Professional Documents
Culture Documents
Integrated Flood Risk Analysis and Management Methodologies FLOODsite
Integrated Flood Risk Analysis and Management Methodologies FLOODsite
net/publication/272827005
CITATION READS
1 158
4 authors, including:
Paul Sayers
University of Oxford
79 PUBLICATIONS 1,482 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Paul Graham Samuels on 27 February 2015.
Co-ordinator: HR Wallingford, UK
Project Contract No: GOCE-CT-2004-505420
Project website: www.floodsite.net
FLOODsite Final Report: Volume 1
Contract No:GOCE-CT-2004-505420
DOCUMENT INFORMATION
DOCUMENT HISTORY
Date Revision Prepared by Organisation Approved by Notes
17/12/08 0_0_P1 Paul Samuels HR Wallingford First partial draft
17/02/09 1_0_P1 Paul Samuels HR Wallingford First full draft for comment
14/04/09 2_0 Paul Samuels HR Wallingford Final version with comments
integrated
ACKNOWLEDGEMENT
The work described in this publication was supported by the European Community’s Sixth Framework
Programme through the grant to the budget of the Integrated Project FLOODsite, Contract GOCE-CT-
2004-505420.
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.
EXECUTIVE SUMMARY
FLOODsite was the largest ever EC research project on flood risk management, with an EC grant to
the budget of nearly €10 Million complemented by supporting national funds. The project, which
started in 2004 and was completed in February 2009, involved over 200 researchers from 13 countries
including many of Europe’s leading research institutes and universities. The project was
interdisciplinary integrating expertise from across the environmental and social sciences, as well as
technology, spatial planning and management. All professionals involved in flood risk management
practice now need to consider the question of what actually constitutes “integrated” flood risk
management and to prepare society at large for the change in policy from one of flood defence to flood
risks being managed, but not eliminated. The circumstances in which the research will be
implemented are changing with firstly the entry into force of the European Union Directive on the
assessment and management of flood risks and secondly the concern of the potential for increased
flood hazards arising from climate change as set out in the IPCC Fourth Assessment Report.
The scope of flood risk management is broad with FLOODsite providing incremental contributions to
knowledge and understanding. FLOODsite has examined specific aspects of flood risk management
on risk analysis and assessment methods, appropriate policy and instruments, and event management
and decision support. The research concentrated on topics identified in the original call for research
and took account of the broader context of national and international research projects. During the
research the project team made links with over 80 other research projects and programmes. The
research tasks in FLOODsite were grouped in four of the project “Themes” and covered:
1. Flood risk analysis methods
2. Sustainable Flood Risk Management - Innovative Mitigation and Flood Risk Management
3. Frameworks for technological integration
4. Pilot studies
This final report on the project science serves as a summary of all the activities and results over the
full duration of FLOODsite, with the first volume covering the results of Themes 1 to 3 and the second
volume covering the pilot studies in Theme 4. The content of many of the project science reports are
outlined in task Fact Sheets which are available for download from the website pages that describe
each task. In addition, most tasks have produced an Executive Summary report on their area of
investigation providing an overview of their activities, principal results and remaining gaps in
knowledge. The project partners have been active in presenting and publishing project advances to
scientific, technical and professional communities; the Final plan for using and disseminating the
knowledge identifies 650 publications at the end of the FLOODsite project:
• 155 Journal papers
• 16 Contributions to books
• 300 Conference papers
• 29 Institutional reports and theses
• 31 Posters
• 119 FLOODsite reports on the project science.
FLOODsite strengthened the world-leading position of Europe in knowledge and practice for flood
risk management, which was clearly demonstrated at the FLOODrisk 2008 Conference in Oxford; the
project pilot studies have drawn together the project knowledge and provided feedback from flood risk
managers of rivers, estuaries and coasts as well as from local stakeholders. The use of pilot sites and
collaboration with executive agencies in several countries helped to ensure that FLOODsite results are
of real value, practicable and usable. Exploitation of the project results will follow primarily through
the subsequent development and production of professional standard software to encapsulate the
results. This is beyond the scope of FLOODsite.
CONTENTS
Document Information ii
Document History ii
Acknowledgement ii
Disclaimer ii
Executive Summary iii
Contents v
1. Introduction 1
1.1 Background 1
1.1.1 Terms of reference - The FP6 call and research proposal 1
1.1.2 Scope and ambition of FLOODsite 1
1.1.3 Organisation and governance of the research 2
1.1.4 Specific Objectives of the FLOODsite Themes 3
1.2 The European “Floods” Directive 4
1.3 Scope and layout of the report 6
REFERENCES 72
Tables
Table 2.1 Categorisation of measures and instruments (after Olfert, 2007) 11
Table 2.2 Categorisation of flooding hazards and potential impacts (after Samuels, 2005) 12
Table 3.1 The framework of criteria and indicators for a full assessment 30
Table 5.1 Hierarchy of knowledge and activity for flood risk management 50
Table 5.2 Future flood management strategies for the Netherlands 54
Table 8.1 Projects with links to three or more FLOODsite tasks 70
Figures
Figure 1.1 Overall project structure 2
Figure 1.2 Project governance structure 3
Figure 1.3 Implementation of the Floods Directive 5
Figure 3.1 Integrated frameworks for flood risk analysis and management 14
Figure 3.2 Water level statistics and extrapolation to extremes using Bayesian (Bayes) against
Maximum Likelihood (ML) methods 16
Figure 3.3 Methodology to generate a coastal flood hazard map 17
Figure 3.4 Different types of flood defence assets 18
Figure 3.5 Photograph taken during the large-scale model tests on dike breaching due to wave
overtopping 19
Figure 3.6 Components of the RELIABLE software 20
Figure 3.7 Schematic for mapping risk to life (Priest et al., 2007) 22
Figure 3.8: Three dimensions of strategies for FRM 26
Figure 3.9 Schematic overview of the method for developing and assessing long-term flood risk
management strategies in view of uncertain futures 28
Figure 3.10 Sample graphical system output for FFG and FFT 32
Figure 3.11 Application of the ORCHESTRA web tools to the Gard Region of France 35
Figure 3.12: Framework of flood risk management (Schanze, 2005) 37
Figure 3.13 Overview of processes steps within the methodological framework 38
Figure 3.14 DSS Tools for long term planning: Example screen shots from the three pilot sites 39
Figure 3.15 Example of multi-staged decision process used within the Thames DSS 39
Figure 3.16 Methodological framework for flood event management DSS 41
Figure 3.17 Technological framework of the DSS for the Thames and Schelde Estuaries 41
Figure 3.18 Display of evacuation times for Thamesmead 42
Figure 3.19 ESS screen showing the time of inundation for a breach scenario ‘Rilland’ 43
Figure 3.19 Flood Hazard mapping for the project site in Nice 44
Figure 3.20 Routes of evacuation during flooding event 44
Figure 3.21 Uncertainty analysis framework – documenting, recording and challenging uncertainty 46
Figure 3.22 The generalized framework of the UNEEC method. 47
Figure 5.1 Links between project science in Theme 1 and the pilot sites 51
Figure 5.2 Links between project science in Themes 2 and 3 and the pilot sites 52
Figure 5.3 Thames Estuary flood risk for a range of climate change scenarios from Task 24 56
1. Introduction
1.1 Background
1.1.1 Terms of reference - The FP6 call and research proposal
In late 2002, the European Commission Directorate General for Research issued the first call for
research in the Sixth Framework Programme priority on Global Change and Ecosystems with the Call
Identifier FP6-2002-Global-1. The work programme, paragraph 1.1.6.3.IV.2.b, called for one of the
two “new instruments” - an Integrated Project (IP) or Network of Excellence (NoE) - to tackle the
following priority topic:
“Integrated strategies and tools for hazard, vulnerability and risk assessment, prevention and
mitigation of flood risks in the river basin, coastal zone and the estuaries. Development of
innovative design of sustainable flood defences and risk mitigation measures. Operationalisation of
methods and technologies developed as well as their efficiency and cost of implementation.
Understanding and prediction of coastal flood related extreme events, their interaction and
synergetic effects with coastal morphodynamics. Exchange and dissemination of related
information to user communities.”
The FLOODsite partners submitted a proposal in April 2003 for an IP to respond to this call and
formed a consortium to undertake the research. The Consortium negotiated a programme of work
with DG Research and the FLOODsite project was one of the first IP’s to commence work on 1st
March 2004, with project duration of 5 years.
In May 2005, a major international conference was held on flood management at Nijmegen the Third
International Symposium on Flood Defence (van Alphen et al, 2006). This drew an audience of over
300 researchers, experts and professionals engaged in flood management worldwide. The move from
flood protection and defence to integrated flood risk management and “making room for rivers” was
evident in many national contexts and in the policy of the EU as presented by DG Environment. This
change in philosophy is at the heart of the FLOODsite objectives and research plan. FLOODsite has
explored the proper system representation and societal understanding of flood risk management as the
context within which the scientific and technological research advances will be implemented. During
our research progress has been made on many issues including:
• Identification, design and appraisal of sustainable flood risk mitigation measures
• Improved understanding of complex flood defence systems, their failure modes and their
interaction with morphodynamic processes.
• Consistent and integrated flood risk assessment and management procedures for all sources of
flooding, including reviewing methodologies for the preparation of a European Flood Hazard Atlas
• Understanding and statistical appraisal of weather and marine extremes which generate flood
hazards and the hydrometeorology of flash flood hazards in small basins
• Improved understanding of the vulnerability of the public and assets to flood damage
• Improved disaster preparedness, evacuation and emergency management procedures and social
resilience
The research has also provided methods that have enabled the exploration during the project of how
flood risks may develop over time due to changes in the climate, demographic, economic and social
development and of the possibilities to influence these changes. This has been achieved through
scenario-based analyses in the context of real-life applications.
As described in Section 1.2 below a fundamental change in the external context over the past year is
the entry into force of the EU “Floods” Directive. The wording of the Directive allows for
considerable flexibility in its implementation in national law, respecting the national and regional
context in which flood risk management occurs through the Subsidiarity Principle. Accordingly
FLOODsite has provided a set of Integrated Methodologies for use in flood risk management practice
not a single methodology, thus respecting the diversity of practice in each Member State
The project governance structure included a Management Team responsible for the day-to-day running
of the project, and three boards to provide assessment, review and advice, see Figure 1.2. There were
two expert advisory boards (on scientific and technical issues and on application and implementation)
and an oversight project board under independent chairmanship which reported annually to DG
Research. In addition the FLOODsite Consortium reported annually on the scientific progress of the
project to DG Research who commissioned an independent evaluation of the reports. The Scientific
and Technical Advisory Board (STAB) reviewed the project science on several occasions, taking
reports from and questioning the task leaders; the STAB then recommended means of improving some
Task outputs. The Applications and Implementation Advisory Board (AIB) also met annually, making
recommendations on the take-up of the project science. The AIB recommendations resulted in the
Management Team organising a dissemination meeting with WG-F (see section 1.2 below) and
producing fact sheets which provide an easy entry into the project outputs for practitioners.
Figure 1.2 Project governance structure
The Floods Directive applies to the whole Community territory, and therefore to flood risk
management in both rivers and coastal areas. The new Directive is aligned with the Water Framework
Directive (Directive 2000/60/EC; or the “WFD”). It sets out the need for assessments, maps and plans
that cover the river basin district including the borders of the river basins, sub-basins and where
appropriate associated coastal zones through:
• Preliminary flood risk assessment
• Flood risk maps
• Flood risk management plans
It is clear that FLOODsite is directly relevant to the needs of the Floods Directive and this was
identified in the explanatory memorandum to the Commission’s proposal for the Directive:
“European research policy has been supporting research into different components of flood risk
management since the early 1980s through successive Framework Programmes. The Sixth
Framework Programme is supporting the largest ever EU flood research project, “FLOODsite”,
which is developing integrated flood risk analysis and management methods. The proposed 7th
Framework programme will continue to support research on flood risk assessment and
management.”
Working Group F (WG-F) has been constituted by DG Environment under the Common
Implementation Strategy (CIS) of the WFD and has two primary tasks:
• information exchange for example on research outcomes and current good practice, and
• support for the implementation of the Floods Directive within the CIS framework.
The FLOODsite coordinator has participated in WG-F meetings as one means of communicating the
project results to those with responsibility for flood risk management in practice.
Implementation Timetable
It is recognised that the Floods Directive requires management in similar geographical units to the
Water Framework Directive (WFD) and its river basin management plan process. Thus the
Commission is encouraging a harmonisation of reporting processes between the two directives where
this is appropriate. The implementation timetable is set out in the Floods Directive is:
• the preliminary flood risk assessment should be complete by 22 December 2011;
• the flood hazard and flood risk maps are to be complete by 22 December 2013
• the flood risk management plans are to be prepared and published by 22 December 2015
Article 14 requires the revision and updating of the preliminary flood risk assessment by 22 December
2018, the flood hazard maps and of the flood risk maps by 22 December 2019 and revision of plans by
22 December 2021 and then all updated on a six-yearly cycle. This is illustrated in Figure 1.3 below.
1
This Figure was provided by DG Environment (Unit D.2)
In the Floods Directive, the phrase “flood risk” means the combination of the probability of a flood
event and of the potential adverse consequences for human health, the environment, cultural heritage
and economic activity associated with a flood event. This definition is now included in the second
edition of the FLOODsite Language of Risk (FLOODsite Report: T32-04-01) as the first edition did
not specifically define the phrase. The nature and analysis of the means of this “combination” of
probability and consequences of a flood formed part of the research agenda for FLOODsite.
The interactions can be assessed, for example, in the DPSIR framework as adopted by the European
Environment Agency (EEA) and as used in the UK Flooding Foresight review (OST, 2004). The EEA
website gives the definition as follows (see http://glossary.eea.europa.eu/EEAGlossary/D/DPSIR ):
The causal framework for describing the interactions between society and the environment adopted
by the European Environment Agency: driving forces, pressures, states, impacts, responses
(extension of the PSR model developed by OECD).
The research in FLOODsite has challenged the current approaches to risk analysis, assessment and
management and highlighted the need for:
• Risk analysis – that includes hazard definition; whole system model integration; software
integration; the identification and handling of uncertainty.
• Risk assessment – that includes normative perspectives; multi-criteria assessment; acceptability and
tolerability; disciplinary integration;
• Risk management - that implements a portfolio of measures from pre-event, during event and post
event actions in association with supporting activities of monitoring and resourcing, and embedded
within wider societal plans.
• Exploration of the future – through the use of scenarios and consideration of strategic alternatives;
structured approaches to the assessment of sustainability through consideration of decision
robustness and flexibility.
However, the multiplication “x” is really a combination across all floods, and so an alternative
description in the Second Edition of the Language of Risk (FLOODsite Report: T32-04-01, 2nd Ed.)
risk is defined as:
Or more generally:
The interpretation of this simple relationship is more complex and much richer than first appears. For
example:
• Probability (of a flood event) - is the chance (or in Bayesian terms the measure of our strength of
belief) that a given event will occur (for example a flood depth exceeding 1m at a given place in
the floodplain). We consider probability over a specific time-frame (1 tide, 1 month, 1 year or an
average human life-time etc). In determining the chance of the flood event both a source of the
flood and pathway, from the source to the given place on the floodplain, must exist. The concept
of probability can be further extended to consider the chance of a receptor being exposed to
flooding and experiencing adverse consequences.
• The consequence term can be sub-divided into two key components – exposure and vulnerability.
o Exposure - Quantification of the receptors that may be influenced by a hazard (flood), for
example, number of people and their demographics, number and type of properties etc
o Vulnerability – Characteristic of a system that describes its potential to be harmed.
Vulnerability can, however, be further sub-divided into a combination of:
In understanding the likely consequences of a flood it is therefore important to understand the nature
of the receptor and how it will be impacted by a flood. For example, some receptors, such as
residential properties, can be considered “static”, whereas receptors such as people and cars may be
“dynamic”, and may or may not be present at the time of a flood. This may reflect actions taken to
evacuate or simply due to the time of day the flood occurs (rush hour, night time etc). This dynamic
behaviour can change the chance of a receptor being present and hence the exposure to a flood. Often
receptors can initiate secondary sources of risk. For example, pollutants may be released from a
flooded sewerage works (leading to public health issues), water supply maybe disrupted, roadways
blocked etc. More elaborated methods are starting to emerge that can deal with these interactions;
however these are not considered here (Tapsell, 2008). The valuation of socio-economic flood
damages has been examined in FLOODsite Task 9, which has provided guidance on the assessment of
socio-economic flood damages across Europe (Messner et al, 2007). Finally the overall consequences
of flooding on receptors will depend also their “resilience” which covers the capacity to return to
normal after the flood. FLOODsite Task 11 explored factors that comprise social resilience
(FLOODsite Report: T11-07-12).
FLOODsite took place in a time of change in policy and practice in flood risk management. Thus the
research was delivered into a different context from that existing at the conception of the FP6 work
programme and the research project proposals. Professionals involved in flood risk management are
being forced to consider the question of what actually constitutes “integrated” flood risk management
and to prepare society at large for the change in policy from one of defence to flood risks being
managed but not eliminated. The Floods Directive provides a framework for doing so and has
particular reference to the need for collaboration in transnational basins.
Within flood risk management “risk” has generally been understood as a combination of probability
and consequence, interpreted in a comprehensive way (see Section 2.2 above). A more formal and
detailed syntax describing “risk” enables a more subtle and useful understanding of the drivers of risk
to be developed. In particular, although current approaches seek to be support “risk-based” and
“sustainable” decision making they are often limited in their consideration of the sources, pathways
and receptor impacts of flooding and often fail to integrate within broader spatial planning and social
policies. “Risks” are also typically evaluated in deterministic terms with limited effort devoted to
understanding uncertainty, a position that runs counter to robust decision making.
The Floods Directive and its close association with the WFD mean that the impacts of flood risk
management measures have to be considered broadly, rather than focussing narrowly on the
immediate effects on water level. Integrated flood risk management (IFRM) is a comprehensive and
continuous process of analysis, assessment and action. It considers the external pressures placed upon
the flood risk system by climate and societal change; the state of the flooding system (including all the
sources of the flood hazard and the various pathways that link them through to the receptors); as well
as a full range of potential impacts and the possible responses to mitigate them. Most importantly
IFRM demands an integration of the flood risk management process with wider societal demands and
aspirations. As such, IFRM can be seen as distinct from the primarily reactive approaches that have
often characterised traditional flood defence based paradigms and the often sectorial context of current
flood risk management approaches.
The challenge of achieving IFRM in practice can not be underestimated. It will depend upon
improved and more efficient tools and techniques (providing improved functionality to explore risk
and a richer, more useful and useable evidence on risk). It will also crucially depend upon the
common desire across all stakeholders (researchers, practitioners and policy makers) to improve their
performance. The science and development within FLOODsite (as summarised in this report) provides
an important step towards this goal.
Instruments are, in contrast, not direct physical interventions in the environment. Policy instruments
are intended to influence the attitude or actions of others than the immediate responsible authorities’
managers themselves. Three main groups of instruments can be distinguished, namely
communication, financial and regulation instruments. Communication may, for example, enhance the
people’s risk awareness and preparedness. Financial instruments may influence people’s investments
or may encourage them to flood-proof their property. Regulatory instruments, such as land-use
regulations, allow or prohibit certain activities at all.
In addition to these measures and instruments, actions taken in a flood event can also reduce the harm
and damage experienced and so be part of the overall risk reduction.
Table 2.2 Categorisation of flooding hazards and potential impacts (after Samuels, 2005)
Type of flood Type of event Number of Geographic Type of damage Type of mitigation
or frequency properties affected Distribution
River Flash Summer 1-500+ • Small steep river • Inundation damage to buildings and contents, vehicles • Flood warning service
Flood storms catchments written off • Community
• May trigger other • Buildings and bridges destroyed preparedness
hazards e.g. • Campsites and caravan parks particularly vulnerable • Land use regulation
landslides and • Possible significant loss of life (e.g. 300 in Sarno IT, • Land management
mudflow 1998). practice
Lowland river Winter season 1-500+ • River flood plain • Inundation damage to buildings and contents, vehicles • Hard defence measures
flood (Basins rainfall, spring Major towns and • Flooding in one or written off, possible structural damage. • Flood storage
up to 10000 snowmelt cities may give more river • May lead to locally significant displacement of • Flood warning service
km2) larger numbers catchments at any population • Land use regulation
one time • Deep flooding possible behind raised defences. • Emergency evacuation
• Disruption to critical national infrastructure (water,
power, telecommunication, transport)
Major river Seasonality 10,000+ • River flood plains • Inundation damage to buildings and contents, vehicles • Hard defence measures
basins up to depends upon Major towns and • Trans-national basins written off, possible structural damage. • Flood storage
250000 km2 basin and cities give large • Flooding in one or • Major infrastructure disrupted • Flood warning service
climate numbers more river • May lead to large displacement of population (approx • Land use regulation
catchments at any 250,000 for Rhine floods in 1995). • Emergency evacuation
one time • Deep flooding possible behind raised defences.
• Flash floods in • Economic loss for event significant in terms of national
headwaters economic growth rate (> 0.1% GDP)
Coastal (surge Winter season 1-500+ • Coastal fringe / • Deep flooding possible behind raised defences. • Hard defence measures
and wave) Major towns and estuary / mapped • Inundation damage to buildings and contents • Soft engineered
cities may give flood risk zone • Possible loss of life (2000+ in 1953) defences
larger numbers • Flooding possible • Major disruption possible from storm surges • Storm-tide warning
over large lengths of • Single storey dwellings, campsites and caravan parks services
coast (e.g. 1953 particularly vulnerable • Land use regulation
North Sea storm) • Vehicles written off • Emergency evacuation
Tsunami Any time but 1-10,000+ • Coastal zones and • Inundation damage to buildings and contents, vehicles • Contingency planning
infrequent depending on hinterland written off, • Spatial planning
location • Widespread agricultural damages • Warning systems
• Destruction of buildings, major infrastructure disrupted • Emergency evacuation
• Potential significant loss of life
Type of flood Type of event Number of Geographic Type of damage Type of mitigation
or frequency properties affected Distribution
Groundwater Prolonged Small clusters • Certain geologies • Inundation damage to buildings and contents • Flood proofing
seasons of (limestone, chalk, • Especially basements • Pumping from
rainfall sandstone etc) • Long duration inundation damage (weeks or months) properties
• Outside main river • Land use regulation
floodplain
Pluvial floods - Prolonged Isolated or small • Hill slopes • Inundation damage to buildings and contents • Flood proofing
overland flow heavy rainfall clusters • Outside main river • Damage to agriculture • Pumping from
floodplain • Possibly vehicle accidents properties
• Land use regulation
Pluvial floods - Intense storms Isolated or small • Anywhere adjacent to • Inundation damage to buildings and contents • Routine maintenance
storm and – especially in clusters roads or urban areas • Possibly vehicle accidents of watercourses and
highway summer Urban areas drainage systems
drainage • Sustainable drainage
systems
Pluvial floods - Common 1-20+ • Outside main river • Inundation damage to buildings and contents • Land use regulation
minor water- floodplain • Hard defence measures
courses • Blocked culverts and • Routine maintenance
bridges of watercourses
Dam break Any time but 1-1,000+ • Specific river valleys • Inundation damage to buildings and contents, vehicles • Monitoring on-site.
infrequent • Outside main river written off, • Risk management by
floodplain • Destruction of buildings, destruction of bridges, major owners
infrastructure disrupted • Contingency plans
• Potential significant loss of life within 25 km of dam • Emergency evacuation
Water mains Any time Small clusters • Anywhere • Inundation damage to buildings and contents • Asset inspection and
burst • Especially basements renewal by water
service providers
Sewerage e.g. Any time Isolated or small • Anywhere • Inundation damage to buildings and contents • Asset inspection and
by blockage or clusters • Outside main river • Foul sewage contamination renewal by water
collapse floodplain • Especially basements service providers
Figure 3.1 below shows the interaction of these concepts in a simplified way.
Figure 3.1 Integrated frameworks for flood risk analysis, assessment and management
The remaining sub-sections summarise some key results following the risk source-pathway- receptor-
consequence model; further information is available from the FLOODsite website pages for the
various tasks, see www.floodsite.net.
Task 1 aimed to improve understanding of the hydrometeorological processes associated with flash
floods. The approach adopted was to undertake both a detailed analysis of storm events and of their
hydrological responses. Meteorological analyses were based on both observations (radar, raingauge
data) and results of high-resolution simulated rain fields. In total, four different storms producing flash
flood were analyzed: three in Southeast France and one in the Eastern part of the Italian Alps. These
analyses highlighted some common features and explained the steadiness of the storms that lead to
locally intense precipitation: the role of the orography, favourable synoptic conditions (low level
convergence; slow evolving convective system). This work identified the need to develop and
generalize radar observation for flash-flood understanding and the capacity of the high-resolution
meteorological model to reproduce such intense precipitation events. Several catchments (in Italy,
Netherlands and Spain) were selected to study hydrological processes associated with flash-floods; see
report T01-07-01 2. The role of initial soil moisture has identified and based on theoretical simulation
with the physically based model HsB, the link between the characteristics of the hillslopes (slope,
depth …) and the dynamic of the flood has been identified. Efforts were also focused on the
conceptualization and the modelling the hydrometeorological processes involved in flash flooding for
ungauged catchments.
FLOODsite Task 2 dealt with extremes and has used available methods (Bayesian techniques, trend
analysis, Singular Spectrum Analysis, Joint Probability Analysis, Regional Frequency Approach,
Canonical Correlation Analysis, Bootstrapping, Neural Network) to analyse limited samples of
extreme data, to find correlations (both spatial and in time), and to provide improved probability
2
Annex 1 lists the FLOODsite reports; these are available from www.floodsite.net
density functions for extreme data. By their nature only limited observations of extremes were
available and so advanced analysis methods were needed and these were adapted and further
developed. The work of Task 2 encompassed both data preparation and the evaluation of fit
parameters (those probabilistic model parameters estimated from comparing the available observations
against model predictions). The research has specifically included the criteria to define extremes and
how to select a homogeneous sample to perform the corresponding statistical analyses. This becomes
particularly important in view of the limited number of extreme events actually recorded, due to the
exceptional character of such events. The work in this task also compared critically the various
probabilistic distributions available to characterize these extremes.
The use of joint probabilistic distributions for extreme values is considered to be an important topic
since many of the recently observed “failures” of both man-made and natural systems have been
associated to the simultaneous occurrence of extreme (or even relatively high) values of more than one
variable or “driver”. Some suggestions on the common use of such joint probability distribution
functions and some advanced guidance for a better assessment have also been produced within the
task. The uncertainty should be estimated explicitly, including as many sources of uncertainty as
practical, to avoid misunderstandings in associated decision making.
The assessment of risk (due to flooding or erosion) in regional terms requires the joint distribution of
extremes at different (neighbouring) locations. The total hazard level depends on the degree of spatial
correlation between the considered points. Inclusion of this spatial dimension is also recommended to
capture better the dynamic structure of a storm event; this results in a more reliable prediction of
regional risks. This could also lead to a reduction in uncertainty for marginal (point-wise) extreme
distributions and the resulting estimates. The joint probability distribution concept should also be
applied to evaluate hazards in many natural (e.g. beaches and dunes) or man-made (e.g. dike) barriers.
This is because failure modes or the corresponding limit state equations are normally dependent on
more than one variable (e.g. waves and storm surges for coasts or river level and velocity for rivers).
This involves the use of joint distributions with suitable consideration of correlations. The
applications described in the guidance document from Task 2 illustrate some of the difficulties
associated with analysis of extremes, particularly for cases with more than one variable.
Understanding the assumptions and interpreting the obtained results are important for extremes
analysis.
Task 3 of FLOODsite reviewed existing methods for mapping flood hazards in rivers and at the coast.
From that, a methodology for mapping (coastal) flood hazards was drafted (Figure 3.3), including
information on the uncertainties of flood mapping. Examples from other hazard mapping initiatives
and projects have been considered (e.g. DEFRA, 2006; FEMA, 2003) for drafting this concept and
links have been made to parallel European activities (EXCIMAP, 2007). The concept was tested and
further developed using data from the pilot site ‘Ebro delta coast’. The recommendations (FLOODsite
Report T03-08-03) are particularly relevant for coastal sedimentary environments, where the impact of
the storm can produce a significant morphodynamic response that will interact with the storm and can
affect the scale of the flooding (enhancing or reducing its effect). The guidelines concentrate on
coastal flood hazard mapping (because other EC Actions have dealt with fluvial flooding) and were
implemented for a practical case in the Ebro delta pilot area, where the emphasis was on the impact of
including the coastal morphodynamic response.
morphodynamic
feedback
wave data water level
data
Runup
R2%
overtopping
Q
inundated area
surface & depth
The principal result from Task 4 is FLOODsite Report T04-06-01: Failure mechanisms for flood
defence structures; however, other outputs from the individual areas of research also have direct
relevance to industry practice. A comprehensive review of failure modes of different flood defence
assets (Figure 3.4) has been performed and limit state equations (LSE) for further use in reliability
calculations have been formulated. By collating a definitive collection of failure modes, this work
supports system risk modelling, by helping to ensure common agreement on failure modes, their
representation and use; the report contains over 70 separate failure mode descriptions. The document
is structured to allow for easy updating and addition of failure modes hence will provide a working
framework that can be used and updated into the future as knowledge on failure modes develops.
Information is presented using a standard template format for each type of structure and load
combination. The scientific knowledge from Task 4 has been delivered through the FLOODsite work
on flood defence infrastructure reliability in Task 7.
The research in Task 5 on coastal morphology has led to a number of new developments. These
include a stochastic model of beach plan shape variability, a regional model for regional scale
changes, a rapid coastal evolution model, beach overwash and dune erosion models. These have all
been used on actual sites, in order to illustrate their application. The stochastic model has been applied
to Christchurch Bay on the south coast of the UK. The regional model methodology for assessing the
coastal vulnerability to storm impacts has been applied to the Catalan coast, Spain. Storms on the
Catalan coast have been classified in terms of their inundation and erosion potential. The methodology
has been applied to the longest existing wave record in the Catalan coast for two different coastal
types. The rapid coastal evolution model is a fully integrated, dynamically linked coastal management
tool, GTI-SEAMaT, which is illustrated through an application to the shoreline of Calabria in Italy.
For beach overwash and dune modelling an analytical model to simulate dune erosion and dune foot
retreat during severe storms was further developed and tested. Four different data sets on dune
erosion, originating from the laboratory and the field, were employed to validate this model. An
analytical approach was formulated to describe the response of a dune to wave impact and overwash.
The approach has been applied to sites in the USA and the Ebro delta, Spain. The principal
deliverable from Task 5 is the FLOODsite report T05-07-02.
Task 6 deals with breaching of sea dikes and river embankments. For coastal areas, based on a detailed
review of existing literature, detailed models have been given to describe the breaching process from
both sides of sea dikes either under wave overtopping or wave impact loads, respectively. Large-scale
hydraulic model tests on breaching have also been performed to support the developed models and
provide further knowledge on breaching processes under different types of loading (Figure 3.5).
Numerical models have been developed or further advanced to describe the breaching processes
(Stanczak, 2008; D’Eliso, 2007). For river embankments, analysis of embankment soil state and
associated processes have led to the development of the next generation of HR BREACH and BRES
predictive models. The BRES model was further developed to address cohesive as well as non
cohesive materials (Zhu, 2006); the HR BREACH model (Morris et al, 2008) was refined and
extended to include multiple zoned structures (both cohesive and non cohesive, head-cut and surface
erosion).
Figure 3.5 Photograph taken during the large-scale model tests on dike breaching due to wave
overtopping
As part of this work, a detailed review of the earlier European IMPACT project breach field and
laboratory data was undertaken. Model development in FLOODsite builds upon that and other data.
Development and testing of the HR BREACH model was also undertaken in conjunction with the
CEATI Dam Safety Interest Group breach modelling project team, and hence links wider international
expertise and practice to the FLOODsite European programme for breach. The FLOODsite Report
T06-06-03 provides a comprehensive review of the current state of the art for modelling breach
initiation and growth processes. The review addresses current state of the art for modelling breach
initiation and growth, including wave induced initiation. Definitions of breaching processes are
provided and issues of relevance to end users identified. Integration of breach models with reliability
or system risk models is also considered.
Flood defence systems consist of many components or “elements”, and the overall reliability of the
flood defence system depends upon the reliability of each component. In Task 7 flood defence
systems of flood-prone areas have been represented by fault trees which provide the potential chain of
events leading to overall failure of flood defence function. An example is given in Figure 2.4. Failure
of the subsystems (dike, dune sluice, levee) of the system leads to flooding of the polder area. The
subsystems all consist of “elements” (e.g. sea dikes or river embankments can be divided in different
sections). Failure of any of these elements of the subsystem “dike 1” leads to flooding of the
hinterland. All failure modes of each element have the potential to lead to the overall failure of the
flood defence system as a whole. The most important failure modes have been addressed and modelled
in Task 4 of FLOODsite. Decision-making for flood risk management relies on an understanding of
the likelihood of existing or proposed flood defences failing under given hydraulic loadings. This is
often referred to as ‘fragility’. For probabilistic risk assessment methods, a probabilistic measure of
the structural performance is required. This is typically expressed as a fragility curve relating
‘loading’ to ‘probability of failure’. Combined with descriptors of decay or deterioration, fragility
curves can be generated that enable the likely future performance of the structure to be described.
Task 7 has worked on improving reliability methods for flood defences. This has been initiated by
comparing existing reliability models for three pilot sites (Scheldt estuary, Thames estuary, and
German Bight coast) and subsequently using different tools for the same site to identify
inconsistencies. Task 7 has implemented the new knowledge for failure modes of flood defences into a
new reliability tool (Figure 3.6) which is available for use within flood risk assessment systems.
In Task 8, five software packages commonly used for flood inundation modelling were compared by
developing models for different pilot sites. The models also were used to answer a number of
additional research questions, for instance related to the importance of modelling breach growth and
the impact of wind. The results were used to list a number of conclusions and to derive guidelines on
model choice and appropriate application of models. Typically, 1D models are used to accurately
represent linear features such as channels using detailed cross sections. 2D models are more
commonly used to solve problems of flow over broad shallow areas such as flood plains. Task 8
considered several modelling issues including:
• the spatial resolution of the model,
• the influence of breaches in flood defences,
• the effects of strong winds on water levels
• the effects of land use and land cover on the hydraulic resistance to the flow
The guidance document FLOODsite report T08-09-03) gives the practitioner an understanding of the
most appropriate model types for various applications and the way in which 1D and 2D can be linked
together to solve channel and floodplain flow problems.
Task 10 has worked on innovative methods to understand, model and evaluate intangible flood
damages and other flood losses. This comprises a loss of life model, effects of flood warnings,
ecological issues, and multi-criteria analyses (MCA). Most of the activities in Task 10 deal with
riverine flooding since more data is available in this field. Loss of life modelling is calibrated with
data from UK and German rivers.
The Risk to Life research (Figure 3.7) took as a starting point the Flood Risk to People model
developed in the UK and assessed the applicability of this model for fluvial flood events in
Continental Europe, which tend to be more severe and life threatening. Data on flood events were
gathered from 25 locations across six European countries as well as data from an additional case study
in the UK. The most significant achievements include: identification of problems with the current UK
Risk to People model, e.g. when applied to cases with high Hazard and People Ratings, and
suggestions for refining the model; better understanding of the process of loss of life through analysis
of causes of flood-related deaths in Europe; reconfirmation of important factors leading to loss of life
that were statistically significant e.g. depth, velocity, building collapse, human behaviour; and
highlighting of the importance of systematic data collection. It suggests that a model for use
throughout Europe to predict numbers of deaths/injuries is problematic due to the lack of available and
reliable data. Overall, the research has increased the understanding of the factors surrounding
fatalities from flood events in the broader European context including the potential roles of building
collapse and human behaviour.
Task 10 also developed two modelling methodologies for quantifying the effects of different flood
warning lead times and mitigating actions with the associated reduction in damage costs. Firstly,
survey data was recalibrated using the original UK model which assesses the potential damages that
could be saved at an individual household level. This model was then refined for application within
the European context. Refinements were made on the basis of new data since the model’s inception
and the addition of the effect of evacuation of people on flood damages which was previously
excluded. This first method is for use with the detailed calculation of domestic household damage
savings. The second method, the “Flood Warning Response Benefits Pathway Model”, provides a
more flexible approach to incorporate several mitigation strategies. These methodologies are a
significant step forward in European flood risk management, since they allow the justification of
optimum flood warning systems on the basis of reduced flood damage costs. They also allow
quantification of the effectiveness of mitigation strategies for reducing damage costs. Guidance for
improving flood warning response among receptors is also provided alongside the methodologies.
This is a frequently overlooked but vital component of a successful flood warning system.
Figure 3.7 Schematic for mapping risk to life (Priest et al., 2007)
The research on toxic stress was based around the Optimal Modelling for Ecotoxicological
Assessment (OMEGA) framework. The framework was further enhanced and tested for use against
multiple pollutants in aggregating pollutant effects. The methodology allows the integration of water
quality and sediment modelling with the ecotoxicological response of receptors. The highest level of
output is given as a Potentially Affected Fraction (PAF) of the total population of interest. The
methodology is therefore intended as a tool for identifying areas particularly susceptible to flood
induced pollution. Significant achievements include:
• new knowledge on the application of Qualitative Structure-Activity Relationships (QSAR) in
environmental chemistry for acute aquatic toxicology;
• new software for combining flooding with sediment transport and sedimentation and combining
the prediction of spreading of different sources of toxic components;
• new insight on the specific toxic risks for groups of organisms;
• a better ecotoxicological model of acute and chronic toxic stress for a mixture of toxicants;
• a proof of concept model of a flooding of an dike ring in the Netherlands, calculating the
spreading and sedimentation of toxic components from a variety of sources (including flooding
specific sources); making it possible to take measures for high risk situations before a flooding
occurs or helping in determining where a clean up of contaminated soil is necessary after a flood.
The research Activity on GIS methodology for mapping risk to life focused on the development of a
GIS-based Multi-Criteria Analysis (MCA) methodology for the assessment and decision-making
process within flood risk management. The methodology focuses on economic, social and
environmental measures of risk, which are aggregated together to form a single risk metric. A raster-
based GIS-dataset for social, economic and environmental risk criteria was developed and various
MCA approaches were applied including: a disjunctive approach and an additive weighting approach.
Both these approaches were shown to be appropriate for use within the framework of MCA risk
mapping. An additive weighting approach was also developed which is applicable to show the spatial
distribution of benefits of certain risk reduction measures. The new methodology addresses
shortcomings in existing decision-making techniques and provides the possibility for evaluating
monetary and non-monetary flood risk in an integrated way, as well as showing their spatial
distribution and the uncertainties included. The research developed the FloodCalc software tool which
supports calculations and mapping of different damage and risk criteria as to how uncertainties can be
documented and dealt with.
Although in Task 11 the concept of risk perception was prominent in its title this concept proved to
have shortcomings. An important factor is the impossibility of perceiving a mental construct like
‘risk’. A more useful concept proved to be risk constructions, a broader term which also comprises
risk awareness and subjective perceptions of risk-related issues. The fact that people take any risk-
reducing measures implies that those people are both aware of the risk of being flooded and that they
attribute significance to the measures they take. In other words, they regard the measure they take as
meaningful in the hope that it will be effective. These perceptions and behaviours relate to people’s
social constructions of risk. Risk is neither simply attributed to a natural hazard nor an objectively
given constant. Rather, perceived risk is to be understood as being socially constructed in the sense
that people’s views, values, and belief systems all influence and possibly define it.
The research has led us to question many apparently compelling (and, from the perspective of flood
risk management, desirable) linear relationships. For example there are supposed links between
“being affected” and “having higher risk awareness” or “being affected” and “applying precautionary
measures”; however, real-life is much more diverse, contradictory and difficult to explain. On
investigating public involvement in flood risk management, we found a large number of residents
maintained that technocratic top-down approaches were dominant in flood management and judged
that “technicians” are the most influential actors in decision processes. Most of those surveyed did
not feel involved in the decision processes and tended to delegate responsibility – to agencies in
charge of flood prevention and mitigation. Thus, precautionary measures and flood defence are
primarily regarded as pertaining to public institutions. Public involvement allows the incorporation of
a factor that is often forgotten: local knowledge. Decrease in peoples’ adaptive capacity is triggered
by the progressive erosion of local knowledge; this knowledge concerns not only the locality
(morphology, dangerous areas, etc.), but also its management. It is fundamental for the residents not
only to understand how to behave in emergencies, but also to foresee and mitigate dangerous
occurrences. Residents lose the skill to recognise environmental signals, so that they become less
able (or likely) to undertake self-protection. From this perspective, risk assessment becomes an
integrated activity, which is open to different types of knowledge, be it disciplinary and not.
Moreover, effective flood risk management needs public involvement. Whilst the paradigm of flood
risk management is now firmly established within the scientific community and among flood-risk
managers, the population at risk retain more “traditional” assumptions about flood protection. Thus
placing greater responsibility on the general public is likely to be ineffective without new partnerships
and synergies between citizens and public institutions. Flood risk management is a great challenge for
practitioners, policy-makers and researchers who increasingly are required to talk to and with the
people at risk and, perhaps most importantly, to listen to them if the actual outcome is to include them
in decision-making processes and overall risk governance. Report T11-07-14 is the main deliverable
and provides recommendations from the research in Task 11.
Flood risk management aims at preventing losses and damages by preventing flooding and/or by
preventing the exposure of people and property to flooding. This includes lowering the probability of
flooding as well as reducing the vulnerability of the society in flood-prone areas. Consequently, flood
risk management may involve a large number of measures, for example flood defence measures, flood
control measures, but also spatial planning and measures aimed at reducing the vulnerability of people
and susceptibility of property to damage. This is because single-measure management approaches do
not take advantage of the way that various measures can interact to reinforce each other. For example,
better spatial planning to keep urban and other vulnerable development out of hazard zones may mean
smaller scale engineering works to protect towns and villages. And adequate emergency response
during floods can reduce flood damage and thereby lower insurance premiums.
Flood risk management is not a one-off activity, such as building an embankment or a dam. It is a
continuous process, characterised by repeated activities: analysis of the flood risk, consideration of
measures and policy instruments to reduce the risk, making policy decisions, implementing measures
and instruments, monitoring their effects, etc. This permits constant adaptation to changing
circumstances and changing societal requirements.
Flood risk management (FRM) is essentially preventive, as it focuses on all possible floods – both
frequent and rare – in contrast to flood event (or incident) management (FEM or FIM), which is about
dealing with floods that are happening or are about to happen. But flood risk management (FRM) does
of course involve the development of flood warning systems or of insurance schemes that are essential
for flood event management, as these should operate all the time and not just during flood events.
All Tasks focussed on methodologies rather than on empirical findings, and this is reflected in the advances
made. The degree of interdisciplinarity and integration of these six tasks differed substantially, related to
the aim of the task and the partners involved. We shall discuss the main outcomes for each task.
3.2.2 Learning from the past by evaluating measures and instruments in place:
a methodology
The main aim of Task 12 was to learn from the past by the ex-post evaluation of the performance of
individual structural and non-structural measures and policy instruments. Ex-post evaluation can close
the knowledge gap between past / current practice and future decisions in flood risk reduction.
To allow such an evaluation, a classification of all possible measures and instruments was needed, and a
thorough investigation and discussion on criteria to assess these by. The ex-post evaluation methodology is
the key outcome of this Task, but more concrete outcomes on the performance of individual measures and
instruments was gained by its testing in various pilot cases in Central Europe (Elbe, Odra and Tisza). The
main achievement is, therefore, academic, rather than practical at present. Only after thorough evaluation of
a wealth of existing measures and instruments, i.e. by adding lots of empirical data, can practical results be
expected.
The testing of the methodology in case studies proved that it could be applied, but that the interpretation of
the relevant assessment criteria is not without problems. And it was also confirmed that the performance of
measures and instruments depends on the context in which they are applied (even if all fluvial). This fact
hampers drawing general conclusions too soon. Still, the main achievement of the task is that measures and
instruments of widely varying character, from changed agricultural practice, via insurance to flood-
proofing of individual houses and retention polders, to flood walls and huge storm surge barriers can now
in principle be assessed by the very same criteria.
The task was interdisciplinary in the sense that it drew knowledge and approaches from technical sciences,
economy, ecology and geography. The quantification of risk was according to the methods described in the
former section. The lists of measures and instruments and some results were shared with task 14.
The long-list of possible measures and instruments was based on earlier research and transferred into a
database structure. Most attention, however, was given to the criteria by which to evaluate the past and
present performance of measures and instruments. These included hydraulic, economic, ecological and
social criteria, but the prime evaluation criteria were:
• Effectiveness (relation to risk reduction objectives)
• Efficiency (like the former, but then in relation to costs)
• Robustness (in light of uncertainty)
• Flexibility (adaptability, in light of changing circumstances and global change)
Effectiveness describes the extent to which the objectives of an intervention are achieved. It is thus
dependent on the availability of objectives for an intervention. Being also the traditional focus of
evaluation, effectiveness analysis constitutes a core element of ex-post evaluation.
In the light of changing societal priorities and considering financial limitations, the question of
economic efficiency is raised. In simple terms can be determined by relating economic (financial) costs
and benefits to obtain the benefit/cost ratio. The benefit/cost ratio is a measure of how economically
resources are converted to results. Benefits are represented by observed intended economic (financial)
effects. Costs are expressed by intended and unintended as well as direct and indirect realisation,
operation and maintenance costs of the intervention.
Robustness describes the ability of a measure or instrument to sustain the intended serviceability over
a wide range of known and unknown changes of conditions. Interventions are supposed to deliver their
intended effects under different magnitudes or dynamics of flooding or climatic and societal changes.
For example, a scheme introduced to protect against floods up to the 100 years event in future due to
different reasons may be hit more frequently by a design event. Or, encroachment into the floodplain
may substantially increase vulnerability and thus the risk in a flood prone area. An intervention
applied to reduce risks should be able to ensure the tolerability of risk in different situations unless the
investment it becomes obsolete.
Flexibility describes a project’s operational and long term adaptability. Changes of conditions
modifications may require adaptation of applied interventions. The reasons for these conditional
changes may be various ranging from a new hazard situation to changed risk perception and connected
expectations of ‘safety’. Thus, flexibility can be an important criterion also to decide on most
appropriate options in many cases.
The different language (jargon) used in this field of research as well as the more academic and less
practical character of it hampered discussions or uptake by the natural and technical scientists who
formed the majority of FLOODsite’s community. And indeed, one might judge that the main outcome
of the task lies in the thorough exploration and discussion of which theoretical ‘modes of planning’
apply best to the real-life situation of flood risk management planning in practice. And whereas the
research does not reveal any really new facts about flood risks or their management, it does give
insight in the many different views among scientists on the complex issue of planning: the task’s
innovation lies in taking another perspective to FRM.
Task 13 highlights challenges of strategy-making as linear and adaptive process of politicians and
officials (key decision-makers). It complements the content-oriented approach of Task 14, the focus
on ex-post evaluation of measures in Task 12, and the citizen-oriented approach in Task 11. Research
results fed into different tasks, Task 21 (Pilot study Elbe River) in particular.
A first important outcome of task 13 is a set of related concepts, which in conjunction may help to
bridge the gap between academic theory and practical experience. This involved four concepts:
1. a multidimensional understanding of strategy for FRM,
2. the distinction between a linear and an adaptive model of strategy for FRM,
3. strategic planning to bridge different spatial levels and policy fields, and
4. learning as complex process to use existing knowledge (“exploitation”) and to generate new
knowledge (“exploration”) to cope with uncertainty
In textbooks Strategic planning is often described as a linear process of designing and assessing
strategic alternatives to realize overall goals, aims and specific targets. Based on a multidimensional
understanding of strategy and the distinction between linear and adaptive models of strategy as well as
case study results, Task 13 developed the following two statements to deploy strategic planning for
long-term FRM:
• Strategic planning is an umbrella concept for content- and process-oriented approaches to long-
term FRM.
• Strategic planning facilitates travelling across spatial levels and between policy fields. Therefore,
it contributes to FRM as holistic and continuous process.
Secondly, a several principles for long-term FRM strategy making were formulated, namely:
1. Nurture diversity in decision-making: This notion of “requisite variety” in complexity science can
be applied to some extent to FRM. Nurturing diversity is important for the content, processes, and
context conditions of long-term FRM: Diversity of contents relates to the range of aims and targets
and the range of measures (e.g., structural and diverse non-structural measures like spatial planning
to control development on flood plains). Nurturing diversity in process patterns involves, among
others, using strategic planning as an adaptive process of highly political character. Nurturing
diversity in context conditions includes sharing of management power and responsibility and
involving multiple institutional linkages. This holds true especially for very complex governance
conditions like in the Thames Estuary.
2. Embrace uncertainty and change: Long-term planning in the context of FRM necessarily involves
dealing with considerable uncertainties. Effective risk reduction measures like spatial planning
require long-term planning horizons. One major challenge therefore seems to be making best use of
learning about uncertainties from previous experience or future scenarios. Perhaps the most
important result of Task 13 is that embracing uncertainty and change in long-term planning of
FRM is not only about looking at distant futures. It is also about a continuous effort to balance
exploration and exploitation within a complex social process which is prone to manifold
interruptions and limitations.
3. Combine the range of existing knowledge systems into the decision-making process: Decision-
makers focus their attention through formulating and implementing strategies for FRM. However,
strategies should not dilute, homogenise or diminish the diversity of existing knowledge systems
for management (knowledge with regard to physical processes, engineering, social processes,
strategic alternatives, and so forth).
Thirdly, Task 13 investigated whether and to what extent these principles were applied in case studies,
incorporating the Dresden / Weisseritz River, the London / Thames estuary and the Trento/ Adige
River. Based on these cases, Task 13 concluded that the four principles to make strategy for long-term
FRM are quite easy to understand, but obviously difficult to implement. Difficult because strategy
making is not only a rational science-and-technology-driven process based on a common “language”
and common interests, but instead a political process based partly on different perceptions and world
views (e. g., between local flood risk managers and planners) and diverging interests (e. g., embracing
uncertainty vs. restoring order). As a consequence, understanding and steering governance becomes
important for FRM.
In many research projects for policy planning strategic alternatives for long-term policy making are
being developed and evaluated. Their assessment nowadays often involves assessing their
performance in different future scenarios. Task 14 of FLOODsite aimed to provide methodological
guidance on how to perform the design/ development of long term strategic alternatives for flood risk
management and their assessment.
The main outcome of the task research is, therefore, the procedure depicted in Figure 3.9, which can
be regarded as the graphic summary of the key methodological guidance on how to proceed when
developing and assessing long-term flood risk management strategies. In Figure 3.9, the blocks
represent the different steps within the method. The arrows represent the connections between
different steps).
Analysis of current
flood risk Current flood risk analysis Future flood risk analysis
management & assessment & assessment
strategy
Figure 3.9 Schematic overview of the method for developing and assessing long-term flood risk
management strategies in view of uncertain futures
The criteria for assessing the alternatives’ sustainability have been given due consideration which led
to a better understanding of the relevant assessment criteria. Especially on the criteria robustness and
flexibility new insights have been gained; this can be regarded a second key outcome.
This procedure was followed in three case studies in order to test it, namely on the Thames Estuary, on
the Western Scheldt Estuary and finally on the Elbe river. The results for these case studies have direct
policy relevance for the flood risk managers of those areas. They can be regarded a third key outcome.
Because this kind of research has huge societal relevance (perhaps it is the most relevant of all
FLOODsite research), the approach was also (partly) adopted for some large national research
projects, such as Thames2100 in the UK, investigating the future of the FRM policy for the Thames,
and the Netherlands 2nd Sustainability Outlook and the Water Safety 21st Century project in the
Netherlands. This can be regarded as direct spin-off of the work of FLOODsite.
Finally, the procedure strongly affected the work in Task 18 on technological integration, where the
procedure was transformed into a computer-based Discussion Support System with prototypes for
again Thames, Scheldt and Elbe.
The three key elements scenarios, strategic alternatives and assessment, are incorporated in these
steps. The most important findings on the use of scenarios, the development of strategic alternatives
and their full assessment are discussed below.
Although, in reality, management planning is a cyclic process, figure.1 depicts it as a linear process
which starts with an exploration of the flood risk system and ends with a full assessment of strategic
alternatives. In practice, alternatives may be adapted when unfavourable results are obtained or
additional alternatives may be added halfway. Also, various iterations between design and analysis are
likely. However, in order to clearly show the main procedure and relationships the figure is kept
simple.
The first two steps in this procedure, obviously, closely relate to things discussed in earlier chapters:
the definition of the relevant flood risk system (Ch. 2) and the analysis and assessment of the flood
risk applying the methods described earlier (Section 3.1). A difference is that the focus is not only on
the present flood risk to people and property, but also on possible future risks as a function of
changing climate, demography and economy. To cover for these ‘global’ and local changes in time
outside the influence of the flood risk manager, a scenario approach is advocated, based on thorough
literature review and practical experience (EU, UK Foresight, NL amongst others). This is the first
innovative point in comparison to anything discussed before.
The next step is about the design of strategic alternatives; by content (in contrast to what task 13
focused on). If the flood risks are not acceptable now or in the future – or when the current strategy
has other major disadvantages –, various strategic alternatives for flood risk management may be
developed. The research produced guidance on how to design those alternatives, comprising technical,
regulatory, financial and communicative instruments.
economic and ecological functioning of the system, as well as criteria which specifically pertain to
‘dealing with uncertainties’.
Table 3.1 The framework of criteria and indicators for a full assessment
In the context of long-term planning for global change, robustness and flexibility are both very
important criteria since they reveal the sensitivity of strategic alternatives to uncertain events and
gradual or sudden changes. Robust strategic alternatives are less sensitive to uncertain events such as
very extreme water levels, malfunctioning of structures, malfunctioning communication systems,
unforeseen behaviour amongst the inhabitants, etc. Flexible strategic alternatives function well across
a range of future scenarios or can be easily adapted if future developments differ from the ones
anticipated. Future regret is thus less likely when such flexible strategic alternatives are being adopted.
Although important progress has been made on understanding what robustness and flexibility are, their
operationalisation still requires more thought.
In order to show possible ways of coping with flood risks different strategic alternatives need to be
defined and assessed. These strategic alternatives must be visionary and clearly different. Therefore, it
is advocated to define them top-down from guiding principles to select combinations of measures and
instruments. As guiding principles, for example, ‘world views’ (‘perspectives’) or the concepts of
resilience and resistance may be used.
The trials in Thames and Schelde Estuaries also showed that strategic alternatives are a good means to
explore possibilities for long-term flood risk management. Important findings are, e.g. that a FRM
policy which functions well in one scenario, does not necessarily do so in another with, for example,
less economic growth. And some alternatives never score the best, but neither do they hold the risk of
terrible failure if developments are other than expected. This is highly relevant for policy makers
trying to define a sustainable policy in the context of uncertainty.
This makes flood warning the primary policy instrument for the management of flash floods during
the event. Through the images from national weather forecast services in most countries and on many
public internet sites, many people are acquainted with satellite imagery of cloud cover as well as radar
rainfall fields. These show where rainfall can be expected, but estimating how much water is likely to
fall is more complicated and difficult. The characteristics of flash floods provide a substantial
challenge flood forecasting, mainly because operational atmospheric models are too generally coarse-
scaled (typically around 10km grid resolution), and because many small basins are not equipped with
rain gauges. It is therefore unrealistic to expect high forecast reliability for localised thunderstorms
occurring on small and medium size watersheds. However, flash flood forecasts do not necessarily
need to be precise to be of practical use. Indeed, Task 16 has demonstrated that effective flash flood
warnings may be issued without the use of detailed rainfall-runoff models.
In Task 15 a set of improved methods has been developed and piloted for QPE from radar imagery,
building on earlier intensive observational campaigns (Delrieu, 2008). Improvements in the
interpretation of the radar imagery have been made on several fronts including:
• a technique to identify the “ground clutter” caused by reflection of radar beams from locally high
ground
• adaptation of two rain-type separation algorithms
• estimation, conditional on the rain type, of the vertical profile of radar reflectivity
• rainfall estimation at ground level from corrected reflectivities above
• a stochastic model of range profiles of rain-drop size distributions to test the sensitivity of
attenuation correction schemes to the variability in drop-size.
These contributions of Task 15 have led to improvements in the interpretation of ground-based radar
data which directly affect the work of Météo France, who co-operated in the research.
There is the basic problem with satellite data which derives from sensors that can only scan the top of
clouds and are unable to see which processes go on further inside and below a cloud. Therefore the
rain rates have to be estimated from the temperature and reflectivity of the cloud tops. Currently used
algorithms were also developed for other satellites or regions and are not directly applicable to
Meteosat-6/-8 without adjustments to its different spectral responses and spatial and temporal
resolution.
Differences between the radar and satellite rain rates may also result from the different spatial
resolution of the radar data and Meteosat-6/-8 as well as from the different scanning time. Task 15
considered sources of error and uncertainty causing discrepancies between radar and satellite derived
rain rates. The work revealed (Görner et al, 2008) that the satellite based rainfall estimation still
includes several uncertainties when compared with ground-based observations. These differences
concern the size and location of the rain field area as well as the amount of the estimated rain rates.
However, the structure of the radar rain field is partly distinguishable in the satellite estimated rain
fields. All of the algorithms used show a tendency to underestimate the rain rates measured by radar
for the grid cells having rain rates over 10 mm/h. For the areas with less or no radar derived rain a
tendency of overestimation is evident for all techniques. Both tendencies are partly a consequence of
the relocation of the satellite rain field. This relocation may result from the different scanning angle of
the two systems radar and satellite, the different points in scanning time and processes and influences
(wind, moisture etc.) between the cloud top and the earth surface. Differences between the radar and
satellite rain rates also result from the different spatial resolution as well as from the different points in
time of the measurement and scan. Nevertheless the satellite derived rain rates already can provide
helpful additional information on rainfall, particularly in regions low or not covered by ground
measurements.
In this context, it investigated whether the methods of Flash-Flood Guidance (FFG), an approach
developed in the US, might help increase the reliability of flash flood warning, in comparison to
rainfall-runoff modelling and in comparison to various other approaches. The FFG approach is a
relatively straightforward comparison of rainfall amount with pre-determined thresholds which are
known to cause flooding. The FFG is the depth of rain of a given duration, taken as uniform in space
and time on a certain basin, necessary to cause some flooding at the outlet of the basin concerned.
This rainfall depth, which is computed based on a lumped hydrological model, is compared to either
real-time observed or forecast rainfall of the same duration and on the same basin. If the predicted
rainfall depth is greater than the FFG (i.e., the Flash Flood Threat – FFT - is greater than a given
threshold), then flooding in the basin is likely (see Figure 3.10). Thus direct flood alerts can be
triggered from rainfall measurement or prediction. Based on the comparison of predicted rainfall
amount and threshold, a warning can be issued.
Figure 3.10 Sample graphical system output for FFG and FFT
The study provided an assessment of this technique based on data from eleven basins (six nested
included in five larger parent basins) located in north-eastern Italy and central France. The results
show improvement in terms of probability of detection and false alarm rate (Borga, 2008) over a static
rainfall threshold alone to trigger an alarm. The FFG approach has been implemented into the
operational system for flood forecasting in the Adige River Flood Forecasting System in Italy.
Overall the outcome of Task 16 is improved insight into the advantages and disadvantages of various
approaches to flash flood forecasting. Not each technique was equally suited for each basin, because
of different availability of data and basin characteristics, but the Task yielded useful guidance on
which approach to favour under different circumstances. In particular, the concept of Flash Flood
Guidance can help considerably in the communication between hydrologists, meteorologists and
decision makers.
The work encompassed a review of flood event management practice in Europe; a review of the
requirements for evacuation planning based on end user consultation; an overview of evacuation
modelling; and the development and testing of evacuation and traffic management models relevant to
flood event management in the Thames Estuary in the UK, the Schelde Estuary in the Netherlands and
the Gard Region of France.
The study focused on the many possible ways of modelling of evacuations and aimed to assess which
models are most suitable to support evacuation planning. The various approaches to modelling were
tested by applying them to case study areas along the Thames Estuary in the UK (Canvey Island,
Thamesmead embayment), and along the Western Scheldt in the Netherlands. The testing of the tools
encompassed not only their validation but also their functionality and the usefulness of the results that
they provide to emergency responders and flood event managers. This yielded clear guidelines on
which approach to use under which circumstances. For the Gard Region in France a prototype system
to forecast road inundation was developed.
Review and testing of evacuation models
The first key outcome of Task 17 is the result of the review of evacuation models used worldwide for a
range of hazards (including technological hazards such as nuclear accidents). These included at three
spatial scales: micro-scale, meso-scale and macro-scale:
• Traffic simulation models;
• Evacuation behaviour models;
• Time-line and critical path management diagrams.
It was found that there has been little work undertaken in Europe, or the rest of the world, related to
evacuation modelling for flood event management. Previous work carried out in Europe has been
generally limited in scope and carried out a macro- and meso- scale, never at a micro-scale. Three
evacuation models were tested in the Thamesmead embayment and Canvey Island, these were:
• BC Hydro Life Safety Model (LSM) – a micro level model;
• A non-linear optimisation technique – a meso scale model;
• A simple spreadsheet based model – a macro scale model.
This revealed that the LSM has the potential to be used to inform emergency plans for heavily
defended areas (e.g. London, some coastal areas) and dam failure risk assessments. It allows the
assessment of location and the number of safe havens; effect of road closures; rate of dissemination of
warning; time of day and the location of the population; number of fatalities and injuries. The BC
Hydro LSM micro level model combines evacuation modelling with estimates of loss of life and
injuries to people, as well as the number of buildings that could collapse.
Such a model provides information that is useful to two different sets of stakeholders, in that it
provides information that is of use to the emergency planners and it also allows flood risk managers to
estimate the residual risk behind flood defences or downstream of dams in terms of the risk to people
and buildings.
The LSM was found to be a powerful tool. However, it was tested to its limits and at present cannot
deal with more than around 50,000 individual receptors. This is the first time a micro-level evacuation
model that dynamically links the movement of the receptors (e.g. people and vehicles) with the flood
wave has been applied to tens of thousands of receptors in a European situation.
For the area bounding the Westerschelde Estuary, a number of evacuation models were also tested;
these were EC, ESCAPE and INDY. None of the models model individual movement of persons,
instead they model the flows of total number of inhabitants from each postal code zone. The large
number of inhabitants in this area requires such meso or macro-scale modelling.
The models were compared on spatial scale they can be applied on; suitability for evacuation
planning; resulting evacuation times; and reliability of the results. This translates into guidelines on
their use. As the model results could not be compared with actual evacuation times from a historical
event, it is difficult to conclude which model is the most reliable., but all three models are expected to
underestimate the evacuation time as they rely on the assumptions that no congestion occurs from
cross flow traffic, that no accidents will occur, and that residents will respond to an evacuation call in
an ideal and optimal way.
The review and testing revealed that evacuation modelling for flood events in most EU countries falls
outside the remit of flood risk managers. The lessons learnt from the piloting of the model indicates
that emergency managers often require micro level tools to provide them with the information they
require to formulate and improve their emergency plans. In some EU countries (e.g. UK and the
Netherlands) the data required for such models are readily available. However, there will be many
countries and regions where such data are not available.
The blue area in Figure 3.11 depicts one of the submerged roads, based on the output of the prototype
model on flash floods developed for the Gard Region. The red line shows the road section that has
become unavailable as a result of the flooding. The points labelled “A” and “B” indicate the start and
the end point of the route. These can be selected by the user. After the user has chosen their start and
endpoint they can request a “route calculation” and, if the route intersects with a submerged road the
routing algorithm will calculate the next best route.
B
B
A
A
Shortest route between two points Shortest route between two points
closed by a flash flood
B
A Start point
A
Revised route found by the web based
algorithm between the two points
B End point closed by a flash flood
Figure 3.11 Application of the ORCHESTRA web tools to the Gard Region of France
The work achieved by the collaboration between the ORCHESTRA and FLOODsite projects
illustrated that it was possible to form a link between two models that had been developed
independently. The combination offers a useful risk management solution. The work also
demonstrates that if open standards are applied, added value can be created for the end user in terms of
an easy “plug-in” of other modules.
The prototype flash flood forecasting model set up for the Gard Region of France allows the
emergency services to have a better idea of which roads will remain open and which will be shut
before a flood occurs. Its success depends on having a database of historical road inundations with
which the model can be validated. In the case of the Gard Region of France, from an emergency
planning point of view, forecasting the time at which the road becomes impassable accurately proved
actually more important than predicting the peak flood flow accurately.
• Macro-scale evacuation models are useful for obtaining first order estimates of evacuation times
for relatively large areas.
• Meso and micro scale models are needed for detailed evacuation planning.
• From the work carried out it would appear that micro-scale models, although more time consuming
to set up, provide emergency planners and other end users with more insight into the areas at
greatest risk and also provide decision makers with other risk metrics (e.g. number of buildings that
have collapsed loss of life, inundation of escape routes). However, to be effective such models
should be applied to the whole area at risk.
Although Task 17 has demonstrated the potential of evacuation modelling for more effective
emergency management, more work needs to be carried out to make evacuation models easier to use
by non-specialists to encourage their widespread use.
In doing so it is recognised that different decisions have different constraints and opportunities and
hence are specific tools. Hence within FLOODsite two specific decision types are considered:
• Frameworks for long term planning (Task 18)
• Flood event management (Task 19)
Although these decisions are separated according to their context, they do share many common
features. Indeed within FLOODsite the conceptual and associated discussion support system (DSS)
tools share many similarities; with each adopting the Source-Pathway-Receptor concepts and similar
modes of user interaction with the prototype tools developed.
Underlying all good decision making is however the notion that our knowledge is not prefect and the
world around exhibits significant natural variability. Good decision making is that about recognising
this uncertainty and making wise robust choices. Uncertainty is not therefore an abstract concept but
one that is fundamental to good decision making:
“Uncertainty is the rationale doubt as which choice to make” (Colin Green, personal
communication.)
FLOODsite recognises this and advances our ability to understand the uncertainty in the decision
making (supporting robust long term planning) and provides advances techniques for the propagation
of uncertainty through complex forecast models. The supporting research is provided through Task 20
on uncertainty frameworks.
The activities within Theme 3 built upon and extend the science developed in other Tasks and were:
• User orientated – focusing of key decisions of interest
• Case study based – throughout case studies have been used to aid the development of practical
science
• Innovative – integrated, risk and systems based concepts challenging concepts. Floodsite has made
significant advances towards future implementation in practice.
The scientific advances within each of Tasks 18, 19 and 20 as well as the overarching concept of
integrated flood risk management, that has characterised the work within FLOODsite, are also
explored and summarised below.
The conceptual framework is a high-level view of the long-term planning process which seeks to
understand the full range of questions that different stakeholders may pose to support their long-term
planning needs. As illustrated in Figure 3.12, this includes:
• development of a knowledge base on flood risk issues;
• an overview of the relationship between flood risk management and the development of natural and
societal systems (landscapes, social acceptance, local economy);
• a review of the organisational structures and institutions across Europe together with the regulatory,
economic and socio-cultural conditions in which they operate;
• an overview of societal needs and long term perception of risks;
• an overview of available strategic options (including mixed resistance and resilience based
strategies).
Measures and
Risk Analysis Risk Assessment instruments
Risk perception
The methodological framework is a translation of the conceptual framework into tangible algorithms,
methods and component interactions (see Figure 3.13 below). Implicit within this is the need to
develop future climatic and socio-economic scenarios and to characterise the management response
through time (strategic alternatives) and hence the evaluation of these responses in an uncertain future,
building on the Task 14 work. In Task 18, specific methods are set-out to address option robustness,
flexibility and sustainability and the integration of these concepts through time and across ‘all’
possible future scenarios. Uncertainty is addressed in terms of that arising from methods, models and
data as well as the more gross uncertainty associated with the future. Emphasis is also placed on how
best to present the information (including uncertainty) to form a useful evidence-base for the decision
maker.
Source Module
External Driver Module
Management Response
Cost of interventions
DSS input databases
Pathway Module
Uncertainty throughout process
Receptor Module
Consequence
Risk Module
Decision Support
The technological framework considers the development of prototype tools used to enact the methods
and algorithms through a specific application which may involve development of specific software
modules or computer-based tools to execute the methods and display the output risk metrics. In
support of the development of the FLOODsite tools, existing tools were reviewed and compared.
Three prototype tools were developed in FLOODsite. Each tool is specifically tailored towards the
needs of the local end users in the Thames, Scheldt and Elbe, see Figure 3.14
own needs. This is typically linked to robustness and flexibility as well as social, ecological and
economic considerations.
• Resilience: the ability of the system to withstand hazards or extreme events (larger than the
design criteria) or shocks i.e. performance under abnormal events.
• Uncertainty: recognition and representation of uncertainty due to date, methods and models as
well as long-term uncertainty associated with the future.
Figure 3.15 below illustrates the concept of a decision “pipeline” as explored in the long-term
planning on the Thames Estuary pilot to support a formal assessment of “flexibility”.
Figure 3.14 DSS Tools for long term planning: Example screen shots from the three pilot sites
Legend
Portfolio 3:
Resilient e.g.
Decision point
Flood Portfolio 6:
awareness Resis/resil.
Decision path
Portfolio 1: raising; Build flood
abc Portfolio of measures Resilient e.g. Relocation; barrier;
evacuation Damage Managed
Strategic alternative planning; compensation realignment;
flood Evacuation
Decision time / epoch warning;
planning.
floodable
basements.
Portfolio 4:
Resis/resil.
Maintain
Existing defences;
Ongoing
Build more Portfolio 7:
System maintenance
retention; Resilient e.g.
Portfolio2: Flood Maintain
Resistant e.g. warning defences;
Raise Build
defences; retention;
Build Portfolio 5: Managed
retention; Resilient e.g. realignment
Improve Raise
conveyance defences;
Build flood
barrier; build
flood gates.
Figure 3.15 Example of multi-staged decision process used within the Thames DSS
Consequence module
Damage to inhabitants (affected people,
Damage and exposed people, casualties)
casualties Damage to livestock, property and utilities
model (expressed in the value lost)
Risk module
1. Base data
- Population data
- Background mapping
- Digital Terrain Model (DTM)
6. Evacuation times
- Import evacuation times from models
Figure 3.17 Technological framework of the DSS for the Thames and Schelde Estuaries
3
This map is reproduced from the OS map by HR Wallingford with the permission of the Controller of Her
Majesty's Stationery Office, Crown Copyright. Unauthorised reproduction infringes Crown Copyright and may
lead to prosecution or civil proceedings: Licence Number 100019904
Figure 3.19 ESS screen showing the time of inundation for a breach scenario ‘Rilland’
Figure 3.19 Flood Hazard mapping for the project site in Nice
The practice of uncertainty analysis and use of the results of such analysis in decision making is not
widespread, for several reasons:
• There is a bewildering proliferation of uncertainty methods published in the academic literature.
• Uncertainty analysis takes time, so adds to the cost of risk analysis, options appraisal and design
studies.
• The data necessary for quantified uncertainty analysis are not always available, so new data
collection campaigns (perhaps including time-consuming expert elicitation exercises) may need to
be commissioned.
• The additional requirements for analysis and computation are rapidly being (more than)
compensated for by the availability of enhanced computer processing power.
However, computer processing power is only part of the solution, which also requires a step change in
the approach to managing data and integrating the software for uncertainty calculations.
Task 20 has sought to make a significant contribution to the methods and application of uncertainty
analysis by targeting novel areas of uncertainty analysis and decision support. The research fell into
four sub-tasks:
• Development of an overall framework for uncertainty analysis in flood risk management decisions.
• Development of new methods to deal with the uncertainty issues associated with composite model
construction – in particular associated with whole system modelling
• Propagation of uncertainty through composite (hybrid) models using efficient computational
methods – in particular applied to flood forecasting
• Development of methods for robust decision making under severe uncertainty – in particular
associated with long term planning and linked to Task 18.
Figure 3.21 Uncertainty analysis framework – documenting, recording and challenging uncertainty
In the presented approach an uncertainty model is built to estimates total uncertainty of a deterministic
(process) model by analyzing its residuals (errors). The following assumptions are made:
• the process model error (mismatch between the observed and modelled value) is a proper indicator
of model uncertainty;
• the data which are close (e.g. in terms of Euclidean distance) in the input space correspond to
similar real-life situations and will have similar values of model errors;
• “optimal” process model is used to produce a deterministic forecast.
The concept of optimal model, however does not exclude the possibility of using an optimal
combination (ensemble) of “good” models having the same structure but different in the parameters –
which could result from a Monte Carlo exercise.
The UNEEC method has been tested to estimate uncertainty of runoff simulations by conceptual
rainfall-runoff model with application the Brue catchment in the United Kingdom and other
catchments. It has been applied in collaboration with FLOODsite partners to the Alto Adige. It was
also compared with other uncertainty estimation methods (GLUE and meta-Gaussian approach).
Figure 3.22 illustrates the UNEEC method, which has three steps:
• clustering,
• estimation of the probability distribution of the model error, ep, and
• building model U for probability distribution of the error.
Once the model U is trained in the calibration data set (Xu), the model can be used to predict the
probability distribution of the model error in a new or unseen data set (e.g., validation data).
Robustness analysis was one of the topics of frequent communication between FLOODsite Task 20
and Task 18 (Development of Framework for Long-term Planning).
At some place in this document it might be worth mentioning the dissemination activities in
summarising how many research papers have been published. Furthermore, all PhDs with some
relation to FLOODsite could be mentioned.
However, the existence of raised defences may be taken as an indication that risk maps and plans need
to be prepared. In that case the contribution of the science in Tasks 4, 6 and 7 will be directly in these
subsequent activities. Likewise it is not yet clear whether the flood damage estimation guidelines
developed in Task 9 or the estimation of loss-of-life developed in Task 10 will be needed in
preliminary assessments. If the preliminary flood risk assessment requires initial drafting of flood
inundation and consequence maps, then other FLOODsite results are also relevant as discussed in the
paragraphs below on the preparation of Flood Risk Maps.
4.3 Potential Support for Flood Hazard and Flood Risk Maps
As for flood hazard mapping, Task 3 has developed and tested a procedure for coastal areas based
upon the FLOODsite methodology together with a review of existing mapping technologies for
riverine and coastal areas. The objectives of the research were limited to a better understanding of the
coastal area to avoid duplication with other initiatives (such as EXCIMAP set up DG Environment)
that considered river hazard and risk mapping.
The project has developed a deeper understanding of the elements of risk and this will be available to
support the flood risk mapping process. FLOODsite has in Task 2 contributed knowledge leading to a
(joint) probability density function for extreme values such as water levels, river discharges, and sea
state parameters. In addition, the understanding of reliability of flood defences from Tasks 4, 6 and 7
has enabled a fuller assessment of risk to be prepared through factoring the likelihood of defence
failures into the assessment of risk. In Task 8, models for flood inundation have been benchmarked,
and this has led to guidance on the suitability of hydrodynamic modelling approaches for hazard and
risk mapping as well as to the identification of relevant hazard characteristics and their classification
on maps.
The work in Tasks 9 and 10 is of direct relevance to flood risk mapping since this work supports the
evaluation of the consequences of flooding. In particular, the risk mapping may use the flood damage
estimation guidelines, the estimation of loss-of-life model, and the GIS-based multi-criteria evaluation
of risk developed in Task 10.
4.4 Potential Support for the preparation of Flood Risk Management Plans
The purpose of the flood risk management plans is to identify means of reducing the impacts of
flooding. In addition to the using the project knowledge outlined above which supports the
preliminary assessments and flood risk mapping, several other tasks are researching areas which
support the preparation of flood risk management plans. These are:
• The understanding of community preparedness and resilience from Task 11
• Identification, design and appraisal of sustainable flood mitigation measures from Tasks 12 to 14
• The rôle of flood event warning systems from Tasks 15 and 16
• Emergency evacuation planning, coupling inundation and traffic models from Task 17
• Decision support for long-term planning and the selection of a portfolio of measures and
instruments for flood risk management from Task 18
It should be recognised that the process models and decision support software will require further
development for application in practice. The FLOODsite project will prototype methods and pilot
their application but only to a pre-competitive level.
The activities of flood risk management and the FLOODsite research can be considered at a series of
hierarchical levels:
Table 5.1 Hierarchy of knowledge and activity for flood risk management
Level Description Examples of policy, activity or knowledge
1 Societal context Sustainability policy, environmental objectives, regional
development planning,
2 Flood risk system National policies and activity to implement the Floods Directive,
investment planning for flood risk management …
3 Structural clustering of Risk analysis, risk mapping, flood forecasting and warning, civil
activities contingency planning, emergency management, national appraisal
of flood risk …
4 Domain specific Radar hydrology, runoff modelling, embankment reliability, flow
knowledge modelling, loss-of-life estimation, flood damage estimation …
5 Detailed supporting Atmospheric physics, scattering of radiation, sediment transport
technical knowledge processes, joint probability statistics, wave shoaling processes,
hydraulic resistance of vegetation, soil mechanics, public
consultation methods, GIS, …
This hierarchical assessment of knowledge and activities may not be precise in that the distinction
between levels 4 and 5 above may be questioned. However the hierarchy indicates the broad context
for the contributions to knowledge from the FLOODsite tasks and facilitates the integration of the
project knowledge.
The scope of FLOODsite as a project was at Level 2, with the Themes 1 and 2 focussed on and
combining knowledge at Level 3. The individual FLOODsite Tasks in Themes 1 to 3 then operated
mostly at Level 4 with some detailed components of research at Level 5 (for example research on
wave impact forces in Task 6). The FLOODsite Tasks were substantially self-contained and so
connections between them was either sequential (like Tasks 4 and 6 feeding to Task 7) or from one
hierarchical level to another (for example Task information from Themes 1 and 2 into the Decision
Support work in Theme 3).
The key to transferring knowledge within the project tasks is for close collaboration at a working
level. The linkage between the science and the Pilots is a direct mechanism for integration of the
project and for the pilot sites to advance the project science as well as the pilots using more of the
FLOODsite tools in their areas. The pilot studies acted in some respects as knowledge integrators
operating at Levels 3 or 2. The nature of the links was “hard”, that is a direct dependency, or “soft”
through the passage of information to enhance an activity without it being on the critical path for that
activity. Figures 5.1 and 5.2 below provide an overview of this inter-linkage. These linkages show
that although no research Task from Themes 1 to 3 is tested in all pilots, all Tasks in Themes 1, 2 and
3 are tested in the context of at least two pilot areas (taking account of Task 23 covering four distinct
Flash-flood basins).
Figure 5.1 Links between project science in Theme 1 and the pilot sites
Pilots T 21 T 22 T 23 T 24 T 25 T 26 T 27
Thames Schelde Ebro River German
Elbe River Tisza River Flash flood River River Delta Bight
Basin Basin Basins Estuary Estuary Coast Coast
Besos
Adige
Mulde
Moldawa
Ardennes
Cevennes
Lower Elbe
Task Methods/Knowledge
Theme 1 Advancing scientific knowledge …
Sub-theme 1.1 Hazard (Risk Sources)
T1 Hydrological models for extreme flash floods in X X X
ungauged basins
T2 Models for (statistical)/joint probability analysis of X X X
extreme event data
T3 Techniques for mapping flood hazards X X
Sub-theme. 1.2 Hazard (Risk Pathways)
T4 Failure mode analysis X X X X
T5 Model tools for morphological changes at coasts X X
T6 Improved models for breach initiation and growth X X X X
T7 Methodology for reliability analysis X X
T8 Numerical inundation models X X X X
Sub-theme. 1.3 Vulnerability (Receptors, consequences)
T9 Flood damage evaluation methods X X X X X X X X
T 10 Methodologies for GIS-based multi-criteria evaluation of X X X X X X X X X X
flood damages
T 11 Risk perception, community behaviour and social X X X
resilience
Note
T4, T5 and T8 mainly delivered technologies through other FLOODsite tasks
Figure 5.2 Links between project science in Themes 2 and 3 and the pilot sites
Pilots T 21 T 22 T 23 T 24 T 25 T 26 T 27
Tisza Thames Schelde Ebro River German
Elbe River River Flash flood River River Delta Bight
Basin Basin Basins Estuary Estuary Coast Coast
Besos
Adige
Mulde
Moldawa
Ardennes
Cevennes
Lower Elbe
Task Methods/Knowledge
Theme 2 Innovative mitigation …
Sub-theme 2.1 Pre-flood measures
T 12 Methodology for ex-post evaluation of pre-flood and X X X
flood event measures and instruments (ex-post EFM)
T 13 Strategies for pre-flood risk management in case studies X X X
T 14 Scenario development and analysis X X X
Sub-theme 2.2 Flood event measures
T 15 Theoretical and operational assessment of the SAS X X X X X
efficiency
T 16 Methods for evaluation flash-flood at regional level X X X
T 17 Open system model integration framework for 2-D X X
inundation modelling
Theme 3 Framework for Integration
T 18 Developing specification of a prototype DSS (long-term X X X
planning)
T 19 Developing t specification of a prototype DSS (event X X
management)
T 20 Uncertainty propagation software X X X
FLOODsite Theme 3 explored various frameworks of integration of the knowledge domain in flood
risk management. This takes place in three ways Conceptual, Methodological and Technical.
The conceptual framework is directed at the activities of decision makers and other actors in the
process of flood risk management. It sits mainly on Level 2 of the hierarchy of Table 5.1 (with some
interaction with Level 1 drivers) and this was the basis of the project design. It covers
• Processes for the analysis and assessment of risk
• Public attitudes to risk and its acceptability
• Flood risk management interaction with natural and social systems
• Process for developing risk management strategy and associated measures and instruments
The methodological framework covers both the influences on flood risk and the methods for
understanding the flood risk system. Hence this sits mainly at Level 3 of the hierarchy of Table 5.1
and encompasses
• Representation of external drivers and responses
• Process modelling and risk-based integrations
• Uncertainty estimation
• Assessment criteria: flexibility, robustness, sustainability
A framework for technological integration operates at Level 4 of the hierarchy of Table 5.1 and
envisages a computer-based coupling of tools and technologies that support the enactment of the
methodological framework for a particular user community. It covers:
• Identifying and coupling models in risk simulation
• Integration of spatially distributed and temporally varying data
• Techniques to visualise simulation results
• Operational support in terms of advice, training, maintenance etc
explored through compatible or at least plausible scenarios of climate (mean and variability), GHG
emissions, environmental response to the climate, social systems (including attitudes, affluence,
institutions and governance) and policy. Many Tasks in Theme 1 advance the science needed to
analyse flood risk – and these can be applied in the context of the current climate or any future
scenario. Some examples are: the methods for estimation of extremes in Task 2, the morphological
response of coasts and rivers from Task 5, understanding the failure modes of defences from Task 7 or
the economic and social impacts of flooding from Tasks 9, 10 and 11.
FLOODsite (Task 14) focused on methods of designing and assessing alternative flood risk
management strategies, as well as on their assessment against different future scenarios: all possible
and equally likely. In Task 14, FLOODsite established that the design of strategic alternatives for the
far future is preferably done according to a top-down approach, by defining clear and opposite guiding
principles, such as resistance versus resilience, and / or by specifying different objectives (economy
versus ecology, people versus material damage, etc.). One might also distinguish flood risk
management strategies analogous to war strategies, such as attack, defence, regroup and retreat. These
might well apply as societal adaptation strategies with different contributions of either land use
planning and engineering. In the Netherlands, Partner 2 (Deltares) applied this approach on the 2nd
Sustainability Outlook for the Netherlands under the title ‘The Netherlands Later’. For 2020 and 2040
projections of population and economic development were translated into land use projections under
various policy strategies. The following flood risk management strategies were explored.
It was found that climate change is no larger an issue of adaptation than economic growth is; both
result in a doubling to tripling of the flood risk between the years 2015 and 2040. It was also found
that the present strategy is not so bad but has flaws, especially as the whole flood risk system –
including socio-economy – becomes less robust and less flexible. Alternative strategies perform better
on some points and are now being seriously considered for some regions in the context of the national
debate on Water Safety 21st Century. This tempts us to conclude that studying Integrated Strategies
does influence public debate and policy making.
5.3.4 Long term planning and risk assessment for the Thames flooding system
From early in TE2100 project it was clear that flood risk planning would need to be based on a sound
understanding of the estuary and its various flood risk management components acting as a system
over the long term. Some of the main elements include: fresh water inflows, tide and surge levels, the
channel itself, the Thames Barrier and other moveable barriers, fixed defences both upstream and
downstream of the barrier, and numerous tide gates, pumps and other components. It is also necessary
to consider the possible impacts of flooding including flood inundation, and people, property and
environmental assets at risk. A comprehensive system model has been developed to be able to answer
questions about levels of risk now and in the future, and to assess the robustness of various risk
management policies (including their benefits and costs). The model uses the Source-Pathway-
Receptor approach and is a development of the so-called “RASP High Level Methodplus” used for the
annual UK National Flood Risk Assessment.
A sophisticated system model of the Thames Estuary has been developed that includes joint
probability calculations, defence reliability, through a Monte-Carlo sampling approach – this has been
allied to inundation estimates and economic damage model to create a sophisticated modelling tool.
This model was extended under the FLOODsite project to produce a ‘Rapid’ system model. This
enables the fast exploration of potential strategies in the context of different socio-economic and
climatic futures, at multiple epochs. It also provides the means for a more robust ‘variance-based’
uncertainty analysis, building confidence in the underlying methods and enabling prioritisation of data
collection and quality. A further innovation is the development of methods for building long-term
coherent storylines and assessing these in terms of their sustainability, robustness and flexibility. The
model has been run for a number of different management scenarios. These include a “do nothing”
and maintain existing defence standards under a variety of climate change scenarios on the Thames
Estuary. Figure 5.3 illustrates the economic damage estimates in the “do-nothing” case for current
conditions and some climate scenarios.
35,000
30,000
25,000
20,000
Damage
15,000
10,000
5,000
0
1
10
100
1000
10000
Present Day (river level capping) Defra (Year 2100) Medium high (Year 2100) High+ (Year 2100)
Figure 5.3 Thames Estuary flood risk for a range of climate change scenarios from Task 24
Results indicate temperature and precipitation changes in the Mulde River catchment with higher
precipitation changes in the lowlands in summer. Extreme precipitation (> 50 mm) is not covered by
climate projections (like 2002) which may change the probability density functions. However, based
on the calculations historical 100 year return period will probably be 65 year (1day), 85year (2day), 97
year (3day)… by the end of the 21st century. Changes in pre-event moisture and snow melt floods are
possible which should be further investigated.
Although all participants in FLOODsite (from the Czech, Dutch, English, French, German, Greek,
Hungarian, Italian, Spanish and Swedish language areas) could communicate very well in English, in
their home country they communicated on flood risk management in their own mother tongue. From
experience with previous projects (e.g. IRMA-SPONGE) it appeared that certain English terms were
interpreted differently by people from different countries, and sometimes even by people from the
same country with different scientific backgrounds. In fact, interpretation differences also exist within
language areas: in some cases it can be difficult to agree on translation of terms for region-specific
concepts (e.g. certain measures) into or from Dutch, German, French or whatever language.
In some cases there is a “many to one” correspondence of words in translation will mask the richness
of meaning in an original language, thus both the French words “crue” and “inondation” would
probably be translated in English as “flood” which misses out the different scale of the events implied
in the native language. The English equivalent to “crue” (a small flood) might be “freshet” but this
regional dialect word (from the England-Wales border area) is not in common national usage and so
would fail to communicate to much of the UK. “Freshet”, however, is in use in North America but
here is describes the flow from a spring thaw resulting from snow and ice melt in rivers located in the
northern latitudes. Literal translation by non-experts may also be misleading, such as “lit mineur”
being the “minor bed” of a river, which in English technical usage actually means “main channel”.
Furthermore in many instances technical jargon is common, such language can make quite an impression
(it sounds scientific), whereas in reality it causes mystification and misunderstanding of what was actually
done. Real language problems (among various European languages), disciplinary jargon, etc. all hamper
true understanding. These may arise from the need for more technical precision in definitions than words
offer in common, vernacular usage, an example is afforded by the words “hazard” and “risk”, which in
common usage are often interchangeable as synonyms. In other cases a jargon word might be taken from
one discipline into another and its common “natural” meaning obscures the intended concept; an example
might be the word “fragility” in the context of “fragility curves” for components of flood defences which
express their probability to fail under a particular flood condition. A further problem with the word
“fragility” in this context is that a flood risk manager does not naturally communicate in public about the
fragility of a flood embankment, but rather about its reliability, strength and safety. Another case which
has been highlighted in FLOODsite comes from importing and developing a method from US practice, the
so-called “Flash Flood Guidance” method. This is not “Guidance” in the meaning of a document of
accepted good practice; rather it is a method to compare the forecast rainfall quantity with a threshold
quantity (adjusted according to recent rainfall) as a basis for issuing flood warnings.
All these examples emphasise not only the need to define the terms (like “Flood”) unambiguously, but
also the need to keep in mind that, some of those we communicate with will have another notion when
hearing this word due to the “common” translation in the language they use at home. Thus an early
output from FLOODsite was the document “Language of Risk” (FLOODsite Report T32-04-01) for
use within the research team as an aid to cross-disciplinary and cross-cultural communication between
the many European partners on FLOODsite. Nevertheless “the risk of language” remains as a
potential barrier to understanding.
6.2 Cooperation and interaction takes time
FLOODsite has involved working together with many people from different institutes, backgrounds,
disciplines and nationalities. To be effective this requires an understanding through meetings and
discussion about basic concepts, but it is easy to underestimate the time needed for this with the potential
consequence that insufficient time is scheduled for this interaction. When this occurs the tendency is for
different parts of the research team to proceed without checking whether all are moving in the same
direction. In Tasks 14 and 18, for example, several face-to-face team meetings took place between the
institutes involves, but still documents produced after a meeting showed discrepancies in understanding
which required further iteration. Although one of the advantages of FLOODsite was that it assembled a
team from many institutes to bring several differing perspectives on each task, this meant the available
budget to each institute per task required team members to work within their institute on more than one
project (although these were often related) and the competing obligations complicated co-operation at a
long distance.
The research investigation treated this as a linear chain; difficulties with one element (the simulation
of the flood discharge response of the Elbe) had the effect of delaying subsequent appraisals both
within the Elbe pilot and in other tasks which were linked to these results.
In retrospect the difficulty with the models could have been regarded more as a means to (re)formulate
research questions, through seeking alternative approaches possible through a dialogue of scenarios.
Thus the sequential chain can be broken without losing sight on the end research and management
question: what if …? Detailed appraisal of the catchment response through runoff modelling is then
pursued “off-line” as a separate (and valuable) topic of research.
Many analyses can run in parallel and only depend to other models to a certain extent. A tiered
approach would be to use simple models with potentially large uncertainties where little knowledge is
available and adopt sophisticated models with less uncertainty on a higher tier where more detail and
complexity are required to represent the processes and phenomena. This could be argued to be an
application of the principle of Ockham’s razor – that of avoiding unnecessary complexity.
A related issue is that for some FLOODsite topics the research involved making advances in depth in a
relatively narrow area of science (for example the work of Task 6 on wave impact forces on embankment
and Task 15 on improving the quantitative estimation of precipitation from radar imagery). This detailed
understanding, however, needs differing skills and experience for it to be set within the broader context of
flood risk management and integrated into practice as a whole through stepping back from details.
Again these issues were identified, monitored and resolved through the project quarterly progress reporting.
These two approaches are reflected in the following sections, with the broader contextual research
needs being treated in Section 7.2, with the more specific issues in the subsequent sections.
One question of widespread interest to policy-makers, water management agencies and the public
alike is how flood risks will change in the future, especially as the climate changes. FLOODsite has
explored methods for analysing flood risk which may be applied in current or future meteorological
conditions and for developing risk management strategies and exploring choices between strategic
alternatives. As the distance of the horizon of the future conditions being explored increases from
present conditions, the nature and source of uncertainty change. However, FLOODsite did not address
directly the future development of the climate and the consequent changes intensity of the
hydrometeorological forcing on the flood risk system; this is the domain of other research projects in
the area of global change.
The Floods Directive makes specific mention in Article 7 that “… flood risk management plans
established in one Member State shall not include measures which, by their extent and impact,
significantly increase flood risks upstream or downstream of other countries…”. FLOODsite included
pilot studies which covered two or more countries (Elbe, Tisza and Schelde). The FLOODsite
methodologies do enable the upstream and downstream influences of flood risk management measures
and instruments to be assessed, but only limited transnational application was possible in the context
of the project. It is recommended that:
R1 In order to support the transnational aspects of the Floods Directive international cross-
comparisons are undertaken of flood risk development and its management through longitudinal
studies of major rivers systems / basins / estuaries and how they have changed over the years.
The Floods Directive makes specific mention in Article 7 of “non-structural initiatives” and of “water
retention and controlled flooding”. In paragraph 14 of the preamble to the Directive it states: “Flood
risk management plans should focus on prevention, protection and preparedness. With a view to
giving rivers more space, they should consider where possible the maintenance and/or restoration of
floodplains, as well as measures to prevent and reduce damage to human health, the environment,
cultural heritage and economic activity”. Thus there is need to understand the real performance of
non-structural measures; the FLOODsite research (e.g. the Elbe Pilot) has demonstrated that the
effectiveness of retention measures varies with the severity of the flood. One of the aims of
FLOODsite was to provide an ex-post evaluation of instruments and measures to inform the
development of new and sustainable strategies. The methodology is now available from the work of
Task 12, but a comprehensive cross-comparison on equal grounds has not yet been performed. Hence
it is recommended that:
R2 The ex-post evaluation of measures and instruments must be applied to the many measures and
instruments in place in different Member States in order to learn from empirical results as many
preferences still rely on belief and prejudice for certain measures and instruments.
The Floods Directive makes specific mention in Article 7 of “potential adverse consequences of
flooding for human health, the environment …”. Amongst the factors which produce such adverse
consequences are the liberation transport and deposition of pollutants in floods including heavy
metals, organics and pathogens. FLOODsite Task 10 did consider pollution from floods, but further
research is needed on the spatial spread of pollutants (good data does not exist from past events) and
their persistence in the environment. It is recommended that:
R3 Further investigation is undertaken to understand the rôle of pollutants in flood risk and its
management. Issues include whether specific groups of organisms suffer from toxic stress,
change in soil chemistry caused by flooding, testing of ecotoxicological relationships based on
field observations, fluxes to and from the soil during flooding and the collection of validation
data on the observed ecological impact of toxic compounds.
It is clear that the development of flood risk management plans for the Floods Directive will require a
method which allows making decisions on when to change to another flood risk management strategy
and the effects of deciding on options for the future. The implementation of alternative and innovative
flood risk management strategies which are more sustainable in the long-term may in some cases have
obvious disadvantages for some stakeholders in the short term. It is recommended that:
R4 Further research is undertaken jointly between natural and social sciences on the transition
processes between the short and long-term view of flood risk management strategies. In
addition, for long-term decisions, research is needed on how to incorporate ‘decision-pipelines’
in the analysis of assessment and management of flood risk; on improving knowledge of
‘breakpoints’ (when developments are such that a certain strategy does not perform adequately
any more); on whether analysis in-time is able replace the assessment criterion of ‘flexibility’;
and on further development of the concept of robustness of strategic alternatives for flood risk
management.
The Floods Directive makes specific mention in Article 7 that: “Flood risk management plans shall
take into account relevant aspects such as costs and benefits …”. FLOODsite Task 9 has produced
guidelines on the evaluation of flood damages which are an important component of the appraisal of
costs and benefits and Task 10 has explored multi-criteria analysis (MCA) of flood risk management
activities. The Directive requires the use of such appraisal tools in developing flood risk management
plans; the FLOODsite research has identified issues that need further investigation and it is
recommended that:
R5 Further research is undertaken on assessing indirect economic damages rather than the
traditional approach to assess direct economic damages only; on the evaluation of intangible
losses such as loss of life, human injuries, environmental and cultural losses; and on
methodologies to integrate all losses over the intended life time of the flood defence system
with associated tools to define tolerable flood risk. In addition further development is needed
on the number of criteria be used in MCA to show a more complete picture of flood risk, MCA
of flood risk reduction measures including the spatial distribution of their impacts and
development of value functions for integration into the decision rule for criteria standardisation
within MCA.
The Floods Directive does not make any specific mention of the public perception of flood risks, of
what are tolerable or acceptable risks or, of how decisions are made on what proportion of national
expenditure should be applied to flood risk management. However, in paragraph 10 of the preamble
to the Directive it states: “Throughout the Community different types of floods occur … The damage
caused by flood events may also vary … Hence, objectives regarding the management of flood risks
should be determined by the Member States themselves and should be based on local and regional
circumstances.” This embodies the principle of subsidiarity in flood risk management policy and
practice. The FLOODsite research did not tackle the issue of how people compare flood risks with
other individual and societal risks, but there is a well known aversion to high consequence events.
We pose the question whether there is a government obligation to be open about standards of risk
managed and mitigated through public expenditure. Does the “public” really want to know what the
true risk is and if they knew would their behaviour or priorities change? The FLOODsite research on
risk perception community behaviour and social resilience has identified several important gaps in
understanding relating to the development and acceptance of flood risk management as a policy
framework rather than traditional flood defence. To make flood risk management become a real part
of people’s lives, not just for policy-makers, it is recommended that:
R6 Further research is necessary on the consequences of the increasing ‘privatisation’ of risks with
regard to social vulnerability and coping with flood risk; on individual, household and
institutional dimensions of social vulnerability; on how to encourage people at risk from
flooding to undertake private precautionary measures; on how to adequately address and explain
urban-rural differences with regard to social vulnerability and resilience; and on what impact
demographic changes (e.g. age, population density, occupancy rates, ethnicity, etc.) have on
preparedness, coping and long-term recovery.
Experience shows that failures in flood defence infrastructure tend to occur due to a combination of
more that one driver. Thus there is a need to consider more than one driving factor and this implies
joint probability analyses. In the analysis of past flood related observations it is necessary to develop
methods to remove seasonality from the samples and any trend due to climatic change. Within the risk
sources, which essentially consist of the combination of extreme events such as storm surge water
levels and waves or river discharges and rainfall, it is recommended that knowledge gaps and
modelling weaknesses are addressed regarding the assessment of the probability of hydro-
meteorological extremes as follows.
R7 Further research is necessary on the physical basis of extrapolation fitted statistical distributions
to very rare extremes; on temporal dependencies (essential for time dependent failure modes
and stepwise determination); on the probabilistic analysis of time-series that show long-term
trends at yearly and decadal scales; on joint probability estimation, particularly for failures in
man-made or natural systems resulting from a combination of factors that exceed a threshold;
and on estimation of error bounds and the explicit presentation of such intervals in the
assessment of extreme values for failure of structures.
The strength and intensity of the risk sources determine the loading transmitted to the risk pathways
and in some cases large uncertainties remain in the transformation models up to the point of
application of defence failure probability functions, for example from uncertainties in the assessment
of morphological change. For the risk pathways, most of the knowledge gaps and modelling
weaknesses were found to be related to the assessment of the performance under environmental
loading of the defence infrastructure. There are weaknesses in conventional fault tree analysis,
particularly for cases where the load and resistance parameters are time dependent and the time
duration of the failure mechanisms as well as their time sequencing and links may be crucial for the
final outcome. It is recommended that knowledge gaps and modelling weaknesses are addressed as
follows.
R8 Further research is necessary to identify the relative importance of failure mechanisms which
have implications for the design and construction of flood defence embankments and “resistant
dykes”. Specific topics include research on the physical basis of overflow and wave
overtopping of embankments (3D-structure, temporal and spatial distribution); on limit state
equations for specific flood defence structures (e.g. single point structures) to include in
reliability analysis; on the influence of structure transitions and damage caused by animals at
earth structures; and on techniques to represent the non-homogeneity of loads and resistance
parameters along a defence line.
The Floods Directive requires the preparation of flood risk maps which show inundated areas, water
level and depth and potentially other information such as velocity. Although computational modelling
of flood movement is a relatively mature area of science, some research challenges remain especially
where hydrodynamic modelling is required as a component of more complex analyses. This may
occur in the context flood event management where the state of the system changes (e.g. with the
failure of an embankment) or in risk analysis and attribution where large numbers of realisations of the
model are needed to sample and explore a highly dimensioned risk system. It is recommended that
knowledge gaps on modelling weaknesses are addressed as follows.
Modelling complex flood risk systems to support long-term decisions is a developing rather than
mature area of practice. Significant outstanding research requirements persist; not least in the
development of structured approaches to appraisal, comparison of alternative adaptive management
options and representation of future scenarios as continuous rather than discrete futures. There is no
consensus on what uncertainty can be tolerated in making flood risk management decisions and there
are issues of scaling both for the spatial scale and resolution appropriate for the decision in hand and
how to up- and down-scale data and nested models. More comprehensive demonstrations of the
overall frameworks for uncertainty analysis will aid their translation from the research domain into
practice.
FLOODsite has made initial but significant steps towards a more structured approach to decision
making. In theory rational decisions are based upon a comparison of the costs and benefits, in terms
of risk reduction, of a range of structural and non-structural interventions; also in the past few years
the importance of robust long term planning has been increasingly highlighted. However, the
evidence used to estimate risks (often against multiple criteria) and costs is almost invariably uncertain
for a variety of reasons. FLOODsite has also made progress in advancing decision support for long
term planning, but before such tools become routine significant future research and development will
be required. As decision making context becomes more complex tools will need to adapt and remain
in advance of the user need. This will include advancing beyond risk information to useful data on
uncertainty and the key sources of the uncertainty. It is concluded that:
R10 Further research is necessary on the areas of decision optimisation, uncertainty analysis and
sensitivity analysis which continue to remain far from routine practical implementation.
Research is needed on genetic algorithms, neutral networks and other system component
emulators and on innovative use of computational processing power to enable whole systems
risk based modelling for whole life system based decisions where there is a trade-off between
model complexity (accuracy) and runtime.
Long-term decisions have to be made in the context of severe uncertainty on the social, economic and
environmental drivers of the flood risk system. The approach adopted and exemplified in the
FLOODsite project has been to use scenarios and story-lines to explore the future. An appropriate
coupling is made of socio-economic scenarios with global emissions scenarios used in climate
simulations. Climate modelling undertaken for the IPCC Fourth Assessment Report has expressed the
range of outcomes of multi-model ensembles in terms of probability distributions. The use of
ensembles to provide information on the range of the uncertainty in future outcomes needs to be taken
through to subsequent impact assessments in a more comprehensive way than has been possible in the
applications of the FLOODsite examples of exploring climate impacts as reported in Section 5.3
above. It is likely that the IPCC Fifth Assessment Report currently under preparation will explore
further the issue of uncertainty in climate predictions and its use in impact assessments.
R11 Research on radar networking, the use of additional parameters (e.g. polarimetry) and additional
data sources and the characterisation of radar QPE uncertainty all have the potential to improve
the quality of precipitation forecast information for flash floods.
More generally FLOODsite has explored methods for forecasting flash flooding in small and medium
basins and this has identified the need for protocols for flash flood observations and the potential value
of systematic archives of flash flood events across Europe. The flood forecasting approaches would
benefit from better methods for defining soil moisture conditions at the onset of the event and the
remote sensing of soil moisture status by means of satellite observations may represent a promising
data source for this purpose, at least for some European hydro-climatic environments. The threshold-
based approaches developed in FLOODsite could be further refined to incorporate knowledge about
flash flood related damages in a Bayesian utility function minimization.
The management of flood emergencies will draw upon resources both physical and institutional
developed through contingency planning. Currently evacuation models such as those developed in
FLOODsite are rarely used to inform flood emergency plans developed by emergency planners in
Europe. It is essential that flood risk managers work more closely with emergency planners in order to
achieve this but it is acknowledged that more work needs to be carried out to make evacuation models
more “user friendly” so that they are used more widely. For development of these technologies:
R12 Research and development of real-time decision support could concentrate on forecasting flood
risk through agent-based approached as other flood forecasting information becomes available
(rather than just forecasting floodwater levels) and thus enable the emergency management
agencies to prioritise their response based on real-time risks. The decision support should
incorporate uncertainties in the flood defence infrastructure performance as well as the
uncertainties from the real-time hydro-meteorological situation. Such a support tool could also
be developed into flood event scenario training environment for both decision-makers and
emergency managers.
This section of our final report is a reflection on the Consortium’s experience in FLOODsite as a
multi-disciplinary, multi-institute project and is set out for others to consider in undertaking similarly
sized projects in the future. These issues have not been subject to rigorous testing or analysis but by
their nature spring from our collective experience.
The project also had two advisory boards one on scientific and technical issues and the other providing
an external stakeholder view on application and implementation. Members of the Management Team
participated in one or the other of these Boards, which also drew members from outside the immediate
research team. The Scientific and Technical Advisory Board (STAB) reviewed the project science on
several occasions, taking reports from and questioning the task leaders; the STAB then recommended
means of improving some Task outputs. The Applications and Implementation Advisory Board (AIB)
met annually, making recommendations on the take-up of the project science. The scrutiny and
advice offered to the project team through this governance structure was helpful and made a positive
contribution to the final outcome of the work.
In addition to this project-led scrutiny of the project progress, DG Research organised an independent
external evaluation of the project annual reports.
Amongst the issues considered in Young FLOODsite were the most suitable research conditions,
facilitating and strengthening the exchange of information and experiences. It also provided a
platform for discussion of research results motivating and elaborating dissertations underpinning
Flood Risk Management. Perhaps most critically it established a basis for long term co-operation
between young researchers within an international framework. Separate web pages were created to
facilitate the networking of Young FLOODsite members and they also arranged an occasional
newsletter “Floodflash”.
The main activities were centred on the annual workshop where time was set aside for a network
meeting and an evening social event. In addition the Year 4 workshop included a Master Class with
two senior professors held an open forum on ‘Flood Forecasting and Warning’. The Master Class was
a unique opportunity for students to address senior scientists, meet peers working on similar topics,
and identify possible collaborations for the continuation of their PhD work.
Overall the activities of Young FLOODsite enhanced the experience of the younger researchers on the
project and were possible because the scale of funding of the IP allowed a “critical mass” of early
stage researchers to be engaged in the research area. In retrospect, it would have been even better for
this activity to have been included as part of the original research plan, with specific institutional
resources negotiated from across the consortium at the outset.
• including, on advice from our AIB, the Project Executive summary in 15 languages,
• translation of pages for some of the pilot study in the relevant national language
• the development of a paperless electronic newsletter (the “e-zine”).
Horizontal integration also has taken place within the project with the chain of flow on information
between tasks for example the work of Tasks 4 and 6 leading into Task 7 and the close coupling of
Tasks 14 and 18, Tasks 16, 17 and 19, and the tasks on flash flooding. The approach of Theme 3 was
to provide a mechanism for the horizontal integration of the FLOODsite knowledge across disciplines
in the context of long-term planning and flood event management.
The scope of FLOODsite research was to make advances in knowledge and understanding on
particular aspects of the overall flood risk system, concentrating in the areas of the original FP6 call
for proposals. For example, FLOODsite was primarily concerned with the flooding sources of rivers,
estuaries and the coast as opposed to from inadequate drainage in urban areas. Nevertheless the
FLOODsite results have a broader application than for just these specific sources, for example the
work on economic damages from flooding. Likewise although there was no specific Task within
FLOODsite on flooding and climate change, our methodologies for flood risk analysis from Theme 1,
for flood risk management in Theme 2 and Frameworks for integration in Theme 3 all can be used to
explore risks and policy responses in a changing climate. This was discussed in more detail in Section
5.2 above.
The pilot sites in Theme 4 were all embedded in areas of local or regional concern in the countries
involved with data and information coming from external agencies and the project results informing
other work in these pilot areas in a variety of ways. Most of the pilots involved close cooperation with
agencies and authorities responsible for flood management in their areas.
8.4.3 Discussion
One of the difficulties encountered in identifying the research links is that there is no comprehensive
central point of information on current and previous projects on flood risk. Although, the ERA-NET
CRUE has now established the CRUISE database of national programmes within their network
(http://www.crue-eranet.net/cruise.asp), this taken alongside the FLOODsite linked project will
necessarily still be incomplete.
The FLOODsite database of linked projects is available to the public from the project website, but
inevitably the information in this and CRUISE will degrade over time as project based websites and
URL’s often cease to be maintained on project completion. As Coordinator of FLOODsite, HR
Wallingford has undertaken to maintain the project website available for 20 years after the project
completion, although in a static form. This will ensure the research results remain accessible for their
useful lifetime to researchers and practitioners.
The linking between FLOODsite and external stakeholders in the research, the research management
and flood management communities was a distinctive feature of the Integrated Project. In the
planning of the research “Networking” was identified as a specific project component with specific
allocation of budget. This networking outside the project team will assist the uptake and
implementation of the research results.
REFERENCES
Allsop, N.W.H.; Buijs, F.; Morris, M.W.; Hassan, R.; Young, M.J.; Doorn, N.; Van der Meer, J.W.;
Kortenhaus, A.; Van Gelder, P.H.A.J.M.; Dyer, M.; Redaelli, M.; Visser, P.J.; Bettess, R.;
Lesniewska, D. (2006): Failure mechanisms for flood defence structures. FLOODsite - Integrated
Flood Risk Assessment and Management Methodologies, Report T04-05-01, available from
www.floodsite.net
Borga M (2008), Real time guidance for flash flood risk management, FLOODsite report T16-08-02,
available from www.floodsite.net
D'Eliso, C. (2007): Breaching of sea dikes initiated by wave overtopping - a tiered and modular
modelling approach. Ph.D. thesis, Dissertation, Civil Engineering Faculty, Leichtweiß-Institut für
Wasserbau, Technical University Braunschweig, University Florence, Braunschweig, Florence,
142 p.; 3 Annexes.
DEFRA (2006): Flood risk to the people. Defra / Environment Agency, R&D Technical Report
FD2321/TR2.
Delrieu G (2008), Radar observation of storm rainfall for flash-flood forecasting - radar structured
algorithm system, FLOODsite report T15-08-02, available from www.floodsite.net
European Commission (2003) “Provisions for implementing integrated projects”, FP6 Instruments
Task Force, available at: http://cordis.europa.eu/documents/documentlibrary/66622311EN6.pdf,
Document dated 12-05-03
EXCIMAP (2007): Handbook on good practices for flood mapping in Europe. Prepared by EXCIMAP
(a European exchange circle on flood mapping), 45 p.; 1 Annex.
FEMA (2003): Guidelines and specifications for flood mapping partners. Appendix D: Guidance for
coastal flooding analysis and mapping. Federal Emergency Management Agency (FEMA), 177 p.
Galiatsatou, P.; Prinos, P.; Sanchez- Arcilla, A. (2008): Estimation of Extremes: Conventional versus
Bayesian techniques. Journal of Hydraulic Research, vol. 46, Extra Issue 2, pp. 211-223.
Görner C, Jatho N, Bernhofer C & Borga M (2008), Satellite observation of storm rainfall for flash-
flood forecasting in small and medium-size basins, Paper presented at FLOODrisk 2008 - The
European Conference on Flood Risk Management, Oxford, U.K, proceedings published by CRC
Press, ISBN 978-0-415-48507-4
Hall J W, Harvey H, Solomatine D & Shresta D L (2009), Development of framework for the
influence and impact of uncertainty, FLOODsite report T20-08-04, available from
www.floodsite.net
Hawkes, P.J.; Gonzalez-Marco, D.; Sánchez-Arcilla, A.; Prinos, P. (2008): Best practice for the
estimation of extremes: a review. Journal of Hydraulic Research, vol. 46, Extra Issue 2, pp. 324-
332.
Meyer, V.; Messner, F. (2006): National flood damage evaluation methods – a review of applied
methods in England, The Netherlands, the Czech Republic and Germany. FLOODsite - Integrated
Flood Risk Assessment and Management Methodologies, Task 9, Braunschweig, Germany, 44 p.
Morris, M.W.; Morris, E.M.A.; Samuels, P.G.; Van Os, A. (2006): www.FLOODsite.net Using the
Web for Research Dissemination, Team Building and Project Management, 7th International
Conference on Hydroinformatics, Sept 2006, Nice, France, 8 pp.
Morris, M.W.; Kortenhaus, A.; Visser, P.J.; Geisenhainer, P. (2008): Modelling breach initiation and
growth. Paper presented at FLOODrisk 2008 - The European Conference on Flood Risk
Management, Oxford, U.K, proceedings published by CRC Press, ISBN 978-0-415-48507-4
Olfert, A, (2007), Guideline for ex-post evaluation of measures and instruments in flood risk
management, FLOODsite - Integrated Flood Risk Assessment and Management Methodologies,
report T12-07-03 available from www.floodsite.net .
Priest, S.; Wilson, T.; Tapsell, S.M.; Penning-Rowsell, E.C.; Viavattene, C.; Fernandez-Bilbao, A.
(2007): Building models to estimate risk to life for flood events. FLOODsite - Integrated Flood
Risk Assessment and Management Methodologies, no. T10-07-10, Task 10, 146 p., 6 Appendices.
Samuels P G (2005), The European Perspective and Research on Flooding, Chapter 2 of River basin
modelling for flood risk mitigation, Eds D W Knight and A Y Shamseldin, Taylor & Francis,
London, ISBN 0415383447, (2005)
Sánchez-Arcilla, A.; Gonzalez-Marco, D.; Doorn, N.; Kortenhaus, A. (2008): Extreme values for
coastal, estuarine, and riverine environments. Journal of Hydraulic Research, vol. 46, Special
Issue.
Stanczak, G. (2008): Breaching of sea dikes initiated from the seaside by breaking wave impacts.
Ph.D. thesis, Dissertation, Civil Engineering Faculty, Leichtweiß-Institut für Wasserbau,
Technical University Braunschweig, University Florence, Braunschweig, Florence.
Van Alphen J, van Beek E & Taal M Eds. (2006), Floods, from Defence to Management, Symposium
Proceedings, Taylor & Francis, ISBN 0415391199
Zhu, Y. (2006): Breach growth in clay-dikes. Ph.D. thesis, Technical University Delft, Civil
Engineering Department, Delft, The Netherlands, 231 p.
*CO = Co-ordinator
CR = Contractor
[NB Names in this table are not necessarily the same as the legal entity names on the contract. These names are
as translated to English or the more commonly used organisation names.]