Paul Sayers

Proceedings of ICECivil Engineering 150 May 2002

Pages 36–42 Paper 12803

Keywordsfloods & floodworks; management;

risk & probability analysis

is head of the Engineers Systemsand Management at HR Wallingford

Jim Hall

is civil engineering lecturer at theUniversity of Bristol and RoyalAcademy of Engineering Post-

Doctoral Research Fellow

Ian Meadowcroft

is risk analyst at the EnvironmentAgency’s National Centre for Risk

Analysis and Options Appraisal

Management of risk is a topic thatreceived increasing attention in the lat-ter half of the twentieth century and isset to be a dominant issue in the twenty-first century. Indeed, it has been sug-gested that risk will be the primary issuethat occupies our modern societies1—anassertion that often seems to be borneout in the media. Specific areas of atten-tion that will be familiar to all civil engi-neers are the management of health andsafety risk,2–4 construction risk,5,6 envi-ronmental risk7 and business risk.8

Moreover, because river and coastalengineering has traditionally been lesscodified than other sectors of the civilengineering industry, risks have alwayshad to be addressed in a fairly explicitway. For example, it has always beenacknowledged that however extreme thedesign loading, more severe conditionscould be encountered.

This paper explores the concept of‘risk’ and related issues such as ‘perfor-mance’ and ‘uncertainty’ as key compo-nents of decision making in floodmanagement and planning. The authorsdiscuss current best practice and look tothe future when risk will form a centralpart of a more integrated ‘whole system’approach to flood management.

Risk in the flooding contextIt has long been recognised that ‘risk’

is a central consideration in providingappropriate flood defences. TheWaverley report following the devastat-ing east coast floods of 1953 recom-mended that flood defence standardsshould reflect the land use of the pro-tected area, noting that urban areas

could expect higher levels of protectionthan sparsely populated rural areas.Today, the term ‘flood risk’ is used in anumber of ways. A range of meanings,derived from either common language orthe technical terminology of risk analy-sis, are in use. These different meaningsoften reflect the needs of particular deci-sion-makers—there is no unique specificdefinition for flood risk and any attemptto develop one would inevitably satisfyonly a proportion of risk managers.Indeed, this very adaptability of the con-cept of risk is one of its strengths.

In all of these instances, however, riskis thought of as a combination of thechance of a particular event, with theimpact that the event would cause if itoccurred. Risk therefore has two com-ponents—the chance (or probability) ofan event occurring and the impact (orconsequence) associated with that event.Intuitively, it may be assumed that riskswith the same numerical value haveequal ‘significance’ but this is often notthe case. In some cases, the significanceof a risk can be assessed by multiplyingthe probability by the consequences. Inother cases it is important to understandthe nature of the risk, distinguishingbetween rare, catastrophic events andmore frequent, less severe events. Forexample, risk methods adopted to sup-port the targeting and management offlood warning represent risk in terms ofprobability and consequence, but lowprobability/high consequence events aretreated very differently to high probabil-ity/low consequence events. Other fac-tors include how society or individualsperceive a risk (a perception that is

Flood engineering in Britain is evolving from traditionalapproaches based on design standards to the development ofrisk-based decision-making, which involves taking account of arange of loads, defence system responses and impacts offlooding. Adopting such a ‘whole systems’ approach enablescombinations of structural and non-structural approaches to beassessed and compared so resources can be targeted to besteffect. Building on methods and know-how already in widespreaduse in river and coastal engineering, risk-analysis techniques arenow being adopted in all areas—from high-level planning basedon outline analysis to detailed designs using high-resolutionsimulation models.This paper explores some of the newtechniques of this emerging approach to flood risk management.

Towards risk-based flood hazardmanagement in the UK P. B. Sayers, J.W. Hall and I. C. Meadowcroft

36 C I V I L E N G I N E E R I N G

TOWARDS RISK-BASED FLOOD HAZARDMANAGEMENT IN THE UK

influenced by many factors including,for example, the availability and afford-ability of insurance) and uncertainty inthe assessment.

Benefits of a risk-based approachThe benefit of a risk-based approach,

and perhaps what above all distinguish-es it from other approaches to design ordecision making, is that it deals withoutcomes. Thus, in the context of flood-ing it enables intervention options to becompared on the basis of the impactthat they are expected to have on thefrequency and severity of flooding in aspecified area. A risk-based approachtherefore enables informed choices to bemade based on comparison of theexpected outcomes and costs of alterna-tive courses of action. This is distinctfrom, for example, a standards-basedapproach that focuses on the severity ofthe load that a particular flood defenceis expected to withstand.

The focus on outcomes coincides withthe move towards performance-basedengineering that has taken hold in theseismic engineering community in theUS9 and which is increasingly recog-nised as a model for efficient provisionof infrastructure. In performance-basedengineering, the range of demands thatmay be placed on a system are explicitlyrecognised and targets are set for theperformance of the system under each ofthese demands. For example, in a moder-ate earthquake the performance targetmay be that structures suffer only superfi-cial damage, whereas in a very severeearthquake it is recognised that somebuildings will be rendered uninhabitablebut essential emergency facilities, such ashospitals, must continue to function. Thisrepresents a much more subtle approachthan the conventional crude engineeringclassification of a system as either ‘failed’or ‘not failed’, or indeed the conventionalextension to consider two performancecriteria: ‘serviceability’ and ‘ultimate’limit states. This conventional engineer-ing distinction between failure and non-failure has, for many years, beentranslated to the design of flood defencesby way of the concept of a ‘design load’,and is almost always expressed in termsof a return period (in years). Within thisrather simplistic engineering paradigm,

design is proceeded by

• establishing the appropriate standardfor the defence (e.g. the ‘100 yearreturn period’ river level), based onland use of the area protected, con-sistency and tradition

• estimating the design load, such asthe water level or wave height withthe specified return period

• designing (that is, determining theprimary physical characteristics suchas crest level or revetment thickness)to withstand that load

• incorporating safety factors, such asa freeboard allowance, based onindividual circumstances.

Over the last decade the limitations ofsuch an approach in delivering efficientand sustainable flood defence solutionshave become clear and act as a barrierto the large scale, long-term planning offlood defence that is now desired. Inaddition, there has recently been muchmore emphasis on the process ofappraisal in order to make choicesbetween options (Fig. 1). These optionscan be quite diverse and offer quite dif-

ferent approaches for managing floodrisk and, again, the simple paradigm of‘design loads’ is rather limiting.Consideration of loads above a designstandard encourages development ofrobust designs that have appropriate lev-els of residual strength and that do notfail in catastrophic modes.

The challenge of flood riskmanagement

Modern flood risk management isaimed at managing whole flooding sys-tems, be they catchments or coastlines,in an integrated way that accounts forall of the potential interventions thatmay alter flood risk. In support of thisaim, the science and technology of floodmanagement has made tremendousprogress in the last half century. Process-based, parametric and statistical modelsdescribing key elements of the floodingsystem (loads, defence response, inunda-tion and impacts) are now available andare continuing to be developed. Thepotential exists for an integrateddescription of the whole flooding systemfrom causes to consequences, includingrisk management measures.

37C I V I L E N G I N E E R I N G

Set out perceived problems

Decision process

Typical actions

Problemidentification

Reviewperformance

Objectivesetting

Optiongeneration

Decisionmaking

Implementation

Monitoring

Riskassessment

Optionappraisal

Valuemanagement

Assess past performanceagainst agreedperformanceindicators

CostLocal monitoring(pre/post storm)Regional monitoringSocial monitoring(public satisfaction)

Consider preferencesWeigh up' and comparesafety, environmental andeconomic performance

Do nothing (walk away)Do something Environmental enhancement Flood warning and evacuation Change policy objectives Capital works Maintenance works Development control

Identifyobjectives/policiesIdentify measurableperformance indicators

Fig. 1. The cyclic option appraisal process—problems are identified, objectives set, options gener-ated and appraised, decisions made, implemented and monitored, leading to improved under-standing and better decisions. The outer oval gives an indication of issues considered at eachstage; the centre circle reflects the weighting-up of risk and value in appraising options

‘‘ ’’The potential exists for an integrated description of the whole flooding systemfrom causes to consequences, including risk management measures

SAYERS, HALL AND MEADOWCROFT

Without appropriate decision supporttools, however, ‘flood risk managers’understandably struggle to handle thecomplexities inherent in integrated man-agement of the flooding system and thehazard it represents. Some of the chal-lenges faced by modern day flood riskmanagers are set out below.

Loading is naturally variableThe loads such as rainfall and marine

waves and surges on flood defence sys-tems are not forecastable beyond, at themost, a few days into the future. Fordesign purposes, loads have to bedescribed in statistical terms. Extremeloads that may never have beenobserved in practice form the basis fordesign and risk assessment.Extrapolating loads to these extremes isfundamentally uncertain, particularlywhen based on limited historical dataand in a climate that may be changing.

Load and response combinations areimportant

The severity of flooding is usually aconsequence of a combination of condi-tions. So, for example, overtopping orbreach of a sea defence is usually a con-sequence of a combination of highwaves and surge water levels, ratherthan either of these two effects in isola-tion. The severity of any resultant flood-ing will be governed typically by thenumber of defences breached or over-topped, as well as the vulnerability ofthe assets and preparedness of the peo-ple within the floodplain. Therefore,analysis of loads and system response isbased on an understanding of the prob-ability of combinations of random load-ing conditions and the systemresponses.

Increasing understanding of systembehaviour has illustrated the importanceof increasingly large combinations ofvariables. For example, the Easter 1998and autumn 2000 flood events were botha consequence of complex spatial/tempo-ral distributions of rainfall rather than asimple, single rainfall–runoff event.10,11

Statistical techniques are commonly usedto address joint loading and responseconditions12 and research into this areacontinues.

Complex and uncertain responses mustbe accommodated

The response of river, coastal andman-made defences to loading is highlyuncertain. Research into topics such asembankment stability and overtoppinghas provided engineers with some toolsfor addressing specific aspects of systemresponse. However, experimental resultsthat these tools are based upon oftenshow that a great deal of scatter andfield verification is scarce. Whilstresearch in the UK and Europe isaddressing the breaching process, predic-tive models will always be limited by theavailability of data on the condition andperformance of defences that may havebeen modified over hundreds of years.

Spatial interactions are importantRiver and coastal systems show a

great deal of spatial interactivity. It iswell recognised that construction offlood defences upstream may increasethe water levels downstream in a severeflood event. Similarly, construction ofcoastal structures to trap sediment andimprove the resistance of coasts to ero-sion and breaching in one area may

deplete beaches down-drift. These inter-actions can be represented in systemmodels, but engineering understandingof the relevant processes, particularlysedimentary processes over long-timescales, is limited. Even where we have adetailed understanding of the physicalprocesses, there may be fundamentallimits to our ability to predict behaviourdue to the chaotic nature of some of therelevant processes and loading.

Flooding systems are dynamic over arange of time-scales

River and coastal systems are subject tochange. Change may impact upon theloads on the system, the response to loadsor the potential impacts of flooding. Itmay be due to natural environmentalprocesses, for example, long-term geomor-phological processes, evolution in ecosys-tems, or intentional and unintentionalhuman interventions in the flooding sys-tem. Social and economic change willhave a profound influence on the potentialimpacts of flooding and the way they arevalued. All of these futures are difficult topredict. Fig. 2 illustrates how variouschanges may influence the consequences

38 C I V I L E N G I N E E R I N G

Defence deteriorationClimate change

Capital ormaintenance works

Development

Improved hazardwarning andpreparedness

Consequences

ProbabilityRepresents degree of uncertainty

Futurerisk

Presentrisk

Futurerisk

Futurerisk

Futurerisk

Fig. 2. Some factors that can change flood risk19

TOWARDS RISK-BASED FLOOD HAZARDMANAGEMENT IN THE UK

and/or probability of flooding, with anindication of their associated uncertainties.

Rivers and coasts are valued in differentways by a range of stakeholders

The aim of risk reduction is central tothe UK government’s Department forEnvironment, Food and Rural Affairs(DEFRA). Environment Agency andDEFRA guidance on appraisal of projectsidentifies the economic, environmentaland social dimensions as relevant to deci-sion making, but there are no well-estab-lished criteria for what might represent‘acceptable’ risks from flooding.13–15

Risk-based approaches provide a sub-tle and adaptable framework for sup-porting decision makers in addressingthese uncertainties. The aim is not toreplace the judgement and expertise ofdecision makers by prescribing preferredoptions, but to make sense of some ofthe complexities and uncertainties out-lined above, in appropriate ways, thatreflect the needs of specific decisionproblems. The concept of appropriate-ness—finding the balance between unin-formed decision-making and paralysis by

analysis, depending on the circum-stances and consequences of any partic-ular decision—is well established in riskmanagement. Within flood management,this concept is being translated into atiered risk assessment methodology. Theremainder of this paper explores howthis range of appropriate risk assessmentmethods is being used in practice.

Applying risk-based techniques toflood hazard management

Analysis of flood risk involves consid-eration of the whole flooding system andacknowledgement of the uncertaintieswithin predictions. Within this process itis necessary to consider each of thestages that lead from the extreme climat-ic conditions that initiate a flood to theindividuals and assets in the floodplainwho suffer the consequences. The disci-pline of this procedure provides tremen-dous insight into the way floodingoccurs and how flood risk may be effi-ciently reduced (Figs 3 and 4). Unlike astandards-based approach, schemes canbe assessed and compared in caseswhere factors other than geometry affectthe effectiveness of the defence—such as

39C I V I L E N G I N E E R I N G

Hazard identification

Source identification(rain, surge, wave)

Pathway identification(flood extent and depth)

Receptor identification(housing and habitat)

Identification of consequences (harm)(material damage, loss of life)

Magnitude of consequences(economic, social or conservation terms)

Quantification of risk(expected annual damage, loss of life)

Significance of risk

Combined probability ofconsequences occuring

Fig. 3. The flood risk assessment process

Load

(w

ater

leve

l, w

ave

heig

ht)

-log [p (exeedance)]

Probability ofloading

Systemmodel +

Dam

age

(£k)

Probability

Integrate togive expecteddamage (i.e. risk)

Loss probability

Floo

d de

pth

(m)

-log [p (exeedance)]

Probability ofsystem response

Dam

age

(£k)

Flood depth (m)

Depth-damagerelation

Fig. 4. How economic flood losses are calculated using risk-based techniques

‘‘’’

Analysis of flood riskinvolves consideration ofthe whole floodingsystem andacknowledgement of theuncertainties withinpredictions

SAYERS, HALL AND MEADOWCROFT

flood gates, which depend on accurate,timely flood warnings and operationalresponse.

Figures 5 and 6 illustrate a risk assess-ment undertaken by HR Wallingford inpartnership with West Dorset DistrictCouncil at West Bay on the south coastof England. The site is small, but com-

plex from the point of view of a systemsanalysis. The low-lying areas at the sitehave been labelled as basins 1–6 in Fig.5. These are also the areas that are occu-pied by people and economic assets atrisk from flooding. These basins may beflooded by the sea and/or fluvial waterin the River Brit. Other uncertainevents, such as failure to open the sluicegates, will modify the flooding process.Fig. 6 provides a cause–consequencechart, which illustrates the ways inwhich flooding can occur. The arrowsdenote the direction from cause to con-sequence. Note that a cause will onlyresult in a given consequence if thecause is sufficiently severe. Thus, forexample, lowering of East Beach willonly result in overtopping of the beachand flooding if the beach lowering andmarine storm are sufficiently severe.

Figure 6 has been further subdividedbased on the commonly usedsource–pathway–receptor model.7 For a‘risk’ to be present there must be asource of the hazard (atmosphericstorms), a receptor (people and proper-ties) and a pathway between the sourceand the receptor. As shown in Fig. 6, aparticular atmospheric storm may causeheavy rainfall and hence high fluvialflows, surge water levels in the sea, highwaves, or any combination of the three.

Therefore, to analyse the flooding sys-tem, a joint probability analysis of all ofthe load conditions was required. Thiswas then combined with numericalmodelling of all of the pathways (systemresponses) to generate probability distri-butions for the depth of flooding inbasins 1–6. Flood depth was then usedin an economic analysis of the risk offlooding.

Once a risk model had been estab-lished, it was then possible to test andoptimise a range of potential designsand management strategies (providedthe model could resolve their influenceon the system behaviour). Thus, it waspossible to examine the relative meritsof investing in new capital works orimproving the maintenance of existingdefences. Scenarios of future land usecould also be tested in order to informthe planning process.

Alternative designs and managementstrategies can be compared on the basisof their cost and risk profiles. There aremany reasons for different options tohave different risk profiles, includinghigher sensitivity to extreme loads aswell as different degrees of uncertaintyabout future performance. Risk-basedanalysis can help differentiate betweenoptions with similar expectedbenefit–cost ratios, for example. As

40 C I V I L E N G I N E E R I N G

Sources Pathways Receptors Harm

Waves

Surge

Rainfall Fluvialstorm

Marinestorm

Lowering ofWest Beach

Collapse of WestBeach seawell

Lowering of EastBeach crest

Blockage ofharbour entrance

Overflow ofRiver Brit

Breach of RiverBrit flood plainembankments

Failure ofsluice gates

Increased waveovertopping/overflow

Increased waveovertopping/overflow Flooding

Residentialproperties

Commercialbusinesses

People

Damage toproperty

Damage tobusinesseconomy

Distress healthimpacts, potential

loss of life

Wave overtoppingof West Beach

Wave overtoppingof East Beach

Loss of widthof East Beach

Failure ofharbour piers

Fluvial flow inRiver Brit

Fig. 6. Cause–consequence diagram for flooding in West Bay—a probability analysis of sources was combined with numerical modelling of pathways togenerate probability distributions of flood depth in each basin

Fig. 5. Aerial photograph of West Bay on thesouth coast of England, indicating low-lyingareas or ‘basins’ at risk from flooding

TOWARDS RISK-BASED FLOOD HAZARDMANAGEMENT IN THE UK

shown in Fig. 7, option B has a greaterexpected benefit–cost ratio than optionA. However, given the greater spread ofuncertainty of the expected value, it ismore likely to fail to achieve a minimumbenefit–cost ratio of unity if pursued.

Development of hierarchical riskassessment and decision making

Table 1 provides an overview of a tieredassessment methodology being developedin the UK. The principle is to provide con-sistent approaches at each level but withincreasing detail of analysis and reducinguncertainties. For each tier of analysis, theappropriate level of detail is based on con-sideration of the type of decision in handand the availability of the required dataand analysis, or its expected cost if it is not

available. Thus, if high-resolution data andanalysis are available at little or no cost,then it is appropriate that it is used in thehigh-level methodologies to reduce uncer-tainty. Insights into the uncertainty associ-ated with a given level of analysis can thenbe obtained by comparing the results ofthe analysis from progressively moredetailed levels.

The high-level method in Table 1 isbased on the minimum information thatis currently available from nationaldatasets (including the EnvironmentAgency’s new national flood and coastaldefence database). Topographical, landuse and property data are now availableat quite high resolutions on a nationalbasis. Information on flood defencesincludes an inventory of assets, a visualcharacterisation of the defence conditionand an estimate of the return period ofthe flood event in which the defence isexpected to be overtopped. Based on thislimited dataset, the high-level methodaims to provide a national estimate offlood risk that can be used to monitorchanging risk from year to year, buildingon methods used in the recent assessmentof national economic assets at risk.16,17

More site-specific decisions areinformed by progressively more detailedanalysis. At the intermediate level, a mod-elling and decision support framework iscurrently being developed for DEFRA andthe Environment Agency to support the

new catchment flood management planprocess.18 At the most detailed level arehigh-resolution simulation models thatcombine simulated time series of loads(for example, rainfall) with high-resolu-tion models of runoff, channel con-veyance and floodplain inundation. Boththe intermediate and detailed level analy-sis have recourse to high-resolution topo-graphic data and each considers multipledefence failure mechanisms.

A look to the futureThe techniques and application of

flood risk assessment and management inthe UK are in the midst of a period ofsignificant change. This paper representsa snapshot taken at an instant when thedirection of change has been clearlyestablished and some of the componentsof future techniques are in place but thefull systems approach has yet to be imple-mented in practice. At the moment,aspects of UK practice that represent bestpractice internationally, for example mul-tivariate statistics, are being disseminatedmore widely to practitioners.19

Meanwhile, best international practice,for example recent Dutch work on relia-bility assessment of dyke rings,20,21 isbeing identified and adapted for the par-ticular conditions in the UK. New dataacquisition, analysis, communication andpresentation techniques are being testedand applied. Emerging issues, such as theimpact of climatic and other long-termchange, are being assimilated into therisk assessment framework. This work isbeing conducted with the overall aim ofintegrated risk assessment and continu-ous active management of whole rivercatchments and coastal flooding systems.

Integrated flood risk management is acomplex endeavour, which will need to besupported by computer-based tools thatenable analysis of the whole system, eval-uation of the consequences of strategicintervention and coordination of interven-tion activities. A flood risk managementdecision support system will include

• an inventory of data, distributeddatabases and metadata to enableusers to judge the pedigree of data

• knowledge bases that guide decisionmakers through key processes, such asselecting risk assessment methodologies

41C I V I L E N G I N E E R I N G

Level Decisions to inform Data sources Methodologies

High • national assessment of • defence type • generic probabilities of economic risk, risk to life • condition grades defence failure based onor environmental risk • standard of service condition assessment and

• prioritisation of expenditure • indicative floodplain maps crest freeboard• regional planning • socio-economic data • assumed dependency• planning of flood warning (including Ordnance Survey between defence sections

Address Point/Focus datasets, • empirical methods toaggregated agricultural and determine likely flood extenttransportation datasets)

• land use mapping

Intermediate As above plus As above plus • probabilities of defence• flood defence strategy • defence crest level and other failure from reliability

planning dimensions where available analysis• regulation of development • joint probability load • systems reliability analysis• maintenance management distributions using joint loading

• floodplain topography conditions• detailed socio-economic data • modelling of limited

number of inundation scenarios

Detailed As above plus As above plus • simulation-based reliability • scheme appraisal and • all parameters required analysis of system

optimisation describing defence strength • simulation modelling of• synthetic time series of inundation

loading conditions

Table 1. The tiered risk assessment methodology being developed in the UK

Pro

babi

lity

dens

ity

Performance indicator (Benifit-cost ratio)

Expectation option A

A B

Expectation option B

321

Fig. 7. Risk profiles for different options—option B has a greater expected benefit–costratio than option A but is more likely to fail toachieve a minimum benefit cost ratio of unity22

‘‘ ’’Integrated flood risk management is a complex endeavour, which will need to besupported by computer-based tools

SAYERS, HALL AND MEADOWCROFT

• tiered risk assessment methods, frombroadly applicable high-level methodsto powerful simulation tools, whichenable a wide range of possible riskmanagement actions to be evaluated

• techniques for representing uncer-tainty in data, models and predic-tions, and accounting for it indecision making

• open architecture modelling supportsystems that help decision makersapply best practice methodologies

• a geographical information systems

(GIS) interface that presents riskinformation to a range of decisionmakers and stakeholders, includingthe public, in an appropriate format.

Technology on its own will not result inintegrated flood risk management. Itneeds to be accompanied by continuedevolution of existing expertise and experi-ence. It will require organisational changeand the building of capabilities within theflood risk management industry, and somechanged behaviours on the part of the

general public. Finally, if integrated floodrisk management is to be a product ofcomplementary actions of many individu-als and organisations, then the generalunderstanding of the objective, processand best practice of risk managementamong those individuals and organisa-tions will have to be enhanced. It is hopedthat this paper will contribute to that aim.

AcknowledgementsData for the West Bay case study were

provided by West Dorset District Council.

42 C I V I L E N G I N E E R I N G

What do youthink?

If you would like to comment onthis paper, please email up to

500 words to the editor at simon.fullalove@ice.org.uk.

If you would like to write a paper up to 2000 words about your own

experience in this or any relatedarea of civil engineering, the

editor will be happy to provide any help or advice

you need.

References 1. BECK U. Risk Society:Towards a New

Modernity. Sage, London, 19922. INSTITUTION OF CIVIL ENGINEERS. The

Management of Health and Safety in CivilEngineering.Thomas Telford Publishing,London, 1995.

3. BARNARD M. Health and Safety for Engineers.Thomas Telford Publishing, London, 1998.

4. FINK S. E. Health and Safety Law for theConstruction Industry.Thomas TelfordPublishing, London, 1997.

5. GODFREY P.S. Control of Risk: a Guide to theSystematic Management of Risk fromConstruction. CIRIA, London, 1996, SpecialPublication 125

6.HALL J.W., CRUICKSHANK I. C. and GODFREY

P. S. Software-supported risk managementfor the construction industry. Proceedingsof the Institution of Civil Engineers CivilEngineering, 2001, 144, 42–48.

7. DEPARTMENT OF THE ENVIRONMENT,TRANSPORT AND THE REGIONS,ENVIRONMENT AGENCY AND INSTITUTE OF

ENVIRONMENTAL HEALTH. Guidelines forEnvironmental Risk Assessment andManagement, 2nd edition.The StationeryOffice, London, 2000.

8. INSTITUTE OF CHARTERED ACCOUNTANTS IN

ENGLAND AND WALES. Internal control: guid-ance for directors on the combined code.Institute of Chartered Accountants inEngland and Wales, London, 1999.

9. STRUCTURAL ENGINEERS ASSOCIATION OF

CALIFORNIA. Vision 2000—a framework forperformance based design. SEAOC,Vision2000 Committee, Sacramento, California,1995.

10. BYE P. AND HORNER M. Easter 1998 Floods:Report by the Independent Review Team tothe Board of the Environment Agency, vol-umes 1 and 2. Environment Agency,Bristol, 1998.

11. ENVIRONMENT AGENCY. Lessons Learned:Autumn 2000 Foods. Environment Agency,Bristol, 2001.

12. HR WALLINGFORD. The joint probability ofwaves and water levels JOIN-SEA: a rigorousbut practical new approach. HRWallingford,Wallingford, 1998, HRReport SR 537.

13. DEPARTMENT FOR ENVIRONMENT, FOOD AND

RURAL AFFAIRS. Flood and Coastal DefenceProject Appraisal Guidance—Approaches toRisk.The Stationery Office, London, 2000,Report FCDPAG4.

14. ENVIRONMENT AGENCY. An EnvironmentalVision. Environment Agency, Bristol, 2000.

15. ENVIRONMENT AGENCY. Framework forChange—Reducing Flood Risk. EnvironmentAgency, Bristol, 2001.

16. DEPARTMENT FOR ENVIRONMENT, FOOD AND

RURAL AFFAIRS. National appraisal of assetsat risk from flooding and coastal erosion.TheStationery Office, London, 2000, HRWallingford Report.

17. DEPARTMENT FOR ENVIRONMENT, FOOD AND

RURAL AFFAIRS. National appraisal of assetsat risk from flooding and coastal erosion,including the influence of climate change.The Stationery Office, London, 2001,Halcrow Report.

18. HR WALLINGFORD. Catchment flood man-agement plans—a best practice approach.HR Wallingford,Wallingford, 2001, HRReport.

19. HR WALLINGFORD. Risk, performance anduncertainty in flood and coastal defence, areview. HR Wallingford,Wallingford, 2002,HR Report SR 587 (second draft).

20.TECHNICAL ADVISORY COMMITTEE FOR

WATER DEFENCE. From ExceedanceProbability to Flooding Probability. TAW, DenHaag, 2000 (in Dutch).

21.VRIJLING AND VAN GELDER. Policy implica-tions for uncertainty integration in

design. Proceedings of the 8th InternationalSymposium on Stochastic Hydraulics, 2000.

22. MEADOWCROFT I. C., HALL J.W. AND

RAMSBOTTOM D. M. Application of risk meth-ods in flood and coastal defence—a scopingreport. HR Wallingford,Wallingford, 1997,HR Report SR 483.

Bibliography1. INSTITUTION OF CIVIL ENGINEERS. Learning to

live with rivers. ICE, London, 2001, FinalReport of the ICE’s presidential commis-sion to review the technical aspects offlood risk management in England andWales.

2. MORRIS M.AND SIMM J. Construction Risk inRiver and Estuary Engineering.ThomasTelford Publishing, London, 2000.

3. SAYERS P. B., GOULDBY B. P. AND SIMM J. Risk,performance and uncertainty in flood andcoastal defence—a review. HR Wallingford,Wallingford, 2001, HR Report SR 587.

4. SIMM J. AND CRUICKSHANK I. Construction Riskin Coastal Engineering.Thomas TelfordPublishing, London, 2000.