Nat. Hazards Earth Syst. Sci., 6, 485–503, 2006www.nat-hazards-earth-syst-sci.net/6/485/2006/                             ...
486                                                 B. B¨ chele et al.: Flood-risk mapping:...extreme events and associate...
B. B¨ chele et al.: Flood-risk mapping:...extreme events and associated risks    u                                        ...
488                                                  B. B¨ chele et al.: Flood-risk mapping:...extreme events and associat...
B. B¨ chele et al.: Flood-risk mapping:...extreme events and associated risks    u                                        ...
490                                                                                                                       ...
B. B¨ chele et al.: Flood-risk mapping:...extreme events and associated risks    u                                        ...
492                                                 B. B¨ chele et al.: Flood-risk mapping:...extreme events and associate...
B. B¨ chele et al.: Flood-risk mapping:...extreme events and associated risks    u                                        ...
494                                                 B. B¨ chele et al.: Flood-risk mapping:...extreme events and associate...
B. B¨ chele et al.: Flood-risk mapping:...extreme events and associated risks    u                                        ...
496                                                   B. B¨ chele et al.: Flood-risk mapping:...extreme events and associa...
B. B¨ chele et al.: Flood-risk mapping:...extreme events and associated risks    u                                        ...
Risks reading1
Risks reading1
Risks reading1
Risks reading1
Risks reading1
Risks reading1
Upcoming SlideShare
Loading in …5
×

Risks reading1

229 views

Published on

0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
229
On SlideShare
0
From Embeds
0
Number of Embeds
2
Actions
Shares
0
Downloads
3
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Risks reading1

  1. 1. Nat. Hazards Earth Syst. Sci., 6, 485–503, 2006www.nat-hazards-earth-syst-sci.net/6/485/2006/ Natural Hazards© Author(s) 2006. This work is licensed and Earthunder a Creative Commons License. System SciencesFlood-risk mapping: contributions towards an enhanced assessmentof extreme events and associated risksB. Buchele1,3 , H. Kreibich2 , A. Kron1,3 , A. Thieken2 , J. Ihringer1 , P. Oberle1 , B. Merz2 , and F. Nestmann1 ¨1 Institutefor Water and River Basin Management, Technical University of Karlsruhe, Germany2 GeoForschungsZentrum Potsdam, Section Engineering Hydrology, Potsdam, Germany3 Center for Disaster Management and Risk Reduction Technology (CEDIM), Karlsruhe, GermanyReceived: 5 January 2006 – Revised: 22 March 2006 – Accepted: 28 March 2006 – Published: 12 June 2006Abstract. Currently, a shift from classical flood protection as 1 Introductionengineering task towards integrated flood risk managementconcepts can be observed. In this context, a more conse- Within the framework of the Center for Disaster Man-quent consideration of extreme events which exceed the de- agement and Risk Reduction Technology (CEDIM, a co-sign event of flood protection structures and failure scenarios operation between the University of Karlsruhe and the GFZsuch as dike breaches have to be investigated. Therefore, this Potsdam), the project “Risk Map Germany” aims to providestudy aims to enhance existing methods for hazard and risk improved methods and data for the quantification and map-assessment for extreme events and is divided into three parts. ping of risks. Here, in particular the developments concern-In the first part, a regionalization approach for flood peak dis- ing flood risk assessment are presented. The damages in Ger-charges was further developed and substantiated, especially many due to severe flood disasters in the last decades amountregarding recurrence intervals of 200 to 10 000 years and a to billions of Euro. Examples are the Rhine floods in 1993large number of small ungauged catchments. Model com- with 530 M , and in 1995 with 280 M , the Odra flood in 1997parisons show that more confidence in such flood estimates with 330 M , the Danube flood in 1999 with 412 M , the Elbefor ungauged areas and very long recurrence intervals may and Danube flood in 2002 with 11 800 M (Kron, 2004). Thebe given as implied by statistical analysis alone. The hy- need of specific research efforts and spatial data, particularlydraulic simulation in the second part is oriented towards haz- hazard or risk maps, for an improved risk assessment andard mapping and risk analyses covering the whole spectrum prevention on regional and local level is evident.of relevant flood events. As the hydrodynamic simulation is In this project, “hazard” is defined as the occurrence of adirectly coupled with a GIS, the results can be easily pro- flood event with a defined exceedance probability. “Risk”cessed as local inundation depths for spatial risk analyses. is defined as the potential damages associated with such anFor this, a new GIS-based software tool was developed, be- event, expressed as monetary losses. It becomes clear thating presented in the third part, which enables estimations of hazard and risk quantification depend on spatial specifica-the direct flood damage to single buildings or areas based on tions (e.g., area of interest, spatial resolution of data). Withdifferent established stage-damage functions. Furthermore, regard to flood risk, the local water level is decisive for thea new multifactorial approach for damage estimation is pre- occurrence of damage (e.g. Smith, 1994; see below). There-sented, aiming at the improvement of damage estimation on fore, a high level of detail, i.e. an appropriate scale of floodlocal scale by considering factors like building quality, con- maps is a fundamental precondition for a reliable flood risktamination and precautionary measures. The methods and assessment. Detailed spatial information on flood hazard andresults from this study form the base for comprehensive risk vulnerability is necessary for the development of regionalanalyses and flood management strategies. flood-management concepts, planning and cost-benefit anal- ysis of flood-protection measures and, extremely important, for the preparedness and prevention strategies of individual stakeholders (communities, companies, house owners etc.). Moreover, the mapped information that an area or object is potentially endangered due to a given flood scenario directly implies legal and economical consequences such as com-Correspondence to: B. B¨ chele u petencies of public authorities for flood control and spatial(buechele@iwg.uka.de) planning, owner interests, insurance polices, etc.Published by Copernicus GmbH on behalf of the European Geosciences Union.
  2. 2. 486 B. B¨ chele et al.: Flood-risk mapping:...extreme events and associated risks u In Germany, the federal states (Bundesl¨ nder) are respon- a in practice, it can be noted that flood-risk assessment remainssible for flood management and for the generation of flood a quite challenging task, especially regarding the uncertain-maps. Many state authorities have been working for years on ties related to extreme events exceeding the design flood or tothe delineation of inundation zones with map scales of up to the damage due to failures of flood control measures (Apel et≥1:5000 in urban areas in order to recognise the flood haz- al., 2004; Merz et al., 2004). For example, uncertainties asso-ard for discrete land-parcels and objects. Increasingly, public ciated with flood frequency analyses are discussed by Merzflood-hazard maps are available on internet platforms (e.g., and Thieken (2005). However, the visualisation and com-Nordrhein-Westfalen, 2003, Rheinland-Pfalz, 2004, Sach- munication of uncertain information in hazard maps shouldsen, 2004, Bayern, 2005, Baden-W¨ rttemberg, 2005). In u be optimised in a way, that non-experts can understand, trustthe next years, with respect to amendments in legislation, re- and get motivated to respond to uncertain knowledge (K¨ mpfagional significance and technical possibilities, flood-hazard et al., 2005).maps with high spatial resolutions can be expected succes- In contrast to hazard mapping, the assessment of dam-sively for all rivers in Germany. Details of procedures and age and its visualisation as risk maps is still far from be-mapping techniques vary from state to state and due to local ing commonly practised in Germany. Risk maps, however,concerns (e.g., data availability, vulnerability, public funds). help stakeholders to prioritise investments and they enableAn overview of different approaches is published by Klee- authorities and people to prepare for disasters (e.g., Takeuchi,berg (2005). For example in Baden-W¨ rttemberg, two types u 2001; Merz and Thieken, 2004). Good examples for risk as-of flood-hazard maps will be provided for all rivers with sessments and maps are – among others – the ICPR Rhine-catchment areas >10 km2 . The first map will show the ex- Atlas (ICPR, 2001), the programme of flood-hazard mappingtent of inundation zones of the 10-, 50- and 100-year event, in Baden-W¨ rttemberg (UM Baden-W¨ rttemberg, 2005), the u usupplemented by an “extreme event” being in the order of integrated flood management conception in the Neckar rivermagnitude of a 1000-year event and, as documented, infor- basin (IkoNE, 2002), the DFNK approach for the city ofmation on historical events. The second map will provide Cologne (Apel et al., 2004; Gr¨ nthal et al., 2006), and the uthe water depths of the 100-year event. The basic require- risk assessment in England and Wales (Hall et al., 2003).ments and typical features of these upcoming nation-wide Since flood risk encompasses the flood hazard and the con-mosaic of flood-hazard maps can be drafted as follows (com- sequences of flooding (Mileti, 1999), such analyses requirepare e.g. UM Baden-W¨ rttemberg, 2005, MUNLV, 2003): u an estimation of flood impacts, which is normally restricted to detrimental effects, i.e. flood losses. In contrast to the – Representation of present flood-relevant conditions (up- above discussed hydrological and hydraulic investigations, dating after significant changes) flood damage modelling is a field which has not received much research attention and the theoretical foundations of – Representation of inundation zones for flood events of damage models need to be further improved (Wind et al., different recurrence intervals up to generally 100 years, 1999; Thieken et al., 2005). for large rivers 200 years (e.g., Rhine) A central idea in flood damage estimation is the con- – Representation of inundation depths, potentially flow cept of damage functions. Most functions have in common velocities that the direct monetary damage is related to the inundation depth and the type or use of the building (e.g. Smith, 1981; – Representation of extreme, historical events (exceeding Krzysztofowicz and Davis, 1983; Wind et al., 1999; NRC, the 100-year event, as available) 2000; Green, 2003). This concept is supported by the ob- servation of Grigg and Helweg (1975) “that houses of one – Representation of flood-protection measures, poten- type had similar depth-damage curves regardless of actual tially local hazard sources value”. Such depth-damage functions, also well-known as – Level of detail for local analyses and planning purposes stage-damage-functions, are seen as the essential building blocks upon which flood damage assessments are based andFollowing these requirements on flood-hazard assessment they are internationally accepted as the standard approach toand related purposes on local scale, one can rely on a set of assess urban flood damage (Smith, 1994).methods for the quantification of hydrological and hydraulic Probably the most comprehensive approach has been theparameters and their spatial intersection with digital terrain Blue Manual of Penning-Rowsell and Chatterton (1977)models (DTM) and land-use data. A number of investiga- which contains stage-damage curves for both residential andtions (e.g. Uhrich et al., 2002) showed that the quality of commercial property in the UK. In Germany, most stage-flood maps strongly depends on the quality of the DTM used. damage curves are based on the most comprehensive Ger-Uncertainties in DTMs are more and more overcome by an man flood damage data base HOWAS that was arranged byincreasing availability of high-resolution digital terrain mod- the Working Committee of the German Federal States’ Wa-els from airborne surveys (e.g., Laserscanner or aerial pho- ter Resources Administration (LAWA) (Buck and Merkeltographs). In spite of these technical standards and advances 1999; Merz et al., 2004). But recent studies have shownNat. Hazards Earth Syst. Sci., 6, 485–503, 2006 www.nat-hazards-earth-syst-sci.net/6/485/2006/
  3. 3. B. B¨ chele et al.: Flood-risk mapping:...extreme events and associated risks u 487that stage-damage functions may have a large uncertainty proposed by Ihringer (2004) based on statistical analyses of(e.g. Merz et al., 2004). downscaled regional climate-model outputs. The investigation concept within the CEDIM working On the other hand, it is important to bear in mind thatgroup “flood risk” is based on the main goal to improve the flood-estimation procedures in practice mainly rely on ob-flood-risk assessment on local scale in different modules of served discharge data. In the field of hydrology, it has longthe quantification procedure. Special attention is given to been recognised that many annual maximum flood series areextreme events. This was realised in pilot areas in Baden- too short to allow a reliable estimation of extreme events,W¨ rttemberg, where a good data and model basis is given. u leading to the conclusion that instead of developing new Following this modular concept, the quantification proce- methodologies for flood-frequency analysis, the comparisondure can be divided in three major steps: of existing methods and the search for other sources of in- formation have to be intensified (e.g., Bob´ e et al., 1993). e – Regional estimation of flood discharges (basin-, site- This is especially true for small catchment areas where the specific hydrological loads) availability of flow data is generally worse (numerous un- – Estimation of flow characteristics in potential inunda- gauged areas or rather short periods of records). Accord- tion areas (local hydraulic impacts) ing to this, the need of regional analyses to compensate the lack of temporal data and to introduce a spatial dimension – Estimation of the resulting damages (area- or object- in flood estimates is evident. Beside flood-frequency analy- specific risk assessment) sis, regional analyses can help to identify physical or mete- orological catchment characteristics that cause similarity inAn overview of suitable approaches for these steps of hazard flood response. Considerable uncertainties, although beingand vulnerability assessments is given in Table 1. The left an intrinsic part of extreme value estimations, can be man-column states minimum requirements on data and methods aged by the complementary use of different methods (e.g.,for a standard quality of hazard and risk assessment on local flood-frequency analysis and rainfall-runoff models). In fact,scale in Germany. The right column lists more sophisticated a stepwise approximation from different directions, involv-approaches which require more spatial information and more ing both statistical theory as well as knowledge of catchmentcomplex calculations up to fully dynamic simulations of un- characteristics and flood processes seems to be the most vi-observed extreme flood situations. Parts highlighted in bold able way to build confidence in flood estimates, to identifyletters are further addressed by the present study, without giv- and exclude implausible values and thus, to reduce uncer-ing priority to any of the listed possibilities. The present pa- tainties to a smaller bandwidth.per is structured in the Sects. 2, 3 and 4 according to the Hence, the specific goal here is to discuss a regional-above mentioned steps, respectively. ization method for state-wide flood probabilities in Baden- W¨ rttemberg. Emphasis is given to comparisons among u2 Estimation of extreme flood events and their proba- models for recurrence intervals from 200 to 10 000 years. bilities 2.2 Method and data2.1 Basis and objectives The first regionalization methods for flood estimates inThe estimation of flood frequencies is well-known as a key Baden-W¨ rttemberg were developed in the 1980’s (Lutz, utask in flood hazard assessment. Actually, the availabil- 1984). In 1999, the following regionalization approachity of reliable and spatially distributed event parameters for for the mean annual peak discharge (MH Q) and peak dis-extreme floods is a fundamental prerequisite for any com- charges (H QT ) for recurrence intervals T from 2 to 100prehensive flood-risk management. For instance, peak dis- years, partially 200 years was published (LfU, 1999), fol-charges for recurrence intervals up to T=100 years (corre- lowed by an updated version on CD in 2001 (LfU, 2001).sponding to an exceedance probability of one percent per The approach is based on flood frequency analyses at 335year) are commonly accounted in flood mapping and flood- gauges which cover catchment areas from less than 10 km2protection planning. Peak discharges for larger events with (∼7% of all gauges) to more than 1000 km2 (∼7%) and pe-recurrence intervals up to 1000 or even 10 000 years are re- riods of records varying from a minimum of 10 to morequired for dam safety analyses (cf. DIN 19 700), hazard map- than 100 years (average 45 years). At large, these statis-ping for extreme cases, related risk analyses and emergency tical analyses at single gauges involved 12 types of theo-planning purposes. For example in Baden-W¨ rttemberg, a u retical cumulative distribution functions (cdfs); the param-guideline gives specific technical recommendations for the eter estimation was done using the method of moments anddimensioning of flood-protection measures (LfU, 2005a). the method of maximum likelihood. The final selection ofThese recommendations already include the preventative cdfs was supported by regional comparisons (e.g., for neigh-consideration of potential impacts of future climate change boured gauges) to avoid inconsistencies especially for higheron peak discharges by a so-called “climate change factor” recurrence intervals.www.nat-hazards-earth-syst-sci.net/6/485/2006/ Nat. Hazards Earth Syst. Sci., 6, 485–503, 2006
  4. 4. 488 B. B¨ chele et al.: Flood-risk mapping:...extreme events and associated risks uTable 1. Overview of methods and data for high-resolution flood-risk mapping in Germany. The left column (basic approaches) indicatesminimum requirements on data and analysis expenses, the right column stands for more detailed approaches (additional requirements, onlyin particular cases needed/possible). Parts in bold letters are addressed in this article. Basic approach More detailed approach (higher spatial (minimum requirements) differentiation, inclusion of dynamic aspects) Regional estimation of flood discharges – Extreme value statistics (mainly – Rainfall-runoff simulation for for gauged basins) extreme flood events – Regionalization of flood param- – Long-term simulation of flood eters (for ungauged basins) variability (probabilistic evalu- ation, e.g. climate trends) Estimation of local hy- draulic impacts – Documentation of historical wa- – assessment of flow situation based ter levels/inundation zones (vary- on 2-D HN models (e.g., spa- ing availability/quality) tial distribution of flow veloci- – Calculation of water levels and ties/directions) inundation zones/depths based – unsteady hydrodynamic simula- on hydraulic models and digi- tion of extreme flood scenarios tal terrain models (local scale, (e.g., impact of dike failures) e.g. 1:5000) – Simplified approaches (only large-scale, e.g. 1:50 000) Damage estimation – stage-damage curves for aggre- – stage-damage curves for indi- gated spatial data (e.g., ATKIS- vidual buildings or land-units data) (e.g., ALK-data) – consideration of further damage-determining factors (e.g., flow velocity, precau- tionary measures, warning time) Beside the gauge-specific flood quantiles, the following These eight parameters, especially hN G and LF , wereeight parameters as spatial data sets for approximately 3400 identified as significant for the peak discharge. LF is an em-catchment areas are taken into account by the regionalization pirical factor and represents all kind of regional influences,approach. particularly geological characteristics. Together, they are AE0 catchment area [km2 ] taken into account in the following multiple linear regres- S percentage of urban area [%] sion equation, which is used asapproach for flood quantile W percentage of forest area [%] estimation (i.e. MH Q and H QT ) especially for ungauged Ig weighted slope [%] sites: L length of flow path from head of catchment to mouth [km] ln (Y ) = C0 + C1 × ln AE0 + C2 × ln (S + 1) LC length of flow path from center of catchment +C3 × ln (W + 1) + C4 × ln Ig + C5 × ln (L) (1) to mouth [km] +C6 × ln (LC ) + C7 × ln (hNG ) + C8 × ln (LF ) hN G mean annual precipitation depth [mm] LF landscape factor [–]Nat. Hazards Earth Syst. Sci., 6, 485–503, 2006 www.nat-hazards-earth-syst-sci.net/6/485/2006/
  5. 5. B. B¨ chele et al.: Flood-risk mapping:...extreme events and associated risks u 489 with: 0.20 0.5 Y, YT dependant variable: C8 Y =MH q for regionalization of MH Q 0.15 0.0 C7 YT =HqT /MH q for regionalization of H QT MH q mean annual peak discharge per unit area 0.10 -0.5 [m3 /(s×km2 )] C8 C7 H qT annual peak discharge per unit area 0.05 -1.0 [m3 /(s× km2 )] of recurrence interval T C0 ...C8 regression coefficients 0.00 -1.5 The regression coefficients C0 ...C8 are estimated basedon the gauge-specific flood estimates and the above men- -0.05 -2.0 1 10 100 1000 10000tioned spatial data sets (available in LfU, 2005b) using recurrence interval T [a]the method of least squares. The application of this ap-proach requires two steps. First, MH q is estimated using Fig. 1. Progression of regression coefficients C7 and C8 for recur-Eq. (1). Subsequently, H QT in unit [m3 /s] is determined rence intervals T from 2 to 10 000 years.using YT =H qT /MH q in Eqs. (1) and (2):H QT = YT × MH q × AE0 = YT × MH Q (2) for this specific gauge it may be noticed that the regional- ization approach is able to reproduce the shape of the sta-Recently, this approach was extended to recurrence intervals tistical distribution. At the same time, the 95%-confidenceof 200 to 10 000 years using a selection of 249 gauges and interval for the statistical distribution (dashed lines) is indi-applied to a more detailed spatial data set (6200 locations cating substantial uncertainties, especially in the area of ex-of the river network, LfU 2005b). The selection of gauges trapolation (for this sample: about 25% for all H QT withwas done considering the record length and the quality of T≥100 years). Summing up, the deviations between region-the flow series in order to achieve more reliable model ad- alization and flood-frequency analysis vary among the men-justments for low-frequency events. The present regionaliza- tioned 249 gauges as presented in Fig. 3. The mean devia-tion approach is thus consisting of 13 regression equations, tion is <2.5% at approx. 40% of the gauges, <7.5% at ap-i.e. one equation for MH Q and each H QT for T from 2 to prox. 75% and <12.5% at approx. 90%. The deviation is10 000 years. The coefficients (C0 –C8 ) of these equations >20% at approx. 3% of the gauges where generally humanare fully documented in LfU (2005b), at which the corre- activities (e.g., urban drainage systems) or karst conditionssponding coefficients of determination are R 2 >0.99 for all are present. Figure 4 illustrates a sample map of the region-single recurrence intervals (logarithmic analysis). As Fig. 1 alization approach for H q1000 in Baden-W¨ rttemberg. Ac- uexemplifies for C7 and C8 (compare Eq. 1), the coefficients cording to this sample, the highest peak discharges per unitshow a homogeneous progression upon the whole spectrum area occur in the mountainous regions of the Black Forestof recurrence intervals, although they are estimated sepa- (Upper Rhine Basin) and the upper Neckar Basin.rately for each recurrence interval. To enable user-specificestimates, the complete spatial data sets and a calculation To substantiate the regionalization approach especially fortool for the regionalization approach are integrated in a ge- small ungauged catchment areas, the results can be comparedographical information system (LfU, 2005b), which is dis- to outcomes of rainfall-runoff (RR) models which are sup-tributed as stand-alone software to local authorities and plan- posed to build on a better representation of catchment char-ning companies. By these means, regionalized MH Q and acteristics. This was done here for the Fils catchment, aH QT are provided for any user-defined location of the river tributary to the Neckar River (707 km2 , see Fig. 4), wherenetwork in Baden-W¨ rttemberg, completed by analogous in- u a RR-model (software see Ihringer, 1999) is available fromformation at 375 gauges and longitudinal profiles for 163 ma- a hydrological study on local flood problems. Within thejor rivers. Furthermore, the extension of the regionalization RR-model, the catchment is represented by 907 subareasapproach to very high recurrence intervals supports the ongo- and 1501 nodes for the drainage network, considering a to-ing state-wide elaboration of flood hazard maps and regional tal urban area of 92 km2 , 331 stormwater holding tanks ofdam safety analyses. urban drainage systems and 7 flood-retention basins. As input of the RR-model, rainfall statistics provided by the2.3 Model results, comparisons and discussion German Weather Service (DWD, 1997) were used; these rainfall statistics cover recurrence intervals from 1 to 100A comparison of at-site and regional flood-frequency anal- years for different duration classes from 0.5 to 72 h. For theysis is exemplified in Fig. 2 for the Fils river in Plochin- assessment of higher peak discharges, the mean precipita-gen showing that the regionalized H QT (squares) are very tion depths for the different duration classes were extrapo-similar to the statistical distribution for the annual maximum lated. However, different precipitation characteristics (dura-flood series 1926–2002 (solid line). Beyond the agreement tion classes) cause a set of different flood peaks. Therefore,www.nat-hazards-earth-syst-sci.net/6/485/2006/ Nat. Hazards Earth Syst. Sci., 6, 485–503, 2006
  6. 6. 490 B. B¨ chele et al.: Flood-risk mapping:...extreme events and associated risks u 0 .0 1 1 0 0 0 0 5 0 0 0 0 .0 5 g a u g e P lo c h in g e n / F ils 2 0 0 0 N o v 1 9 2 6 - O c t 2 0 0 2 ; 7 6 v a lu e s [% ] 0 .1 0 1 0 0 0 [a ] 5 0 0 e x c e e d a n c e p r o b a b ility 0 .5 0 2 0 0 r e c u r r e n c e in te r v a l T 1 .0 0 1 0 0 2 .0 0 5 0 5 .0 0 2 0 1 0 .0 0 1 0 2 0 .0 0 5 5 0 .0 0 c o n fid e n c e in te r v a l 2 = 9 5 % L O G - N O R M A L -2 9 5 .0 0 P L O T T IN G P O S H Q R e g io n a l. 1 2 3 4 1 0 1 0 1 0 1 0 a n n u a l p e a k d is c h a r g e H Q [m ³/s ]Fig. 2. Comparison of H QT from regionalization (as squares) with flood-frequency analysis at Plochingen/Fils (location see Fig. 4, annualpeak discharges 1926–2002 as plotting positions, Log-Normal distribution as solid line, 95%-confidence interval as dashed lines). 70 MHQ percentage of gauges (n = 249) [%] 60 HQ2 HQ5 50 HQ10 HQ20 40 HQ50 HQ100 30 HQ200 HQ500 20 HQ1000 HQ2000 10 HQ5000 HQ10000 0 -5 ± 2,5 < -30 0 ± 2,5 5 ± 2,5 -25 ± 2,5 -20 ± 2,5 -15 ± 2,5 -10 ± 2,5 10 ± 2,5 15 ± 2,5 20 ± 2,5 25 ± 2,5 > 30 classified deviation [%] (regionalized HQT vs. HQT from statistical analysis at gauges)Fig. 3. Comparison between regionalization and flood-frequency analysis at 249 gauges in Baden-W¨ rttemberg: classified deviations [%] ufor MH Q and H QT for T =2, ..., 10 000 a (bars from left to right).the maximum peak of a set representing the relevant spec- eas (especially for H Q1000 <10 m3 /s). The variation aroundtrum of precipitation characteristics was chosen to estimate the bisecting line, standing for a perfect agreement of boththe 1000-year quantile according of this approach. This value models, can be understood as residual uncertainty of the mu-is compared to the quantile estimated from the regionaliza- tual application of both models for these specific catchmenttion approach based on observed flood peaks. Figure 5 shows areas. The deviations may be further discussed taking morea comparison of 1000-year peak discharges of both mod- knowledge on local characteristics into account, that was notels (H Q1000 from regionalization and RR). According to the yet used in one or both models (e.g., outlets of urban drainagespatial discretization of the regionalization approach, 265 lo- systems). Mathematically, the deviations between both mod-cations of the drainage network are plotted. The axes are log- els amount to <7.5% for 66% of all 265 plotted locations,arithmic in order to visualise the small values better. It can be <12.5% for 82% and >30% for 3% of these locations. Theconcluded that the results match fairly well with tendencies latter belong to smaller areas respectively peak dischargesto higher deviations between both approaches in smaller ar- (e.g., 6.96 m3 /s from regionalization versus 11.02 m3 /s fromNat. Hazards Earth Syst. Sci., 6, 485–503, 2006 www.nat-hazards-earth-syst-sci.net/6/485/2006/
  7. 7. B. B¨ chele et al.: Flood-risk mapping:...extreme events and associated risks u 491Fig. 4. Regionalized H q1000 in Baden-W¨ rttemberg; location of sample gauge in Plochingen (Fils catchment, 704 km2 ). uRR, Fig. 5) where the RR-model – in general – may recog- 1000nise local influences better. Therefore, more confidence maybe given to the regionalized H QT as implied by statistical HQ1000 from RR-model [m³/s]analysis alone (Fig. 2). 100 In view of the needs of practitioners, a coherent and ro-bust approach for regional flood estimates is thus availableand broadly established in a state-wide sense. Uncertainties 10of this approach concerning local distinctions call for hydro-logical justifications of the plausibility of flood estimates onlocal scale. The model comparison strategy seems to be the 1 1 10 100 1000logical way for model validation and practically the unique HQ 1000 from regionalization [m³/s]chance to reduce uncertainties effectively in areas where theavailability of flow records is scarce. This is valid not only Fig. 5. Comparison of peak discharges calculated by regionalizationin spatial sense (ungauged areas) but also for the extrapola- and RR-model: H Q1000 at 265 locations of the drainage networktion to very long recurrence intervals. Apart from its practical in the Fils catchment.use for regional flood estimates, the regionalization approachleads from at-site flood frequency analysis to distributed hy-drologic modelling of flood events, enabling a vice-versa re-view and mutual enhancement of these methods. the flow patterns in a given river section are characterised by compact and coherent streamlines, 1-dimensional (1-D) HN-models are considered as adequate for the estimation3 Hydrodynamic modelling for extreme floods of flood-water levels and delineation of inundation zones (e.g., Baden-W¨ rttemberg, 2005). In cases with more com- u3.1 Basis and objectives plex river geometries and flow patterns (e.g., at river con- fluences or other complex flow conditions), 2-dimensionalTo quantify flood hazard and risk in urban areas or at (2-D) models are used for a spatially differentiated hydraulicindividual locations, flood discharges (e.g., H Q100 ) have analysis, especially when local parameters like flow direc-to be transformed into hydraulic parameters like water tion, flow velocity, shear stress, etc. are requested. Depend-levels, inundation depths or flow velocities by means of ing on the intended purposes, both types of models (1-D, 2-hydrodynamic-numerical (HN) models. In many cases when D) may be applied for stationary flow conditions (e.g., hazardwww.nat-hazards-earth-syst-sci.net/6/485/2006/ Nat. Hazards Earth Syst. Sci., 6, 485–503, 2006
  8. 8. 492 B. B¨ chele et al.: Flood-risk mapping:...extreme events and associated risks uassessment for a certain H QT ) or unsteady flow conditions tion and validation for a spectrum of (flood) discharges is(e.g., for impact analyses of dike failures). possible. A detailed description of the system is given by At the Neckar river (see Fig. 4), the pilot area of this part Oberle (2004).of the study, a complex flood-information system was set up Areas that contribute to the retention volume of the Neckarsince the late 1990’s (Oberle et al., 2000), consisting of a se- river during a flood event are taken into account by a func-ries of 1-D- and 2-D-HN-models which are interactively con- tion of storage capacity depending upon the water level. Thisnected with a geographical information system (GIS). This function can be determined from the digital terrain model bysystem enables the simulation of different flood scenarios in means of several GIS-functionalities and can be verified byorder to evaluate, for example, effects of river engineering comparing calculated with surveyed flood hydrographs.measures on flood waves. Through its GIS-interface, the hy-draulic results can be superimposed with a high resolution 3.2 Hydraulic modelling of extreme floodsDTM (grid size: 1×1 m) to determine inundation zones andrespectively the boundaries. The DTM is based on eleva- In most cases, HN-models are applied to documented floodtion data from different data sources, i.e. terrestrial and air- events from the last decades (in order to assess the presentborne surveys. Apart from topographical information, flood- hydraulic conditions) or to statistical flood events with recur-relevant spatial data like flood marks, flood impact area, re- rence intervals up to 100 or 200 years (e.g., for delineationtention zones and legally defined flood areas, are integrated of inundation zones or as design events for flood protectionin the GIS. Linkups to aerial photographs of recent flood measures). With regard to the increasing process complexity,events complete the volume of spatial data sets. for instance in the case of overtopping or even destruction With respect to the main target parameters of flood-risk of a flood protection structure, and to the lack of measure-analysis and mapping (water levels, inundation zones/depths) ments for the calibration and validation of model parametersand to the flow characteristics along the Neckar river, a gen- for such cases, the application of HN-models to larger floodserally 1-dimensional HN procedure was chosen. The choice is rarely practised. However, despite of the uncertainties, itof this procedure was supported by the fact that the handling is necessary to apply HN-models for floods that exceed theof the system and the computing time should match with the 100- or 200-year level, as they are the most relevant situa-size of the study area (approx. 220 river kilometres) and the tions in terms of residual risks, causing severe damages andgoal, to install the system as operational tool for daily work- fatalities. In particular with respect to residual risks, it ising practice in the water management authorities. Finally, obvious that model parameters should also be valid for ex-the calculation of a flood event and the visualisation of in- treme events. Only the reflection of all physically plausibleundation depths in the GIS only takes minutes with this sys- hazardous situations from the occurrence of first inundationstem, so that analyses can be realised also based on actual to the maximum possible water levels yields to a compre-flood forecasts. Some river sections with more complex flow hensive hazard and risk assessment. This applies equally forconditions (e.g., tributary mouths) could be assessed only in- flood situations below the design event, as required e.g. forsufficiently by means of an 1-dimensional approach. Here, cost-benefit-analyses of protection measures or for the as-local 2-D-HN-models were additionally applied. However, sessment of residual risks due to other failures of technicala stationary calculation on the base of a 2-dimensional HN or non-technical measures (e.g., late installation of mobileprocedure requires several hours even using a powerful com- protection elements).puter. Often, historical flood marks indicate much higher water The hydrodynamic method of the above mentioned 1- levels than current flood protection level and thus, shoulddimensional procedure is based on the solution of the serve as realistic reference scenarios for extreme events. InSaint-Venant-Equations by an implicit difference scheme the upper part of the Neckar, the flood with the highest ever(Preissmann-scheme, compare Cunge et al., 1980). Under recorded water levels occurred in 1824. The water marks canthe normal flow conditions of the Neckar river, this approach be found at several buildings in flooded communities giv-is valid and very efficient even for large river sections with ing impressions of the severity of historic floods. They canrespect to data handling, model build-up, model calibration be taken into account for all flood related planning. Withand validation as well as sensitivity analyses and, finally, adapted HN-models (discharge-relevant areas, roughness co-studies of variants. The functionality of the system includes efficients, etc.) it is possible to assess if similar flood watermodelling schemes for looped and meshed river systems as levels could appear in the present situation.well as for river-regulating structures (e.g. weirs, groins, wa- The present HN-calculations at the Neckar river have con-ter power plants). The system geometry of the HN-model, firmed that the historical event 1824 was much higher thani.e. the discharge area of the main channel and the flood- today’s design flood. Figure 6 shows the calculated maxi-plains, is represented by modified cross sections. The model mum water levels of the Neckar river for the 100-year floodcalibration is done by comparing calculated water levels with (H Q100 ) and the historical flood of 1824 (under actual hy-surveyed ones. In most sections, water level measurements draulic conditions). For example, around the community ofof different recent flood events are available, thus a calibra- Offenau, 98 km upstream of the confluence of the NeckarNat. Hazards Earth Syst. Sci., 6, 485–503, 2006 www.nat-hazards-earth-syst-sci.net/6/485/2006/
  9. 9. B. B¨ chele et al.: Flood-risk mapping:...extreme events and associated risks u 493River with the Rhine River, the historical water level of 1824 155.0 Flood marks 1824was approximately 2.5 m higher than the dikes that have been 154.0 HW 1824 HQ200built for a 100-year flood. 153.0 HQ100 It has to be emphasised that the consideration of extreme 152.0 HQ50 water level [m asl] HQ20historical events can not only support flood awareness as re- 151.0 HQ10alised scenarios (under historical conditions), but also used 150.0as reference for the analysis of potential extreme cases un- 149.0der present conditions. In this regard, the intention here was 148.0not to reconstruct historical hydraulic conditions or to ver- 147.0ify historical information in terms of peak discharges. The 146.0results shown can help, for example, to assess the proba- 145.0 106.0 104.0 102.0 100.0 98.0 96.0 94.0bility of flood events that cause comparable water levels in Neckar-kmthe actual situation. In terms of a reconstruction of histori-cal discharges, a further investigation on historical hydraulic Fig. 6. Maximum water levels along the Neckar river for statisticalboundary conditions is required (Oberle, 2004). However, flood events H QT with T =10, 20, ..., 200 years) and the historicaldue to the limited historical data availability and quality, ma- event 1824 (HW1824), the latter as reconstructed from water-leveljor uncertainties are expected. marks. Note location of the community Offenau at the river Neckar (98 km upstream of the confluence with the river Rhine).3.3 Hazard mappingThe above presented hydrological and hydraulic models, and Green, 2000), the largest part of the literature on floodi.e. the regionalization approach for the estimation of ex- damage concerns direct tangible damage (Merz and Thieken,treme events (H QT ) as well as the GIS-based flood informa- 2004). The present study is limited to direct monetary floodtion system for the Neckar river served as basis for the gen- damage to buildings and contents of private households.eration of hazard maps with prototype character in a state- As outlined above, stage-damage-functions for differ-wide sense. For example, hazard maps for the lower Neckar ent building types or building uses are an internationallyriver (Figs. 4 and 6) are published on the -internet platform accepted standard approach for flood damage estimation.(Baden-W¨ rttemberg, 2005). u While the outcome of most stage-damage functions is the absolute monetary loss to a building, some approaches pro- vide relative depth-damage functions, determining the dam-4 Flood-damage estimation age e.g. in percentage of the building value (e.g. Dutta et al.,4.1 Basis and objectives 2003). If these functions are used to estimate the loss due to a given flood scenario property values have to be prede-Based on the knowledge of accumulated values in the areas at termined (Kleist et al., 2004, 2006). However, using theserisk and relationships between event parameters and resulting functions, one has to be aware that the damage estimation isdamage, flood risks can be identified and quantified, i.e. ex- generally associated with large uncertainties as recent stud-pected damages for a given flood scenario can be calculated. ies asserted (Merz et al., 2004). One approach to reduce theThis information about flood risk for individual buildings, uncertainty connected with stage-damage functions is theirsettlement areas and river basins is indispensable to inform specific adjustment to the area of interest (Buck and Merkel,the population and stakeholders about the local flood risk, 1999). This strategy was followed here, supported by inten-for planning of flood control measures and for benefit-cost sive on-site investigations of the building structure in someanalyses of these measures. pilot areas along the Neckar river. The comprehensive determination of flood damage in- Recent flood events have shown that during slowly risingvolves both, direct and indirect damage. Direct damage is river floods the maximum water level during the flood eventa damage which occurs due to the physical contact of flood is responsible for the resulting damage. In these cases, thewater with human beings, properties or any objects. Indirect gradient of the flood wave is small and for this reason theredamage is a damage which is induced by the direct impact, are no damaging effects due to flow velocity impacts. Ma-but occurs – in space or time – outside the flood event. Ex- jor damages are caused by wetting of contents and buildingamples are disruption of traffic, trade and public services. structure in the cellar and the ground floor. This does notUsually, both types are further classified into tangible and in- apply for flash floods e.g. in mountainous areas where, duetangible damage, depending on whether or not these losses to high flow velocity, buildings may collapse partly or to-can be assessed in monetary values (Smith and Ward, 1998). tally. Therefore, it is obvious that flood damage depends,Although it is acknowledged that direct intangible damage or in addition to building type and water depth, on many factorsindirect damage play an important or even dominating role which are not considered using stage-damage functions. Onein evaluating flood impacts (FEMA, 1998; Penning-Rowsell factor is the flow velocity, but there are also others like thewww.nat-hazards-earth-syst-sci.net/6/485/2006/ Nat. Hazards Earth Syst. Sci., 6, 485–503, 2006
  10. 10. 494 B. B¨ chele et al.: Flood-risk mapping:...extreme events and associated risks uduration of inundation, sediment concentration, availability 4.2 Flood damage estimation on local scale based on stage-and information content of flood warning, and the quality of damage curvesexternal response in a flood situation (Smith, 1994; Penning-Rowsell et al., 1994; USACE, 1996). Although a few stud- 4.2.1 GIS-based damage analysis (tool)ies give some quantitative hints about the influence of someof the factors (McBean et al., 1988; Smith, 1994; Wind et As discussed above, it is commonly required in flood-risk as-al., 1999; Penning-Rowsell and Green, 2000; ICPR, 2002; sessment to locate accessible information about hazard andKreibich et al., 2005), there is no comprehensive approach vulnerability at a high spatial resolution (e.g. for cost-benefitto consider these factors in a loss-estimation model. Using analyses, for local protection measures, rating of risks for in-actual flood damage data from the 2002 flood in Germany, surance purposes). In view of these practical requirements, awe followed this idea here and developed a multifactorial ap- GIS-based tool for damage estimation was developed in theproach for damage estimation. present project. This tool supplements the above mentioned flood information system at the Neckar river, i.e. it builds The flood-damage estimation can be undertaken on differ- directly on the water level information for individual endan-ent levels of spatial differentiation: gered objects based on hydrodynamic calculations. The GIS-based tool for damage estimation on local scale uses the following procedure. – On local scale, the damages can be estimated based on spatial data and stage-damage-functions for individual – Selection of the project area (spatial, postcodes or areas buildings or land parcels. In Germany, commonly the of communities). Automated Real Estate Map (ALK) is used for these as- – Identification and categorization of each building in the sessments. The ALK data show the base-area of the sin- project area (based on ALK-Data) gle buildings and give their specific use (e.g. residential building, commercial building, stable, garage). – Estimation of the flood-sill for each structure (lowest damaging water level). – On a more aggregated level, the approach can be based – Estimation of the ground-floor elevation (floor above on statistical information about population, added val- the cellar). ues, business statistics or capital assets for land-use – Estimation of the values for building-structure and con- units. These values are published yearly by respon- tents (fixed/mobile inventory). sible state authorities (statistical offices). Commonly data from the Authoritative Topographic-Cartographic – Estimation of the stage-damage-functions, differenti- Information System (ATKIS) is used for this approach ated for different types of buildings, cellar/floor, build- in Germany. The ATKIS data differentiate more than ing structure/contents. 100 types of land-use (e.g. residential area, power plant, sports facilities). – Calculation of the water-level for each object in the area. – Estimation of the damages to buildings and contents for different water-levels based upon the type and use of – Large-scale analyses may be carried out for larger land- each building. use units, like communities or ZIP-code areas, consid- ering that they may be only partially flooded. These The tool provides the selection of the project area on the base analyses are often based on the CORINE land cover of different spatial or administrative areas: barrages, commu- data (Coordinated Information on the European Envi- nities or postcodes. The area of interest or spatial objects can ronment). The CORINE data differentiates 45 different be selected from tables or as graphical selection in the GIS. types of land-use (e.g. continuous urban fabric, indus- For the damage estimation, the water depth close to trial or commercial units, agro-forestry areas). or inside the object is the determining factor. With the HN-modelling in connection with the digital terrain model (DTM) the water depths above the terrain is calculated. TheDuring the last years, the computational power increased in assumption that the damaging water depth inside the objecta way that today flood damage analyses even for larger river is the same as the depth over terrain is correct if the groundcourses can be undertaken with a high level of detail. In this floor elevation has the same elevation as the surrounding ter-context, the question of spatial scale of damage analysis is ritory and if there are no protection measures. In this case,moving from limitations concerning the area size to limita- the relevant elevation of the object basis can be calculated ontions concerning the quality respectively the level of detail of the base of the DTM as the mean value of the terrain altitudeavailable spatial data sets. on the buildings base. A second option in the tool is to enterNat. Hazards Earth Syst. Sci., 6, 485–503, 2006 www.nat-hazards-earth-syst-sci.net/6/485/2006/
  11. 11. B. B¨ chele et al.: Flood-risk mapping:...extreme events and associated risks u 495the ground floor elevation and the height of the flood-sill for 1.0each single object. Thus, local object features and protection 0.9measures can be considered. 0.8 0.7 The damage estimation is based on the general assump- damage ratio [-] 0.6tion that the monetary damage depends on the type and use 0.5of the building. One of the basic studies was performed by Linear Polygon Function 0.4Penning-Rowsell and Chatterton (1977). In the Blue man- Square-root Function, b=0.5 0.3ual, stage-damage functions for residential buildings in the Square-root Function, b=0.4 0.2 Point-based Power Function C=0.8UK were derived for age and type of the buildings, the du- Point-based Power Function C=3.0 0.1ration of the flood event and the social class of the inhabi- User-defined points 0.0tants. The damages are differentiated for building fabric and 0.0 0.5 1.0 1.5 2.0 2.5contents. Other similar international studies were done by water depth [m]Wind et al. (1999); Smith (1994); Parker et al. (1987). InGermany, in the HOWAS database, some 3600 single dam- Fig. 7. Examples of function types used in the GIS-based damageage cases for different objects are included. The damage data analysis tool.were collected after different flood events in Germany. Anal-yses by Buck and Merkel (1999) showed that for practical the damage estimation tool as second function type, whereuses, damage estimation with a root function provides rea- the parameter b is user-defined:sonable results. √ Due to the fact that the absolute damage depends on a S =b× h (4)variety of factors being specific for every single buildingor land parcel, a meaningful damage estimation can be ex- with S=estimated damage, h=water depth, b=user-definedpected from the application of such stage-damage functions parameter.and their adaptation for individual objects or – in terms of The parameter b characterises the damage for h=1 m.exposure and vulnerability – uniform spatial units. For that Hence, using Eq. (4), the damage progression can be de-reason, the possibility to apply different functions was imple- scribed with only one parameter. For the sample damage es-mented in this software module, where the user can choose timation in this paper (see below), the damage functions forat least one of the three following function types: 1. Lin- different building types in the project area were chosen basedear Polygon Function, 2. Square-Root Function, or 3. Point- on the flood-damage database HOWAS (Buck and Merkel,based Power Function. 1999). – Point-based Power Function – Linear Polygon Function In some cases, damage does not occur until the stage rises toThe user interface allows to enter 5 pairs of variates (hi/Si) a threshold height in the building. For example in rooms orof water-depth and damage, which are interpolated sectional storeys, where floors and walls are tiled, damage can be neg-with linear functions. Between the minimum (i=1) and max- ligible until the water level affects the electrical installationimum (i=5) pair, the function can be noted: (power sockets). On the other hand, the maximum damage is often obtained, when the contents are submerged; a fur- Si+1 − Si ther rise of the water level does not increase the damage inS = Si + × (h − hi ) , hi ≤ h < hi+1 (3) hi+1 − hi a relevant manner. For these cases, a power function can be chosen in the tool, where the points of first and maximumwhere S = estimated damage, Si , hi = user-defined nodes of damage as well as the exponent C determining the gradientthe function, h=water depth. of the function can be individually defined. The first pair of variates (h1 /S1 ) defines the minimum wa- 1ter depth below which the damage is zero. The last point S0 S0 h − h0 C(h5 /S5 ) sets the possible maximum damage; for water depths S = Smax × + 1− × (5) Smax Smax hmax − h0above the damage stays constant. The Polygon function (3)allows a simple adaptation to individual damage symptoms with S=estimated damage, h=water depth, (h0 /S0 )=pointof different types of objects. of first damage, (Smax / hmax )=point of maximum damage, C=user-defined exponent. – Square-Root Function The three function types (Eqs. 3–5) are exemplified in Fig. 7.In practical view, square-root stage-damage functions pro- The creation, editing, and choice of these three functionsvide good results for damage estimation (Buck and Merkel, are realised by different masks that allow the user to con-1999). Therefore, a square-root function is implemented in veniently handle the input. For the Polygon and the Powerwww.nat-hazards-earth-syst-sci.net/6/485/2006/ Nat. Hazards Earth Syst. Sci., 6, 485–503, 2006
  12. 12. 496 B. B¨ chele et al.: Flood-risk mapping:...extreme events and associated risks uTable 2. Standardized damage to buildings and inundated areas for the test community. Exceedance Affected buildings [–] Inundated area [m 2 ] Standardized probability [%] residential total residential settlement damages [%] 10 4 8 4000 6000 0.002 5 14 40 15 000 29 000 0.01 2 71 126 41 000 71 000 53 1 127 242 81 000 129 000 100 0.50 223 378 143 000 220 000 194 0.20 237 403 156 000 237 000 215 0.10 266 447 174 000 267 000 255 0.05 296 496 192 000 298 000 304 0.02 328 540 203 000 326 000 364 0.01 342 571 207 000 341 000 400Fig. 8. GIS-based damage analysis tool (screenshot of graphical user interface). The damage or the general involvement can be calculatedby selecting the area of interest, land-use and event information and damage-relevant factors.function types, the damage can be calculated in absolute a range of events, beginning from the flood causing the firstmonetary units (EUR) or percentages of damage. damage up to the 1000-year event. The damage values in Fig. 9 are standardized to the 100-year event. That means for Before starting the calculation, a flood event must be se- example, that the damage caused by the 1000-year flood islected. According to the coupling of the damage estimation approximately 2.6 times higher than the one caused by thetool to the flood information system for the Neckar river, 100-year event.the outcomes of the hydraulic calculations, i.e. water sur-faces, can be directly used as input of the damage estimation. Furthermore, the tool includes functionalities to cope withThe implementation of the damage estimation tool in a GIS- cases where detailed land-use data (e.g. ALK) are not avail-software environment is realised in four dialogue modules able or where the assessment could be simplified. Asshown in Fig. 8. revealed in Table 2, it is possible to make assumptions Hence, the GIS-based damage estimation tool enables the e.g. about the number of affected houses in flood-prone ar-user to assess the flood damage to single buildings in flood- eas, in order to give an overview on flood risk without ex-prone areas and the spatial aggregation of the event-specific plicitly calculating monetary damage. For damage calcula-damage for a defined group of buildings or areas. Figure 9 tions on a more aggregated spatial level, the values at risk canshows the calculated damage values for a test community for be derived from statistical data for administrative districtsNat. Hazards Earth Syst. Sci., 6, 485–503, 2006 www.nat-hazards-earth-syst-sci.net/6/485/2006/
  13. 13. B. B¨ chele et al.: Flood-risk mapping:...extreme events and associated risks u 497and related to their spatial unit (EUR/m2 ). In this case, the 10.00damage estimation can be delivered by spatial intersectionof flood-hazard information (inundation zone) with land-use exceedence probability [%]data (e.g. ATKIS) in order to calculate the extension of the 1.00inundated settlement area (see columns 4 and 5 in Table 2). Usually, the flood-damage calculation is provided for cost-benefit analyses of flood protection measures. For this pur- 0.10pose, the costs of a flood-protection measure can be com-pared to its benefit, i.e. the avoided damage up to the designevent respectively the residual risk after the implementation 0.01of the measure, normally expressed as mean annual damage 0.0 1.0 2.0 3.0 4.0 5.0(MAD). For the above mentioned sample community, a dike standardized damage [-]designed for a 100-year event provides a significant reductionof MAD, but the residual risk due to a larger flood event still Fig. 9. Standardized damage to residential buildings in a test community for annual exceedance probabilities from 0.1 to 0.0001accounts for approximately 40 percent of the original value. (i.e. damage to a 100-year flood=1, situation without flood protec- The main advantage of the presented damage estimation tion measure).on a local scale is that the damage-determining factors aregiven both on the hazard side (being in general the waterdepth) as well as on the side of vulnerability (stage-damage a building-specific random sample of households was gener-functions for individual objects). For estimations on an ag- ated, and always the person with the best knowledge aboutgregated spatial level, where areas of the same building type the flood damage in a household was interviewed. An in-may be defined (e.g. ATKIS-data) but no information on in- terview comprised around 180 questions and lasted aboutdividual objects is available, one has to make assumptions on 30 min. The computer-aided telephone interviews were un-the spatial distribution of buildings and building types. Fur- dertaken by the SOKO-Institute, Bielefeld. Detailed descrip-thermore, as the water depth in an inundated area varies in tions of the survey were published by Kreibich et al. (2005)space, the definition of the damage-determining water level and Thieken et al. (2005).gets more uncertain with increasing spatial units. Thus, us- Statistical analysis was undertaken with the software SPSSing stage-damage functions, one has to define the damage- for Windows, version 11.5.1. and Matlab, version 7.0.1.relevant depth or use a statistical approach to estimate the Since a big share of the resulting data show skewed distribu-spatial distribution. tions, the mean and the median are given. Significant differ- ences between two independent groups of data were tested4.3 Development of a multifactorial approach for damage by the Mann-Whitney-U-Test, for three or more groups of estimation data the Kruskal-Wallis-H-Test was applied. For all tests a significance level of p<0.05 was taken.4.3.1 Damage data of the extreme flood in August 2002 4.3.2 Factors influencing the flood damageSince flood damage is also influenced by other factors be-sides the water depth, more knowledge about the connections Flood damage influencing factors can be divided into im-between actual flood losses and damage-determining factors pact factors like water depth, contamination, flood duration,is needed for the improvement of damage estimation. There- flow velocity and resistance factors like type of building, pre-fore, during April and May 2003 in a total of 1697 private ventive measures, preparedness, and warning (Thieken et al.,households along the Elbe River, the Danube River and their 2005).tributaries, people were interviewed about the flood damage During the extreme flood in August 2002, for exam-to their buildings and household contents caused by the Au- ple, contamination led to significantly higher damage ratiosgust 2002 flood as well as about flood characteristics, precau- (fraction of the flood damage in relation to the total value)tionary measures, warning time, socio-economic variables, to buildings and contents (Fig. 10). The damage ratio ofregional and use-specific factors. The 2002 flood was an ex- contents was increased by 93% for high contamination intreme event, e.g., with a discharge return period of 150–200 comparison with no contamination. For building damage ityears at the river Elbe in Dresden and with a return period was increased by more than 200%. During the 1999 floodof 200–300 years at the river Mulde in Erlln (IKSE, 2004). in Bavaria oil contamination on average led to a three timesDetailed descriptions were published by e.g. DKKV (2003), higher damage to buildings, in particular cases even to totalEngel (2004), IKSE (2004). The total damage in Germany is loss (Deutsche R¨ ck, 1999). uestimated to be 11.6 billion . The most affected federal state On the resistance side, private precautionary measures sig-was Saxony where the total flood damage is estimated to be nificantly reduce the flood loss even during an extreme flood8.6 billion (BMI, 2002; SSK, 2003). In the affected areas, like the one in 2002 (Fig. 11). The damage ratio of contentswww.nat-hazards-earth-syst-sci.net/6/485/2006/ Nat. Hazards Earth Syst. Sci., 6, 485–503, 2006

×