You are on page 1of 71

Integrated Flood Risk Analysis and Management Methodologies

Guidelines on Coastal Flood Hazard Mapping


Date Report Number
Revision Number Deliverable Number: Due date for deliverable: Actual submission date: Task Leader

June 2008 T03-08-02


3_0_P13 D03.1 February 2008 June 2008 UPC-LIM

FLOODsite is co-funded by the European Community Sixth Framework Programme for European Research and Technological Development (2002-2006) FLOODsite is an Integrated Project in the Global Change and Eco-systems Sub-Priority Start date March 2004, duration 5 Years Document Dissemination Level PU Public
PP RE CO Restricted to other programme participants (including the Commission Services) Restricted to a group specified by the consortium (including the Commission Services) Confidential, only for members of the consortium (including the Commission Services)

PU

Co-ordinator: Project Contract No: Project website:

HR Wallingford, UK GOCE-CT-2004-505420 www.floodsite.net

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

DOCUMENT INFORMATION
Title Lead Author Contributors Distribution Document Reference Guidelines on Coastal Flood Hazard Mapping Jos A. Jimnez Andreas Kortenhaus, Markus Anhalt, Christoph Panayotis Prinos, Wojciech Sulisz Public T03-08-02

Plogmeier,

DOCUMENT HISTORY
Date 21/06/08 24/06/08 24/06/08 25/09/08 23/11/08 Revision 1.0 1.1 1.2 2.0 2.1 Prepared by Paul Samuels Paul Samuels Organisation HR Wallingford HR Wallingford Approved by Notes Convert from .docx to .doc format Convert to correct deliverable template Contents updated including A. Kortenhaus comments Contents updated including uncertainty Jos A. Jimnez UPC

Jos A. Jimnez UPC Andreas Kortenhaus LWI

ACKNOWLEDGEMENT
The work described in this publication was supported by the European Communitys Sixth Framework Programme through the grant to the budget of the Integrated Project FLOODsite, Contract GOCE-CT2004-505420.

DISCLAIMER
This document reflects only the authors views and not those of the European Community. This work may rely on data from sources external to the FLOODsite project Consortium. Members of the Consortium do not accept liability for loss or damage suffered by any third party as a result of errors or inaccuracies in such data. The information in this document is provided as is and no guarantee or warranty is given that the information is fit for any particular purpose. The user thereof uses the information at its sole risk and neither the European Community nor any member of the FLOODsite Consortium is liable for any use that may be made of the information.

FLOODsite Consortium

T03_08_02_D03_1_V3_0_p13.doc ii

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

SUMMARY
This report reviews main technical aspects involved in Coastal Flood Hazard Mapping. It recommends methods and techniques to evaluate the different variables required to characterize the flood risk sources, pathways and receptors. They are suited for coastal sedimentary environments where to properly estimate the extension of coastal flooding, the coastal response has to be also calculated to assess the morphodynamic feedback between both two processes, erosion and inundation. Within this context, the report provides general guidelines on major technical aspects on data acquisition (requirements and techniques), data analysis, processes to be considered and how to evaluate them and, finally, type of flood hazard maps to be produced. Most of these guidelines have been implemented in a practical case of coastal flood hazard mapping in the Ebro delta. This case is included as an appendix at the end of the report with the emphasis being put on the estimation of the impact of including or not the coastal morphodynamic response to a 100 year return period storm on the extension of the flood hazard area.

T03_08_02_D03_1_V3_0_p13.doc iii

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

T03_08_02_D03_1_V3_0_p13.doc iv

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

CONTENTS
Document Information Document History Acknowledgement Disclaimer Summary Contents Tables Figures 1. ii ii ii ii iii v vi vii

Introduction ...................................................................................................................... 1 1.1 Scope of the report............................................................................................... 1 1.2 Flood Hazard Mapping in the European Flood Directive.................................... 2 1.3 Structure............................................................................................................... 3 1.4 Definitions ........................................................................................................... 3 Data requirements ............................................................................................................ 4 2.1 Introduction ......................................................................................................... 4 2.2 Risk Sources ........................................................................................................ 4 2.3 Risk Pathways...................................................................................................... 5 2.4 Risk Receptor ...................................................................................................... 8 Calculation methods....................................................................................................... 10 3.1 Introduction ....................................................................................................... 10 3.2 Risk sources....................................................................................................... 10 3.3 Risk pathways.................................................................................................... 12 3.4 Risk receptor...................................................................................................... 19 Mapping ......................................................................................................................... 21 4.1 Data Collection .................................................................................................. 21 4.2 Sources............................................................................................................... 22 4.3 Pathway Inundation ........................................................................................... 23 4.4 Map Content ...................................................................................................... 24 Summary ........................................................................................................................ 30 References ...................................................................................................................... 32

2.

3.

4.

5. 6.

Annex I. Run-up and overtopping formulas................................................................................ 37 Annex II. Coastal flood hazard mapping in the Ebro delta ......................................................... 41 Annex III Sources of Uncertainty ............................................................................................... 50

T03_08_02_D03_1_V3_0_p13.doc v

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Tables Table 2.1: Main characteristics of data requirements in coastal flooding analysis to define risk receptors and pathways. Table 2.2: Overview of different land use data types for flood hazard analysis 7 9

T03_08_02_D03_1_V3_0_p13.doc vi

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Figures Figure 1.1. Source-Pathway-Receptor model for Coastal Flood Hazard Mapping. 1 Figure 2.1: Domains for data acquisition in coastal flooding analysis. 6 Figure 3.1: Flow chart for statistical approach to be used in the response method (Fema, 2005). 11 Figure 3.2: Definition sketch for run-up calculations. 13 Figure 3.3: Qualitative hazard scale of coastal changes during storms as a function of water level 14 Figure 3.4: Morphological changes in the dune/beach morphology affecting coastal flooding. 15 Figure 3.5: Dune erosion profile 16 Figure 3.6: Geometric foredune erosion model 17 Figure 3.7: Delineation of the coastal area to be inundated for a given water level 20 Figure 4.1: Combination of geographical survey data and laser scan data 22 Figure 4.2: Input data for hydraulic calculation 23 Figure 4.3: Intersection of hydraulic results with the digital terrain model 23 Figure 4.4: Possible map contents of flood hazard maps 24 Figure 4.5: Delineation of flood hazard areas to a water level of 50 years return period 25 Figure 4.6: Example of maximum flood inundation depth caused by sea flooding 26 Figure 4.7: Example of Flood Hazard Map with a hazard scale 27 Figure 4.8: Example of Flood and Erosion Risk Maps 27 Figure 4.9: Example of digital flood hazard map generated through the Fema web application. 29 Figure 5.1: Sketch of the methodology to generate a coastal flood hazard map. 30 Figure 5.2: Changes in Flood hazard maps in the Ebro delta for a 100 year return period storm 31

T03_08_02_D03_1_V3_0_p13.doc vii

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

1. Introduction
1.1 Scope of the report

Flood Hazard mapping is a worldwide concern to all the nations with low-lying areas subjected to the impact of flood events. Flood management agencies usually have to build and maintain coastal and flood defences and associated infrastructures such as barriers and gates to protect the hinterland. Due to this, different programmes for identifying flood prone areas have been launched. They require welldefined procedures to produce hazard maps twhich can be used to identify protection needs and to make decisions. Some recent examples can be found in DEFRA (2001) for the UK and FEMA (2003) for the US. These programmes are usually based on a series of procedures that can be grouped into a sequence of activities such as: identifying mapping needs, generating flood data, establishing the extent of area affected and assets lying within those areas and its attached value. Due to the relatively broad spectrum of activities included and the level of detail that they require to be properly described, it might prove to be impossible to try to elaborate a single set of guidelines covering all the aspects related to Flood Hazard Mapping. Thus, this report will cover the most common aspects of flood hazard mapping in coastal areas according to the generic FLOODsite approach of source-pathway-receptor (figure 1.1).

Hazard receptor pathway source

damages tangible & intangible

coastal erosion lowering flattening

coastal storm extreme waves storm surge

Figure 1.1. Source-Pathway-Receptor model for Coastal Flood Hazard Mapping. The scope of this report is to provide general guidelines to Coastal Flood Hazard Mapping. These guidelines will cover major technical aspects such as data requirements, processes to be considered and type of maps to be produced. Although flood hazard mapping in coastal domains should cover any coastal type with any level of protection, these guidelines mainly mention coastal sedimentary environments without coastal protection structures, i.e. those highly dynamic environments sensitive to the impact of extreme events that will impulsively respond to the impact of the storm generating a morphodynamic feedback. In any case, these recommendations will also be valid for coastal areas protected by infrastructures such as dikes, since most of the procedures are also directly applicable and, the morphodynamic response is simply substituted by the probability of failure of the structure.
T03_08_02_D03_1_V3_0_p13.doc 1 26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

1.2

Flood Hazard Mapping in the European Flood Directive

Flood Hazard Mapping is an issue directly covered in the Directive 2007/60/EC of the European Parliament and of the Council of 23 October 2007 on the Assessment and Management of Flood Risks.Which, in Chapter III: Flood hazard maps and flood risk maps states that

1. Member States shall at the level of the river basin district, or unit of management, prepare flood hazard maps and flood risk maps, at the most appropriate scale for the areas identified under Article 5.1 [it refers to areas lying within their territory for which it is concluded that potential significant flood risks exist or might be considered likely to occur]. 3. The flood maps shall cover the geographical areas which could be flooded according to the following scenarios: (a) floods with a low probability, or extreme event scenarios; (b) floods with a medium probability (likely return period 100 years); (c) floods with a high probability, where appropriate. In addition to this, the Directive also states that Member States may decide that, for coastal areas where an adequate level of protection is in place, the preparation of flood hazard maps shall be limited to the scenario (a). For each scenario set out in the first subparagraph the following elements shall be shown: (a) the flood extent; (b) projected water depths; (c) where appropriate, the flow velocity or the relevant water flow. According to the Directive the main characteristics of the maps to be produced are: (1) Regarding probabilities, with the exception of the scenario (b) which is already given, the other two scenarios can be defined as a function of the objective of the study. In those cases in which the coastal zone is protected by dikes, the extreme probability to be employed in the flood hazard mapping should be derived from the safety level of the structures. (2) Regarding the variables to be mapped, an adaptation to the characteristics of coastal floods should result in coastal flood hazard maps including the extension of the area to be inundated and projected water depths. In addition to this, areas along the coast sensitive to the event to be mapped (i.e. showing a projected erosion significant enough to affect/enhance inundation of the hinterland) should be indicated. (3) Regarding the damages, the directive recommends to prepare maps including all the types of potential damages: inhabitants, economic and environmental affectations. These maps should be similar to those prepared for river basins although adapted to the specific characteristics of the coastal zone.

T03_08_02_D03_1_V3_0_p13.doc 2

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

1.3

Structure

This report is structured as follows: Chapter 2 introduces main data requirements and techniques to gather them to characterize risk sources, pathways and receptor in the coastal domain; Chapter 3 presents main calculation methods to evaluate variables required to characterize risk sources, pathways and receptor. This includes the estimation of the probability of occurrence of the event, the induced wave runup, overtopping discharges, the coastal response and the inundation of the floodplain. Chapter 4 includes different aspects regarding map production. Chapter 5 summarizes the main points of the recommendations. Chapter 6 lists the references cited throughout the text. Annex I includes a specific method to calculate wave runup and overtopping discharges. Annex II presents an example of application of the guidelines to Flood Hazard Mapping in the Ebro delta coast. Annex III gives a brief and general overview of the sources of uncertainties

1.4

Definitions

In what follows, the definitions of main terms used in this report are given. Flood: water temporary covering land normally not covered by water. This shall include floods from rivers, mountain torrents, Mediterranean ephemeral water courses, and floods from the sea in coastal areas, and may exclude floods from sewerage systems. Flood risk: the combination of the probability of a flood event and of the potential adverse consequences to human health, the environment and economic activity associated with a flood event. Flood plain maps indicate the geographical areas, which could be covered by a flood according to one or several probabilities: floods with a very low probability or extreme events scenarios; floods with a medium probability (likely return period >= 100y); floods with a high probability. Flood hazard maps are detailed flood plain maps complemented with: type of flood, the flood extent; water depths or water level, flow velocity or the relevant water flow direction. Flood risk maps indicate potential adverse consequences associated with floods with several probabilities, expressed in terms of: the indicative number of inhabitants potentially affected; type of economic activity of the area potentially affected; installation which might cause accidental pollution in case of flooding; other information which the Member State considers useful.

T03_08_02_D03_1_V3_0_p13.doc 3

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

2. Data requirements
2.1 Introduction
A critical issue in building flood hazard maps is the collection of necessary data on the different aspects involved in the process to be mapped. According to the general scheme followed in FLOODsite, this implies to acquire data to properly describe the risk-sources, risk-pathways and risk-receptor. In what follows a general guide, with respect to the requirements, for each type of data and recommendations for data collection is presented. In all the cases and as a general rule followed in these guidelines, the recommendations are given in terms of the best possible option.

2.2

Risk Sources

The main input data for creating coastal flood hazard maps is the occurrence probability of the total water level at the shoreline. This total water level, t, can be expressed as the sum of a series of components

in which as, is the astronomical tide level, ss, is the storm surge (meteorological tide), Ru is the waveinduced run-up (which also includes the wave set-up at the shoreline), LF is a component accounting for the contribution of low frequency forcing such as seiches and, LT, is a long-term component representing eustatic and/or local land-elevation (isostatic and/or subsidence) changes. This last component, LT, is usually given as a magnitude associated to a climate scenario based on projections of climate change (Solomon et al., 2007) plus a local component accounting for the relative vertical movement of the land due to processes such as subsidence. This component should be included in any long-term flood hazard analysis. A recent example of how to deal with this component in coastal flood risk analysis from a probabilistic standpoint can be found in Purvis et al (2008). The astronomical tidal level, as, is a deterministic process that can be easily estimated provided water level data do exist. This data consists of water level data recorded by tide gauges that are usually located in major harbours and waterways worldwide. The analysis of existing data permits to estimate the local tidal components of a given site which can be used to reconstruct the expected tidal curve at any moment. There exist different databases for tidal components for many locations worldwide that can be use to reconstruct the astronomical tide in a given site or, to feed tidal models to simulate the tidal propagation and to estimate the tidal wave in large coastal domains. This is the usual approach when no tidal gauges at the site of interest are available. Detailed descriptions of tidal data analysis can be found in Pugh (1987) and Godin (1991) among others. The meteorological tide or storm surge, ss, is also obtained from recorded water levels by tide gages and it is the residual water level after subtracting the astronomical tide from the recorded water level (see e.g. Pugh, 1987). This component integrates all the meteorological effects but waves, i.e. atmospheric pressure and wind set-up. If no water level data, to estimate this component directly, does exist, it can be calculated by using a storm-surge model which, fed by wind data, simulates the generation and propagation of storm surge for any coastal domain. An example of the use of this kind of models (including a wave generation model) can be seen in Cheung et al (2003). The long-frequency component, LT, can be obtained from recorded water levels by retaining the component associated to a range of frequencies longer than the ones associated to waves (25 sec) and shorter than the astronomical frequencies (see e.g. Pugh, 1987).

T03_08_02_D03_1_V3_0_p13.doc 4

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

The wave-induced runup, Ru, is usually not a measured variable but a calculated one. To calculate it, wave data and beach profile characteristics are required. In regard to wave data, most of the existing runup formulas require simultaneous information of wave height and period (see chapter 3). This information is usually obtained from wave records acquired by using coastal buoys. To calculate runup the required information of the beach morphology is the beach slope, although in extreme dissipative beaches, the runup is usually considered to be independent of the slope. This information should be obtained from topographic and bathymetric data acquired in the coastal fringe (see section 2.3). One of the most important aspects to be considered is the duration of the period covered by these data sets. As they are going to be used to define water levels associated to long return periods (see section 3.4), very long time series are required. In Snchez-Arcilla and Gonzlez-Marco (2007) different aspects on quality and length of time series required to make a reliable estimation of extremes are reviewed. If a long time series of this information is not available elsewhere, it can be replaced by simulated data. This data consists of time series of water level and wave conditions generated by wave and surge generation models fed by meteorological time series (wind and atmospheric pressure fields). For the European coast there exist different data sets of hindcasted conditions for periods longer than 40 years, which is significantly longer than most of the existing recorded data sets. Examples of these hindcasts are the obtained data sets within the Wasa and the Hipocas projects (Wasa, 1998; Guedes Soares et al., 2002).

2.3

Risk Pathways

Data required for risk pathways make reference to those used to characterize the intensity of the processes taking place, i.e. coastal erosion, overtopping and flooding. Data required for defining risk receptors make reference to those used to characterise the consequences of the inundation. To define both risk pathways and receptor, three main types of data are required: topographic, bathymetric and land-use data. Topographic data of the coastal zone will cover all the data to be gathered for the analysis. Due to the strong spatial gradients in dynamic conditions across the coastal zone, they can be separated in two major domains: (i) the floodplain and the (ii) coastal fringe (figure 2.1 and table 2.1). Topographic data of the floodplain are needed to define the relief of the land to be inundated. The accuracy of these data will control the extension of the flood prone area for a given water level scenario and, in this sense, they should control the accuracy of the flood hazard map. The general way to provide these data is through a digital terrain mode, DTM, or digital elevation model, DEM, which is a mathematical 3D representation built from a data set composed by a collection of points of known elevation and spatial co-ordinates. The presently most common way to obtain these topographic data in an efficient manner, in terms of accuracy and cost, is the LIDAR (LIght Detection and Ranging). Eessentially it is a remote sensing technology that derives the elevation of the terrain by measuring the time delay between the transmission of a laser pulse and the detection of the reflected signal from a plane. This technology permits to obtain a high density of points per scanned surface, although it has to be considered that not always more is better. Thus, point spacing (distance between points) and density (coverage of points within the area) are critical considerations in LIDAR mapping. Their optimum selections depend on the desired vertical accuracy, and type of terrain (slope) and land cover. Their number will control the cost and time of acquisition time and data processing (see e.g. Anderson et al., 2005; Liu et al., 2007). Once these data are obtained, they have to be filtered in order to remove all the noise in the signal which is not corresponding to the real terrain topography. The whole process can efficiently be
T03_08_02_D03_1_V3_0_p13.doc 5 26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

done by the service (Lidar) supplier but it is important to know all about the process involved and, also to be sure about the accuracy obtained. With the so measured data, a DTM or DEM has to be derived. The accuracy of this DEM will determine the accuracy of the flood mapping (see e.g. Haile and Rientjes, 2005). The required DEM resolution to assure proper map accuracy will depend on the type of terrain to be analyzed. However, a reasonable DEM resolution should be of the order of 5 m cell (for raster DEMs). Because the size of cells will also determine the (computational) costs of inundation models, they can be later re-sampled to larger cells provided the relief is simple enough to do not soften the real topography nor mask existing canal networks. In any case, before selecting the definitive cell size to be used in the analysis, it is recommended to test different options with the selected inundation model and to analyze the differences. Because this domain is usually of very low intensity dynamics, its updating needs are relatively small (figure 2.1 and table 2.1). Thus, it is not expected to significantly change in its relief unless a high energetic event should take place (e.g. a flood) or, the human influence should produce a major footprint such as building of infrastructures or artificial changes of the topography that should affect floodwaters paths in the area. Once an accurate DEM is available, it is recommended to update the information, and later the DEM, of only those areas subjected to any change of any origin. Depending on the scale of these modifications, the proper acquisition method should be chosen (conventional surveying techniques such as total stations, DGPS or Lidar). The new data should be used to update the DEM in the corresponding area whereas the rest can remain as it was.

floodplain low variability topography updating small requirements

coastal fringe high variability topo-bathymetry updating large requirements

nearshore bathymetry low variability bathymetry updating small requirements

Figure 2.1: Domains for data acquisition in coastal flooding analysis.

T03_08_02_D03_1_V3_0_p13.doc 6

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Table 2.1: Main characteristics of data requirements in coastal flooding analysis to define risk receptors and pathways. Domain Floodplain Coastal fringe Nearshore Type Topograhy Top-bathym Bathymetry Response-time Slow Rapid Slow Updating needs Low High-Medium Low Update recommended Y years *1 Six months X Years *2

Topographic data of the coastal fringe are needed to characterize the morphology of the border between the sea and the land. This relatively narrow zone includes a subaerial (topography) part and a subaqueous one (bathymetry). It will act on the one hand as a barrier for the flooding, i.e. the main element of protection for the hinterland and, on the other hand, it will control the intensity of the flood by modifying wave and surge propagation. In addition to this, the flooding intensity can be affected by the changes in the morphology of the beach (e.g. run-up modification due to changes in beach slope, overtopping variation due to changes in dune and beach crest changes, etc.). Topographic data in this domain can be easily acquired when the topography of the floodplain is being gathered and, then, the same methodology as recommended before should be used, i.e. Lidar. Moreover, due to the characteristics of this zone in which the terrain is usually free of obstacles masking the signal (forests, buildings, etc.), this measuring system is very efficient in this area and, the application of filters to the recorded signal is not required as frequent as in the previous case. One of the main differences with respect to the floodplain is that the morphology of the coastal fringe is highly variable because it is continuously changing as a response to the littoral dynamics, especially in the case of coastal sedimentary environments. Due to this, one of the main problems in coastal flooding analysis is to have a reliable coastal fringe morphology to represent the pre-storm conditions available. Thus, in a ideal the coastal fringe morphology is continuously updated at a reasonable frequency (table 2.1 and figure 2.1). This reasonable frequency depends on the intensity of the local littoral dynamics, but assuming that most of sedimentary coasts will show a seasonal behaviour we can fit the ideal update scheme in two configurations per year. For those coastal environments protected by rigid structures such as dikes or seawalls or for rocky coasts, the floodplain update requirements should be applicable, i.e. data updating when significant changes are detected. This information update can be done by using the same methodology, i.e. Lidar, or alternative techniques that provide enough resolution. This depends on the extension of the area to be covered and the type of morphology (dunes, low gradient beaches, etc..). An alternative technique to Lidar to efficiently monitor coastal fringe topography in large areas is real time kinematic GPS (RTK-GPS) which allows sampling of the surface along a given path with the selected density and a 15cm vertical accuracy (e.g. Morton et al., 1999). Bathymetric data of the coastal fringe are needed to characterize the underwater morphology of the border between the sea and the land. This is required to describe the coastal processes taking place in this area and it will affect the magnitude of the flooding. As it was mentioned before, this domain is a highly dynamic environment resulting in a domain with bathymetric changes taking place continuously (figure 2.1 and table 2.1). From this, it could concluded that the characterisation of the bathymetry of this area is of relatively small use if it is not conveniently (frequently) updated. And unless an efficient system (in terms of costs and time) is available, this will not be the usual case. The ideal situation should be an area of analysis formed by relatively clear waters. Under these conditions, the gathering of bathymetric data can be combined with topographic data by using the SHOALS (Scanning Hydrographic Operational Airborne Lidar Survey) system. This version of Lidar is able to measure the bathymetry in shallow waters with accuracies up to 15 cm (Lillycrop et al.,
T03_08_02_D03_1_V3_0_p13.doc 7 26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

1996; Guenther et al. 2000). However, in areas with turbid waters the maximum surveyable depth could be very small and, even worse, the obtained accuracy should be unacceptable. Maximum surveyable depths range from around 50 meters in very clean offshore waters to less than 10 meters in murky near-shore waters. For extremely turbid conditions, surveying may not be possible (Guenther et al, 2000). Under ideal conditions, these bathymetric data should be collected simultaneously with the topographic ones and at the same frequency. In those cases where Shoals cannot be used, an alternative technique should be employed. The traditional way is to make bathymetric surveys with echo-sounders as in the nearshore zone although with the additional problem of the existence of very shallow areas where ships cannot easily operate. Due to this and, to keep costs within reasonable limits, representative transects along the study area could be selected. These transects should be monitored at a frequency reflecting the natural seasonal variability as it was the case of the topographic data in this domain. Thus, a six-month frequency should be a reasonable choice. Bathymetric data of the nearshore zone are needed to characterize the underwater morphology of this area where waves and surge propagate towards the coast. The presently most commonly used technology is the Multi-Beam Echosounder Surveying (MBES), which has demonstrated high quality with respect to meeting the IHO standards on depth accuracy (IHO, 1998).

Because the bottom evolution in this zone (significant enough to modify wave and surge propagation) is, usually, very slow, the requirements for data updating are low. Thus, once the bathymetry of the nearshore area is available, it is not expected to require major updates in periods shorter than decades. Of course, this will depend on the intensity of the processes taking place in the area. Thus major changes due to human influence such as dredging will affect wave propagation and, in consequence, will make these updates necessary.

2.4

Risk Receptor

Finally, to fulfil the requirements of the EC Directive, it is necessary to obtain data to prepare flood damage maps. For this purpose, the main type of information to be included is the land-use of the floodplain. This will include all the relevant categories for economic and environmental damage analysis. In addition to this, information on the spatial distribution of the population or, in its absence, urbanization types should also be compiled. This data will also be relevant to gain information on the expected flood intensity such as roughness and permeability of the land. Moreover, in this chapter, major infrastructures must be localized and , any obstacle to flood propagation should be identified to be included in the DEM. The main source of information are data gathered by field surveys, data obtained from remote sensing techniques such as aerial photographs, satellite and/or airborne mounted multi-spectral sensors as well as pre-existing data sources such as the Corine Land Cover (although only valid for large scale analysis). A detailed review on land use data required for flood hazard analysis can be seen in Messner et al (2007) (see e.g. table 2.2).

T03_08_02_D03_1_V3_0_p13.doc 8

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Table 2.2: Overview of different land use data types for flood hazard analysis (Messner et al., 2007).

T03_08_02_D03_1_V3_0_p13.doc 9

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

3. Calculation methods
3.1 Introduction

In this chapter, practical recommendations on how to calculate the different aspects required for coastal flood hazard mapping are given. This involves selecting the methods to be applied to characterize (i) risk sources, (ii) risk pathways and (iii) risk receptors.

3.2

Risk sources

3.2.1 Estimation of the probability of occurrence of a given water level


There are two main ways to calculate the probability distribution of the total water level for coastal flooding analysis: to directly estimate it from existing time series of water levels; to estimate it by analysing the integrated contribution of each component.

In the first case, the procedure to follow is to analyse existing water level time series to obtain its extreme distribution and, to define the probabilities (or return periods) for the water levels required. How extreme distributions can be used in fitting water levels can be seen in Sobey (2005) and Pirazzoli and Tomasin (2007) among others (see review in Snchez-Arcilla and Gonzlez-Marco, 2007). The main problem with following this approach is that water level records usually do not include wave-induced contributions. Thus, its use should only quantify the water level components associated to astronomical and meteorological tides. In the second case, the contribution of each component has to be estimated and the joint probability has to be calculated. Here two main approaches exist: (i) response and (ii) event approaches (see e.g. Fema, 2005; Divoky and McDougal, 2006; Garrity et al, 2006). The event approach is a deterministic approach. It uses one or more combinations of water level and wave conditions (events) associated to a given probability and it computes the resulting flood level (response). Due to the nature of the problem, one or more combinations of water level and wave conditions (events) will result in a certain flood level (response), and as a consequence, to properly assign a probability to such variable, it is necessary to jointly consider all the possible options. Thus, it should not be realistic to work with marginal probability distributions (H, T, ) since the final water level will depend on how waterlevel and wave conditions are combined. A recent paper of Callaghan et al. (2008) proposes a methodology to derive joint probabilities of the main involved variables in extreme wave climate to be used in beach erosion studies and, consequently, also potentially useful for flood analysis. In Snchez-Arcilla and Gonzalez-Marco (2007) different methods to obtain marginal and joint extreme distributions of wave and water level can be seen. The main problem is that, in many cases, different combinations of events with a given probability will not generally result in a response of a different probability. Moreover, the flood associated to the given probability could be produced by many combinations of conditions. On the other hand, in the response approach the water level of interest (associated to a given probability or return period) is directly calculated from a probability distribution of total water levels. The response method is based directly on measured or simulated water levels and waves as they occurred in nature and, the water level of interest (associated to a given probability or return period) is directly calculated from a probability distribution of total water levels. It is especially recommendable
T03_08_02_D03_1_V3_0_p13.doc 10 26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

when the variables (events) determining the flood level (response) are partially or poorly correlated, i.e. when surge and large waves are uncoupled and, for areas in which wave height and periods during storms (both will determine the wave run-up) are poorly correlated. At present, this approach is recommended by Fema guidelines for flooding studies (Divoky and McDougal, 2006). If simultaneous water level and wave records do exist, the best option is to estimate the induced run-up for each wave condition and to add it to the instantaneous water level. This will produce a time series of total water levels at the shoreline which can be fitted to an extreme distribution. From that distribution the water level associated to any probability or return period can be estimated. Because run-up depends on the beach slope, the calculations have to be done for representative beaches in the area to cover the different slopes. Moreover, the run-up formula to be used will depend on the morphology of the coastal fringe (beaches, dikes, etc.) (see next section). The main restriction to the application of this method is that long-time series of simultaneous measured relevant variables (water level, waves) are required to calculate reliable estimates of total water level associated to low probabilities (e.g. 100 years or longer). When these simultaneous time series do not exist or they are too short to be used to estimate a reliable extreme distribution, an alternative technique has to be used. In these conditions, the measurements have to be substituted by simulations. The idea is to simulate all the conditions prevailing on the site and from the simulated combinations to obtain the desired probability distribution (see e.g. Figure 3.1).

Figure 3.1: Flow chart for statistical approach to be used in the response method (Fema, 2005).

T03_08_02_D03_1_V3_0_p13.doc 11

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

An additional element to be taken into account is that, to fully define this transient wave-induced plus storm surge flood event we have to assign a duration to it. This means that is not enough to define a water level associated to a given probability but we have also to give information about the duration of such event. This variable is very important since it will control the amount of water to flow landwards and, in consequence, will control the extension of the affected area as well as the water depth in that area. Thus, once the events are identified to be subjected to extreme distribution fitting, their duration must also be recorded to obtain the relationship between a given level and the duration of the event. This can be done in probabilistic terms (by estimating the joint probability of water level-duration) or in deterministic ones (by obtaining a relationship between event duration for given water levels).
When the analysis is performed at a coast able to dynamically react to the impact of the storm, i.e. to be eroded, the most straightforward approach should be the event one. For this purpose, joint probability distributions of wave and water level conditions should be used to define the event assigned to a given probability or return period, which will be used to calculate the risk pathways. This is due to the fact that erosion and inundation are not necessarily correlated and, in consequence, hydrodynamic conditions resulting in a water level of a given probability and an erosion of the same probability should be different. If the analysis is applied to a coast where one of the two processes (erosion and/or inundation) clearly dominates, the recommended approach should be the response one. For this purpose, joint probability distributions of wave and water level conditions should be used to built a probability distribution of the target variable (e.g. total water level by estimating the contribution of the different components), from which the value associated to a given return period is directly obtained. This should be the usual approach for static flooding analysis, i.e. when the coastal response is not included.

3.3

Risk pathways

In this section, different methods to calculate the different variables required to characterize the risk pathways in a coastal flooding analysis are presented. They cover variables related to the quantification of the two main coastal processes taking place during the impact of a storm in the coast, inundation and erosion.

3.3.1 Runup estimation


The wave induced runup can be simply defined as the height with respect to the still water level reached by the uprush of wave action (figure 3.2). As it was previously mentioned this is a variable that it is usually calculated from wave data and, consequently, its accuracy will be strongly dependent on the model used. There exist numerous models for estimating wave runup depending on the characteristics of the coastal fringe on which the waves impact. For runup estimations for coastal structures, a detailed review of existing formulas for the main coastal structures typologies together their range of application can be seen in Burcharth and Hughes (2003). For analysis of areas protected by coastal structures, the formula recommended for the corresponding structure typology should be used. For wave runup estimations for beaches, there also exist different options in the literature. Here we recommend the use of the recently proposed by Stockdon et al. (2006). This recommendation is due to the fact that this formula has been derived from reanalysis of run-up data obtained in field experiments and large scale laboratories (see Annex I).

T03_08_02_D03_1_V3_0_p13.doc 12

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

R2%

SWL Hs0

R'2%

Figure 3.2: Definition sketch for run-up calculations.

Once this wave run-up has been estimated for a typical profile of the area of analysis, its value can be modified to take into account local factors affecting its magnitude. A typical expression that is based on these factors is the modified runup, R2%, as a function of the original value, R2%, multiplied by a series of correction factors as

where b r p
influence factor for beach slope influence factor for beach roughness influence factor for beach percolation, permeability influence factor for wave obliquity

The coefficient to account for the existence of a back-beach slope, b, milder than the one used to characterize the beach is given by / with a limiting value of 1 which should correspond to an uniform slope and a minimum value of 0.6 (van der Meer and Jansen, 1995). The coefficient to account for the surface roughness of the slope, r, depends on the material composing such surface. In natural beaches, the typical value for sandy surfaces is 1. In Eurotop (2007) different characteristic values for elements composing the surface of coastal structures are given. The coefficient to account for the obliquity of the incident waves, , is generally used in run-up estimations for dikes to account the effects of wave directions departing from the normal incidence. However, runup formulas derived from prototype experiment at beaches should include not only normal wave incidence but also oblique one. Taking this into account as well as the recommendation done by Eurotop (2007) to relate this effect only to angles larger than 20 degrees; we could assume that the corresponding value for this coefficient in beaches is 1. Finally, the coefficient to account for the percolation of water through its path along the emerged surface of the beach, p, depends on the permeability/porosity of the material composing such a surface. This factor will be calculated as 1-p, with p being the factor of percolation.
T03_08_02_D03_1_V3_0_p13.doc 13 26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

3.3.2 Wave overtopping discharge


Wave overtopping will occur when the beach/dune crest height is lower than the calculated potential run-up. Under these conditions, waves will reach and pass over the crest of the beach/dune and will flow into the hinterland. Thus, wave overtopping can be defined as the mean discharge of water per linear meter of width of the beach, q, flowing landwards. One of the problems when calculating overtopping discharges in beaches is that most of the existing formulas have been derived for coastal structures such as dikes and seawalls (see recent review in Eurotop, 2007), so their application to beaches can only be used as a approximation of the potential floodwater discharges occurring during a given event since these formulas are usually empiricallyderived. In Annex I, details to estimate wave overtopping discharges for coastal flood analysis are given.

3.3.3 Coastal response


Once the storm impacts on the coast, two situations can occur: (i) the coast is rigid (e.g. protected by coastal structures) and will be inundated if total water level exceeds the crest of the structure or the structure fails and (ii) the coast is dynamic and reacts to the impact of the coast by being eroded. In the second case the inundation will not only be controlled by the initial beach/dune height but by its evolution during the event. Due to this, risk pathways in the case of coastal flooding analysis must include not only the inundation but the induced coastal changes. The impact of extreme storms on sedimentary coast causes different morphodynamic processes and responses that can significantly affect coastal flooding. Their intensity correlates with the intensity of the storm. As an example, Figure 3.3 shows a qualitative hazard scale of coastal changes during storm impacts on barriers (applicable to any low-lying coast). It serves to illustrate the different regimes of functioning as a function of the water level. These regimes have been formalized in a conceptual model by Sallenger (2000) in: swash, collision-dune erosion, overwash and inundation regimes.

Figure 3.3: Qualitative hazard scale of coastal changes during storms as a function of water level (USGS, 2001). Figure 3.4 illustrates the potential changes, suffered by a dune under the impact of a storm, which are capable of influencing the magnitude of the inundation. Iin this case the dune is lowered by overwash and, the final representative beach slope becomes milder than at the initial stage of the storm. In this scenario, there are two potential sources of change for wave overtopping during the storm, which contribute in opposite ways. The decrease in the dune/beach height, D, during the storm will produce a progressive increase in the freeboard (Ru-D) and, in consequence, will induce larger overtopping. On the other hand, the decrease in the beach slope during the erosion process will induce a lower run-up and, if this factor is solely
T03_08_02_D03_1_V3_0_p13.doc 14 26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

considered should induce a decrease in the freeboard and would induce a lower overtopping. The resulting effect of the beach/dune changes during the event on the volume of water flowing landwards with respect to the initial beach configuration will depend on which will become the dominant factor.
DHIGH = 3 m

(composite/representative slope) tan = 1/10

initial dune/ barrier configuration

DHIGH = 2.5 m (composite/representative slope) tan = 1/15

erosion type I: dune/ barrier lowering and flatening

Figure 3.4: Morphological changes in the dune/beach morphology affecting coastal flooding. The prediction of the evolution of the coastal fringe under the impact of the storm is a question far from trivial and, there is not a single morphodynamic model capable of simulating all the possible conditions at any coast. A discussion on different options for models to be used for coastal flood forecasting is given by HR Wallingford (2003). Although they are not specificly for morphodynamic modelling the given recommendations cover most of the aspects to be considered. There exist different models to simulate the response of beaches and dunes to the impact of extreme storms. Generally speaking, the best model is that model that has been calibrated and/or developed for a specific coastal stretch. In this sense, the most general recommendation regarding this aspect is to work with calibrated and validated morphodynamic models. The most complicated model is not necessarily the best. In many cases they are difficult to calibrate and/or validate, although they give insight into the processes taking place during the storm, One of the most common analytical models for the estimation of the erosion induced by the impact of a storm on a beach profile is that introduced by Vellinga (1986). It is an empirical model based on a series of dune erosion laboratory experiments. It uses the concept of the equilibrium profile, predicting the final erosion profile corresponding to a set of hydrodynamic forcing (defined in terms of wave height and storm surge-water level) acting on an initial beach profile (characterised by the sediment grain size). The final profile is independent of the storm duration and it is given by the equation

T03_08_02_D03_1_V3_0_p13.doc 15

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

with Hs0 being the significant wave height during the storm, w is the fall velocity of the sediment, y is the vertical co-ordinate (elevation) of the profile with respect to the mean water level during the storm and, x is the horizontal co-ordinate (distance to the coast). The method constructs a storm profile and, by horizontally shifting the storm profile over the pre-storm profile until erosion and accretion areas are equal, determines dune erosion (Figure 3.5). This model has been widely used even in sites with completly different characteristics (in terms of coastal morphology and wave and water level climate) to the ones of the Dutch coast where this empirical model was developed.

Figure 3.5: Dune erosion profile (Vellinga, 1986). Recently, van Gent et al (2007) has proposed a modification of the equation for erosion profile by including the wave period (Tp), to take into account the effect of this parameter in the final dune profile configuration, in such a way that the longer the wave period is, the larger the profile retreat will be.

Another simple geometric model to predict the erosion (retreat) of beach foredunes induced by wave runup during extreme storm events has been developed by Komar et al (1999, 2001). The model is similar to the Bruuns rule but applied to a shorter time scale in which the total water level during the storm is the driving force. The dune retreat, Sdune, is given by the relationship (Figure 3.6)

Sdune =

(WL Hj ) + BL tan

with Sdune being the maximum beach or dune recession due to the total water level, WL, associated with an extreme storm (which is a combination of the mean water level and the wave runup); tan is the average slope of the beach face (constant during the event), Hj is the elevation of the dune toe or beach/dune junction; and BL is the vertical shift in the beach profile that results from the presence of a rip current (considered as a safety factor since it will imply an additional sand loss for the profile). This model was developed for the Oregon coast (Pacific) and it is used by the Oregon Department of Land Conservation and Development within a DSS tool to help coastal managers to identify coastal erosion-related risks.

T03_08_02_D03_1_V3_0_p13.doc 16

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Figure 3.6: Geometric foredune erosion model of Komar et al (1999, 2001). As in the previous case, this model is an equilibrium model in the sense that it should predict the maximum erosion induced by a given event without taking into account its duration (storm of unlimited duration). When this type of models is used, the information to be obtained is the final dune/profile configuration after the impact of the storm. Under these conditions no intermediate updates on floodwaters can be obtained. The only possibility is to calculate run-up and overtopping discharges for the initial and final configurations and to estimate their range of variation due to beach evolution. This problem is common to any model providing the final configuration without information of the evolution of the process. Recently, Larson et al. (2004) presented an analytical model to simulate the impacts of storms on dunes. The model quantifies the erosion in terms of recession distance and eroded volume. It uses a transport relationship based on the wave impact theory, where dune erosion is induced by individual swash waves impacting the duneface. The model was validated using different datasets obtained in the laboratory and the field. They conclude that the model produce reliable quantitative estimates of storm-induced dune erosion (retreat and volume loss), provided that the forcing conditions are known and that the geometry of the dune configuration is similar to the one assumed in the model (plane-sloping foreshore backed by a vertical dune). Finally, authors recommended applying the model using a range of transport coefficient values (calibration parameter) to include some uncertainty estimate in the calculated variables. Numerical beach/dune erosion models are more versatile than analytical ones because there are no limitations to describing the initial beach profile and, in most of the cases, are able to include most of the variables characterising the forcing in a realistic manner. The major limitation will be the reliability of the sediment transport model included to simulate the process (see e.g. Schoones and Theron, 1995) and, as it was mentioned before, they need to be calibrated/validated for the area of application. Among the existing beach erosion models, the Sbeach model (Larson and Kraus, 1989, Wise et al. 1994) is the most common. It is an empirically based model that was originally developed using a large data set of cross-shore sand transport rates and geomorphic change obtained in large wave tanks. It includes a module for wave propagation across the beach profile which is used to estimate the crossshore transport rates in different zones from outside the surf zone to the swash zone. These transport rates are included in the conservation equation to estimate beach profile changes. The model requires an initial beach profile and the sediment grain size to characterize the receptor and, the wave height, period furthermore it requires the direction time series and the water level time series to characterize the forcing. The model has been largely validated via field cases and now it can be considered as a
T03_08_02_D03_1_V3_0_p13.doc 17 26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

standard for beach/dune erosion calculations. In fact, it has been used to start beach erosion by overwash to be subsequently complemented with more sophisticated models such as Delft3D (see. e.g. Caizares and Irish, 2008). Due to its simplicity and its robustness, this model is recommendable for estimating the evolution of the beach during the impact of storms. It has to be noted that although coefficients to be used in the model have been calibrated in numerous applications and, default values have been recommended, results of its application without an ad-hoc calibration must be interpreted as an order of magnitude of the expected coastal response. As storm magnitude increases, the dune and the beach are more frequently overtopped and overwash transport begins to be important. However, there are few models capable of dealing with overwash processes in a realistic manner. One of the last attempts to include the transport in a beach profile model has been made by Larson et al (2004) with the before mentioned analytical model for dune erosion. It also has to be mentioned that the last version of Sbeach simulates the effect of overwash transport on the beach profile evolution. However, this is a subject under development and a realistic model is still required (e.g. Donelly, 2008). During very intense storms, the final coastal response, especially in low-lying areas, is breaching. A breach is a new opening in a narrow landmass such as a barrier beach that allows water to flow from the sea to the area behind the barrier -a water body or a low-lying area- (Kraus and Wamsley, 2003). They usually occur during extreme events and, in consequence, they are associated to the inundation stage in the scale of Sallenger (2000). In spite of the potential importance of this process, the number of existing models able to simulate breaching is quite limited (models dealing with dike breaching have been reviewed elsewhere, see e.g. Oumeraci et al., 2005; Allsop et al, 2007). One of the first serious attempts to model the breaching process in sandy barriers was made by Basco and Shin (1999). They developed a one-dimensional model for simulating storm-induced breaching in coastal barriers. The model considers the different (four) phases-stages of the process starting by the beach/dune erosion which is simulated by means of the Sbeach model (Larson and Kraus, 1989). This is followed by a series of hydrodynamic models used to simulate the water flow over the barrier and induced sediment transport. Roelvink et al (2003) assessed the feasibility of using a general morphodynamic model such as Delft3D to model barrier island breaching and storm surge. To do that, they first validated the model against laboratory and prototype experiments of breaching of a sand dike. They concluded that the model was able to predict the occurrence of breaches with a satisfactory level of accuracy. They proposed to use the model to identify weak spots in barrier islands and to identify conditions and locations where a breakthrough might occur. Another process-modeling approach is due to Tuan et al (2006, 2008) who presented a process-based model to simulate the overflow-induced growth of an erosional channel in noncohesive bodies such as sand-dikes and barriers. It includes an overwash model for barriers which makes use of the UnibestTC approach (Tuan et al., 2006) and a model for the lateral growth and morphological development of erosional channels (Tuan et al., 2008). In spite of all of these attempts, there is not yet a morphodynamic model able to accurately simulate the dynamic response of low-lying coast to the impact of extreme events. Due to this, the USACEERDC has launched a program to develop an open-source program to simulate effects of hurricanes on low-lying sandy coasts. The aim of this program, XBeach (Roelvink et al., 2007), is to simulate the different phases taking place during the impact of extreme storms in low-lying coasts, i.e. dune erosion, overwashing and breaching. A first version which has been tested against some laboratory experiments is already available. The main emphasis has been put up to date in numerical robustness and first order accuracy.

T03_08_02_D03_1_V3_0_p13.doc 18

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Once the beach/dune response during the event has been calculated, the run-up level and overtopping discharges will be updated. This can be done by estimating the respective variable for intermediate beach configurations as supplied by the model. This uncoupled way of updating floodwaters can be superseded when using a model able to properly take into account the modifications of such processes. An overview of different morphodynamic models to predict changes in river, coasts and estuaries useful for flood hazard analysis can be seen in Reeve (2007).

3.4

Risk receptor

3.4.1 Inundation
Once the design water level, the associated overtopping discharge and related probabilities of exceedance have been determined, the next step is to determine the extent of the area to be affected by inundation. In some cases, depending on the estimated water level, the coastal area can be subjected to a full inundation in with reformed broken waves able to propagate overland. On these occasions, it should is to estimate the wave propagation to correctly delineate the inland extension of the inundation. For these situations, a wave propagation model capable of simulating the wave breaking and reforming in very shallow waters with artificial roughness (due to vegetation, buildings, etc.) should be used. Examples of such models are the simple approximation followed by Fema (2003) with the Whafis model or, more process-oriented ones, such as the one by Mller et al (1999). Again the selection of the proper model will depend on the characteristics of the study area and on the importance of the hazard analysis. However, before estimating any wave propagation overland, the first estimation to be done is the determination of the surface to be flooded. For this purpose, flood inundation models combined with Digital Terrain Models are used to describe the processes taking place in the flood plain. As in the previous calculation methods, there exist different options to estimate the magnitudes of the inundation. The simplest option is the use of the so called empirical models, which are often described as pure mapping. No physical laws are involved in the simulations performed. They are rather simple methods and of low cost, but they provide only poor estimates of flood hazard in large low lying or extensive areas where flows through a breach may be critical in determining the flood extent. They are usually applied to assess flood extents and flood depths on a broad scale. In essence they are just GIS routines to delineate areas to be inundated applied to a DTM. This can be done in different ways: (i) determining the extension of the coastal domain and its height below the target water level or (ii) distributing the estimated water volume entering the hinterland depending on the internal topography. The first case should be equivalent to assuming a situation in which the target water level is maintained for a certain time period long enough to supply the required water volume to fill the entire area. This option should be acceptable when analyzing the flooding associated with the relative sea level rise (RSLR)-induced water level scenarios and/or when the shoreward extension of the area to be inundated is relatively small (figure 3.7). The second case should be applicable in situations when the beach topography is not as simple and the event involves the overtopping of the beach/dune crest with a flow of water entering to a domain with a lower elevation than the beach crest. The application of the before mentioned method could result in projected inundated areas with a very unrealistic dimension landwards. In this case, the approach

T03_08_02_D03_1_V3_0_p13.doc 19

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

should be to estimate the total volume of water overtopping the beach along the coast and, to distribute it in the floodplain according to the existing topography.
Limit of the flood hazard area

SWL

Target total water level

Figure 3.7: Delineation of the coastal area to be inundated for a given water level for the simplest case. However, in this second situation, i.e. time varying water levels pumping water towards the hinterland, the best way to determine the flood extension should be the use of inundation models. As in all the cases involving a model selection, there is no general rule for accepting the advantage of the use of one specific model over a different one other than an ad-hoc analysis of the case study. The best model appears to be one including the processes and conditions of the specific case, for which the required data is available and which has been calibrated/validated. In Woodhead (2007) and Prinos (2007) different available inundation models are presented together their range of validity.

T03_08_02_D03_1_V3_0_p13.doc 20

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

4. Mapping
Once the properties of the flood in the study area have been determined, the final step is to produce the Flood Hazard map. This map should provide information on the flood conditions that harm people during a flood. It can include the hazards associated with a single event or a combination of events. Flood mapping in river domains is a well known task, which can be easily adapted to the coastal domain when the boundary conditions are known. The workflow described below is based on the workflow, which is used in Baden-Wrttemberg, but can be seen as a standard workflow for producing flood hazard maps in river domains (Ministry of the Environment Baden-Wrttemberg, 2005).

4.1

Data Collection

The first step of building flood hazard maps should be the collection of necessary data. The general way to provide all required data is to build a digital terrain model. A digital terrain model is usually built from a data collection that contains laser scan data as well as geographical survey data. Laser scan data are collected by flying a plane over the territories which are at risk to be flooded. The plane contains a rotating laser that scans the land at right angles to the direction of flight. The scan should be made during the winter month, when vegetation is sparse. Afterwards, the scanned data have to be processed with the assistance of, for example orthophotos, in order to eliminate vegetation or manmade objects like cars, bridges or buildings. This is used to calculate regular, comprehensive dot grids with grid spacing of for example 1 m or 5 m. Any gaps in the terrain point data set (e.g. due to buildings) are filled in using interpolation. To increase accuracy of the details with respect to buildings and terrain structures near to water bodies, the laser-derived digital terrain model is combined with geographical survey data. This process is especially important for adding cross section data for water bodies. Due to the lack of the existence of detailed geographical survey data it could be necessary to acquire data by an on-site survey. All buildings in and around a water body should be surveyed. Furthermore, cross sections of the water bodies should be surveyed at appropriate intervals for hydraulic requirements. In the end the laser-scan data can be combined with the geographical survey data to produce a digital terrain model which, in this case, reach an accuracy of about +/- 50 cm in position and +/-25 cm in elevation for laser-scan data and about +/- 5 cm elevation accuracy for surveyed data. The combination of the data is shown in figure 4.1.

T03_08_02_D03_1_V3_0_p13.doc 21

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Figure 4.1: Combination of geographical survey data and laser scan data for extracting cross section data along the water bodies (Ministry of the Environment Baden-Wrttemberg, 2005)

4.2

Sources

Hydrology The main input data for calculating flood hazard maps is the occurrence probability and the amount of high water discharge in rivers. High water discharge for particular annualities can be calculated by a regionalisation approach for water levels measured by gauges. In individual cases river basin studies need to be carried out. The application of a rainfall runoff model can also be helpful for considering water body system characteristics such as retention reservoirs or long term calculations. The following annualities are considered: HQ10 HQ50 HQ100 HQEXTREM The results of the hydrologic calculation are high water discharges for certain annualities at certain locations along the water body. However, the identification of annulities for coastal domains is not a simple task. Please refer to chapter 3.2.1 for further information. Hydraulic Based on the hydrologic output data and the cross sections derived from the digital terrain model, as shown in figure 4.2, a hydraulic calculation can be made. In so doing, the choice of an appropriate model is essential for getting good results. Whereas for the vast majority of water bodies a 1D model is sufficient there are some areas which cannot be correctly represented by the simplified numeric system of a 1D model. This is particular true for any kind of water body which does not satisfy the boundary conditions of a 1D model, e.g. estuaries or other areas where irregular flow occurs. In those cases a more complex calculation has to be done by a 2D model. Whereas 1D models can only be
T03_08_02_D03_1_V3_0_p13.doc 22 26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

applied for assessing water levels along the axis of the water body, 2D models are able to assess extensive hydraulic dimensions such as flow depth, and speed of current in both directions along the axis of the water body. The results of the hydraulic calculation show the different water levels along the length of the water bodies. The results from a 2D model and a 1D model differ in terms of resolution. Therefore an adaptation on cutting edges has to be done.

Figure 4.2: Input data for hydraulic calculation (Ministry of the Environment Baden-Wrttemberg, 2005)

4.3

Pathway Inundation

With the results of the hydraulic calculations the flood outline can be calculated. The main step is to calculate the inundation area by subtracting the water level plane from the digital terrain model for results produced by a 1D model. The resulting map can be interpreted as a water depth map where negative values denote inundation areas as it is shown in figure 4.3. Results of a 2Dmodel bring out the water depth directly. Other hydraulic dimensions such as speed of current are also available for areas calculated by a 2D model. The use of high resolution data in the digital terrain model can cause a so called salt and peppereffect. This effect is caused by vegetation that adds a random noise to the elevation data. A classification of the map content might be necessary when the effect occurs heavily in the map. These classification leads to a corrected version of the map. Classifications are usually based on boundary criteria which depend on the water depth and noise area.

Figure 4.3: Intersection of hydraulic results with the digital terrain model (Ministry of the Environment Baden-Wrttemberg, 2005)

T03_08_02_D03_1_V3_0_p13.doc 23

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

The calculated map of the water depth is the basis for further work. Another display option is the outline of the flooded area as it is shown in figure 4.4. In Baden-Wrttemberg following maps are created by using the results from the hydraulic calculations: Flood extents (all annualities displayed in one map) Water depth for flooded areas HQ10 (uncorrected) Water depth for flooded areas HQ50 (uncorrected) Water depth for flooded areas HQ100 (uncorrected) Water depth for flooded areas HQEXTREM (uncorrected) Water depth for flooded areas HQ100 (corrected) Water depth for flooded areas HQEXTREM (corrected)

Figure 4.4: Possible map contents of flood hazard maps (Ministry of the Environment BadenWrttemberg, 2005) Other maps can also be built form the water depth that is given by the hydraulic results. For example, an intersection of the water depth and the flood protection buildings can lead to a map in which the protected area is shown. Another example is the identification of bridges that can be used during the flood.

4.4

Map Content

Ideally, according to the EU Directive 2007/60/EC a Flood Hazard map should include the flood extent, projected water depths and, where appropriate, the flow velocity. Due to the characteristics of coastal floods, Flood Hazard maps will mainly include the extension of the area to be inundated and projected water depths. In addition to this, the areas along the coast sensitive to the event to be mapped (i.e. showing a projected erosion significant enough to affect/enhance inundation of the hinterland) should also be indicated. The most simple flood hazard map should be one just showing the extension of the area to be flooded during an event of a given probability or return period (e.g. figure 4.5).

T03_08_02_D03_1_V3_0_p13.doc 24

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Figure 4.5: Delineation of flood hazard areas to a water level associated to a 50 years return period (dashed blue lines, GIOC 1998). A further developed hazard map is one that includes not only the extension of the area to be flooded but also the expected water depth. As an example of such maps, figure 4.6 shows the maximum inundation depth due to sea flooding in the Netherlands. It has to be noted that although expected damages should be related to water depth, this map is not equivalent to a flood damage map since it is only providing information on the physical process. The Flood Hazard Map can also include the information of the different flood-related variables to derive a hazard scale. An example of such an approach is the development of flood hazard maps for events affecting people. Figure 4.7 shows an example of such maps where the hazard is rated in terms of danger to the people according a simple equation that integrates the depth of flooding, the velocity of floodwaters and the debris factor.

T03_08_02_D03_1_V3_0_p13.doc 25

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Figure 4.6: Example of maximum flood inundation depth caused by sea flooding in the Netherlands (Excimap, 2007a).

T03_08_02_D03_1_V3_0_p13.doc 26

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Figure 4.7: Example of Flood Hazard Map with a hazard scale in the UK (DEFRA, 2006). Finally, in coastal sedimentary environments where the coastal fringe is capable of responding to the impact of an event (a storm associated to a given probability), the hazard map can also include an indication of the areas prone to be eroded. This information can later be integrated into the flood hazard to obtain the integrated hazard of the zone. Figure 4.8 shows an example of deriving two different maps associated with the same event but disaggregating the two induced processes, i.e. flooding and erosion.

Figure 4.8: Example of Flood and Erosion Risk Maps in USA (NOAA, 2008).

T03_08_02_D03_1_V3_0_p13.doc 27

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

The maps presented above, summarize the main technical information to be included in flood hazard maps. In addition to the technical information, according to the recommendations produced by Excimap (2007) after analysing different map types in Europe, Flood Hazard maps should include at least the following information: Title (what kind of information on which location is presented). Probability of the event. Location of the map. North and scale (preferably graphic bar as this allows for changes in page sizes). Responsible authority or institute with contact details. Date of publication. Disclaimer, including remarks on the quality of information. All maps should have consistent information (e.g., consistent extents for given event probability), although the content, format and dissemination may differ depending on the purpose and target audience. Explanations should always be given (directly onto the maps) for correct interpretation of maps (e.g. return period or probability, method of development, uncertainty, etc.), as appropriate for the target audience. Public maps should be simple and self-explanatory and include a legend, so that as little supporting or explanatory information as possible is required for correct interpretation. The extent of potential flooding has to be presented as a surface covering the topography for a specified flood level /frequency. For reference, roads, railways, houses and permanent water bodies from which the floods originate may be included. A good option is to overlap the inundation area over an ortho-photo in which elements in the territory can be easily identified. Presently, there are many administrations providing digital versions of flood hazard map so that the user can interactively decide what information will be included (see e.g. figure 4.9). If the Flood Hazard map is produced for organisational users (governments, local authorities, etc.), the map may require more detailed explanatory information to enable the reader to fully understand its development as well as the limitations of the included information. Supporting information for organisational users should include: (a) GIS data available for download, emphasising the use of supporting information, (b) Metadata information (e.g. quality, data sources, inventory of existing information, etc.). As a final remark, publication of flood hazard maps requires the determination of their level of confidence and the likely uncertainty. An easy way to do this is to include the calculated uncertainty as an area of uncertain flooding.

T03_08_02_D03_1_V3_0_p13.doc 28

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Figure 4.9: Example of digital flood hazard map generated through the Fema web application.

T03_08_02_D03_1_V3_0_p13.doc 29

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

5. Summary
In this report the main technical aspects related to flood hazard mapping in coastal areas have been reviewed. The main steps to produce a flood hazard maps are summarized in figure 5.1. They are: Characterization of the source. This is primarily done by specifying wave and water level conditions associated to a given probability to be used in the analysis (which should be the one characterising the flood hazard map to be produced). Characterization of the pathway. This is done by estimating (i) the wave-induced runup on a representative beach transect. To do this, representative beach profiles of the study area are needed. If the area of study is morphodynamically active, it should be necessary to estimate the coastal response to the impact of the event by using a morphodynamic model. This will serve to update the estimated run-up and the beach/dune height during the event from which the overtopping discharges are calculated.

morphodynamic feedback wave data water level data floodplain DEM

beach profile data

Runup R2%

total water level

overtopping

inundated area surface & depth

Figure 5.1: Sketch of the methodology to generate a coastal flood hazard map.

T03_08_02_D03_1_V3_0_p13.doc 30

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Characterization of the receptor. Finally, the estimated overtopping discharges to the hinterland are propagated landwards over a Digital Elevation Model of the floodplain, from which the flood hazard areas are delineated in terms of extension and depth. For each of the mentioned steps, specific recommendations have been specified in this report ranging from data acquisition, data analysis to modelling. These recommendations can be considered as general enough to be applied to most of coastal environments, although they need to be adapted to the specific conditions of the area of study, including the proper selection of techniques and models to be used throughout the process. Finally, as an example of application of these recommendations, in Annex II the presented methodology is applied to the Ebro delta coast to delineate the flood hazard area associated to the impact of a storm with a Tr of 100 years. This case is a clear example of the influence of the coastal morphodynamic feedback on the extension of flooding. The obtained results showed that the selection of a given initial beach profile from an existing dataset to characterize the coastal fringe, can result in variations of the duration of overtopping events of about 300 %. When the beach evolution during the storm is included, the volume of floodwater entering the coastal plain is significantly larger than for any tested static scenario. This means that any flood hazard mapping in coastal sedimentary environments without including the beach response will significantly underestimate the hazard area.

Figure 5.2: Changes in Flood hazard maps (extension of the flood) in the Ebro delta for a 100 year return period storm as a function of the initial beach morphology.

T03_08_02_D03_1_V3_0_p13.doc 31

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

6. References
1. 2. ALLSOP, W., KORTENHAUS, A. and MORRIS, M. (Eds) (2007) Failure mechanisms for flood defence assets. FLOODsite report T04-06-01. ANDERSON, E. S., THOMPSON, J.A. and AUSTIN, R.E. (2005), LiDAR density and linear interpolator effects on elevation estimates, International Journal of Remote Sensing, 26, 38893900. BASCO, D.R., and SHIN, C.S. (1999). A One-Dimensional Numerical Model for StormBreaching of Barrier Islands. Journal of Coastal Research, 15, 241-260. BURCHARTH, H.F. and HUGHES, S.A. (2003). Fundamentals of Design (Part VI). In: Coastal Engineering Manual, US Army Corps of Engineers. CALLAGHAN, D.P., NIELSEN, P., SHORT, A. and RANASINGHE, R. (2008). Statistical simulation of wave climate and extreme beach erosion. Coastal Engineering, 55, 375-390. CAIZARES, R. and IRISH, J.L. (2008). Simulation of storm-induced barrier island morphodynamics and flooding. Coastal Engineering, doi:10.1016/j.coastaleng.2008.04.006. CHEUNG, K.F., PHADKE, A.C., WEI, Y., ROJAS, A.R., DOUYERE, Y.J.-M., MARTINO, C.D., HOUSTON, S.H., LIU, P.L-F, LYNETT, P.J., DODD, N., LIAO, D.S. and NAKAZAKI, E. (2003). Modeling of storm-induced coastal flooding for emergency management. Ocean Engineering, 30, 13531386. DEFRA (2001), National Appraisal of Assets at Risk from Flooding and Coastal Erosion, including the potential impact of climate change. Final Report. Flood Management Division. Department for Environment, Food and Rural Affairs, London, UK. DEFRA (2006), Flood Risk to the People. FD2321/TR2 Guidance Document. Department for Environment, Food and Rural Affairs, London, UK. DELFT HYDRAULICS (1983). Wave Runup and Overtopping at Dunes During Extreme Storm Surge. Report M1819, Part II. Delft. DIRECTIVE 2007/60/EC OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL on the assessment and management of flood risks, 23/10/2007, Official Journal of the European Union, L 288/27 - L288/34 DIVOKY, D. and McDOUGAL, W.G. (2006) Response-based coastal flood analysis. Proc. 30th ICCE, ASCE, 5291-5301. DONELLY, C. 2008. Coastal overwash: processes and modelling. Ph D Thesis, University of Lund. EXCIMAP 2007. Handbook on good practices for flood mapping in Europe. 57 pp. EXCIMAP 2007. Atlas of Flood Maps. 197 pp. EUROTOP. 2007. Wave Overtopping of Sea Defences and Related Structures: Assessment Manual. Environment Agency (EA), Expertise Netwerk Waterkeren (ENW), Kuratorium fr Forschung im Ksteningenieurwesen (KFKI), www.overtopping-manual.com. FEMA. 2003. Guidelines and specifications for Flood Hazard Mapping Partners. Appendix D: Guidance for Coastal Flooding Analyses and Mapping. FEMA 2005. Wave runup and overtopping. FEMA Coastal Flood Hazard Analysis and Mapping Guidelines Focused Study Report, 51 pp.

3. 4. 5. 6. 7.

8.

9. 10. 11.

12. 13. 14. 15. 16.

17. 18.

T03_08_02_D03_1_V3_0_p13.doc 32

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

19.

GARRITY, N.J., BATTALIO, R., HAWKES, P.J. and ROUPE, D. (2006) Evaluation of the event and response approaches to estimate the 100-year coastal flood for Pacific coast sheltered waters. Proc. 30th ICCE, ASCE, 1651-1663. GIOC (1998) Atlas de inundacin del litoral peninsular. Direccin General de Costas, Ministerio de Medio Ambiente. GODIN, G. (1991) Tides, Cicese, Ensenada, Mexico. GUEDES-SOARES, C., WEISSE, R., CARRETERO, J.C. and ALVAREZ, E. (2002). A 40 years hindcast of wind, sea level, and waves in European waters. 21st Int. Conf. OMAE, Oslo, 17. GUENTHER, G.C., CUNNINGHAM, A.G., LAROCQUE, P.E. and REID, D.J. (2000) Meeting the accuracy challenge in airborne lidar bathymetry. Proceedings of EARSeL Symposium, Workshop Lidar, Dresden. HAILE, A.T. and RIENTJES, T.H.M. (2005). Effects of Lidar dem resolution in flood modelling: A model sentitivity study for the city of Tegucigalpa, Honduras. ISPRS WG III/3, III/4, Workshop "Laser scanning 2005". Enschede. HR WALLINGFORD. (2003). Guide to Best Practice in Coastal Flood Forecasting. R&D Technical Report FD2206/TR2, DEFRA

20. 21. 22.

23.

24.

25.

26.
27.

IHO. (1998) IHO Standard for Hydrographic Surveys, IHB Sp Pub 44, 4th Ed, International Hydrographic Bureau, Monaco, 23pp.
KOMAR, P.D., MCDOUGAL, W.M., MARRA, J.J. and RUGGIERO, P. (1999). The Rational Analysis of Setback Distances: Applications to the Oregon Coast. Shore & Beach 67, 41-49. KOMAR, P.D., MCDOUGAL, W.M., MARRA, J.J. and RUGGIERO, P. (2001). The rational analysis of setback distances: Applications to the Oregon Coast. Journal of Coastal Research 17, 407419. KRAUS, N. C. and WAMSLEY, T. V. (2003). Coastal barrier breaching, Part 1: Overview of breaching processes, ERDC/CHL CHETN IV-56, US Army Engineer Research and Development Center, Vicksburg. LARSON M. and KRAUS N.C. (1989). SBEACH: Numerical Model for Simulating StormInduced Beach Change, CERC-89-9, US Army Corps of Engineers, Vicksburg. LARSON, M., ERIKSON, L. and HANSON, H. (2004). An analytical model to predict dune erosion due to wave impact. Coastal Engineering 51, 675 696. LIU, X., ZHANG, Z., PETERSON, J. and CHANDRA, S. (2007) The effect of LiDAR data density on DEM Accuracy. In: MODSIM07 International Congress on Modelling and Simulation, Christchurch, New Zealand. LILLYCROP, W.J., PARSON, L.E., and IRISH, J.L. (1996). Development and Operation of the SHOALS Airborne Lidar Hydrographic Survey System. SPIE Selected Papers, Laser Remote Sensing of Natural Waters: From Theory to Practice, 2694, 26-37. MESSNER, F., PENNING-ROWSELL, E., GREEN, C., MEYER, V., TUNSTALL, S. and VAN DER VEEN, A. (Eds) (2007) Evaluating flood damage: guidance and recommendations on principles and methods. FLOODsite report T09-06-01. MLLER, I., SPENCER, T., FRENCH, J.R., LEGGETT, D. and DIXON, M. (1999) Wave transformation over salt marshes: a field and numerical modelling study from North Norfolk, England. Estuar. Coast. Shelf Sci., 49, 411-426. MORTON, R.A., LEACH, M.P., PAINE, J.G. and CARDOZA, M.A. (1999). Monitoring beach changes using GPS surveying techniques. Journal of Coastal Research, 9, 702720.

28.

29.

30. 31. 32.

33.

34.

35.

36.

T03_08_02_D03_1_V3_0_p13.doc 33

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

37. 38. 39.

NOAA. (2008). Hazard Analysis. http://www.csc.noaa.gov/products/nchaz/htm/case2.htm (consulted in September 2008) OUMERACI, H., DELISO, C. and KORTENHAUS, A. (2005) Breaching of coastal dikes: state of the art. FLOODsite report, LWI 910. PIRAZZOLI, P.A. and TOMASIN, A. (2007). Estimation of return periods for extreme sea levels: a simplified empirical correction of the joint probabilities method with examples from the French Atlantic coast and three ports in the southwest of the UK. Ocean Dynamics, 57, 91107. PRINOS, P. (Ed) (2007) Review on Flood Hazard Mapping. FLOODsite Report T03-07-01. PUGH, D.T. (1987). Tides, Surges, and Mean Sea-level. Wiley,New York. PURVIS, M.J., BATES, P.D. and HAYES, C.M. (2008). A probabilistic methodology to estimate future coastal flood risk due to sea level rise. Coastal Engineering, doi: 10.1016/j.coastaleng.2008.04.08. REEVE, D. (Ed.) (2007) Predicting morphological changes in rivers, estuaries and coasts. FLOODsite report T05-07-02. ROELVINK, J.A., VAN KESSEL, T., ALFAGEME, S. AND CANIZARES, R. (2003). Modelling of barrier island response to storms. Proc. Coastal Sediments 2003, ASCE. ROELVINK, J.A., RENIERS, A. VAN DONGEREN, A., DE VRIES, J.V.T., LESCINSKI, J. and WALSTRA, D.J. (2007). Modeling hurricane impacts on beaches, dunes and barrier islands. Proc. 10th workshop on Waves and Coastal Hazards, North Sore, Oahu SALLENGER, A. (2000). Storm impact scale for barrier islands. Journal of Coastal Research, 16, 890-895. SANCHEZ-ARCILLA, A. And GONZALEZ-MARCO, D. (Ed.) (2007) Report on best suitable models for a statistical analysis of joint probabilities of extreme event data. FLOODsite report T02-07-03. SOBEY, R.J. (2005) Extreme low and high water levels. Coastal Engineering, 52, 63-77. SCHOONEES J. S. and THERON A. K. (1995). Evaluation of 10 cross-shore sediment transport/ morphological models. Coastal Engineering , 25, 1-41. SOLOMON, S., D. QIN, M. MANNING, Z. CHEN, M. MARQUIS, K.B. AVERYT, M. TIGNOR and H.L. MILLER (Eds) (2007) Climate Change 2007: The Physical Science Basis. Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge. STOCKDON, H.F., HOLMAN, R.A., HOWD, P.A. and SALLENGER, J.A.H. (2006). Empirical parameterization of setup, swash, and runup, Coastal Engineering, 53, 573588. TUAN, T.Q., VERHAGEN, H.J., VISSER, P. and STIVE, M.J.F. 2006. Numerical modeling of wave overwash on low-crested sand barriers. Proc. 30th ICCE, ASCE, 2831-2843. TUAN, T.Q., STIVE, M.J.F., VERHAGEN, H.J. and VISSER, P.J. 2008. Process-based modeling of the overflow-induced growth of erosional channels. Coastal Engineering, 55, 468-483. USGS. 2001. Mapping Coastal Change Hazards. Coastal Change Hazard Scale. http://coastal.er.usgs.gov/hurricanes/mappingchange/scale.html VAN DER MEER, J. W. and JANSSEN, P. F. M. (1995) Wave Run-up and Wave Overtopping at Dikes. In: DEMIRBILEK, Z. (ed.), Wave Forces on inclined and vertical wall structures, ASCE, 1-27.

40. 41. 42.

43. 44. 45.

46. 47.

48. 49. 50.

51. 52. 53.

54. 55.

T03_08_02_D03_1_V3_0_p13.doc 34

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

56.

VAN GENT, M., COEVELD, E.M., DE VROEG, H. and VAN DE GRAAF, J. 2007. Dune erosion prediction methods incorporating effects of wave periods. Coastal Sediments 2007, ASCE, 612-625. VELLINGA, P., 1986. Beach and dune erosion during storm surges. PhD thesis, Delft Hydraulics Communications No. 372, Delft Hydraulics Laboratory, Delft, The Netherlands. WASA group (1998). Changing waves and storms in the Northeast Atlantic? Bull. Amer. Met. Soc. 79, 741-760. WISE, R.S., SMITH, S.J. and LARSON, M. (1994). SBEACH: Numerical model for simulating storm-induced beach change, CERC, US Army Corps of Engineers, Vicksburg. WOODHEAD, S. (Ed.) (2007) Evaluation of inundation models. FLOODsite report T08-0701.

57. 58. 59. 60.

T03_08_02_D03_1_V3_0_p13.doc 35

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Page intentionally blank

T03_08_02_D03_1_V3_0_p13.doc 36

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Annex I. Run-up and overtopping formulas

T03_08_02_D03_1_V3_0_p13.doc 37

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Page intentionally blank

T03_08_02_D03_1_V3_0_p13.doc 38

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Run-up on beaches
For wave runup estimations on beaches, there exist different options in the literature. Here we recommend the use of one the recently proposed by Stockdon et al. (2006). This recommendation is due to the fact that the formula has been derived from reanalysis of run-up data obtained in field experiments, large scale laboratories and, in consequence, in prototype conditions. Moreover the employed datasets cover most of the expected conditions occurring at natural beaches. The method predicts the runup, R2%, which is exceeded by 2% of the number of incident waves as

and for very dissipative beaches ( )

where:
0 Hs0 Tp L0 f Iribarren Number using Tp and Hs0 [-] deep-water significant wave height [m] wave peak period [s] deepwater wavelength associated to Tp [m] beach slope [-]

This expression also accounts for the wave set-up at the shoreline.

Wave overtopping discharge


As an example of formulas for overtopping discharges in coastal flood analysis, Fema (2003) proposes to use a mean overtopping discharge rate formula derived for sloping surfaces and irregular waves which is given by

Q* = 810 exp[3.1(rR* -Rc / Hs)] Q = Q* (g Hs3)1/2 where Q* is the dimensionless mean overtopping discharge rate, Q is the dimensional discharge 0.5 rate in m3/m/s, R* = [1.5 m/( Hs /L0) ], up to a maximum value of 3.0, is an estimated extreme runup normalized by Hs, for a beach/barrier slope given as the tangent m, Rc is the freeboard of the beach/barrier and r is a roughness coefficient. This formula is applied when the freeboard, Rc < 2 Ru. There are cases where empirical derived formulas for ad-hoc estimations of overtopping rates do exist. In such cases, it is recommended to use the corresponding proposed method. An example of this is the formula derived to estimate the mean overtopping discharge for dunes along the dutch coast associated with a return period of 100 years. This formula was derived by Delft Hydraulics (1983) based on a series of laboratory experiments and it predicts the mean overtopping discharge (m3/m/s), Q, as a function of the freeboard, Rc, of the dune remnant above that water level
T03_08_02_D03_1_V3_0_p13.doc 39 26 September 2008

-5

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Q = 0.489 exp (-0.0771 * Rc)

A recommendation of the corresponding formulas for mean overtopping discharge rates for different typologies of coastal structures can be found in Eurotop (2007). Once the mean overtopping discharge is calculated, the following step is to convert it to overtopping volumes, or, in other words, in the water volume entering to the hinterland, V. In this case, this can be simply done by extending the calculated mean discharge rates to the duration of wave conditions, , i.e. V = Q . .

T03_08_02_D03_1_V3_0_p13.doc 40

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Annex II. Coastal flood hazard mapping in the Ebro delta

T03_08_02_D03_1_V3_0_p13.doc 41

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Page intentionally blank

T03_08_02_D03_1_V3_0_p13.doc 42

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Flood Hazard Mapping for Coastal Storms in the Delta Ebro1


D. Alvarado-Aguilar & J. A. Jimnez*
Laboratori dEnginyeria Martima, ETSECCPB, Universitat Politcnica de Catalunya, c/Jordi Girona 1-3, Campus Nord, ed. D1, 08034 Barcelona, Spain (Email: dagoberto.alvarado@upc.edu, jose.jimenez@upc.edu) * Corresponding author (voice: +34 934016468, fax: +34 934011861)

Abstract
Flood hazard mapping is a critical issue in coastal low-lying areas due to its intrinsic sensitivity to the impact of extreme storms. The estimation of the beach response during the event is a key point, not only to calculate the induced erosion, but to properly evaluate flood hazard areas since the magnitude of the flooding can be affected by a morphodynamic feedback. Here, we present a methodology to delimit the uncertainty in flood hazard mapping associated to natural beach morphodynamics. This methodology has been tested in the Ebro delta coast for the impact of a storm with a Tr of about 100 years. Results showed that the selection of a given initial beach profile, from the different recorded ones, can result in variations of the events overtopping durations of about 300 %, which are amplified for overtopping rates. Along the study area, the uncertainty introduced by the use of a single representative profile for the entire coast is much lower than the one associated with the selection of the beach profile shape. Finally, when the beach evolution during the storm is included, the volume of floodwater entering the coastal plain is significantly larger than for any of the tested static scenarios. This means that any flood hazard mapping in sedimentary coastal environments without including the beach response will significantly underestimate the hazard area.

Introduction
Flooding in coastal areas is a result of the combination of a driving agent, the storm, normally defined in terms of a water level -storm surge plus wave induced runup- acting on a receptor, the coast, which is mainly defined in terms of elevation. However, in sedimentary environments, the impact of the storm will produce a significant morphodynamic response that will interact with the storm and should affect the intensity of the flooding (enhancing or reducing). When applied to low-lying areas, as deltaic coasts are, the estimation of this interaction is crucial due to the expected magnitude of the induced impacts in these environments. Tthese areas are characterized by a very low relief, and consequently easily flooded. They are protected from the sea by sandy beaches and barriers which freely respond to storm action and, in consequence, are easily modified. A review of the potential factors influencing the response of low-lying coasts to storms can be seen in Morton (2002). Although the number of factors is relatively large, a conclusion of the study is that the most important variables controlling the coastal response are the difference in elevation between the water level during the storm and the beach/ barrier/ dune crest and the duration of the flooding. This is common conclusion of most of studies of the response of low-lying coasts to extreme storms and, in fact, Sallenger (2000) proposed a hazard scale for barrier islands (that can be considered a paradigm of sensitive low-lying coasts) based on the ratio of run-up to barrier height. This has also been included

Paper presented at the FLOODrisk 2008 conference, 30 September 2 October 2008, Oxford, UK.
26 September 2008 43

T03_08_02_D03_1_V3_0_p13.doc

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

in most of the existing studies in overwash and breaching due to storms (e.g. Larson et al., 2005; Donnelly et al., 2006; Jimnez et al., 2006). During these events, main induced changes of the beach profile morphology without considering the retreat itself- can be simplified in terms of lowering (decrease of the beach height) and flattening (decrease in the beach slope). These two morphological changes will induce a morphodynamic feedback with acting processes. As an example, the magnitude of the induced inundation should be affected in oppositing terms when they are separately considered (see e.g. Jimnez et al., 2006). On the one hand, beach lowering will tend to increase the water discharge towards the hinterland because for a given water level the lowering of the beach/dune height will increase the freeboard. On the other hand, the profile flattening will tend to reduce inundation because for a given set of wave characteristics, the induced run-up will decrease due to the decrease in the beach slope. The resulting effect (inundation increase or decrease) will depend on the type and magnitude of beach changes during the storm. With these antecedents, it is clear that the estimation of the beach morphodynamic response during the storm is a key point not only to calculate the induced erosion, but to properly evaluate flood hazard areas. This is especially relevant for non-protected coasts where the main barrier to flooding is the beach itself which can hardly be considered as a rigid boundary. Thus, if we are going to predict the flood hazard area in a costal stretch associated to the impact of a given storm, one of the key points we have to face is the selection of the pre-storm coastal morphology. Moreover, once we have solved this question, the next question is what will happen during the event. Within this context, the main aim of this paper is to analyze the influence of the inclusion of the natural variability of beach morphology in coastal flood hazard mapping in non-protected areas. First we present the developed methodology to define the uncertainty in the extension of the flooded area induced by the definition of the beach, which should be the element protecting the hinterland from floodwaters. Second, the methodology is applied to map flood hazard areas in the Ebro delta under the impact of extreme coastal storm.

Methodology
The methodology used in this work is outlined in figure 1.

Forcing
The first step consists in the estimation of a total water level at the shoreline. This is done by using the response-method approach, which is based directly on measured or simulated water levels and waves as they occurred in nature and, the water level of interest (associated to a given probability or return period) beeing directly calculated from a probability distribution of total water levels. This method is specially recommended when variables determining the flood level are partially correlated, i.e. when surge and large waves are uncoupled and, for areas where wave height and periods during storms (both will determine the wave run-up) are poorly correlated (see e.g. Divoky and McDougal, 2006; Fassardi, 2006). Since our analysis is done at a coast without any protection but natural beaches, the run-up model proposed by Stockdon et al (2006) has been selected, since this formula was derived by using run-up data obtained in field and large scale experiments on beaches. The run-up (R2%) is calculated for each beach profile scenario (according to each beach profile definition method, see below), with differences in run-up magnitude being controlled by the use of a different beach slope since wave conditions are

T03_08_02_D03_1_V3_0_p13.doc 44

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

the same in all the cases. The obtained values are then added to simultaneous water level data (m) to build the total water level time series (t). The total water level data are then fitted to an extreme distribution to estimate the water level associated to given probabilities or return periods.

f (t) in dyn. appr.

Hs Tp

Be. Pro. Z (x)

R2%

DEM

Inundated area

Figure 1. Outline of used methodology for coastal flood hazard mapping. Once the target total water level has been estimated, the following step is to calculate overtopping rates (Q) for those cases in which the run-up exceeds the beach/barrier crest. This will determine the volume of floodwater penetrating the hinterland and, in consequence, determining the extension of the flood hazard area. The overtopping volume has been calculated following the method used by Fema (2003) to estimate the inundation in low-lying coasts. In essence the method estimates the mean overtopping rate for smooth slopes based on the former works of Owen (1980).

Beach configuration
A critical issue of mapping coastal areas prone to be inundated during storms is how to properly characterize the beach configuration. Beach configuration takes part in the process by controlling the magnitude of the run-up (via beach slope) and, by controlling the overtopping (via beach/dune crest height). Thus, in contrary to the quasi-static case of dikes, beaches are continuously reacting to coastal dynamics and, especially during the impact of storms, they are significantly modified. This means that to properly map coastal flood hazard areas beach dynamics have to be incorporated to the analysis. In this work, we have tested three different approaches to define the beach configuration during the storm and their impact on the inundation of the hinterland as already outlined in Alvarado and Jimnez (2007).
T03_08_02_D03_1_V3_0_p13.doc 45 26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

The first approach uses a fixed beach profile. This should be equivalent to the case in which the only data available is a pre-storm coastal configuration (taken at any moment and not necessarily just before the impact). This beach morphology is used throughout the analysis which means that run-up will be controlled by the corresponding beach slope and the overtopping by the beach/dune crest which is maintained fixed for the storm duration. This type of the approach is equivalent to considering the beach to be a fixed protecting structure. The second approach introduces some information on the natural beach variability. In essence in this approach the beach is characterized by an envelope of the possible configurations instead of a single value. It can only be used when information about past beach morphologies is available. Thus, for each representative transect along the coast, the morphology is represented by the envelope of all the existing data (ideally covering a period of several climatic periods and, thus, showing the natural changes in the beach morphology to wave action). This will allow the estimation of all the relevant variables (run-up and overtopping) for the different configurations - bounded by the maximum and minimum profiles . This should be equivalent to add some kind of confidence band to the calculated floodwater and, in consequence, to the potentially flooded surface.
2 1 0 -1 -2 -3 -4 0 100 200 300 400 500

elevation (m)

cross-shore distance (m)

Figure 2. Envelope of beach profiles taken during a 4 years period in the Northern part of the Ebro delta. Figure 2 shows an example of the beach profile envelope used in this which was study obtained from 4 years of beach profile data where a significant variation in the beach configuration is observed. The third approach to describe the beach morphology consists in simulating the beach profile response during the storm action recovering all the intermediate configurations from the pre-storm situation to the post-storm one. These intermediate configurations will allow to update the wave-induced run-up and overtopping rates according to the time-dependent beach slope and crest height. Beach profile evolution during the storm has been calculated by using the SBEACH model (Larson and Kraus, 1989; Wise et al., 1996). The model has been previously used to simulate the dune lowering before the inundation of the hinterland during the impact of extreme storms (see e.g. Caizares and Irish, 2008). The model is applied to selected profiles taken along the outer coast of the Ebro delta and, it was assumed that they are representative of the coastal response along a given stretch (alongshore uniform stretches). The model simulates the beach profile response due to storm wave action by assuming that the sediment transport is due to cross-shore processes only.
T03_08_02_D03_1_V3_0_p13.doc 46 26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

An example of the application of the model to simulate beach profile changes is shown in figure 3. As it can be seen, if the beach evolution during the storm is incorporated, there is a significant difference in beach morphology a decrease in beach height of about 1 m-. If this knowledge is incorporated in the flood mapping, results significantly vary with respect to those obtained for the initial configuration.

2 1 0 -1 -2 -3 -4 -100 0 pre-storm profile

elevation (m)

post-storm simulated profile

cross-shore distance (m)

100

200

300

400

500

Figure 3. Beach profile evolution during the impact of a storm simulated by using Sbeach.

Inundation
Once water levels and beach configurations are known, the final step is to determine which part of the coastal plain is flooded. To calculate the potential flood zone we used the LISFLOOD-FP inundation model (Bates and de Roo, 2000). This is a raster grid based model that has been successfully employed to simulate the inundation in fluvial and coastal areas (see e.g. Bates and De Roo, 2000; Bates et al., 2005). The model predicts water depths in each grid cell for each time step, and hence can simulate the dynamic propagation of flood waves over fluvial, coastal and estuarine floodplains. In our analysis we use a time series of water flow at the shoreline bordering the deltaic plain (calculated through the overtopping rates) as input.

Area of study and data


The Ebro delta is located on the Spanish Mediterranean coast about 200 km south of Barcelona. It has an approximate sub-aerial surface of 320 km2 and a coastline length of about 50 km excluding the inner coast of the two semi-enclosed bays (Figure 4). It is microtidal environment with an astronomical tidal range of about 25 cm although storm surges clearly exceeding such magnitude are not infrequent in the area (Jimnez et al., 1997). As many other deltas, it is an ecologically rich environment, with areas of habitats of the highest interest, being composed by freshwater, brackish and saline lagoons, salt marshes and coastal and small dune sandy areas. At the same time, it is actively exploited by means of agriculture; mainly for rice production (about 66% of the total subaerial surface is devoted to rice production).

T03_08_02_D03_1_V3_0_p13.doc 47

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

It is a very low-lying area with a maximum elevation above the MSL of about 4 m. Table 1 shows the percentage of the delta surface below the given elevations. These data clearly show that the system is potentially highly vulnerable to floods since about 49 % it is below an elevation of +0.5 m. These figures were calculated from a DEM of the Ebro delta derived from LIDAR data obtained by the Institut Cartogrfic de Catalunya. The data used in this study has a spatial resolution of 1 m and a vertical accuracy of 15 cm and has been acquired in the year 2004.

Table 1. Distribution of deltaic surface per elevations. Elevation (m ab MSL) < 0.5 0.5-1.0 1.0-2.0 2.0-3.0 3.0-4.0 > 4.0 Surface (ha) 16,296 7,539 5,968 2,495 754 141 (%) 49.09 22.71 17.98 7.52 2.27 0.42

Figure 4. The Ebro delta. The rectangle indicates the area for flood hazard mapping and points indicate profile locations (see figure 9). To simulate the inundation of the delta at a reasonable computational cost, the DEM was aggregated to a 10 m resolution. This size was selected after testing the influence of the cell size by masking the present effects of canals and small dikes separating rice pads.
T03_08_02_D03_1_V3_0_p13.doc 48 26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

In addition to this, a database of beach profiles was also available. It is composed of a series of profiles taken during a 4 years period at a spatial interval of 1-1.5 km along the coast. Although beach profiles were not taken simultaneously with the DEM, they can be used to characterize the beach profile temporal variability along the coast. Moreover, since the deltaic plain dynamics is much slower than beach dynamics (and especially in this case because it is highly regulated by human action) both datasets were integrated to optain a dynamic boundary of the Ebro delta. Thus, the DEM is used to characterize all the elevations along the deltaic surface and the beach profiles are used to estimate the flooding of the area as it is described in the methodology. Finally, to characterize the forcing, different data sets have been used: (i) a 17-year wave time series recorded by a directional wave buoy. The buoy was located at a point off the coast of Cap Tortosa with a water-depth of 50 meters and has been recording since 1990; (ii) a 43-year (1958-2001) time series of hindcasted wave conditions obtained within the framework of the Hipocas project (Guedes Soares et al., 2002). The selected event to test the described methodology was the one associated with a return period of about 100 years. This event was selected because on the one hand it is extreme enough to produce a significant flooding, on the other hand we have a detailed record of a storm of such characteristics. The storm was recorded in November 2001 (figure 5) and produced a significant erosion and inundation along the Ebro delta coast (see Jimnez et al. 2005; 2008).
6 14 5 Hs 12 Tp 10 8 6 4 11/11/01 13/11/01 15/11/01 17/11/01

Hs (m)

3 2 1 09/11/01

Time (dd/mm/yy)
Figure 5. Wave conditions during the target storm recorded by a wave buoy off the Ebro delta (adapted from Jimnez et al., 2005).

Results and discussion


Scenarios for the analysis Here, we have restricted the analysis to the simulation of flooding of the northern part of the Ebro delta, which is one of the areas most sensitive to the impact of storms which frequently experiences temporary inundation of the deltaic plain (Jimnez et al, 2005; 2008). Since the main objective of the paper is to test the methodology, we have spatially restricted the analysis to reduce the computational cost and we have concentrated in an area where we observed significant overwash during the simulated storm (Jimnez et al. 2005, 2008) (see figure 6).

T03_08_02_D03_1_V3_0_p13.doc 49

Tp (s)
26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Along this coastal stretch we have information on the natural variability of four representative beach profiles. The available data cover the evolution of these four profiles (named as 31 to 34 from south to north, see locations in figure 4) during four years during which a significant storm (return period of about 10 years) impacted the coast. The first task was to obtain three representative states for each beach profile: mean, maximum and minimum. These states will be hereinafter called as profile scenarios and they have been obtained by a statistical analysis of all the existing data for each profile. Thus, the mean corresponds to a simple averaging of all the existing data of a given profile; the maximum is a hypothetical configuration given by the upper limit of the envelope of all the data of the corresponding profile and minimum is the hypothetical configuration given by the lower limit of the envelope. These three scenarios will be used to test the influence of the beach morphology on the different parameters controlling flooding.

Figure 6. Northern part of the Ebro delta coast (La Marquesa beach) just after the impact of the storm of November, 2001.

Floodwater
Once we have the different scenarios defined in terms of beach profile configurations, we estimate the variation in the period of beach overtopping during the storm due to the different definitions of prestorm beach morphology. Table 2 shows the obtained exceedence times along the coast defined as the storm time period during which the total water level (here limited to the wave run-up, R2%) exceeded the beach height. The first aspect to be highlighted is that, for all locations along the coast (specified by profile names), there is a very significant variation in the duration of overtopping conditions. Table 2. Periods of effective beach overtopping (R2% > beach height) for the different profile scenarios along the Northern part of the Ebro delta coast. Profile Max (hours) 31 10 32 29 33 29 34 29
T03_08_02_D03_1_V3_0_p13.doc 50

Min (hours) 160 114 111 160

Mean (hours) 38 69 70 100


26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Thus, by measuring the differences in overtopping conditions in terms of a pseudo-coefficient of variation ([Tmax-Tmin]100 / Tmean), we find that the uncertainty introduced of the flood analysis (only referred to the duration of overtopping conditions) should vary from a minimum value of 117 % (profile 33) up to a maximum of 395 % (profile 31). Taking into account that this duration is a critical issue when calculating the total volume of water entering the hinterland, it becomes evident what an impact that the proper selection of the pre-storm profile will have on the final results. These calculated variations are strongly dependent on the natural profile variability, decreasing in importance for low- response coasts. However, low-lying sandy coasts, as the one studied here, are highly dynamic environments and, in consequence, although obtained values are strictly valid for analyzed conditions; they can be used as an order-of-magnitude of the expected variations. Finally, with respect to the spatial variations along the studied coastal stretch, obtained results show a relatively low variability. Thus, obtained coefficients of variation for each scenario range from a minimum value of 20.1% for the worst scenario (profiles defined with the minimum shape) to 39.2 % for the safest one (profiles defined with the maximum shape). This should indicate that the morphological consequences of coastal dynamics on beach configuration are relatively alongshore uniform, especially when high energetic events (those determining the minimum profile scenario) are considered. Moreover, these results should also indicate that the uncertainty introduced in the analysis due to the use of a single profile to represent the entire coastal stretch should be much lower than the associated to the selection of the configuration of such profile. Once the duration of overtopping events was determined, the next step was to compare the magnitude of such events. Table 3 and figure 7 show calculated mean overtopping rates (averaged for the duration of the storm) for each profile scenario. Table 3. Mean overtopping rates (averaged along the storm duration) estimated for the three beach profile configurations in the Northern part of the Ebro delta coast. Profile Q Max Q Min Q Mean -3 3 -3 3 (10 m /m/s) (10 m /m/s) (10-3 m3/m/s) 31 32 33 34 0.292 0.733 0.853 0.811 14.356 3.947 2.918 5.345 1.038 1.978 1.796 2.506

For the overtopping periods obtained results show a behavior similar to the calculated values. Thus, when different profile scenarios are compared, a large variability is detected again, with the pseudo coefficient of variation ranging from a minimum value of 115 % (profile 33) to a maximum of 1,355 % (profile 31). This huge increase in the variation for profile 31 is due to the dependence of the overtopping formula on the beach height (freeboard). Because the worst scenario (minimum configuration) for profile 31 corresponds to a strongly eroded and low profile with overtopping rates dramatically increasing. This indicates that the above estimated uncertainty for the overtopping period could be amplified when converted to volume of floodwater. When the mean overtopping rates of a given scenario are compared along the coast, they also show a behavior similar to the observed ones regarding the overtopping periods, i.e. a much lower variability than the observed as a function of the profile configuration. Thus, calculated coefficients of variation are of the same order of magnitude ranging from a minimum value of 33.31 % to a maximum value of
T03_08_02_D03_1_V3_0_p13.doc 51 26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

78.86 %. Again, there is an increase for the worst scenario (minimum profile) due to the already mentioned sensitivity to the beach height of the overtopping formula. In practical terms, these results show that, similarly to the case of overtopping periods, the uncertainty introduced in the analysis due to the use of a single representative profile for the entire coast will be much lower than the associated to the selection of the beach profile shape. Moreover, it has also to be stressed than for the same conditions, the uncertainty in overtopping rates will be larger than in the period of exceedence for overtopping.
15 minimum mean maximum 10

Q mean (10-3 m3/m/s)

beach profiles Figure 7. Mean overtopping rates (averaged along the storm duration) estimated for the three beach profile configurations in the Northern part of the Ebro delta coast.
Once we have determined the range of variation for the different scenarios, one of the remaining question is what beach morphology must be used throughout in the flooding analysis?

31

32

33

34

Although previous results serve to delimit the uncertainty of the analysis, it is obvious that whatever the selected beach configuration should be, it will change during the event.
As a final test, we have evaluated the previous two variables (period of exceedence for overtopping and overtopping rates) for one location along the coast (profile 32) including the simulated beach response during the event (Table 4). In addition to the mean overtopping rates (averaged during the event), we have included also the peak discharge during the event. Table 4 Calculated overtopping rates during the storm of November 2001 Profile scenario Q peak Q mean Flooded (10-3 m3/m/s) (10-3 m3/m/s) area (ha) Mean Maximum Minimum Evolving Beach response 16.811 6.653 32.717 55.736 1.978 0.073 3.947 7.403 78 36 110 155

Data in Table 4 is from along the Marquesa beach for the different scenarios (peak: at the storm peak; mean: storm-averaged; Flooded area: after 10 hours of continuous overtopping).

T03_08_02_D03_1_V3_0_p13.doc 52

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

As it can be seen and as it was expected, the inclusion of the beach evolution during the event results in an increase of overtopping rates. The rates increase by about 70% when compared to the ones calculated for the scenario previously defined as the worst scenario (minimum configuration). This means that even when selecting the worst scenario for static-oriented (fixed beach profile) flood hazard mapping in low-lying coasts, the volume of floodwater entering the coastal plain would be significantly underestimated. This is illustrated in figure 8 through the comparison of the run-up during the storm (assuming beach slope changes are small enough to modify it) with the beach height for the static approach (a given profile maintained fixed throughout the analysis) and the dynamic one (simulated beach evolution).
2.5 R2%

2.0

elevation (m)

1.5

pre-storm beach height

1.0

0.5

storm evolving beach height

0.0

time (hours) Figure 8. Run-up vs beach elevation during the target storm for static and dynamic beach configurations.

50

100

150

200

250

300

350

Inundation
Once the floodwater at the shoreline was estimated for the different scenarios, the remaining step was to delineate the part of the deltaic plain prone to be flooded during the event. The first practical task was to extend the calculations done for representative profiles along the coast. To do this this according to the existing information we have assumed the existence of an alongshore uniformity in morphology and response. Thus, each profile is considered to be representative of a coastal stretch of about 1 km. With this assumption, the total floodwater entering the hinterland across the beach will be the overtopping rates estimated for a given profile extended along the corresponding stretch. Although there are morphological evidences of alongshore uniform response of the study area under the impact of extreme storms (Jimnez et al., 2008), this introduces some uncertainty in the final extension of the flooded area. This is inherent to coastal flood analysis in sedimentary environments (coasts naturally protected by natural beaches/barriers) since (accurate) morphodynamic modeling of overwash and breaching processes is still an open question (e.g. Kraus and Hayashi, 2005; Tuan et al., 2008; Caizares and Irish, 2008). These spatially-integrated overtopping rates also need to be time-integrated over the duration of the storm. If the main source to induce flooding is the wave-induced run-up, there is some uncertainty about how to extend this pulsating process during a large period. Since the main objective of this paper is to analyze the effects of the morphodynamic interaction in relative terms we have just extended the overtopping rates to a total duration of 10 hours. The resulting integrated water flow was introduced as boundary conditions in LISFLOOD-FP in 100 seaward edge cells (1 km along the delta coastline).

T03_08_02_D03_1_V3_0_p13.doc 53

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

The potentially flooded areas for each scenario are shown in figure 9 whereas the corresponding calculated flooded surface can be seen in table 4. The first aspect to be highlighted is that when a static approach is followed, i.e. the beach is represented by a constant profile throughout the duration of the event, very significant differences are found. The estimated inundated area with a protecting beach represented by the mean profile (figure 9a) is 78 ha. However, when the beach is represented by the extreme profiles, i.e. the recorded minimum and maximum profiles, the surface will vary between 110 ha and 36 ha respectively (figures 9c and 9b). These calculated values stress on the one hand the influence of the pre-existing morphology on the extension of the area to be inundated and, on the other hand, the influence of the selection of the initial configuration. This last point is especially critical since in most of the occasions the use of a just-intime pre-storm morphology is just a matter of (very good) luck. The normal situation should be to have a given beach morphology taken in any moment that will not necessarily reflect the real beach morphology subjected to the impact of the storm. With respect to the topography plain data included in the DEM-, this issue is not too relevant provided major features controlling the extension of the flood have not changed (e.g. canals network, dikes, etc.). The range of calculated values should serve to estimate the uncertainty in the calculations associated to the selection of the pre-storm morphology. In this case, where we were fortunate enough to have a collection of beach profiles representing their natural variability, the flood hazard area for the target storm is delineated not as a single surface but as an area with a confidence band, i.e. an average value given by figure 9a that could vary between a minimum value given by figure 9b and a maximum one given by figure 9c. The decision on what surface has to be selected would depend on the purposes of the analysis and in the level of safety to be imposed. As an example, if we want to be in the conservative side, the situation associated to the minimum beach profile (maximum affected surface) should be selected. Even including this statistical beach evolution, the extension of the flood has been calculated assuming the beach is not modified during the storm impact. When this response is incorporated as described in section 2, the surface of the flood hazard area increases up to 155 ha (figure 9d and table 4). If the previous calculated values showed the importance of the proper selection of the pre-storm morphology, this result stresses the importance of including the morphodynamic feedback in coastal flood hazard mapping. In this case, the inclusion of the coastal response during the vent determined a significant increase in the volume of water entering to the deltaic plain and, in consequence, the potentially affected surface also increased with respect to the previous calculated worst scenario.

Conclusions
This work has presented an analysis of the influence of the definition of the beach morphology on flood hazard mapping in sedimentary coastal environments. To do this, a methodology to delimit the uncertainty associated to the beach natural dynamic variability was introduced. Results showed that the selection of a given initial beach profile from the ones recorded in 4 years, can result in durations of overtopping events varying by more than 300 %. Regarding overtopping rates, the estimated variation is even larger due to the dependence of their magnitude on the freeboard, i.e. the uncertainty is amplified for the volume of floodwaters (we found a peak variation of 1,355%). With regards to the spatial variability, results (for the study area) showed that the uncertainty introduced in the analysis due to the use of a single representative profile for the entire coast is much lower than the one associated to the selection of the beach profile shape. When the beach evolution during the storm is included, the volume of floodwater entering the coastal plain is significantly larger than for any of the tested static scenarios. This means that any flood hazard
T03_08_02_D03_1_V3_0_p13.doc 54 26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

mapping in sedimentary coastal environments without including the beach response will significantly underestimate the hazard area. Thus, for the tested case, the extension of the flood hazard area will vary between 36 ha and 110 ha for the static case and, will increase up to 155 ha when considering beach erosion during the storm.

Flooded Area=78.7 has

Flooded Area=36 has

Flooded Area=110.4 has

Flooded Area=155.3 has

Potencial Flooded Areas with diferents beach profiles a) Mean beach profile b) Maximum beach profile c) Minimum beach profile d) Coastal Response beach profile

Figure 9. Delineation of flood hazard areas in the northern part of the Ebro delta as a function of the definition of the beach morphology
T03_08_02_D03_1_V3_0_p13.doc 55 26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Acknowledgements
We are indebted to Paul Bates for his help and advice in the implementation of the LISFLOOD-FP. We thank DPTOP (Generalitat de Catalunya), Institut Cartogrfic de Catalunya and Puertos del Estado (Spanish Ministry of Public Works) for supplying the data used in this study. In particular, we thank to Xavier Marti (DMAH, Generalitat de Catalunya) for helping us with Ebro delta data. This work has been partly done in the framework of the EU-funded FLOODsite research project (GOCE-CT-2004-505420). First author was supported by CONACYT and the second author was partially supported by a University Research Promotion Award for Young Researchers from the Autonomous Government of Catalonia.

References
Alvarado-Aguilar, D. & Jimnez, J.A. 2007. A pseudo-dynamic approach to Coastal Flood Hazard Mapping. Proc. CoastGIS 2007, Santander. Bates, P.D . & De Roo, A.P.J. 2000. A simple raster-based model for floodplain inundation. Journal of Hydrology 236: 54-77. Bates, P.D., Dawson, R.J., Hall, J.W., Horritt, M.S., Nicholls, R.J., Wicks, J. & Hassan, M.A.A.M. 2005. Simplified two-dimensional numerical modeling of coastal flooding and example applications. Coastal Engineering 52: 793-810. Caizares, R. & Irish, J.L. 2008. Simulation of storm-induced barrier island morphodynamics and flooding. Coastal Engineering, doi:10.1016/j.coastaleng.2008.04.006. Divoky D. & McDougal, W.G. 2006 Response-based coastal flood analysis. Proc. 30th ICCE, ASCE 5291-5301. Donnelly, C., Kraus, N.C. & Larson, M. 2006. State of knowledge on measurement and modelling of coastal overwash. Journal of Coastal Research 22: 965-991. Fassardi, C. 2006. The definition of wave setup and runup using a response based approach. Proc. 30th ICCE, ASCE. 1800-1811. FEMA. 2003. Guidelines and specifications for Flood Hazard Mapping Partners. Guedes-Soares, C., Weisse, R., Carretero, J.C. & Alvarez, E. 2002. A 40 years hindcast of wind, sea level, and waves in European waters. 21st Int. Conf. OMAE, Oslo, 17. Jimnez, J.A., Sallenger, A. & Fauver, L. 2006. Sediment transport and barrier island changes during massive overwash events. Proc. 30th ICCE, ASCE, 2870-2879. Jimnez J.A., Snchez-Arcilla, A. & Valdemoro, H.I. 2005. Storm impacts on the Ebro delta coast. FLOODsite Research Report. Jimnez, J.A., Gracia, V. & Valdemoro, H.I. 2008. The Ebro delta coastal response during 2001-2004: a proxy of the potential effects of an increase in storminess? Int Symposium on Effects of Climate Change on the Worlds Oceans. PICES, Gijon. Jimnez J.A., Snchez-Arcilla, A., Valdemoro, H.I., Gracia, V. & Nieto, F. 1997. Processes reshaping the Ebro delta. Marine Geology 144: 59-79. Kraus, N.C. & Hayashi, K. 2005. Numerical morpholotical model of barrier island breaching. Proc. 29th ICCE, ASCE, 21202132. Larson M. & Kraus N.C. 1989. SBEACH: Numerical Model for Simulating Storm-Induced Beach Change, CERC-89-9, US Army Corps of Engineers, Vicksburg. Larson, M., Donnelly, C. & Hanson, H. 2005. Analytical modeling of dune response due to wave impact and overwash. Coastal Dynamics 2005, ASCE. MINISTRY OF THE ENVIRONMENT BADEN-WRTTEMBERG, Flood Risk Maps in BadenWrttemberg Guidelines, Online Publication http://www.hochwasser.badenwuerttemberg.de/servlet/is/1253/HWGK_Leitfaden_ENG.pdf, 2005 Morton R.A. 2002. Factor Controlling Storm Impacts on Coastal Barriers and Beaches A preliminary Basis for Near Real-time Forecasting. Journal of Coastal Research 18: 486-501.

T03_08_02_D03_1_V3_0_p13.doc 56

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Owen M.W. 1980. Design of seawalls allowing for wave overtopping. HR Wallingford, Report EX 924. Sallenger, A. 2000. Storm impact scale for barrier islands. Journal of Coastal Research 16: 890-895. Stockdon, H.F., Holman, R.A., Howd, P.A. & Sallenger, A.H. 2006. Empirical parameterization of setup, swash, and runup. Coastal Engineering 53: 573588. Tuan, T.Q., Stive, M.J.F., Verhagen, H.J. & Visser, P.J. 2008. Process-based modeling of the overflow-induced growth of erosional channels. Coastal Engineering 55: 468-483. Wise, R.S., Smith, S.J. & Larson, M. 1994. SBEACH: Numerical model for simulating storm-induced beach change, CERC, US Army Corps of Engineers, Vicksburg.

T03_08_02_D03_1_V3_0_p13.doc 57

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Annex III Sources of Uncertainty

T03_08_02_D03_1_V3_0_p13.doc 58

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Page intentionally blank

T03_08_02_D03_1_V3_0_p13.doc 59

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Sources of Uncertainty
In these guidelines there are references and explanations of the uncertainties occurring in the process of flood hazard mapping. This part shall give additional information on these uncertainties. One of the major sources of uncertainties in coastal flood hazard mapping has been identified as the coastal response or the coastal changes that occur during an event. But there are many other sources of uncertainties in the modelling process. Thus a brief and general overview of the sources of uncertainties shall be given. The overview is not restricted to coastal-flood-hazard mapping but deals with flood hazard mapping in general so there may be parts that make reference to sources of uncertainties that are more of a concern when creating a hazard map for river basins.

Definition
According to Pappenberger et al. (2005) the term uncertainty is described as follows: Uncertainty - A general concept that reflects our lack of sureness about someone or something, ranging from just short of complete sureness to an almost complete lack of conviction about an outcome. Uncertainty Analysis (Model) - Assesses the uncertainty in model outputs that derives from uncertainty in structure, parameters, boundary conditions and evaluation data

Sources of uncertainty
Uncertainties arising from the creation of flood hazard maps can be divided into three types: data uncertainty, model uncertainty, Parameter uncertainty. From each element, contributing to the flood disaster chain, arise uncertainties, which can be assigned to at least one of the above types of uncertainty. The following table displays the steps of creating flood hazard maps and shows likely sources of uncertainty and the assigned type:

T03_08_02_D03_1_V3_0_p13.doc 60

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Table 1: Sources and types of uncertainty (Source: FLOWS (Homepage), modified) Variable Source of uncertainty Type of uncertainty Annual discharge Rainfall runoff modelling Model uncertainty Wave modelling Model uncertainty Selection of distribution function Model uncertainty Parameters of the statistical distribution Parameter uncertainty Short or unavailable records Data uncertainty Measurement errors Data uncertainty Levee failure Measurement errors of levee geometry Data uncertainty Breach modelling Model uncertainty Levee parameters (geometry, substrate, Parameter uncertainty breach width, turf) Water stage Model selection (1D or 2D model) Model uncertainty Steady or unsteady calculation Model uncertainty Frictional resistance equation Model uncertainty Channel roughness Parameter uncertainty Channel geometry Parameter uncertainty Sediment transport and bed forms Data uncertainty Debris accumulation and ice effects Data uncertainty Flood damage Dependence on water stage Model uncertainty Dependence on water stage Parameter uncertainty Land/building use, value and location Data uncertainty Content value Data uncertainty Structure first-floor elevation Data uncertainty Flood warning time Data uncertainty Public response to a flood (flood evacuation Data uncertainty effectiveness) Performance of the flood protection system Data uncertainty (possibility of failure below the design standard)

T03_08_02_D03_1_V3_0_p13.doc 61

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Dealing with uncertainty


Several approaches have been developed to assess uncertainty. They can be divided by their need for evaluation data. The following figure provides a general guidance of which method can be used, depending on the existent evaluation data:

Figure 1: Decision tree for uncertainty analysis tools (blue boxes represent the questions to derive a decision for an uncertainty method, yellow boxes show the major classifications of several uncertainty methods and orange boxes stand for individual methods or small sub-groups of those) ( from Pappenberger et al.(2005))

For the creation of flood hazard maps usually no evaluation data is available. Hence, regarding the diagram, the applications of Monte Carlo methods or error propagation methods seem to be reasonable. A short introduction of both methods is given as follows:

Error propagation equation


The Error propagation equation deals with the normal distributed errors of the underlying formula. The goal is to asses how the quantified uncertainties in model inputs propagate during the model calculations and how it affects the results. The following simple example shows how it works: Regarding the following equation z = x + y+ With the appropriate deviation of the input parameter x and y the following equation leads to the deviation of z:

This method can be used to quickly evaluate simple calculations. However, this shows its strength and limitation at the same time. While simple calculations are easy to execute, it is hard to apply this method to complex calculations. For this reason the application of error propagation does not seem to be favourable to assess uncertainties in the run of flood hazard map creation.

T03_08_02_D03_1_V3_0_p13.doc 62

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Monte Carlo method


The Monte Carlo analysis can be used to gain information about the distribution of the output parameters. This can be achieved by applying random drawing of input parameters and examine the results. The variation of the input parameters and the repeated application of the model can be a very time consuming task. Therefore models should be simplified in order to reduce computation time.

Method suggested by the FLOWS Project


Another approach to consider uncertainty is shown by the FLOWS Project (2005). Instead of the examination of uncertainty after calculation FLOWS recommends as a best practice to determine a level of accuracy for which appropriate methods should be used to match the accuracy level. Figure 2 shows a sample application for low lying areas which was originally published in the FLOWS Technical Report (2005). It shows the required accuracy depending on the location of the site and the vulnerability to floods.

Figure 2: Required accuracy (for application to low-lying areas only) (from FLOW, 2005)

T03_08_02_D03_1_V3_0_p13.doc 63

26 September 2008

Task 3 Flood Hazard Mapping Guidelies D3.1 Contract No:GOCE-CT-2004-505420

Presentation of uncertainty in flood hazard maps


As mentioned before publication of flood hazard maps requires remarks to the confidence and the likely uncertainty. An easy way to do this is proposed in Figure 3, where the calculated uncertainty is demonstrated as an area of uncertain flooding. This can nowadays easily be done using GIS.

Figure 3: Scheme of cross section and overhead view showing application of error estimates (from Jones (1998))

References
JONES J. L.; HALUSKA T. L.; WILLIAMSON A. K.; ERWIN M. L., (1998) Updating Flood Inundation Maps Efficiently: Building on Existing Hydraulic Information and Modern Elevation Data with a GIS U.S. Geological Survey Open-File Report 98-200. FLOWS Project http://flows.wb.tu-harburg.de/index.php?id=587 (last visit January 2008) FLOWS Technical Report Techniques for Modelling and Mapping Flood Risk in Low-Lying Areas (2005) Internet Resource http://flows.wb.tuharburg.de/fileadmin/BackUsersResources/flows/Downloads/WP1/WP_1b_iii__Technical_Repor t.pdf (last visit January 2008). PAPPENBERGER, F.; HARVEY, H.; BEVEN, K.; HALL, J.; ROMANOWICZ, R.; SMITH, P.; Report: Risk & Uncertainty Tools and Implementation Available on the internet: http://www.floodrisknet.org.uk/ files/2005/2005-frmrc-rpa9-ir.pdf (last visit January 2008)

T03_08_02_D03_1_V3_0_p13.doc 64

26 September 2008