, our approach should be seen only as a first Banse G, Bechmann G. 1998. Interdisziplinäre Risikoforschung—eine Bibliogra-
attempt to deal with these problems. The approach still has phie. Wiesbaden (DE): Westdeutscher Verlag.
some limitations and further research tasks can be identified. Downton MW, Pielke RA. 2005. How accurate are disaster loss data? The case of
First, with regard to uncertainties, our approach documented U.S. flood damage. Natural Hazards 35:211–228.
such uncertainties only for the economic criterion. Further- [DVWK] Deutscher Verband für Wasserwirtschaft und Kulturbau. 1985. Ökono-
more, only 2 sources of model uncertainties were considered: mische Methoden von Hochwasserschutzwirkungen. Arbeitsmaterialien zum
Uncertainties resulting from the choice of the spatial methodischen Vorgehen. In: DVWK-Mitteilungen. Bonn (DE): DVWK.
FLOODsite Consortium. 2005. Language of risk—Project definitions. FLOODsite
allocation key and the set of damage functions. Uncertainties
Project Report T32-04-01. www.floodsite.net.
resulting from the inundation model used were not yet
Haase D, Weichel T, Volk M. 2003. Approaches towards the analysis and
considered. Hydraulic models show some uncertainties assessment of the disastrous floods in Germany in August 2002 and
especially in urban areas, where fluid water flow modeling consequences for land use and retention areas. In: Vaishar A, Zapletalova J,
is highly complex due to the variety of urban structures and Munzar J, editors. Regional geography and its applications, Proceedings of the
thus fluid flow cannot be expressed as overland flow. 5th Moravian Geographical Conference CONGEO’03,15–19 September 2003,
Furthermore, the documentation of uncertainties in flood Brno. p 51–59.
26 Integr Environ Assess Manag 5, 2009—V Meyer et al.

Handmer J. 2003. The chimera of precision: Inherent uncertainties in disaster editors. Flood risk management—Hazards, vulnerability and mitigation
loss assessment. The Australian Journal of Emergency Management 18:88– measures. Berlin (DE): Springer. p 149–167.
97. Messner F, Penning-Rowsell E, Green C, Meyer V, Tunstall S, van der Veen A. 2007.
Hanley N, Spash C. 1993. Cost-benefit analysis and the environment. Cheltenham Guidelines for socio-economic flood damage evaluation. FLOODsite-Report
(UK): Edward Elgar. T09-06-01. www.floodsite.net.
Hansjürgens B. 2004. Economic valuation through cost-benefit analysis— Meyer V. 2005. Methoden der Sturmflut-Schadenspotenzialanalyse an der
Possibilities and limitations. Toxicology 205:241–252. deutschen Nordseeküste. Vom Fachbereich Geowissenschaften und Geo-
[IKSR] International Commission for the Protection of the Rhine. 2001. graphie der Universität Hannover genehmigte Dissertation. UFZ Dissertation 3/
Übersichtskarten der Überschwemmungsgefährdung und der möglichen 2005.
Vermögensschäden am Rhein. Abschlußbericht: Vorgehensweise zur Ermit- Meyer V. 2007. GIS-based multicriteria analysis as decision support in flood risk
tlung der hochwassergefährdeten Flächen, Vorgehensweise zur Ermittlung management. FLOODsite-Report. www.floodsite.net.
der möglichen Vermögensschäden. Wiesbaden (DE): IKSR. Meyer V, Scheuer S, Haase D. 2008. A multicriteria approach for flood risk
Johnson C, Penning-Rowsell E, Parker D. 2007. Natural and imposed injustices: mapping exemplified at the Mulde river. Germany. Natural Hazards (in press).
The challenges in implementing ‘fair’ flood risk management policy in
Nachtnebel HP. 2007. Cost-benefit evaluation of risk reduction options. In:
England. The Geographical Journal 173:374–390.
Schanze J, editor. Flood risk management research—From extreme events
Keeney RL, Raiffa H. 1993. Decisions with multiple objectives—Preferences and
to citizen involvement. Proceedings of European Symposium on Flood Risk
value tradeoffs. Cambridge (MA): Cambridge University Press.
Management Research (EFRM 2007); 2007 February 6–7; Dresden (DE):
Klaus J, Schmidtke RF. 1990. Bewertungsgutachten für Deichbauvorhaben an der
IOER.
Festlandsküste—Modellgebiet Wesermarsch, Untersuchungsbericht an den
Penning-Rowsell E, Johnson C, Tunstall S, Tapsell S, Morris J, Chatterton J, Coker A,
Bundesminister für Ernährung, Landwirtschaft und Forsten. Bonn (DE): BMELF.
Green C. 2003. The benefits of flood and coastal defence: Techniques and
Knight FH. 1921. Risk, uncertainty, and profit. Boston (MA): Hart, Schaffner, and
data for 2003. Enfield (UK): Flood Hazard Research Centre.
Marx.
Proctor W, Drechsler M. 2006. Deliberative multicriteria evaluation. Environment
Köck W. 2001. Rationale Risikosteuerung als Aufgabe des Rechts—Möglichkeiten
and Planning C-Government and Policy 24:169–190.
und Grenzen des Einsatzes komparativer Risikoanalysen und Kosten-Nutzen-
Renn O. 1998. Three decades of risk research: accomplishments and new
Analysen im Rahmen administrativer Risikobewertungen. In: Gawel E, editor.
Effizienz im Umweltrecht. Baden-Baden (DE): Nomos. p 273–304. challenges. Journal of Risk Research 1:49–71.
Kok M, Huizinga HJ, Vrouwenfelder ACWM, Barendregt A. 2004. Standard Sayers P, Hall J, Dawson R, Rosu C, Chatterton J, Deakin R. 2002. Risk assessment
method 2004. Damage and casualties caused by flooding. Client: Highway of flood and coastal defences for strategic planning (RASP)—A high level
and Hydraulic Engineering Department. Delft (NL): DWW. methodology. DEFRA Conference for Coastal and River Engineers; 2002
Mai S, Grabemann I, Eppel DP, Elsner A, Elsner W, Grabemann HJ, Kraft D, Meyer September. London (UK): DEFRA.
V, Otte C, Yu I, Wittig S, Zimmermann C. 2007. KRIM: Methode der Schanze J. 2006. Flood risk management—A basic framework. In: Schanze J,
erweiterten Risikoanalyse. In: Schuchardt B, Schirmer M, editors. Land unter? Zeman E, Marsalek J, editors. Flood risk management—Hazards, vulnerability
Klimawandel, Küstenschutz und Risikomanagement in Nordwestdeutschland: and mitigation measures. Berlin (DE): Springer. p 149–167.
Die Perspektive 2050. Munich (DE): Oekom. p. 75–92. Schanze J, Bernhofer C, Caletkova J, Görner C, Ferger K-H, Franke J, Haase D,
Malczewski J. 1999. GIS and multicriteria decision analysis. New York (NY): John Hutter G, Kodrova Z, Kominkova D, Lennartz F, Luther J, Meister S, Meyer V,
Wiley and Sons. Olfer A, Rode M, Scheuer S, Sequeira M, Wahren A, Wenk G, Zikmund V.
Messner F, Meyer V. 2006. Flood damage, vulnerability and risk perception— 2008. Floodsite pilot study ‘‘Elbe River Basin’’—Executive Summary. Floodsite
Challenges for flood damage research. In: Schanze J, Zeman E, Marsalek J, Report T21-08-01. www.floodsite.net.