A gis based flood risk assessment tool supporting flood incident management at the local scale

A GIS-based Flood Risk Assessment Tool:

Supporting Flood Incident Management at

the local scale

Meghan Alexander, Christophe Viavattene, Hazel Faulkner and Sally Priest Flood Hazard Research Centre, Middlesex University

July 2011

FRMRC Research Report SWP3.2

Project Website: www.floodrisk.org.uk


Supporting Flood Incident Management at the local scale

i

Document Details

Document History

Version Date Lead

Authors

Institution Joint Authors Comments

001 25th July

2011

Meghan

Alexander

FHRC Christophe Viavattene,

Hazel Faulkner and

Sally Priest

Draft - complete

002 26th

July,

2011

Meghan

Alexander

FHRC Christophe Viavattene,

Hazel Faulkner and

Sally Priest

Submitted to

FRMRC for review

and comment

Statement of Use This report is intended to be used by researchers working on decision support tools in flood risk

management. It describes the construction of a GIS-based flood risk assessment tool, trialled with

emergency professionals in the UK. A more detailed methodology is presented in a companion

document (Alexander et al. 2011). This document presents the feedback from emergency

professionals and some practical recommendations for future tool development.

Acknowledgements This research was performed as part of a multi-disciplinary programme undertaken by the Flood Risk

Management Research Consortium. The Consortium is funded by the UK Engineering and Physical

Sciences Research Council under grant GR/S76304/01, with co-funders including the Environment

Agency, Rivers Agency Northern Ireland and Office of Public Works, Ireland.

Disclaimer This document reflects only the authors’ views and not those of the FRMRC Funders. 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 FRMRC

Funders nor any FRMRC Partners is liable for any use that may be made of the information.

© Copyright 2009 The content of this report remains the copyright of the FRMRC Partners, unless specifically

acknowledged in the text below or as ceded to the Funders under the FRMRC contract by the

Partners.



ii

Summary

AIM The original objective of this research was to develop an exploratory tool that addressed the

potential for incorporating vulnerability assessments into risk mapping; ‘bolting’ this information onto

local, urban scale flood modelling developed within Phase 1 of the Consortium and mapping these

data at a relevant scale to emergency professionals (Wilson, 2008). Preliminary discussions with

professional stakeholders (i.e. Category One Responders, as identified under the Civil Contingencies

Act, 2004) supported Wilson’s (ibid) statement that the household scale of vulnerability and risk

mapping was desirable; however, responders acknowledged not only the impracticality of this goal,

but also the dangers of relying on household information given the highly variable and dynamic

nature of vulnerability, (unless derived from up-to-date databases, such as the health service and

utility companies). Furthermore, the Data Protection Act (1998) constrains the storing, sharing and

mapping of personal information. The vulnerability component in this tool is therefore reliant upon

census-derived data. Rather than allowing this to become a restriction, this research sought to

explore the ways in which potential indicators of vulnerability and a composite vulnerability index

(the Social Flood Vulnerability Index, SFVI after Tapsell et al., 2002), could be adjusted and

manipulated to suit the varied tasks carried out by professional end-users, in the context of Flood

Incident Management (FIM). A key research question regarded the utility of social vulnerability

assessment and whether more interactive engagement with vulnerability data could facilitate its

usage within FIM decision making.

METHODS This study sought to engage with emergency professionals throughout the research

process i.e. both pre and post tool completion. Preliminary interviews with professional stakeholders

(i.e. Category One Responders) were conducted to elicit; i) Professional views on vulnerability, its

application in decision making and how it should be assessed; ii) As the target end-user of the tool,

responders were asked to rate a number of design ideas and make further suggestions on how a tool

of this nature might be packaged together. From these discussions, the tool was designed and

constructed and then demonstrated to a sample of responders originally interviewed. Professional

feedback from this has been used to steer a number of more practical recommendations for future

tool-developers.

“THE TOOL” was approached almost as an ‘eggs all in one basket’ to see how different end-user react

and rate different design features, to inform some practical recommendations for how such a tool

might be more specifically tailored in real-life applications. Datasets are stored within a Personal

Geodatabase constructed in ArcCatalogue and the tool itself, has been designed within, and launches

from, the ESRI product ArcMap. Rather than simply allowing layers to be added and removed, this

tool enables users to manipulate these layers to suit their needs and perform calculations on the data

to produce vulnerability and risk profiles from a number of flood scenarios. This interactive nature

seeks to engage the end-user to become actively involved in the assessment process and map

production. The tool provides a means for integrating the informed subjectivities of the end-user,

with the objectivity of the ‘scientific expert’ that is inherently built within the tool. The tool has been

developed for two case studies: Cowes, Isle of Wight and Keighley, West Yorkshire. The tool is

designed with three key interfaces isolating hazard and vulnerability and allowing the user to bring

these together in the calculation of risk.

The Hazard Interface: Utilises the model outputs from 1D-2D modelling developed for Cowes, Isle of

Wight (Allitt et al., 2009) and Keighley, West Yorkshire (Chen et al. 2009) and provides a range of

scenarios for pluvial flooding. Users can adjust hazard thresholds, ‘clean’ the map to view flooding to

road network and property; where the latter offers two hazard models based on risk to life or depth-

damage thresholds. The user can also launch an interactive flood animation.

The Vulnerability Interface: Utilises existing census data (2001) and adapts the original SFVI

methodology, allowing users to adjust the spatial scale at which relative vulnerability is calculated.

Users can view indicators in isolation, alongside rationales, and construct their own vulnerability index

where selection and weighting of indicators is user-defined.



iii

The Risk Interface: Brings together the hazard and vulnerability models at the property scale. User

can define the weighting between hazard and vulnerability and view an automated property and

people count to summarise risk categories.

FINDINGS Options to animate and interact with flood inundation modelling rated highly on the end-

users ‘wish list’ and professionals commented on the application-potential of this feature for

exercising/training, planning and responding to flood events. The indicator/indices approach to

measuring and monitoring social vulnerability is widely acknowledged as limited for informing

emergency response, which requires accurate and timely information at the household scale: Only

during broad-scale flood events requiring strategic decision making (such as 2007 floods), could this

form of assessment be helpful for prioritising stretched resources. For the purpose of planning and

targeting future mitigation strategies vulnerability indicators are deemed valuable, providing there is

a clear expert-declared rationale underlining each potential indicator. Professionals interviewed

highly rated the option to view vulnerability indicators in isolation, rather than as a product score (e.g.

the SFVI, after Tapsell et al., 2002), with the additional option to adjust the spatial scale at which

relative vulnerability is calculated. The ability to ‘build their own’ vulnerability index, incorporating the

user’s informed subjectivities concerning the relative importance of each indicator, was highly

regarded for the right context (i.e. not during their ‘response’ phase of professional activities).

Furthermore, the interactive nature of the Risk interface of the tool allowed the user to explore the

variability in the risk picture, by varying the weighting assigned to hazard and vulnerability for

assessing local risk; this feature was deemed particularly useful for professional training.

RECOMMENDATIONS

� Design flood scenarios (including different flood drivers and return periods) are useful for

planning and training and exercising; however, the worst case scenario is essential.

Visualising flood scenarios is a useful tool for prompting proactive thinking (rather than

merely reactive), and providing a visual example of the possible spatial patterning of the

flood. Tools supporting emergency response should enable an interactive feature for the

user to manipulate river/rainfall levels to mirror incoming real-time information.

� Interactive animation is in invaluable means of depicting the spatial-temporal patterning of

the flood (as it accumulates, recedes and ponds): Where and when are key questions in

planning for and responding to flood incidents. Providing the animation is a slick operation

i.e. can be sped-up/slowed-down/paused etc. Summary tables should be used to support the

visualisation and give a rapid summary of what is being displayed on screen (e.g. flood start,

end, peak).

� Future vulnerability assessment needs to be malleable and flexible to the broad base of FIM

practitioners, with varied demands, responsibilities and professional constraints. Interactive

assessments and map-making is a powerful tool for not only communicating science at the

professional interface, but integrating professional knowledge and supporting the increasing

demands on FIM in the UK.

� Pragmatic flood research requires the stakeholder to become an active participant in the

research process; thereby acknowledging the importance of a two-way knowledge exchange

in facilitating the uptake of new ideas and tools in practice, as well as prompting new

thinking. This constitutes a broader effort to enhance the translation of science at the

practitioner interface.

� The requisite for simplistic tools is tied to a contentious debate concerning the dualistic

meaning of simplicity: Do practitioners require simplistic-user-friendly tools or simplistic-

information tools. It is apparent from the professional interviews reported on in this report,

that simplistic-user-friendly tools are essential. This is due mainly to the varied demands

placed on professionals which mean flood-related matters are but one, small component of

the day-job; and inexperience, and a lack of confidence and self-efficacy in using new



iv

software. There is mixed support for the interpretation that the desire for simplicity reflects

a desire for simplistic-information. For instance, most responders acknowledged the

difficulties of defining and mapping vulnerability and appreciated the nuances in this

concept; on the other hand, while acknowledging uncertainties in flood modelling, some

responders seemed reluctant to engage with it – this may be because they found it difficult

to visualise how uncertainty might be integrated in mapping, or because uncertainty is

common place in the day-job and therefore not regarded as an issue. There is a need for

further research on this matter which implies a greater tension involving the translation of

science to practitioners and the professional context.

REFLECTIONS Although this research illustrates that there are possibilities for extending and utilising

vulnerability indicators within pragmatic decision support tools, the response from professional

stakeholders demonstrate that there are some fundamental limitations of this approach. In particular

there is a mismatch between the scales available for assessing flood hazard and social vulnerability,

which inherently constrain local risk assessments. On one hand, flooding can be modelled and

depicted through space and time and thus easily transformed into useable visualisations (e.g.

animation), capturing the dynamism of the hazard; conversely, vulnerability assessment remains

constrained by a static-snapshot-layering approach. Professionals interviewed in this study heavily

discussed these limitations (e.g. the decadal timescale of the census and issues of accuracy), which

means that this form of vulnerability assessment is considered useful only as a ‘broad brush’ approach

to painting an areas social make-up. The two approaches in hazard and vulnerability appraisal are

divorced in terms of scale (spatial and temporal) and must be equally resolved in order to inform a

meaningful risk assessment.

It has been argued that more meaningful assessments of vulnerability could derive for instance, from

the use of existing social data regarding people’s attitudes and responses to flood risk (Twigger-Ross,

2010). One might question who has the authority to impart this information and define what is

meaningful; is it the professional stakeholder, the academic researcher or the role of the community

under scrutiny? This report highlights potential methods for adapting existing vulnerability

approaches (using the SFVI as an example) and the potential for integrating this flexibility and end-

user control within a decision support tool. Limitations in the area-wide approach, resulting from the

dependency upon existing census data, could be resolved with the inclusion of locally-informed

information; whether this arises from exiting social surveys regarding flood experience and responses

specifically or from more generic discussions with the public in at-risk locations. Social science could

facilitate this process of seeking more meaningful assessments of social vulnerability.



v

Table of Contents

1 Introduction ............................................................................................................................................. 6

2 The professional context: Flood Incident Management in the UK .................................... 6

2.1 A framework for Integrated Emergency Management (IEM)…………………………7

2.2 Extending the “toolkit” for FIM .............................................................................................. 8

2.3 A note on visualisation ............................................................................................................ 10

3 Research design .................................................................................................................................... 11

3.1 Study sites ..................................................................................................................................... 13

3.2 Preminary interviews: The end-user “wish list”........................................................... 17

4 A flood risk assessment tool ............................................................................................................ 23

4.1 The Hazard Interface ................................................................................................................ 25

4.2 The Vulnerability Interface.................................................................................................... 31

4.3 The Risk Interface ...................................................................................................................... 35

5 Evaluating the tool: Professional feedback ............................................................................... 37

6 Recommendations for future tools in practice ........................................................................ 49

7 Reflections………………………………………………………………………………………………………..52

8 Conclusions ............................................................................................................................................. 54



6

1 Introduction

Flood risk is defined in the UK by the probability of flood occurrence and its potential consequences

(Flood and Water Management Act, 2010). The risk management approach adopted in the UK is seen

as a holistic and sustainable strategy (EA, 2009); as such, alleviating risk is both a matter of addressing

the likelihood of flooding (e.g. flood defences and development planning) and the potential impacts

(e.g. forecasting and warning, and emergency management). Risk is thus a function of the hazard (i.e.

frequency and magnitude of the flood) and the vulnerability (susceptibility) of the receptor exposed to

the hazard. Mapping has become the keystone for flood risk management and communication in

representing the spatial relationship between hazard and vulnerability and resulting risk. In light of

growing climate change concerns and the predicted escalation of flooding, the EU Floods Directive

(2007) means member states are now required to develop and utilize flood hazard and risk mapping to

inform flood risk management plans by 2015. The purpose of these plans is to steer strategies towards

prevention, protection and preparedness, in attempts to alleviate future costs from flooding. This

research seeks to build upon this cornerstone of mapping and develop an interactive, GIS-based tool

for local-scale flood risk assessment.

This report establishes the context of Flood Incident Management (FIM) within the UK and describes

some of the current tools which are employed in practice to support decision making. The research

process is reviewed in section 3. Two study sites have been selected based on i) the availability of

existing flood modelling developed within FRMRC Phase 1 and ii) the socially-contrasting nature of

these locations in Keighley, West Yorkshire and Cowes, Isle of Wight Hampshire. Feedback from

preliminary interviews with professional stakeholders (namely Category One Responders: section 3.2)

was used to inform the construction of the GIS-based flood risk assessment tool. The completed tool is

illustrated and explained in section 4 and followed with a discussion of the secondary-feedback from

emergency professionals to whom the tool was demonstrated. A number of practical

recommendations are outlined in section 6 and it is suggested that further tailoring is required to

launch such a tool in practice. Limitations in the vulnerability approach adopted in this study are

acknowledged and it is recognised that this exists partly because the spatial and temporal scales of

vulnerability and hazard assessment are mismatched, making the assessment of local risk problematic.

2 The professional context: Flood Incident Management

(FIM) in the UK

A central objective to this research was to tailor and trial a flood risk assessment tool with potential

professional end-users; namely Category One Responders concerned with flood incident management.

This section reviews this professional context, which is situated within a framework of Integrated

Emergency Management (IEM). There has been a growth of decision support tools, for data storing,

mining, sharing and visualisation and a sample of these are presented in section 2.2. Although the tool

developed here is not intended for real-life application at this stage, an understanding of what is currently

‘on the market’ was deemed useful for steering the design of the tool (alongside stakeholder

recommendations) and highlighting current gaps – in this instance, the lack of interactive assessment for

social vulnerability for flood risk assessment. This section concludes with a note on visualisation, drawing

heavily from Cartography literature to highlight the scope for continuing tool development which utilises

visualisation and interactivity; to facilitate the communication and knowledge transfer across the

scientific-practitioner divide, ownership and prompt new modes of thinking.



7

2.1 A framework of Integrated Emergency Management (IEM) Risk is acknowledged as changing in dynamic ways during the course of an emergency

1 event; ‘New risks

emerge, previously recognised risks recede and the balance between risks change continuously’ (HM

Government, 2004). Balancing risks is a crucial part of planning, training and exercising emergency

situations. While the UK is faced with a number of climatologically-related hazards (snow storms, heat

waves, drought) flooding has arguably been at the forefront of discussions and subject to recent policy

amendments. Emergency management in the UK is organised through the statutory framework of the

Civil Contingencies Act (HM Government, 2004), set within the wider context of integrated emergency

management (IEM). The aspiration of IEM is to facilitate joined-up, multi-agency response from the local,

regional and national scale. Local Resilience Forums (LRF) provide the setting for multi-agency discussion,

planning and exercising for the array of threats posed to civil protection in the UK. LRF membership

consists of representatives from Category One and Two Responder groups (Table 1). Regional Resilience

Forums (RRF) similarly aim to enhance multi-agency coordination for regional-wide emergency

preparedness and mediate new initiatives and policy amendments effected in Central Government, with

the region, and from the region to local responders.

Table 1: Category One and Category Two Responders

Category One Responders Category Two Responders

Local Authority

A county council, district

council; including emergency

planning

Utilities

Electricity; Gas; Water and

sewerage; Public

communication providers

Emergency Services

A chief officer of Police: A chief

constable of British Transport

Police force: A Fire and Rescue

authority: Maritime and

Coastguard agency: Ambulance

service

Transport Operators

Network Rail: Train operating

companies: London

underground and Transport

for London: Airport

operators: Harbour

authorities: Highways Agency

Health authority

A NHS Truest: A Primary Care

Trust (PCT): Health Protection

Agency (HPA): Foundation

trusts: Acute trusts

Health and Safety

Executive

Environment Agency Strategic Health

Authority

Category One responders are central to emergency response and are subject to the full set of civil

protection duties: These include putting into place a number of plans (contingency, emergency, business

continuity); establishing arrangements for sharing information (with other responders, as well as the

public); meeting the responsibilities within the existing remits of the agency, as well as ensuring the

‘joined-up’ working across agencies. Category Two responders function as ‘cooperating bodies’ to the

Category One response and are principally tasked with sharing information and advice with all necessary

responders involved (for full details of civil protection duties consult the CCA, 2004). Integrated

emergency response is coordinated through a tiered command structure; from broad scale events

requiring a strategic response, through to the tactical and operational command required on the ground.

Within the remits of the individual agencies involved these tiers of command and control are referred to

as gold, silver and bronze, respectively.

1 Where emergency is defined as an event of situation which threatens serious damage to human welfare / or the

environment of a place in the UK (Civil Contingencies Act: HM Government, 2004)



8

In the context of flooding, Multi-Agency Flood Plans (MAFP) necessitate this joined-up working to

coordinate flood response; indeed this process of working together is considered equally as important as

the final product (HM Government et al., 2008). Vulnerable groups are identified and accounted for

within MAFP via the mapping of key facilities; such as schools, elderly care homes, hospitals etc. The

impracticalities of creating inclusive repositories of household-scale vulnerabilities (i.e. keeping it up to

date and issues surrounding data protection), means that emergency planners and responders are

required to build inclusive lists, not of vulnerable groups2 per se, but detailing the appropriate agencies

(and databases) responsible for these groups and pathways for accessing these lists as and when required

(HM Government, 2008); of which the Local Authority and constituent departments of Adult and Social

Care, lead. This central list of partners and contact details are used to infer the potential scale and nature

of the response required in the event of an emergency. The guidance document for identifying vulnerable

people further emphasises the importance of employing this methods as a means of ‘pushing’ warning

messages and ‘pulling’ potentially vulnerable individuals towards the authorities in advance of an

emergency. Vulnerability mapping and the use of Geographic Information Systems (GIS) is highlighted as

an invaluable tool for appreciating both the scale of the response required (e.g. location of rest centres)

and potential impact (i.e. nature of aftercare). This guidance document in particular emphasises the

importance of complementing the mapping of vulnerable groups, with estimated people counts and

appropriate response mechanisms required (HM Government, 2008).

While national flood risk management (and coastal erosion) remains within the remits of the Environment

Agency, the Flood and Water Management Act (2010) assigns Local Authorities (LA) as the lead agency in

response to local flood risk management; including flooding from surface water, small watercourse,

canals, reservoirs and groundwater. Under the Flood Risk Regulations (20093) LAs are required to

complete flood risk assessments by June 2011: Social vulnerability mapping is inexplicitly incorporated

within these maps through the mapping of ‘human health’ (risk indicator) and is based on a property

count, estimated people count (i.e. 2.5 people per property) and the mapping of critical services (e.g.

hospitals: Environment Agency, 2010). It is not a requirement to consider the social make-up of these

properties.

These recent policies, the Flood Risk Regulations (2009), the Flood and Water Management Act (2010)

and earlier suggestions from the Pitt Review (2007), place a mounting pressure on Local Authorities to

identify and map localised flood risk, and coordinate appropriate responses to meet the goals of

sustainable flood risk management. Furthermore, this ‘devolution’ of FRM, places responsibility into the

hands of professional stakeholders with less and less formal training in flood science (Faulkner et al., in

press). There is a need to develop new tools with the ability to not only support the shifting roles of local-

based practitioners and decision making, but also in translating flood science to professional end-users of

this knowledge. It is argued that this translation process could not only facilitate the up-take on new

ideas, knowledge and tools in practice, but also augment practitioner engagement in flood science. In the

long-term this form of translation and focus on pragmatic flood research which seeks to adjoin science

and practitioners, could meet the goals for capacity building and improve local flood risk management

(Defra, 2010).

2.2 Extending the ‘toolkit’ for FIM Information mining is a critical task in FIM. The nature of multi-agency working means professionals

must strive toward a commonly recognised information picture (CRIP) before coming to a collective

decision. Web-based portals have been effectively used to centralise incoming, up-to-date information,

as well as acting as a store house for relevant documents. Hazard Manager is one such example and uses

mapping and real-time weather (e.g. links into the Flood Forecasting Centre for extreme weather alerts)

and incident related information, to facilitate CRIP and joint decision making activities between

2 Where vulnerable people are defined as those that are less able to help themselves in the circumstances of an

emergency. This may include the elderly, young children, mobility impaired, minority language speakers etc. (for a full

list please consultant HM Government, 2008) 3 The Flood Risk Regulations 2009 represent the UKs response to the EC Floods Directive 2007.



9

professional partners (http://www.metoffice.gov.uk/publicsector/hazardmanager). This system was

developed by the Met Office and is considered to be a “one-stop information source”.

The National Resilience Extranet (NRE) has similarly sought to establish itself as an information portal to

facilitate multiagency working through shared knowledge, emergency planning and managing real-time

incidents (Cabinet Office, 2010). The main function of the NRE is to serve data sharing (current, archived

and classified documents), but the system also provides an inclusive contact list of all key responders,

updates on good practice, expertise sharing and an open-shared calendar for organising meetings.

Although this system is currently being used by 350 organisations (mainly local authorities: Cabinet

Office, 2010), it has been greeted with mixed responses. During the course of semi-structured interviews

with emergency professionals in this research, it seems the main failing of the NRE was its attempts to

be a ‘one stop’ tool. Rather than offering limitless capabilities, end users felt that the system would have

been effective if it can kept to its original brief as a storehouse of key documents and contacts only. The

implication of this for the flood risk assessment tool developed in this research, is that it must have a

clear focus; indeed supporting a small number of tasks, but supporting the well, is better than trying to

everything to everyone and hence nothing to no one.

Computer technology is a crucial tool for simulating emergency events and facilitating joined-up

working. A number of tools exist for training and exercising purposes. The HYDRA-MINERVA system used

by Fire and Rescue provide a number of scenarios within fire rescue, to chemical incidents and heavy

snowfall events. Each scenario runs in real-time, feeding-in information from multiple sources, allowing

for fast and slow decision making and tactical and strategic levels of command. Decisions are logged and

audited for the purpose of debriefing – flooding is not currently included in the portfolio of emergency

events simulated in this tool (http://www.hydrafire.org/).

Simulating real-life events is an important part of building-up professional skills and expertise, and

trialling new tools to support decision making. FloodViewer© for instance, was trailed in the UK EA-lead,

national Exercise Watermark 2011. This tool enables end users to view flood information in a dynamic

way, employing functions of zoom, pan and animation, whilst visualising the spatial patterning of

flooding onto vulnerable hot spots (e.g. key roads, hospitals etc.). Users of this tool can select a specific

return period for flooding or view a given water level using a slider control bar, and visualise the corresponding

flood extent. Crucially this tool is designed outside off expensively-licensed products (such as ArcGIS) and

is viewed universally (e.g. online) by multiple decision making partners (Halcrow, 2011).

There has been a growth in the development of commercial products designed to support decision

making in flood risk management. UK consultancies such as Gaist Ltd have sought specialism in

emergency management with products such as Inca (Incident command administrator) and a multi-

agency collaborative tool, ‘Gaist emergency’ (Gaist, 2010). This latter system is specifically designed to

facilitate multi-agency working with universal display, an interactive and user-friendly interface and

provides access to a virtual earth server to access Bing maps. This system is compatible with INCA

(Incident command administrator) tailored to meet the demands of the Fire and Rescue Service and

enables on-the-ground risk assessments (‘tough books’) to be electronically sent and logged into the

INCA interface and used to inform the incident commander and control room.

Socio-demographic data has been exploited for commercial products such as MOSAIC, developed by

Experian Ltd. Details regarding the demographics, lifestyles and behaviour of all individuals and

households in the UK are centralised into this one tool and derived from the census, media and market

research. The tool is essentially based on a classification criterion which identifies 146 person types

and groups these into 69 categories based on lifestyle types; these categories are further reduced into

15 main socio-economic groups (according to MOSAIC Public Sector; Experian 2009). This 3-tier

classification system means that the MOSAIC tool can be deployed at the individual, household or

postcode scale to suit the decision maker’s needs. While this tool was originally designed to support

commercial decision making, it has application potential for emergency management. For instance,

MOSAIC is employed by the Fire and Rescue Service for targeting risk communication and fire safety

campaigns. Other products from this line such as MOSAIC Daytime have application potential for

emergencies. This particular tool records the shift in population patterns from day to evening and is



10

currently deployed for targeting home selling; in an emergency situation this information could prove

highly useful for prioritising response into certain areas.

The tool developed within this research aimed to explore how end-users interacted with an interactive

feature of vulnerability assessment; indeed, when given the choice, do professionals value a

vulnerability ‘product’ (e.g. the social flood vulnerability index), over the option to select and weight

indicators and essentially build their own vulnerability index, adjusted to their professional needs.

Given the partnership with the hazard in governing overall risk, this tool also sought to gauge how

professionals rate different approaches to hazard assessment and the calculation of risk at the local

scale. For example, how might end-users negotiate the balance between hazard and vulnerability

when given an option to weight them within the risk equation? The findings from semi-structured

interviews and questionnaires with professional stakeholders were used to inform some practical

recommendations for real-life application (see section 6).

2.3 A note on visualisation There is an ever-expanding reservoir of accessible, georeferenced data, which has arguably

corresponded with changing scientific and societal demands and applications of these data

(Maceachren, 1998). Visualisation techniques have received considerable attention over the past

twenty years with developments in technology opening new windows into previously inconceivable

approaches. The development of Geographic Information Systems (GIS) for instance, has made

possible the handling (from storing, visualising and analysing) of spatially-referenced data in interactive

ways.

Mapping has become the keystone for flood risk assessment and communication. Presentation is a

crucial element to successful map-based communication (i.e. does the user infer the map information

in the way it was intended?). Visualisation decisions have been shown to exert a profound influence on

the effectiveness of information transfer and the receptiveness of the end-user to this information

(Alphen et al., 2009). Alphen et al. (2009) note the use of colour and the role of social conditioning in

map interpretation; for instance, blue is widely recognised as a sign for water and a scale of red,

orange, green as a scaling for danger. While computer technology has increasingly enabled the

rendering of ‘realistic 3D worlds’ (Kot et al., 2005), it has been stressed that the power of visualisation

rests with its ability to abstract reality (Muehrcke, 1990; cited in DiBiase et al., 1992); for instance

maps can be adjusted to varying scales, employ numerous symbols etc. This enables the viewer to

more readily identify patterns in the data; ‘distinguishing pattern from noise’ (Maceachren and Ganter,

1990).

While traditional, top-down approaches to map making have developed map products for end-user

delivery, there is now a shifting agenda towards tailoring the final product to the end-user themselves.

Visualisation is defined within the Cartography literature under two paradigms; a communication

model and a visualisation model. The first perspective for communication has dominated the field of

Cartography (and other disciplines utilising cartographic principals) and emphasises the value of maps

as communicating ‘what we know’ (Maceachren and Ganter, 1990). Conversely the visualisation

perspective on map function emphasises ‘what is yet unknown’; that is to say, the map is a stimulant

or facilitator for new ways of thinking and in this light has profound implications for exciting scientific

creativity (ibid). The work of Maceachren and colleagues in particular has emphasised the importance

of the visualisation perspective and the associated potential for visualisation techniques in prompting

new ideas and modes of thinking. Visualisation should therefore be viewed as a facilitator, rather than

the end product in itself.

Viewpoints from Cartography have been offered and have proposed a shift from seeking ‘optimal

design’ and visual communication to facilitating visual thinking (DiBiase et al., 1992). Flood science has

thus far focussed on the enhancement of communication i.e. to transfer ‘what we know’, but how

might visualisations prompt new thinking in users? Will this new generation of visualisations and

decision support instruments change the data requirements of end-users in the long-term? McCarthy

et al., (2007) show that user-interaction with new visualisation tools and with tools designers can alter



11

initial assumptions and facilitate the want for new techniques; and most importantly the self-efficacy

of the end-user to utilise new techniques. The expressed desire for ‘simplicity’ voiced by end-users

arguably stems from an uneasiness with new and seemingly complicated tools. For new tools to be

fully integrated and not viewed as ‘the shiny new toy’ (only to be neglected soon after) it is crucial that

any new visualisation tool presented is user-friendly and self-explanatory. Engaging with target end-

users in the design process is a necessary step towards end-user satisfaction. Morss et al. (2005) argue

that the traditional, scientific view of the ‘end-user’ as the passive receiver of knowledge should be

replaced by the inclusion of the ‘end user’, enabling them to become active members of the research

process. This strategy is deemed beneficiary to both science and professional stakeholder groups and

can facilitate co-knowledge production, a sense of shared ownership and encourage the uptake of new

ideas and tools in practice. Visualisation tools in their own right offer the potential for user

engagement with the data and can be considered as useful aids to the decision-making process.

3 Research design

This research is centred on the development of a GIS-based flood risk assessment tool (“the tool”). The

methodological stages of this research are illustrated in Figure 1. Although the tool was not designed for

real-life application, Figure 1 recognises the need for multiple iterations between tool developer and end-

users to tailor the tool to professional cultures and requirements. This report reflects on the first stage of

iteration only and discusses the tool’s construction and evaluation with a select sample of emergency

professionals.

STAGE 1 sought to draw upon existing scientific knowledge to infer what should be included, whilst

equally framing the possibilities for what ‘could’ be included in such a tool. In this research, the scientific

input derived from the interdisciplinary contributions from physical and social sciences to represent the

hazard and vulnerability dimensions of the risk equation respectively. Available to this study were

relatively small-scale flood inundation visualisations, produced as outputs from previous 1D-2D

inundation modelling developed under the auspices of the Flood Risk Management Consortium’s research

(FRMRC Phase1). These model outputs were developed for two UK locations; in Keighley, near Bradford in

West Yorkshire (Djordjević et al., 2005) and Cowes on the Isle of Wight (IOW), in Hampshire (Allitt et al.,

2009). The original objective of this research was to utilise these existing 1D-2D model outputs for pluvial

flooding and address the extent to which vulnerability data can be successfully ‘bolted-on’ to give a

calculation of risk at the local scale. A key research question regards the utility of social vulnerability

assessment and whether more interactive engagement with vulnerability data could facilitate its usage

within FIM decision making. In this first stage, expert consultations and literature review helped inform

the scope for the vulnerability interface of the tool (Wilson, 2008).

STAGE 2 involved the preliminary engagement with emergency professionals. Semi-structured interviews

with the Category One Responders chosen for the study were administered to elicit professional

viewpoints on flood risk assessment, how it is currently assessed and how a decision support tool might

aid current practice. These interviews also helped contextualise the roles and responsibilities of

emergency professionals. A total of 18 professionals participated in preliminary interviews, representing

the Police, Fire and Rescue, Ambulance service, Environment Agency, Emergency Planning (county and

district council), Health Protection Agency and a utility company. Interviews were complemented by

structured questionnaires, which asked respondents to rate the design suggestions proposed by the

scientists (concerning tool functionality and presentation).

STAGE 3 concerned the tool’s construction, which was designed to be exploratory in this first stage (i.e.

not a final product). The tool was written in Visual Basic for ESRI and constructed as an interface to the

commercial GIS application, ESRI’s ArcMap (9.3). Datasets are launched from a personal geodatabase and

automatically situated within the tool’s interfaces for hazard, vulnerability and risk assessment. This ‘GIS-

interface’ approach, aimed to facilitate user-friendliness, whilst maintaining the GIS interactive

capabilities (e.g. zoom, pan). As the user’s interacts with this interface, both the global map in ArcMap

and a corresponding ‘summary window’ on the interface itself are updated.



12

STAGE 4 is essentially the ‘tailoring stage’. Firstly, the completed tool was demonstrated to a sample of

Category One Responders (n=8) and each interface and feature was explained. Responders were also

given the opportunity to interact with certain features. Throughout this process, responders were invited

to give their views on what worked well and whether there was application potential for supporting FIM

decision making; this was essentially an open-interview approach. Finally, following this extensive

interaction and discussion phase, professionals completed a short questionnaire to rate each feature in

turn. All interviews were transcribed and analysed in the qualitative data software NVivo using thematic

analysis (via open and selective coding: e.g. Fereday and Muir-Cochrane, 2006), to locate key themes and

identify different and convergent opinions expressed across stakeholder groups. Professional feedback

from this has been used to steer a number of more practical recommendations for future tool-developers

(section 6).

Figure 1: Summary of research stages



13

3.1 Study sites

This research focuses on two contrasting UK locations in Keighley, near Bradford in West Yorkshire and

Cowes on the Isle of Wight (IOW), in Hampshire. Both locations are exposed to pluvial flooding, resulting

from the interaction between heavy rainfall and failings in managed drainage networks. Pluvial flood

event matrices have been modelled for both areas within the research of the Flood Risk Management

Research Consortium (FRMRC), thus providing the ‘hazard’ details required for a flood risk assessment

tool. This section describes the geographical setting of these case studies, the flood histories and

respective modelling work simulating these.

Moreover, there are some interesting socio-economic differences which make Keighley and Cowes an

interesting comparison. This section addresses these differences drawn from the Index and Multiple

Deprivation (IMD) and less clearly from the Social Flood Vulnerability Index (SFVI). These differences could

prove important to understanding how vulnerability is conceptualised by professionals responsible for

these areas.

3.1.1 Keighley, West Yorkshire

The Stockbridge area of Keighley is situated at a confluence between the River Aire and the River Worth

(Figure 1). Significant fluvial flooding in October 2000 caused damage to 370 residential properties

(Wilkinson, 2007) and represented the worst scale of flooding witnessed in the area for fifty years. While

those registered at the time to receive the EA flood warning had one hour to respond, the event occurred

at 5am local time and was therefore a ‘surprise event’. It took 6-12 months for people to be able to return

to their properties. Post event reviews revealed insurance to be a significant issue, with nearly half of the

affected population lacking either contents of buildings insurance (Wilkinson, 2007). Furthermore,

localised storm events over Keighley have since caused localised flooding in July and August 2003

(CBMDC, 2005).

These events sparked an independent enquiry into water management within the Bradford district which

highlighted the importance of joined-up working, information sharing and need for community

engagement to engender an awareness of ownership of responsibilities for dealing with risk and

mitigations. The Flooding Local Action Plan (FLAP) emerged from this enquiry, which sought to promote

community involvement in flood issues (Cashman, 2009); funding cuts has since meant that the ten FLAPs

set up now cease to exist. However, Bradford city council remains committed to gauging its population’s

awareness of flood risk and suggestions for flood risk management (e.g. current Bradford District online

questionnaire as part of the EU project FloodResilienCities www.bradofrd.gov.uk). Over three thousand

properties are at flood risk within Bradford, with an estimated value of £247m; this figure accounts for

fluvial flood risk only and does not include surface water flooding.

Flood defences have been installed in response to the significant 2000 flood in Stockbridge, including a

levee and reinforcements to the river channel; these flood defences are designed to protect Stockbridge

against the 100 year event (CBMDC, 2005). However, the area remains susceptible to surface water

flooding which has been identified as an increasing problem (EA, 2008). As part of the government’s

Making Space for Water (Defra, 2005a), the River Aire in Bradford and Leeds was one of 15 projects

informing the Integrated Urban Drainage Programme (CBMDC et al., 2008). Hydraulic modelling

conducted in this research suggested a potential rise in surface water flooding by 200% by 2085, resulting

from patterns of climate change and urbanisation.



14

Figure 1: Flood hazard map of Keighley, West Yorkshire (Environment Agency Flood Map)

Flood modelling is integrated in this research and utilised to equate the ‘Hazard’ component of the flood

risk assessment tool. 1D-2D modelling in Stockbridge simulates the interaction between urban sewer

systems and overland flow, based on the SIPSON-UIM model (Chen et al., 2009: Djordjević et al.,2005).

Whereas SIPSON is a 1D hydraulic model which records the flow routing in the sewer systems and rainfall-

runoff hydrographs, the 2D UIM model stimulates ground surface inundation. These models are coupled

via discharge through manholes to reflect drainage and surcharge flows. The spatial distribution of

flooding has been modelled for various combinations of rainfall-runoff intensity and duration (pluvial and

fluvial flood matrices), modelled with and without the presence of flood defences and for scenarios of

levee breach and overbanking (for more details on SIPSON-UIM the reader is referred to Chen et al.,

2009).

Several scenarios were considered suitable for inclusion in the tool. Firstly, a range of scenarios with

depth and depth-velocity details were integrated. These simulate pluvial flood events and are modelled

for a series of return period (2 yr, 5 yr, 10 yr, 20 yr, 50 yr and 100 yr) and storm duration (15 min, 30 min

and 60 min). Ultimately each scenario is modelled for 12 hours and therefore illustrates the progression of

flood water, its flow pathways, ponding and ultimate recession. Secondly, a scenario for a fluvial flood

event was used and simulates a breach and overbanking scenario on the levee system. The breached

scenario was calculated by removing the levee system from the model; water overflows when it surpasses

the grid elevation of the surrounding cells. In the overbanking scenario, water overflows when the flood

stage in the channel exceeds the crest elevation of the levee (100 yr flood + 10 cm). This model runs for 10

hours. Thirdly, a scenario for fluvial-pluvial combined flooding is included, based on the 100 year event,

for a 60 minute storm duration (and again, is run for 10 hours). This model also includes scenarios for

breach and overbanking.

There are a number of underlying assumptions which underwrite the uncertainty of this flood model.

Firstly, the rainfall input is assumed to be spatially and temporally uniform. This is a common assumption

made in pluvial flood modelling which is naturally highly variable in space and time and very difficult to

predict. Secondly, the propagation of flooding (flow routing, pooling) is dominated by the underlying

elevation model and is a potential source of error. In this instance, the elevation surface is based on OS

Mastermap and high resolution LiDAR and thus minimises this source of uncertainty. Finally, the

governing equations of any modelling tool can amplify uncertainty. The SIPSON-UIM model has been

benchmarked with the Environment Agency’s 2D model and trialled in eight different types of case

studies; one can be confident at least that the model is producing similar results. Communicating



15

uncertainty in flood science is a contentious issue and current debates surround the responsibility and

ownership of these data (Faulkner et al., 2007). The tool presents these sources of uncertainties as a

series of bullet points in a pop-up window. Arguably there is a need to consider graphical and cartographic

means of visualising this and a case for avoiding the ‘prescription note’ or ‘health warning’ of uncertainty

(Faulkner, pers coms). The cascade of uncertainty in flood modelling means that there are even

uncertainties in uncertainty communication and while professionals were asked whether the uncertainty

information was something they used, it was not a pivotal research question addressed in the tool.

3.1.2 Cowes, Isle of Wight, Hampshire

Cowes is situated at the mouth of the Medina River as it drains into the sea off the Isle of Wight (IOW). In

theory the Island is an extended arm of Hampshire on the mainland, but in practise its isolated nature

means that in terms of emergency management it often relies upon its own resources. Flooding is a

recurring issue Island-wide. In Cowes, the main source of flooding is tidal and the tourist centre of the

town, the High street (Figure 2), is periodically inundated with more significant flood events occurring in

October 2004 and July, 2006.

Figure 2: Flood hazard map of Keighley, West Yorkshire (Environment Agency website)

As with Keighley, the complex nature of pluvial flooding has been modelled for the West Cowes

catchment. Allit et al (2009) have used Infoworks CS 2D model, in conjunction with a 1D sewer model. The

assimilation of 1D-2D modelling aims to capture the dynamics of minor and major systems respectively,

between the underground sewer network and overland flow pathways. For Cowes, the following model

outputs have been supplied to this research through the Flood Risk Management Research Consortium

(FRMRC phase 1). Firstly, the tool integrates a number of design storms for the 2, 5, 10, 20 and 50 year

return periods, modelled at 30, 60 and 90 minutes. These files include the maximum depth and hazard

values obtained in each model run; where the hazard has been calculated according to the recent Defra

guidelines including debris factors (HR Wallingford et al., 2006). Secondly, a 100 year storm event was

used to constitute the animation feature of the tool, based on model outputs at one minute time-steps

and simulating an event over 120 minutes. These results were based on the SIPSON-UIM model (run for

Keighley, by Exeter University); these two models are compared in Allitt et al (2009).



16

The resulting simulations correlate strongly with known sites of flooding and flood records from actual

rainfall events (Allitt et al., 2009). Field observations have also been integrated into the model to capture

the dynamics between surface features (e.g. walled entrances to properties, alleyways) that can divert or

constrain flow pathways. This model also captures the pluvial runoff within the catchment, based off the

surface topography; assuming no infiltration, initial losses or depression storage (Allitt et al., 2009). The

results presented illustrate the model-run with the inclusion of sewers and buildings.

3.1.3 Contrasting Keighley and Cowes

Both towns have a common flood driver in the form of pluvial flooding, which is used within the

respective tools for these locations. It was decided to utilise earlier research developed within FRMRC

Phase 1 and to use these two socially-contrasting locations to explore whether this exerts any effect upon

professionals’ perspectives of vulnerability, or indeed upon the residents in these respective areas.

Social vulnerability is currently recorded within Catchment Flood Management Plans (CFMP) according to

the Index of Multiple Deprivation (CLG, 2008). The IMD calculates a rank of England-wide Lower Super

Output Areas (LSOA), an aggregation of ca. 1000 properties. A rank of 1 indicates the highest deprivation

and 32,482 indicates the lowest; these are based on the aggregation of seven domain indices, including

income, employment, health and disability, education skills and training, barriers to housing and services,

living environment and crime. Keighley displays a wide range in IMD ranks, from a minimum rank of 199

(high deprivation) to a relatively high rank 29,061 (low deprivation), with an average rank of 9296. Cowes

by comparison has an average rank of 17,917 and is a considerably less deprived area than Keighley

(Graph 1).

Graph 1: The Index of Multiple Deprivation for Keighley and Cowes; Descriptive statistics (based IMD,

2007: CLG, 2008)

The predecessor to the IMD was the Social Flood Vulnerability Index (SFVI, Tapsell et al., 2002). In terms of

recorded social vulnerability, both locations have an average SFVI score of 3 i.e. a national average

category. There are noticeably variations between the two locations and the range of SFVI categories

(Table 2), as Keighley displays a wider range of SFVI categories (from low to very high vulnerability),

compared to Cowes which consistently records an SFVI category of 3 or 4 (from average to high

vulnerability). It was originally hypothesised that Cowes and Keighley would provide to socially-

contrasting areas for comparison: For instance, Keighley is regarded as a diverse, multi-ethnic setting,

with 10% of the population of Asian ethnicity (predominantly Pakistani, based on 2001 census). In terms

of the Social Flood Vulnerability Index (SFVI) however, these differences are masked within this

composite, additive model. Each indicator is treated with equal importance in terms of its influence upon

vulnerability (explained further in section 3.3.2). Graph 2 illustrates the break-down of the SFVI and

presents the average percentage of each indicator within Keighley and Cowes. Lone parent households



17

and non-car ownership dominate vulnerability within Keighley; whereas Cowes is dominated by non-car

ownership, non-home ownership and illness indicators.

Table 2: Observations from the SFVI categories in Stockbridge, Keighley and Cowes, IOW (when SFVI is a relative

measure of vulnerability for England and Wales, based on the original dataset from Tapsell et al., 2002)

SFVI statistics

Stockbridge, Keighley Cowes, IOW

Mean

3.17 3.55

Minimum

2 3

Maximum

5 4

Graph 2: The break-down of the SFVI indicators within Keighley and Cowes, based on average percentages recorded

in the 2001 census

The SFVI methodology represents a ‘classical’ approach to vulnerability assessment, based on

assumptions of demographic indicators, all equally as important in governing vulnerability. This research is

interested in adapting this approach and investigating the ways in which it might be adjusted and made

malleable to suit the varied needs of Category One Responders (as identified under the Civil Contingencies

Act 2004). Aside from the tool, fieldwork was conducted in Keighley, using questionnaire surveys to

appreciate the perception of vulnerability amongst those ‘objectively’ labelled as being vulnerable; i.e. is

there a resonance between the SFVI classification system and householder declared vulnerability?

3.2 Preliminary interviews: The end users “wish list” In both case study locations, a sample of Category One and Two responders were interviewed at the

preliminary stage of enquiry to ascertain current views on vulnerability and its application in decision

making. Furthermore respondents were asked to rate a number of initial design suggestions for the tool

and disclose their opinions on the value of integrating hazard and vulnerability data at the local scale.



18

Although all Category One Responders have a statutory requirement to communicate and integrate their

response, the emphasis on responding agencies shifts throughout the course of the emergency

management cycle; from preparation, response and recovery (Figure 3). It is important to remain mindful

of this when interpreting their responses from the semi-structured interviews. Each interview has been

transcribed and analysed in the qualitative data analysis software, NVivo.

Figure 3: Locating the principal roles of Category One Responders within the emergency management cycle

3.2.1 Findings There was a wide agreement across responders as to what constitutes vulnerability and a high level of

awareness regarding its complexities. For those particularly concerned with the ground-response (e.g. Fire

and Rescue, Police and Ambulance), risk to life is at the forefront of decision making.

Vulnerability was considered by some to be an all-encompassing term; from people’s socio-demographic

characteristics, people’s attitudes and awareness towards flooding, human behaviour, to an area’s

infrastructure etc. From the professional’s interviews, this was regarded as a potential limitation; “ It’s

such an umbrella terms….It’s so broad so as to be meaningless” (Fire and Rescue, West Yorkshire).

However, whilst acknowledging the highly variable nature of vulnerability, given their professional focus,

vulnerability is defined as any characteristic which will limit a person’s ability to save themselves; most

notably, people suffering with a limiting long-term illness or disability. The elderly population is also a

recurring indicator, based on the assumption that it correlates highly with illness and disability

perspective



19

While respondent’s cited a number of indicators and discussed the value of the common indicators cited

in the literature, many emphasised the grey nature of these and observed how an indicator can both

indicate both, high or low vulnerability.

“Yeah I was looking at the home ownership one and it’s kind of a mix because a lot of the council tenants

get a lot of support from the local council when their properties flood because it’s the council’s duty to

maintained council-owned properties. And then you’ve got the other side of people who own their own

homes and are slightly richer and can afford insurance. But then you’ve got in the middle of that people

who own their own homes but can’t necessarily afford insurance and they go uninsured and they shouldn’t

really in a flood risk band or wherever they live and their kind of in that middle ground, they’re not rich

enough to help themselves but they’re not in a council owned property so the council can do very little to

support them.” (Environment Agency, West Yorkshire)

Furthermore some responders challenged the assumptions of vulnerability indicators. While indicators

such as elderly suggest a level of social dependence that is not to say that all elderly people will require

assistance. Prioritising certain social groups when responding to an event, could be at the detriment to

others; “There the people who fall through the gaps and you’ve got this list here, yeah we could target

those but actually it’s the single male in the late 40’s who’s on the disability allowance who spends the day

drinking whisky in the middle of the day. So vulnerability is both about lifestyles, about attitude and our

response I think in the main is to take a very equality and diversity sort of attitude to response (West

Yorkshire Fire and Rescue). This comment from Fire and Rescue is very informative. The literature

suggests that vulnerability assessments are potentially useful for prioritising actions. In the context of

flood response in the UK, this comment from Fire and Rescue implies that this goal may conflict with the

need to offer an equitable service. Targeting certain social groups is instead considered a matter of

prevention; “we probably carry out most of our prevention work with people who you would {276} identify

as being vulnerable in one way or another” (ibid).This perspective suggests that vulnerability may be more

useful from a planning perspective as opposed to informing the operational phase of FIM where it is

informed by incoming 999 calls. Similarly, the IOW council reported that; “The compromise that we came

to was vulnerable is anyone who describes themselves as being vulnerable in response to a situation (680)

because they, the people who are affected by flooding in this case, will be able to know more than any

external label that we put on them.(IOW emergency planning).

From these discussions it can be concluded that the role of vulnerability assessment during the immediate

response phase is limited. Immediate response necessitates accessible and accurate information. While

responders indicated that an overview of an area’s social make-up can do no harm and agreed that it

could play a part in wide-scale flood events, ultimately it is the nature of the hazard which influences

priority setting. Where will it flood first? How bad will the flood be? Second to this, who are the

vulnerable people within these areas? Vulnerability is understood as the location of elderly care homes

and critical patients. Information is passed between organisations such as social services, NHS and PCTs,

and utility companies to locate these people and is not stored centrally by any one organisation due to

data protection (DPA, 2004). While some responders commented on the frustration associated with this

(e.g. Emergency Planning Hampshire), it was widely acknowledge that any database that sought to

centralise critical vulnerabilities (i.e. at the household scale), although a ‘dream tool’, could never function

in reality; indeed given the variable nature of vulnerability how could such a database ever be kept up to

date? Including vulnerability within operational response is a matter of pooling information from a

number of agencies (social services, PCT, utility companies). This is of course supplemented with the 999

calls where households have identified themselves as vulnerable and in need of assistance. During the

other phases of FIM the household scale is considered to be too refined to inform decision making, rather

it is the overview of an area that is required. For instance, demographic profiling is a crucial part of

targeting and tailoring community engagement and flood awareness campaigns, currently being trailed

within the EA’s internal project, FloodWise (EA, Hampshire); for which the proportion of elderly, low-

income families and families with young children are the leading indicators.

Responders were asked to rate a number of design suggestions for the tool, relating to the information

and presentation of the hazard, vulnerability and risk (Graph 3; Table 3).



20

0

2

4

6

8

10

12

1 2 3 4 5 6 7

Nu

mb

er

of

resp

on

de

rs w

ho

ra

ted

th

e

fea

ture

1 t

o 5

Question number (see corresponding Table 3)

Graph 3: Category One Responders feedback on design suggestions for the tool, based on a 1 to 5 rating scale of

usefulness in informing decision making (5 being very useful). This graph presents a count of responders rating each

feature 1 to 5.

Question

Table 3: Professionals were asked to rate initial design suggestions (1 to 7)

1 Ability to select a number of flood scenarios (2yr, 10yr, 20yr, 30yr, 50yr and 100yr return period

flood events)

2 Ability to either run the full duration of the event (up to 720 minutes) (i.e. animation), or to

select a given time (snapshot)

3 Ability to zoom in or out of a given area of interest

4 Ability to obtain summary flood statistics at a point location (e.g. depth and velocity details at a

specific manhole, for a specific event and the range across all events)

5 The option to use an enquiry function, whereby the user can ask a specific question (e.g.

identify all areas flooded to depths greater than … within a …specified time… or specified

event/across all events).

6 The ability to view results within a 3D environment (based on terrestrial LiDAR scanning of the

area)

7 Ability to obtain summary statistics for social vulnerability i.e. at the street level

Any information on the hazard, from the extent of different scenarios, depth and velocity details and

ability to view these at given points in time, rate highly on a scale of usefulness; and could support

decisions regarding the allocation of resources (time, equipment), safety and continuity planning. The

hazard posed to the road network was also cited by the blue light services as crucial for flagging-up issues

of access and planning alternative routes. Visualisation is considered to be a powerful communication

tool; however 3D visualisation (question 6) was considered to be less important for decision making but a

potential tool for communicating with the public. What is also clear from these results is that users value

the ability to switch between spatial scales, from the broad overview down to the local picture to mirror

Rating 5: Highly useful

Rating 4

Rating 3: Neither useful nor not

useful

Rating 2

Rating 1: Not useful



21

the nature of a flood event which might occur very locally only, or may become a region-wide problem.

Responders also highly rated the ability to query vast quantities of data very quickly, supporting the rapid

pace of immediate response.

Question 7 asked responders whether it would be useful to view vulnerability at the local scale. Graph 3

illustrates the mixed response to this. Presenting data obtained at the household scale would be highly

impractical, not only in terms of the labour intensive nature of data collection but also in actually using

these data which could change very rapidly. Furthermore, there are issues surrounding data protection as

previously discussed. The semi-structured interviews queried responders about the use of indices, such as

the Social Flood Vulnerability Index (SFVI, Tapsell et al., 2002) and the value of plotting relative

vulnerability across a district, town etc. Responders commented that they would want to know how the

score had been constructed and be able to deconstruct it to view the individual indicators and their

contribution towards vulnerability.

“…this sort of information would be hugely useful but not in the response plan to the flood, but more in an

evacuation plan and a rest centre plan; in the fact that, as is best practice with the CCA and how we’re

meant to plan and respond rather than just chucking ourselves into it, fire fighting and responding…It

would allow us to target our efforts. At the moment we could target efforts if we had the resource to do

so, in the areas that we know are at flood risk, but within that there are areas that perhaps have more

vulnerabilities which would feed into that prioritisation about where you’re going to do first. {talking about

targeting parishes to write their own flood plans: IOW Emergency Planning)

The phase-relevance of vulnerability indicators was also noted. During response the primary concern is

risk to life, therefore indicators such as the elderly, illness and disability are paramount. Vulnerability is

also manifest after a flood event, effecting single parent households for example (EA, West Yorkshire).

This observation suggests that it is inappropriate to apply a universal vulnerability index, stretching across

the phases of a flood event; rather what is required is a flexible system of selecting and weighting

vulnerability indicators according to the context in which they will be applied.

Vulnerability indicators are already held within council databases, such as social services, housing register

etc. These indicators are not specifically flood related. Instead, there is a common view that vulnerability

is generic in nature across hazard events. For example, “our flood management is just part of our overall

crisis management. We do the same things for snow as we do for floods…”(Fire and Rescue, West

Yorkshire). The same indicators that are used to identify social vulnerability for heat waves, snow storms,

swine flu etc., are equally applied to the context of flooding.

The professionals reported that for them vulnerability assessment is a tool for planning response and

longer term mitigation strategies (e.g. tailoring awareness raising campaigns), as opposed to informing

the immediate response phase of FIM. In addition, there is a role for vulnerability assessment in planning

the recovery strategy of an area, in highlighting the nature of support that might be required from local

authorities (IOW Emergency Planning).

The perceived value of integrating vulnerability during response, varied across groups. For the ground

response teams, vulnerability data appeared secondary to the hazard; i.e. responders indicated a

preference towards where not who; with the exception of the ambulance service where this statement is

reversed. Conversely the emergency planning departments of the council display a keen interest in

vulnerability information.

“All we want to know is, where is it going to be wet? That’s all we want to know because that’s all we base

our response on…we can see you know that’s a residential street and we’ll assume there is 2.4 people in

every house and that gives us a rough estimate of we’re going to have to evacuate that many number of

people” (West Yorkshire Police)

“I think we would be very much focused on responding to water, on rescue and water management. The

actual people issues I don’t think would, I think we would work on the advice of other agencies, I don’t

think it is something that we would (716) in first instance be thinking off, we would be reliant upon other



22

agencies and they would be there anyway so why would we need additional information” (West Yorkshire

F&R)

“Obviously knowing information about vulnerability helps us a lot in terms of gearing-up what response

we need. So if we know that there’s an area affected that’s got a lot of people in care homes, that means

that we can save ourselves a whole lot of grief by before someone getting there and realising there’s a

problem, we can say well actually there’s a likelihood that a lot of people will need medical support in our

residential homes, we can make the call before we get there to make sure that we’ve got a building set up

with adequate facilities, people with professional knowledge, whether that’s from NHS, to make sure they

can go in there and care for them”. (IOW, emergency planning)

3.2.2 Some conclusions There are some conclusions that may be made from this initial survey of the target professional group in

relation to their understanding and need for vulnerability assessment. It would seem we can summarise

the view that “there is a time and a place for vulnerability assessment”. It emerged as a consensus that

vulnerability can be more meaningfully used for prevention, planning, recovery and training but has a

limited role during the immediate response phase of a flood incident. Vulnerability is considered to be a

subjective and disputed concept, and ultimately from a response perspective, responders must prioritise

their efforts to where the hazard is and key vulnerable people. This latter information is provided through

data sharing (health services, utility companies and social scare databases) in accordance with the Data

Protection Act 2004. This information is supplemented by those who disclose themselves to be vulnerable

and actively seek assistance (i.e. 999 calls). The rapid pace of response decision making and the need for

accurate information (exact location, up-to-date details on vulnerability) means that the composite, area-

wide index approach for measuring vulnerability would not be used during a response situation. Instead,

responders felt that this type of information could perhaps be more usefully applied to support broad

scale assessments and for painting the social ‘make-up’ of the area. In this light, this form of social

vulnerability assessment could help inform the planning of response.

This tool was designed to facilitate a host of research questions, rather than as an application that could

roll-out in practise; although it is recognised that the feedback from professionals could help inform some

practical recommendations for the future assessment and presentation of hazard, vulnerability and risk

within FIM. ArcGIS was selected as the platform for designing and demonstrating the tool. It is

acknowledged that this software has a number of limitations which would prevent this tool functioning in

real-life situations. Aside from expensive licensing, the desktop application is not suitable for creating and

sharing a ‘common picture’ of the situation as required for multi-agency working and discussed with this

sample of professional stakeholders. The software is also unfamiliar to practitioners and would require

training. The platform for launching a tool of this nature would need to be something that is user-friendly,

not reliant of licensing and open to all potential users. An open source GIS via a secure website would

probably be the closest to meeting these requirements. This would appease the recurring comment that

any successful decision support tool should K.I.S.S.; Keep It Stupid Simple. This In vivo statement (i.e. in

the respondent’s own words) reflects the nature of the professional context, as responders discussed,

flooding is but one part of the day-job and therefore any supporting tool needs to be simple and self-

explanatory to use after potentially long time-gaps. Furthermore it needs to be user friendly and easily

‘fixable’ should it go wrong, without relying upon the tool’s developer.

The requirement for simplicity raises an interesting debate. Is simplicity a requisite of the tool itself, i.e.

synonymous to user-friendly? Or is the desire for simplicity reflective of a desire to disengage with the

complexities of flood science? This might be framed within two simplicity models: (1) simplistic-user-

friendly and (2) simplistic-information tool. The latter is arguably indicative of a greater tension involving

the translation of science to practitioners and the professional context. On the basis of these preliminary

interviews it seems that there is a cross-over between this dualistic meaning of simplicity: It is very

apparent amongst all responders that a simplistic-user-friendly tool is essential, whereas simplistic-

information received mixed views. This is discussed further in section 6 and 7.



23

Multi-agency working means that any system designed to calculate flood risk needs to be flexible to

multiple users, with different data requirements and different emphases on hazard and vulnerability. The

tool’s potential and limitations also need to be apparent. For instance, the IOW council commented that a

tool which seems to portray the ‘answer’ from a series of tick boxes is arguably less informative than a

tool which prompts new thinking.

“… I was concerned that if it would be something that is a bit more scripted, tick a box if (663) we did that,

tick a box of we did this, it would then funnel the mind into being a bit more closed and blinkered, rather

than opening up to the fullest extent of what one’s day job role is, to think outside of the box, because if

they’re thinking that they’ve got a handy tool that does all this for them and seems to be looking right and

giving them the right answers, it’s just too good to be true and they don’t need to think so therefore they

won’t. (IOW emergency planning)

The importance of being able to view the bigger picture was stressed in these interviews. Examples such

as 2007 flooding were used by some responders to demonstrate the complicated nature of decision

making when resources become stretched and priorities need to be set. Therefore, while this tool focuses

on the very local scale of the case study areas it is considered to be an example of a tool that could be

extended to include a number of local ‘hot spot’ flood locations, framed within the district. The

development of the tool, which is described in the following section, is heavily influenced by these views

expressed in the preliminary interviews.

4 A flood risk assessment tool

This section describes the features of the flood risk assessment tool that were developed in response to

the preliminary interviews discussed in the previous section. The tool was approached almost as an ‘eggs

all in one basket’ to see how different end-user react and rate different design features, to inform some

practical recommendations for how such a tool might be more specifically tailored in real-life applications.

The tool itself loads the datasets from a Personal Geodatabase constructed in ArcCatalogue. Rather than

simply allowing layers to be added and removed, this tool enables users to manipulate these layers to suit

their needs and perform calculations on the data to produce vulnerability and risk profiles from a number

of flood scenarios. This interactive nature seeks to engage the end-user to become actively involved in the

assessment process and map production. This moves away from the traditional paradigm of

communicating “what we know” and targeting an ‘information deficit’ model (Lane et al., 2010); to one

that recognises the end-user as an expert in their right and providing the means in which they can

integrate their informed subjectivities from the day-job, with the objectivity of the ‘scientific expert’ that

is inherently built within the tool.

The tool is written in Visual Basic for ESRI and designed with the ESRI application ArcMap where the tool is

launched. This is a commercial product for Geographic Information Systems (GIS). In the preliminary

interviews with professional stakeholders, many reported unfamiliarity with this system or felt that they

‘knew the basics’ but did not feel comfortable using ArcGIS. If this tool was to be used in practise, the

platform from which it is launched would need to be reconsidered. For the purpose of this research it was

considered appropriate but it was consciously decided to minimise the user’s4 interaction with ArcMap as

far as possible. Therefore the tool uses a display window to illustrate what is going on the main screen; if

the user wishes to view the main screen as opposed to this ‘summary window’ then they need only drag

and slide the interface of the tool aside. Further options to zoom in and out on the main screen are

provided on the interface of the tool (Figure 4).

4 Category One Responders are the target user of this tool and the tool has been designed according to the feedback

from preliminary interviews; as such, the terms end-user (or user) and responder are used interchangeably in this

discussion.



24

Figure 4: Screenshot of the upload-page of the tool, with contents page and summary window on the right.

The tool is designed with three key interfaces (i.e. separate pages), isolating hazard and vulnerability and

allowing the user to bring these together in the calculation of risk. Each interface seeks to address a

number of minor research questions. This is summarised in Figure 5.

Ability to zoom in and out on the main screen Ability to clear all layers from the map

Summary window:

Updates as new layers

are added to ArcMap



25

Figure 5: The architecture of the tool

4.1 The Hazard Interface

The hazard interface centralises all the details pertaining to the flood hazard and aims to answer a

number of research questions. Firstly, do responders value a range of flood scenarios or will they always

plan to the worst case event? Secondly, rather than simply viewing the extent and relying on different

shades of blue to indicate various depths, of the flooding, as is standard practise, do responders value the

option to re-colour the flooding according to the hazard posed to life? Taking this another step forward, is

the option to essentially ‘clean’ the map image to the flooding created on the road network or to

properties only, useful? Fourthly, given the option to select from two hazard models (risk to life versus

depth-damages) which model would responders use (and for what purpose) if they were to view the

flooding at the property level? And finally, how informative is an animation and user-control in viewing

the flood inundation in a dynamic form? Figure 6 provides a screenshot of the hazard interface: Each

feature is numbered to facilitate the discussion hereon.



26

Figure 6: The Hazard Interface to the GIS-based flood risk assessment tool

[1] In the first instance, the user can select from a flood scenario and view the spatial extent of the flood

(based on the maximum flood value). Scenarios range from high frequency, low impact events to low

frequency, high impact events (i.e. 2 year to 50 year rainfall events respectively). Cowes offers pluvial

flood scenarios only, whereas Keighley has additional scenarios for fluvial flooding (100 year event),

including scenarios for overbanking or breaching the levee system; furthermore, Keighley stimulates

pluvial-fluvial interaction. Preliminary interviews with professionals indicated that the ability to upload

and view a range of flood scenarios was deemed useful and highly useful.

[2] The tool provides the user with the option to reload the flood scenario of their choice, but this time

viewing the model outputs for depth-velocity interaction (rather than merely depth offered in the

previous option). Expert-declared thresholds are provided adjacent to this, from which the user can either

select go and view the flooding when it is re-coloured according to risk to life thresholds; or the user can

decide whether to manipulate these thresholds and adjust the hazard classification according to their

choice. The expert-declared thresholds are automatically entered into the tool and are set to those based

on Risk to Life research (HR Wallingford et al., 2006).The Flood Risk to People methodology calculates

flood hazard ratings on the principles of equation 1.

Flood Hazard Rating = ((v + 0.5) * D) + DF Equation 1 (HR Wallingford et al., 2006)

Where: v = velocity (m/s), D = depth (m), DF = debris factor (0.5)

1

2

4

3



27

In the Cowes case study, this calculation has already been developed in the (h) scenario files and uses a

debris factor of 0.5; this is consistent across all scenarios and is not adjusted for depths surpassing 0.25 m

as the original Risk to People methodology suggests. The hazard matrix presented in this paper is

illustrated in the background page of this tool and describes the varying levels of danger at crucial hazard

thresholds (Table 4).

Table 4: Risk to life thresholds based on a combination of depth-velocity (Priest et al., 2008)

Depth X Velocity (m2/sec)

Hazard Description

< 0.75

LOW

Caution

Shallow flood water or deep standing water

0.75 < 1.5 MODERATE

Dangerous to vulnerable groups

Deep or fast flowing water.

Fatalities concentrated in vulnerable groups or

the result of human behaviour.

1.5 < 2.5 HIGH

Dangerous to most people

Deep or fast flowing water.

Fatalities due mainly to exposure to the hazard.

2.5 > 7.0 EXTREME

Dangerous for all

Extreme danger from deep, fast flowing water.

Fatalities due to hazard exposure.

> 7.0 EXTREME

Dangerous for all

Extreme danger from deep, fast flowing water

and risk of building collapse.

[3] This third option enables the end-user to essentially ‘clean’ the map image and view the hazard posed

to the road network only. This is based on the significant depth-velocity thresholds outlined in Table 4.

The Safe access and exit section for the Flood Risk Assessment Guidance document for New

Developments (Defra, 2005) identifies a number of situations in which a vehicle should not be used. These

include i) when the presence of water results in engine malfunctioning; ii) point at which vehicle begins to

float; iii) point at which the vehicle becomes difficult to control. For a standard car floating may occur at

depths of 0.5 m (compared to up to 1 m for heavy duty emergency vehicles) in standing water. These

depth values decrease as velocity of water increases. A more detailed hazard matrix from which the risk to

life thresholds are based, is presented in Table 5. According to the Risk to People methodology, safe

access routes should be based on the interaction between depth and velocity in the white boxes only. The

increase in either depth or velocity results in an escalation of risk, the road hazard presented in this tool is

thus based on the depth-velocity matrix presented here.



28

Table 5: Depth-velocity matrix based on danger posed to people. Safe access routes identified d-v interaction within

the white boxes (Priest et al., 2008)

This model is essentially a risk-based model, as it considers both the hazard parameters and the

population characteristics (whether demographic or behaviour-related) which might heighten

susceptibility towards harm in its classification of risk. Nonetheless it is included within the hazard section

of the tool. The reason for this is that the vulnerability component of this model is generic and not based

on data from the area; the vulnerability pages introduce these data and encourage the end-user to

explore a number of indicators to explain the make-up of the population first, before uniting hazard and

vulnerability together in the risk page.

[4] The final feature of this page provides the user with the option to view whether a potential hazard

exists at the property level. Again, this option provides another means of cleansing the map, but it also

provides the user with the option to view two different hazard models. Firstly, the user can view the risk

to life model (HM1) as previously discussed. On the basis of the hazard rating calculated (equation 1) this

option will re-colour the properties according to whether there is a potential danger to life. Hazard model

2 (HM2) classifies the hazard according to depth information only. The thresholds in this instance are

based on depth-damage thresholds (Table 6 and Table 7). Each threshold is assigned a score on a 1 to 5,

according to the degree of hazard posed; this scoring system corresponds to the one used for vulnerability

assessment and was required for the final risk calculation.



29

Hazard Model 1: Risk to Life

thresholds, based on a hazard

factor (m3

s-1

)

Hazard

category

< = 0.01

1

0.02 <= 0.75

2

0.76< 1.5 3

1.5 < 2.5 4

< 2.5 5

The depth-based hazard model uses critical thresholds identified in the Multi-coloured Manual (MCM)

developed at the Flood Hazard Research Centre, Middlesex University (FHRC, 2010), for depth-damages

arising for the average residential property, based on 2010 prices and flood duration of less than 12 hours.

According to the MCM there are 15 categories of depth-damages; these were simply divided by 5 to

obtain hazard thresholds (Table 7). Although more detailed information is provided for types and classes

of property, this is not considered in this tool; however If this tool were to become a real life application it

is suggested that the EA’s property database could be used to inform a more detailed classification. The

rationale for applying a depth-damage based model relates to the nature of vulnerability, which is not

only exhibited during the actual event but manifests during the post-event, recovery phase also. Users

engaged in this phase of emergency management may wish to identify properties that are particularly

susceptible towards damage, may need to evacuate or seek support during the aftermath of a flood

event.

In addition to this option to select from two types of hazard model, the user can further decide whether

to base this hazard classification according to the minimum, maximum or average flood statistics

calculated in the flood model. A 20 m buffer from the central point of each property was used to ascertain

flood statistics (Figure 7). As such, the user can select whether they wish to view the minimum, maximum

or average hazard value or depth value; these values have all been pre-classified according to the hazard

threshold and assigned a score of 1 to 5, representing very low hazard to very high hazard respectively.

Hazard Model

2: Depth

thresholds (m)

Damage estimates; based on

total damages calculated for

the average residential

property (max damage per

threshold, £)

Hazard

category

<= 0.05

£8529.98 1

0.06 <= 0.3

£ 11,952.27 2

0.31 <= 1.2

£33,225.92 3

1.21 <= 2.1

£41,882.69 4

> 2.1

£51,438.33 5

Table 7: Hazard categories based on depth-damage estimates

for the average residential property (FHRC, 2010)

Table 6: Hazard categories based on Risk to Life

thresholds (Priest et al., 2008)



30

Figure 7: Applying a 20m buffer around each property to assign flood statistics (min, max and mean) to the property

[5] The final page for the hazard section (Figure 8) enables the user to select and view a flood scenario

from beginning-to-end; in the case of Cowes this runs from 1 to 120 minutes (with 1 minute time steps)

and for Keighley scenarios run from 1 to 720 minutes (at 6 minute time steps). The user can load a time

step individually (a), can view the full animation (b) or can use a time scrollbar to control the view (c).

Simultaneously, the user can overlay this animation onto key GIS base layers (e.g. location of property,

vulnerability, fluvial flood zones etc.).

Figure 8: Launching animation in the GIS-based flood risk assessment tool. The user can (a) load individual time-steps

in the scenario, (b) play the animation or (c) use a scroll bar to view a snapshot of the flood at a time of their

choosing.

A

A

5

(a)

(b)

(c)



31

4.2 The Vulnerability Interface

Vulnerability is similarly navigated through a series of pages, the main purpose of which is to explore the

end-users views on scale, dynamic indicator selection and the context in which such information might be

applicable. This section of the tool explores a series of questions. Firstly, to what extent do responders

desire only the ‘answer’ i.e. a vulnerability product such as the SFVI (Tapsell et al. 2002) or the Index of

Multiple Deprivation (IMD, 2007)? To what extent can isolated indicators help support decision making?

How do responders regard the usefulness of being able to select, weight and construct their own index of

vulnerability? And finally, at what scale should vulnerability be assessed, and indeed, what activities

within FIM do responders feel vulnerability could support.

The Social Flood Vulnerability Index is employed in the tool to represent a ‘vulnerability product’; that is

to say it has been packaged together by expert academics. The disadvantage of such a product is that it is

based on assumptions that may not be apparent to end users. Perhaps the strongest critique for the SFVI

is the treatment of vulnerability indicators as equal partners in governing risk. The SFVI represents an

aggregated index for social vulnerability, based on an additive model of four variables (Table 8). The

Townsend Index is handled separately in this approach, with its individual components summed

separately and then multiplied by 0.25 before inclusion into the overall SFVI. All variables are firstly

equated as percentages and then transformed (Table 9): Results are then standardised as z-values before

summation. Final scores are classified into five bands; with 1 to 5 representing low to high social

vulnerability respectively. This method was utilised by the UK Environment Agency as a means of

highlighting socially vulnerable areas more sensitive to the adverse impacts generated from flood events,

but has since been replaced by the Index of Multiple Deprivation (discussed below).

Table 8: The Social Flood Vulnerability Index (after Tapsell et al., 2002); with data sources for this research

Variable Measure

Townsend Index for

Deprivation

Including;

Unemployment

Overcrowding

Non-car ownership

Non-home ownership

Unemployed residents 16yrs and over, as a percentage of all economically

active residents

Households with more than one person per room, as a percentage of all

households.

Households with no car as a percentage of all households.

Households not owning their own home as a percentage of all households

75 years + Residents aged 75years and over as a percentage of all residents

Lone parent households Lone parents as a percentage of all residents

Long-term illness Residents suffering from a limiting long-term illness, as a percentage of all

residents



32

Table 9: Transformation methods selected to reduce skewness and kurtosis in the distribution (Tapsell et al., 2002)

Indicator

Transformation method

Lone parents Log natural (x + 1)

Aged 75 + Log natural (x + 1)

Long-term sick Square root

Non-homeowners Square root

Unemployed Log natural (x + 1)

Non-car owners Square root

Overcrowding Square root

Figure 9 presents a snap-shot of the vulnerability interface developed within this research.

Figure 9: The Vulnerability Interface to the GIS-based flood risk assessment tool

[6] The first feature of the vulnerability interface enables users to view the SFVI for the Bradford district or

Island-wide, or focus the SFVI to the census Output Areas (OA) or properties of Stockbridge or Cowes only.

It is noteworthy that although the SFVI can be illustrated at the property level, the score remains based on

the OA (i.e. properties are merely assigned the score of the OA in which they are located); the scaling

challenge for vulnerability assessments is discussed further in section 7. By enabling the user to switch

between the local to district-wide view, this feature targets the preliminary feedback from professionals.

Furthermore, the SFVI scoring system can be adjusted by the end-user to reflect the relative vulnerability

according to different geographical scales. This adjusts the standardisation technique within the original

6



33

method, which produces a score for each variable on the basis of national measures of central tendency

(i.e. z scores are calculated according to the mean and standard deviation of the national dataset). The

original approach thus paints a picture of relative vulnerability across the England and Wales. While

arguably appropriate for broad scale decision making these scores have little relevance to district or local

based decision making. Vulnerability is recorded as a relative measure and therefore the standardisation

technique must reflect the spatial scale at which decisions will be made. This tool provides this flexibility

and enables the end-user to view the original scores standardised to the nation, as well as further scores

based on the region, the district and the local case study area. As figure 10 illustrates, the distribution of

vulnerability alters as a result of the standardisation technique applied. This feature in the tool serves to

demonstrate the relative nature of vulnerability and the importance of scale in informing decision making.

Figure 10: The SFVI for the Isle of Wight when standardised to (A) the Island and (B) England and Wales

[7] Figure 11 presents the next page within the vulnerability interface of the tool. Here the user can view

individual indicators in isolation and their influence on vulnerability across the district or the Island. Each

indicator on this page has been standardised to the district or the island and classified according to the

relative contributions towards vulnerability; scores of 1 represent very low vulnerability, 3 is an average

level and 5 is a very high degree of vulnerability. One of the recommendations from stakeholder

interviews was that this feature would be useful providing that an expert-declared rationale is provided to

explain the ways in which the indicator may contribute towards vulnerability. In response to this, the tool

provides a question mark icon which brings up a pop-up, explanation.

(A) (B)



34

Figure 11: Loading individual vulnerability indicators within the GIS-based flood risk assessment tool.

All indicators are based on 2001 census data. Expert-declared rationales are opened via this icon,

and appear in the top-right hand corner of the screen.

[8] The ‘combine-indicators’ page (Figure 12) enables the user to essentially build their own vulnerability

index. The user is required to rate the indicators listed in terms of the degree of importance (i.e. high,

moderate, low, to no importance). Each indicator has been manipulated on a scale of 0 to 1; the data is

multiplied by 1 in the ‘high’ scenario, multiplied by 0.66 in the moderate scenario and 0.33 in the low

scenario. The “construct index” command combines these indicators together into a simple additive

model. This is then reclassified in the tool to present vulnerability categories (1 to 5) based on the

minimum to maximum spread in the data.

7



35

Figure 12: Building a vulnerability index according to user-declared relevance of each indicator

One of the recurring critiques of the index approach in the literature is the lack of a defensible weighting

scheme which results in the equal treatment of indicators. By enabling the end-user to communicate and

integrate their subjectivities on the weighting of indicators, this removes the responsibility off the

objective-scientist and builds-in an inherent flexibility into the tool to be adapted and adopted by a

number of different end-users who will each consider the matter differently.

4.3 The Risk Interface

Risk is understood as the probability of hazard occurrence (of a given magnitude) and the consequences

of this (economic, environmental, social or otherwise). Risk is therefore widely acknowledged as the result

of the following calculation (Gouldby and Samuels et al., 2005):

Risk = f (hazard, vulnerability)

This final page thus centralises the hazard information (hazard models 1 and 2) and the vulnerability

details at the property level, within the local case study area. This part of the tool examines how

responders negotiate this risk equation and weight the hazard and vulnerability information within this

(Figure 13).

8



36

Figure 13: The Risk Interface of the GIS-based flood risk assessment tool

[9] To address this research question, the tool provides the user with 5 models for aggregating hazard and

vulnerability data. Firstly there is the option to select an equal weighted model, whereby hazard and

vulnerability are treated as equally important in the assessment of risk. A score of 1 for vulnerability and 5

for hazard is regarded as the same as a score of 1 for hazard and 5 for vulnerability. The next two options

present a minor weight on either the hazard or vulnerability (1), the respective scores for the model

selected are firstly multiplied by 2 before multiplied with the other. The final two options extend this and

assign a greater weight (2) to either hazard or vulnerability; in this case, the scores of the chosen model

are multiplied by 4 before being combined. This system of selecting a weighting model (1 or 2) gives the

user the flexibility to decide whether one is slightly more important or significantly more important in

governing risk. This feature of the tool is designed to offer flexibility in the treatment of hazard and

vulnerability which could conceivably change between different phases of emergency management; while

the hazard is paramount during incident response, vulnerability is perhaps more informative for recovery

efforts.

[10] An automatic property count is calculated on the basis of the risk model that the user has selected.

This provides a simple count of all properties which fall within each risk category; the user can select to

view this in terms of people which are crudely measured according to the EA’s assumption of 2.4 people

per property. This result can be stored on a final page for “Notes” so the user can compare between

different combinations of hazard and vulnerability data and weighting criterion.

10

9



37

5 Feedback from Professional Stakeholders

A sample of the professional stakeholders who took part in the preliminary interviews, were re-

interviewed following a demonstration of the tool; these included representatives from Emergency

Planning (Local Authority and County Council), the Environment Agency, Police and Fire & Rescue for

Hampshire and West Yorkshire. Responders were asked to complete a short questionnaire, rating each

design feature of the tool in turn and considering the potential for these to support aspects of FIM. This

section summarises the feedback obtained for the hazard interface (4.1), the vulnerability interface (4.2)

and the risk interface of the tool (4.3), drawing from the questionnaire results and surrounding

discussions.

5.1 The Hazard Interface of the tool The hazard interface of the tool centralises a number of different features for calculating and presenting

flood hazard information, as discussed in section 4.1. This ‘all in’ approach was used to simulate

discussions with emergency responders; while the preliminary interviews sought to ascertain the data

needs of these professionals, the demonstration and visualisation of the tool was used in attempts to

simulate new thinking and encourage responders to consider the potential usefulness of new forms of

data use and presentation, not currently provided in the current ‘tools of the trade’.

Responders were asked to rate each feature of the tool in turn on a scale of usefulness i.e. to what extent

could feature X support professional decision making (Graph 4: Table 10).

Graph 4: Ratings for the Hazard features of the tool; average score and standard deviation, based on responders’

feedback

Question

(corresponding

to Graph)

Table 10: Professionals were asked to rate each feature according to its degree of usefulness in

supporting decision making

1 Option to select a number of flood scenarios (2yr, 10yr, 20yr, 30yr, 50yr and 100yr return period

flood events)

2 Option to select different types of flood events (pluvial, fluvial, combined)

3 Option to view flooding in terms of the hazard posed to life



38

Question

(corresponding

to Graph)



4 Option to adjust hazard thresholds (based on depth-velocity interaction)

5 Option to select a hazard model: HM1 risk to life or HM2 depth-damages.

6 Option to view flood hazard posed to the road network only

7 Option to view flood hazard posed to properties only

8 View an animation of flooding

9 Option to interact with animation, adjust a time bar and manually select a specific time in the

flood event

10

View background pages to explain hazard calculation, model assumptions and uncertainty etc.

As Graph 4 illustrates each hazard feature was consistently rated highly (4 or above) amongst responders;

and in particular, the animation portion of the tool (Q8 & Q9). The option to adjust hazard thresholds

(based on depth-velocity interaction) was regarded as the least useful feature, followed by the option to

focus the flooding along the road network and property scale only. The following sections present the

feedback from practitioners and justify these answers in turn.

5.1.1 Presenting flood hazard The hazard interface of the tool offered a number of different features which responders were asked to

comment on. These included;

i) The option to select from a number of flood scenarios (using pluvial flooding as an example)

ii) The option to re-colour a flood scenario according to the hazard posed to life and the ability to adjust

these risk to life thresholds

iii) Option to clean the map to view the risk to life posed on the road network

iv) Option to clean the map to view the hazard posed at the property level – based on two hazard models:

Hazard Model 1 (HM1) based on risk to life and Hazard Model 2 (HM2) based on depth-damages.

i) Scenario selection was deemed valuable in terms of training and exercising in particular, although it was

acknowledged that planning is always based on the worst case scenario and therefore this feature has

limited application potential in this context.

“Yeah but that would be useful for exercises, because in an exercise you might not want to choose the

worst case scenario. You would always plan you plan to cope with the worst case scenario but in terms of

an exercise you might do it at another scale, to be more realistic. So having the option to pick the range

but for planning you would use the worst case scenario.” (Emergency Planning, West Yorkshire)

Understandably in a response situation this feature is limited and real-time data is clearly required to

steer response. The value of a repository of scenarios in this situation is the visual representation of the

potential flood, which real-time rainfall or river level data cannot portray. In order to tailor this feature

more suitably towards response, it would perhaps need to be tailored to current real-time data

measurements i.e. a slider tool where the user can adjust rainfall or water levels and view the pre-

modelled flood extent. This recommendation has already been seen in recent decision support tools such

as Halcrow’s FloodViewer© and the findings here support the value of utilising similar instruments in real-



39

life applications.

While pluvial flood modelling is the example used in this tool, it is conceivable that it could supplement

scenarios based on other flood types (e.g. fluvial and coastal); responders were therefore asked

(hypothetically) whether this would be an asset to them. Responders involved in Emergency Planning (LA,

CC) and the Environment Agency considered this to be important, particularly for locations of combined

flood drivers and furthermore in gearing-up the response to varied impacts that result from different

types of flooding. On the basis of these interviews, this is considered less useful from the perspective of

blue-light services.

“…we know that the impacts from different types of flooding are very different. For us what we call surface

water flooding has the main impact on the highways. But if we had environment agency, met office or

coastguard warning us about tidal, coastal flooding then we know that’s more about properties. So how

we might react to what we might expect to happen and have to do, would be different to different

flooding.” (Emergency Planning, Hampshire)

“…it’s nice but from the police’s point of view we don’t need to speculate. It is either a flood or it isn’t.

There’s no half way. We have a plan for a flood, regardless of whether it’s a little one, a big one or a

massive one. It’s just how many resources we throw at it that depends on the size. In terms of planning I

wouldn’t say right if it was a 10 year flood I would have this plan, or if it was a 100 year flood I would have

this plan. I’ve just got a plan. But it’s useful, it’s just not essential. It’s nice to have.” (Police, West

Yorkshire)

ii) The tool provides a feature whereby the user can re-colour the flood extent according to the risk posed

to life, based on expert-defined thresholds identified in Risk to Life Modelling (Priest et al., 2008). While

this may have limited value in terms of pluvial flooding which was the example used in this tool,

responders were asked to consider the value of this more broadly in terms of other flood drivers. Some

commented on the value of representing the flood hazard in terms of the potential impact it may cause

and the importance of visualising the hazard according to a RAG rating (Red, Amber and Green), reflecting

high to low hazard posed.

“On a straightforward RAG rating that’s quite good because you can then have that quick judgement on

what you’ve got to do. I don’t think we’d adjust those thresholds [AS agrees], I think we’d take them out.

From my point of view I go with what the experts tell me…” (Emergency Planning, West Yorkshire)

Users of the tool are able to adjust these thresholds of depth-velocity, which automatically re-colours the

resulting map; however this was widely considered to be inappropriate. Indeed, many commented that

they trust the expert view and would be reluctant to manipulate these thresholds, raising issues of liability

and ‘who’s to say we have a right’ (Police, West Yorkshire).

iii) Users have the option to essentially ‘clean’ the map image and view the flooding where it intersects

the road network only. This was widely regarded as a valuable feature in terms of assessing the access for

people and in particular emergency vehicles. From this it was suggested that further tailoring of the tool

should take the interactive feature of adjusting hazard thresholds (described above) and instead place this

with the road network for accessing safe vehicle access; the user could then adjust a specific depth or

depth-velocity to suit the vehicle, or simply select a vehicle from a list box (e.g. Fire appliance, standard

4x4 car) and view a binary Yes/No for safe access on the resulting map. The challenges of using this

information were also mentioned; for instance it does not consider a debris factor and hidden dangers

under the water (Police, West Yorkshire), but nonetheless as a guideline tool for directing resources this

feature was highly rated.

iv) This final feature enables the user to ‘clean’ the map to view the flooding at the property level, based

on the selection of two hazard models; HM1 based on the risk to life thresholds and HM2 based on depth-

damage thresholds, based on the rationale that the first model could help inform response, while the

latter could support recovery efforts. This initial rationale was supported in the feedback from

professional stakeholders.



40

“I think it’s useful, it would be more helpful for recovery planning to be able to look at the impact on

buildings, on properties and depth; rather than, there’s not much that we’re going to be looking at for life

risk, very useful for fire service no doubt and for us to know. So it’s useful to have those two things.”

(Emergency Planning, IOW)

Responders were asked which of these models they would use to visualise the hazard posed to the

property (Table 11). The predominant answer was a selection of both models would be used; however

where responders did select a specific hazard model it seems the decision is divided amongst the police

and emergency planning officers interviewed in this study, as to which model is most appropriate. The

main justification supporting HM2 was that these responders considered flooding in the UK to pose a very

small risk to life and as such, felt that impact is better recorded in terms of depth-damages to properties.

Table 11: Responders’ selection choice for hazard model

Number of responders who selected

this option

Affiliation

Hazard Model 1: Risk to Life 3

EA (Hants), Police (WY), Emergency

Planning (IOW)

Hazard Model 2: Depth-damages 3

Emergency Planning (Hants), Police

(Hants), Emergency Planning (IOW)

Hazard Model 1 and 2

4

Emergency Planning (WY), Fire and

Rescue (WY), Environment Agency

(WY)

It is noteworthy that while some blue-light representatives gave an answer for this question, it was

commented that the resulting map highlighting certain properties is less useful than a map indicating the

flood extent. From this it might be inferred that the property level presentation is useful in partnership

with traditional flood extent, rather than as a standalone product in itself.

“It’s not particularly useful to me to just be able to select property. I just want a map that’s got the

properties on, got the roads on with the flood extent on it.” (Police, West Yorkshire)

Within this final feature, users have an additional option to view the minimum, average or maximum

result obtained in the model (i.e. where HM1 is based on the depth-velocity, and HM2 is based on depth

only). Responders expressed an appreciation of being able to ‘play’ with these options to view best case

to worst case scenarios; but again, while this would be suitable for exercising and training, the worst case

scenario is relied upon for planning purposes.

5.1.2 Animating flood hazard The interactive animation was the best received feature amongst all the professionals interviewed. This

form of dynamic visualisation was regarded as an effective communication tool for illustrating the spatial

patterning of the flood to other partners (e.g. a briefing tool; Emergency Planning, IOW) and painting a

clear picture of when and where for supporting planning and emergency response. Furthermore many

regarded this as a useful tool for exercising and training.

Some limitations of the animation were commented on. For instance, in response where time is of the

essence an animation of what the flood could look like on the basis of a design storm, i.e. not calibrated to

incoming real-time data, may paint an image of accuracy but is in fact bounded by a number of

uncertainties.

“So even if we start with here’s the model, it’s got these uncertainties, when you show an animation you

get caught up in it and you forget this is plus or minus depth, time. I think it’s good because animation



41

engages people better and makes them more interested and anything that makes people more interested

in flood risk is a good thing. But it’s that uncertainty that kind of gets forgotten when you watch a film of

something.” (Environment Agency, Hampshire)

Conversely, it was also noted that any visualisation of the potential flood is in fact a way of reducing the

uncertainty during response and a useful tool for directing resources; reducing it from nothing (Fire &

Rescue, West Yorkshire). It was suggested by some that users within the response context in particular,

might prefer the static maps that can be produced at incremental times (T15 mins, T30 mins etc:

Environment Agency West Yorkshire); or summary tables to provide rapid, digestible information

describing i) flood start, ii) flood end, iii) flood peak, iv) max depth and v) max velocity (EA, Hampshire).

5.1.3 Background information Responders were asked to comment on the usefulness of the background pages provided throughout the

tool. It was commonly accepted that these pages helped support the user in understanding the underlying

data in the resulting map. A recurring theme throughout these interviews, concerned the rarity of flood

events and multiple demands placed on emergency professionals which mean FIM is but one small aspect

of their professional roles; therefore, a tool which is flood-centred needs to provide clear background

pages such as these to not only refresh the users knowledge on the data, but also the capabilities of the

tool itself. The presentation of these pages is important for ensuring these points are clearly and

succinctly communicated and to avoid misunderstandings. Currently, the tool provides a range of

background pages style – detailed descriptions, bullet points, diagrams and illustrations and small pop-up

boxes. Many of the responders interviewed suggested that a successful background page should include a

mix of bullet points and wherever possible a clear diagram or table to explain the information. A further

requirement is to maintain consistency in how these pages appear to the user.

5.1.4 Communicating uncertainty Communicating uncertainty in flood science has been a growing research interest in recent years (e.g.

Faulkner et al., 2011; 2007). In this tool, the underlying uncertainty is accounted for in a simple list of

model assumptions. While some have researched the importance of visualising this uncertainty (e.g. XXX)

and integrating this into decision support tools (Leedal et al., 2010), this was has not been calculated in

the original modelling research utilised in this study, and thus it was not possible to visualise this in this

instance. However, responders were asked their views on how best to represent uncertainty, and if

indeed they considered uncertainty to be an important element to be integrated within decision making.

While some commented that they would like certain or uncertain information to be discussed more (EA,

Hampshire), there were mixed views on how this could achieved. Simple describing the uncertainty and

providing a ‘health warning’ for example (Faulkner pers comms.), could be difficult for users to

understand (EA, Hampshire) and in light of the earlier views on ‘wordy’ background pages, the first

challenge may be in actually getting users to read this information in the first instance. One way forward

may be too simply provide a range of scenarios and summarise the range of depth or depth-velocity

recorded within the model within each scenario (EA, Hampshire). Some responders expressed the risk of

communicating uncertain information, which could otherwise cloud the overall picture and message of

the map.

The thing is it’s always good to have various elements that give a rounded picture of the situation and

uncertainty is one of those elements, but I think if one represented it in graphics and what-not, instead of

just having it as a statement it may end up becoming a larger element than it actually proportionally

needs to be and end up clouding the message you’re trying to get through…(Emergency Planning, IOW)

An important point to take forward from these discussions was that some users did express an interest in

viewing a series of options for communicating certain/uncertainty data and this certainly warrants further

research.



42

5.2 The Vulnerability Interface of the tool Responders were similarly asked to rate the vulnerability features of the tool. Graph 5 illustrates that

these features were consistently rated lower in terms of usefulness in comparison to those demonstrated

within the hazard face of the tool; the average score of 3 shows that responders felt these features were

neither useful nor not useful. That is not to say that those interviewed didn’t have an opinion on these

features and these are discussed in the following sections.

Graph 5: Ratings for the Vulnerability features of the tool; average score and standard deviation, based on

responders’ feedback

Question

(corresponding

to Graph)



1 Option to view vulnerability at different geographical scales (based on the Social flood

vulnerability index)

2 Option to adjust the social flood vulnerability index to different geographical scales (nation,

region, district, town)

3 Option to view vulnerability indicators at the very local scale

4 Selection of a number of potential indicators of vulnerability

5 An expert-declared explanation accompanying each vulnerability indicator

6 Combine indicators of vulnerability and build your own vulnerability index

5.2.1 Manipulating the Social Flood Vulnerability Index (SFVI) The tool utilises the Social Flood Vulnerability Index (SFVI, after Tapsell et al., 2002) and uses this as an

example of a ‘vulnerability product’ The first feature of this interface enables the user to view the SFVI at

3 spatial scales; district or Island-wide and the town (Keighley or Cowes), mapping the score to census-

defined Output Area (OA; ca. 200 households), and at property scale in Cowes and the Stockbridge area of

Keighley. The user also has the option to adjust the SFVI score and can therefore view relative

vulnerability according to the nation, the region, the district or Island or the town itself. There were mixed

views on this form of assessment in terms of its usefulness in decision making; the reasons for this are



43

discussed in more detail in 4.2.3, but within the context of this method it was considered useful to be able

to adjust the score to suit the spatial scale of decision making.

“I think definitely like you said, on the Island they’d want to know relative to the Island and they don’t

really mind whether it’s relative to whatever’s happening in London. So in that way the national one is the

least important and being able to make it more local is more important.” (EA, Hampshire)

Secondly, users can view potential indicators of vulnerability in isolation, along with an accompanying

rationale based on an academic literature review. The original indicators used in the SFVI (after Tapsell et

al., 2002) were used in this tool as a sample of indicators only and it was recognised that a real-life

application could provide a more extensive list, utilising the census data. Responders commented that this

feature could help facilitate an understanding behind the SFVI scoring system, visualising the ‘make-up’ of

the area (rather than necessarily relying of pre-conceived assumptions: Emergency Planning, IOW) and

furthermore tailoring awareness-raising campaigns (EA, West Yorkshire).

5.2.2 Building your own vulnerability index The final page of the vulnerability interface gives the user the opportunity to construct their own

vulnerability index; using the sample of indicators selected and weighting each in turn according to its

relative importance in the user’s decision making. This represents a move away from the original SFVI

method which treated each indicator as equally important in governing vulnerability, which is a recurring

problem highlighted in the literature and results from the lack of defensible, objective weighting schemes

(Wilson, 2008; Cutter et al., 2003). It is argued here, that the search for objectivity is not a key

requirement; indeed it is the subjectivity of the responder, which is informed by their professional roles

and responsibilities that is required to shape more meaningful assessments of social vulnerability.

Furthermore, it is conceivable that an index for vulnerability (its indicators and weightings) may vary

between different applications in FIM; as such, this approach was considered appropriate for

incorporating this flexibility.

There was a mixed response to this feature (Graph 6). Responders representing emergency planning (EP)

and the Environment Agency all rated this option as highly useful (score 5); with the exception of

emergency planning Hampshire. Scores of 3 and below were recorded by the blue-light services

interviewed as part of this research who explained the limitations of this approach in real-time

operational and tactical response, as well as in supporting strategic decision making: These points are

discussed in more detail in the following section.

Graph 6: Responders’ views on constructing their own vulnerability index



44

“I think that’s really good and just to see as you went down, e.g. well unemployment that’s not really so

important to responding to a flood event, but some of them like elderly really are. So to be able to select

them out a bit I think is really important and I’d almost want to go straight to this page if I was looking at

this information and I wouldn’t look at the other pages.” (EA, Hampshire)

It was also noted that this feature could prove useful when comparing between locations: “..I think that’s

where these drop downs could really come into their own, because you would choose these which are

relevant to Cowes, you know unemployment is not that important; however in Ryde you might have other

variables rated high or less in your drop-downs and therefore comparing Cowes to Ryde in the colours

would then show me more… actually seeing it in the round, comparing place to place “ (Emergency

Planning, IOW). The flexibility in this approach could also prove useful if the scope of the tool were

extended beyond the realms of flooding and integrated further hazard types.

One concerned voiced by Emergency Planning and the Police for Hampshire, was that each responder

may use this tool, set it to different weightings and arrive at different results of vulnerability. In practice,

this feature would probably need to be worked-on within the context of multi-agency working to reach a

consensus on the appropriate indicators and aggregation. This is suggested in section 7 as an area for

further research.

5.2.3 Limitations in assessing vulnerability There was a recurring theme of discouragement surrounding the indicator/index approach to

representing social vulnerability. Ultimately it seems that these views reflect a mismatch in scale between

the flood hazard assessment based at the property level and the vulnerability assessment nestled at a

scale above this. For a true local scale assessment of vulnerability, and risk, responders require locally-

informed information. This information is primarily sourced from local authority adult and social care

services (e.g. the SWIFT database) and health services (e.g. PCT) during flood incident response to locate

key vulnerable households (e.g. dialysis patients). However, for reasons of data protection this

information cannot be centrally stored or mapped and requires the joined-up working of agencies, as

established within the principles of integrated emergency management. Broader-scale assessments of

vulnerability based on the census data can be used to provide the background context, but for local-scale

events it is this key household level information that is required to support FIM. Given the decadal nature

of the census, many responders commented on the danger of relying on out-dated information for

informing any aspect of response. From these views it seems that this type of vulnerability assessment

‘sets the scene’ for flood incident response but does not dictate actions. However, it was suggested that

this role could change in the context of wide spread flooding, similar too if not exceeding the scale of

flooding witnessed in 2007; events of this nature, requiring strategic decision making and priority-setting

when resources become stretched, could find a use for vulnerability assessment of this nature. Although

the mapping of area-wide vulnerability could facilitate this scale of decision making, key point-locations of

vulnerability sources are still required; such as the location of hospitals, blue-light service stations etc.

The temptation as a responder is that you go down one of these streets and somebody’s shouting from

their upstairs window and they become embroiled in the first person that they meet, because they’re at a

tactical level, an operational level even. Strategically, when you’re sat back in the command unit it might

be that that person sat in the upstairs window is perfectly safe…What you have to do is get past them and

stop it going to the sub-station. That’s just one example. (Fire and Rescue, West Yorkshire)

“It gives you that general picture because I think to get to that we’d still have our normal links into the

health service, our normal links into our adult and social care, so we’ve got our normal links running as

well but this will give us a .... Actually it’s a fantastic training tool.” (Emergency Planning, West Yorkshire)

“I think I’d only use these indicators in terms of planning and I don’t think you have time to look at it in

terms of response; then that’s the time when you need to be looking more at the local authority contact



45

details for what they’ve got for vulnerable people and trying to contact them.” (EA, Hampshire)

5.3 The Risk Interface of the tool This final major interface of the tool brings together the hazard and vulnerability assessment and allows

the user to integrate these together to infer risk at the property scale. The risk model depends on the

user’s weighting between hazard and vulnerability (each on a scale of 1 to 5; low to high); whether these

two scores are treated equally and simply multiplied together, or whether one is considered to be more

important over the other in governing the user’s decision making. The option to combine the hazard and

vulnerability scores at the property level and the additional option to weight these components both

received an average rating of 3.8 and are nearly considered as being useful amongst those interviewed.

This feature of the tool seemed to create some confusion amongst responders and even when the

weighting system was re-explained some had difficulty in comprehending the resulting map i.e. is the

property highlighted as high risk due to hazard or vulnerability? In some of the discussions this led to ‘risk’

being treated as synonymous to vulnerability. The initial intention was that the resulting map would again

represent a ‘clean’ image and move away from merely overlaying the flood extent with vulnerability

information; however, there was a wide agreement amongst responders that they would continue to

overlay the flood extent data.

“I think it’s interesting to see the risk calculation shown as a map but I would always like to break it back

down into hazard and vulnerability”. (EA, Hampshire)

In terms of weighting, it seems the hazard dominated (Graph 7). This is perhaps understanding given the

view that to be considered as vulnerable, a household must be vulnerable to something. This is perhaps a

short-sighted view which, while required during the response phase of an event, does not consider the

wider impact of the flood on the surrounding community.

I think it would probably be equal to be honest, because we were very much driven to look at the hazard …

but we are expected to know a lot more about the community and what’s going to flood and what the

impact of that flood’s going to be, so I think that into the future vulnerability is going to definitely rise up in

terms of how important [it is] … (EA, West Yorkshire)

Graph 7: Weighting hazard and vulnerability to inform risk

An interesting point made by emergency planning, IOW, was that the ability to adjust these weightings

could prove useful in times of interacting hazards:



46

“…If we were in a situation of pandemic flu already and then went into a period of flooding in the new

year, then perhaps we would weight vulnerability higher because we’re in an abnormal situation anyway.

Or for example at the moment with all the cuts and recession, all that kind of stuff, people’s vulnerability

could be exacerbated from their general deprivation, so the deprived people could be more deprived, the

people who are middle class, second home owners that we have a prevalence of on the Island suddenly

dip-down to become more vulnerable, just from their social level, so therefore the weighting of vulnerable

could be more than equal, because of the current climate that we’re in. So I like that flexibility.”

(Emergency Planning, IOW)

This report has thus far advocated flexibility in vulnerability indicator-selection and weighting between

different phases of a flood event, but this latter finding also supports the need for flexible systems in a

multi-hazard context. The spatial and temporal shape (or ‘etiology’) of the hazard can influence social

vulnerability (Tapsell et al. 2010). Although emergency professionals commented on the generic nature of

vulnerability in terms of responding to rapid onset and short duration hazard events (e.g. flooding, snow

storm, heat wave), vulnerability can manifest differently between the different flow-out characteristics of

these hazards or for slow onset, long-lasting events such as drought. Therefore, it is conceivable that

vulnerability criterion and hazard-vulnerability weighting could change between certain hazard events and

at times of interacting hazards (Emergency Planning, IOW).

Responders also reported on the tasks within FIM that this risk calculation could support. Once more it

seems to have a very limited role in the context of response, but has potential for planning and in

exercising and learning. The resulting counts of properties and estimated people count within each risk

category, ranked very highly in responders feedback and was considered as something that should be

repeated throughout the tool (i.e. in the hazard and vulnerability face also).

“we used to only use property information and most partners like to know the number of people, so having

them both is really important so I think that’s really good. And having a summary for use during a

response is helpful, more helpful than actually looking at the maps.” (EA Hampshire)

“…the property and people count, how many do I need to put into a rest centre, what do I need to do. Give

us a quick clue, a heads up. I do like that.” (Emergency Planning, West Yorkshire)

5.4 Overall views of the tool In the questionnaire responders were asked to rate the tool in terms of its simplicity, user-friendliness,

interactivity and the overall presentation (Graph 8).

Graph 8: General design of the tool (based on the average score and standard deviation)

Interactive User-friendly

Simple General

presentation



47

Keep It Simple Stupid (K.I.S.S) was a recurring theme and essential requirement in the end-user’s ‘wish

list’: “Keep it simple stupid. The simpler it is the more likely it is to get used”. (Emergency Planning,

Hampshire). Furthermore, any tool should be designed to be user-friendly, that is to say it is self-

explanatory, intuitive and requires very little training to use. Indeed many responders commented on the

shortcomings of the National Resilience Extranet which seems to have failed by trying to be ‘the Swiss

army knife’ (Fire and Rescue West Yorkshire).

“…the NRE… started off as it’s going to be a document storing site where all organisations can store their

documents, and it’s gone now to something which has gone so big that’s it’s lost the initial meaning of

why it was created….It had a very clear start, initiation of what people wanted which was a tool to share

information, that they’ve tried to bolt on so many other bits to it that people have actually stopped using

it.” (Police, Hampshire)

The main justification for these criterions (i.e. simplicity and use-friendly) is due to the nature of

professional’s roles; as previously mentioned, flooding is but one small part of the day-job (with the

exception of those interviewed within the Environment Agency). If a tool is flood-centred, then it needs to

be something that can be like riding a bike (Emergency Planning Hampshire). While interactivity was also

highly rated, this was regarded as less important than the goals of simplicity and user-friendliness. It was

also suggested that interactivity was a requirement of a tool supporting planning and exercising, thereby

allowing the user to play with a number of scenarios and features; whereas for response, the user would

require rapid answers and the interactive nature of the tool would rate less highly if it slowed this process.

“Our work is very different, our day job is planning, training and doing exercises; that is very different from

the response. The day job is measured and calm, everything changes with response, you don’t have time to

sit down with a computer and play with a system. Time is always a luxury in emergency response.”

(Emergency Planning, Hampshire)

Many reflected on separate faces to the tool and felt that isolating hazard and vulnerability, and

combining them on a separate page for risk, was logical, clear and appropriate given that different

responders will value these differently.

“It makes sense, it’s good that you can separate hazard, from vulnerability and then putting it all together

because of the multi-agency thing, some people wouldn’t be interested in vulnerability they’d just be

interested in velocity and depth. But other agencies would find vulnerability the most informative to

them.” (EA West Yorkshire)

The current version of the tool uses a summary window to illustrate the main screen in ArcMap and limits

the user’s interaction with the software itself; based on the preliminary feedback that responders felt

uncomfortable using ArcGIS, or indeed any complicated system and required a simple and user-friendly

tool. There was a restriction with the licensing arrangement with ArcGIS which meant that it was not

possible to interact with the summary window (i.e. zoom in/out, pan view); however this is feasible with

the appropriate license. An ideal tool would completely eliminate the user’s interaction with ArcMap and

preferably would exist independently from such a system that requires expensive licensing. While the

desktop application was appropriate for this ‘mock tool’ which was concerned with asking a number of

research questions rather than becoming a real-life application at this stage, in practice the tool would

need to be accessible to all to facilitate the goal for a Common Recognised Information Picture (CRIP).

Open-source GIS via the internet could be a potential step forward in meeting this goal.

Professionals interviewed were asked to rate the each face of the tool in terms of its application potential

for different phases within FIM (Graph 9).



48

Graph 9: Responders’ views on the application-potential of each face of the tool, for different phases of FIM (based

on the average rating)

On average, responders can see the potential of this tool for real-life application to support activities

involved in planning, response, recovery, mitigation strategies and exercising and training. Both hazard

and vulnerability information are regarded as equally as important in terms of emergency planning, i.e.

knowing where will flood and who will be exposed and vulnerable to that flood. Hazard information

understandably dominates emergency response, but as has been previously commented on, hazard

information presented in the form of design events (as this tool utilised) has a limited role if it cannot be

calibrated or supported by incoming real-time data. Limitations in the vulnerability assessment (as

outlined in 5.2.3) are inherently transferred into the risk calculation at the local scale and thus these two

components (vulnerability and risk) have very limited roles to play in informing response; indeed many

responders felt that the form of vulnerability assessment presented here, has a role to play in strategic

decision making only. Vulnerability assessment becomes more valuable for supporting recovery activities

and targeting and tailoring longer-term mitigation strategies.

5.5 Suggestions for further tailoring The tool was designed on the basis of preliminary interviews and was essentially an ‘all in’ tool. The

purpose of these secondary interviews was to ascertain what aspects/features of the tool were valued

and by which professional stakeholder groups, which section 5 has so far addressed. Responders were

also asked for their views on further tailoring the tool. It was widely accepted that any successful tool

must have an apparent purpose. If the tool is to be an inclusive flood risk assessment tool then it needs to

be designed to distinguish between the phases of FIM; from planning, response, recovery and longer-term

mitigation strategies, and exercising. Some agreed that the tool would then further divide to reflect the

operational/tactical and strategic tiers of decision making. A final suggestion was made that the tool

should distinguished between responders (i.e. police, fire and rescue, emergency planning etc.); but many

disagree with this suggestion and concluded that it went against the efforts of multi-agency working.

It is the research team’s suggestion that a real-life application of this tool would be a district-wide version

(within the boundaries of LRF) and would have a drop-down menu for the user to select from a number of



49

‘hotspot’ locations of flooding – of which Keighley and Cowes are but one example. At the local scale

detailed flood inundation modelling could be commissioned, whereas district-wide flood assessments

could be supported by the existing fluvial and coastal flood zones, and broad scale surface water flooding

outputs already available from the Environment Agency. Existing GIS-layers mapping key locations of

emergency services and critical infrastructure etc. could then be centralised and mapped alongside the

census-derived indicators of vulnerability.

6 Recommendations for designing and implementing

future flood risk assessment tools

Flood incident management in the UK is evolving in light of the predicted increase in flood risk into a

structure of multi-agency working; with the aim of facilitating communication, data and resource sharing,

and enhancing overall response, to minimise the overall impact and losses that may ensue from flooding.

Effective emergency management requires forward-thinking and while to some extent the actual

response to a flood event will contain a degree of ‘fire fighting’ (i.e. reactive response), the success of this

will be largely determined by the proactive activities of FIM (emergency planning, targeted and tailored

mitigation and exercising and training of emergency professionals). There is a mounting pressure upon

FIM professionals to not only to communicate and operate effectively with multiple stakeholders, each

with different professional roles, responsibilities and constraints, and various knowledge domains; but to

realise this when resources are stretched (i.e. financial constraints) and meet performance targets.

Rapidly evolving technologies, decision support systems and visualisation could provide ‘tools’ to support

this goal. The challenge then for future ‘tool’ developers, is a matter of tailoring:

� Who is the end-user?

� What tasks does the tool support?

� How practical is the tool (e.g. can new data be easily integrated, updated, shared?)

� What can the tool NOT do?

Clarity is a key requirement and tool developers need to make it explicit what their system can and cannot

achieve. A recurring example cited in professionals’ responses was the failings of the National Resilience

Extranet (NRE) and it was widely acknowledge that the ‘all in’ approach has been the major failing in the

system. Simplicity is essential. This means a tool needs to be not only simple and user-friendly to use, but

needs to be simple in its objectives and intended purpose. User-friendly is a further requirement so from

the tool’s handover from developer to user, it must be as self-explanatory as possible and require minimal

training. This is particularly important given the diverse nature of the typical Category One Responder’s

day job, of which flooding is but one aspect.

The tool developed in this study was not intended for real-life application but sought to ask many

research questions which could help steer the design of future flood risk assessment tools for practical

application. This study is thus considered to be a stepping stone for future pragmatic flood research and

highlights the importance of engaging with the ultimate end-user of the tool from the early stages of its

design, construction and appraisal. This following section bullet points a number of things to take forward.

Figure 14 summarises the features that would be essential for a flood-centric decision support tool;

however, it is noteworthy that the malleable vulnerability and risk features presented in this research

could be adapted for a multi-hazard decision support tool.

� Design flood scenarios (including different flood drivers and return periods) are useful for

planning and training and exercising; however, the worst case scenario is essential. Visualising

flood scenarios is a useful tool for prompting proactive thinking (rather than merely reactive),

and providing a visual example of the possible spatial patterning of the flood. Tools supporting

emergency response should enable an interactive feature for the user to manipulate river/rainfall

levels to mirror incoming real-time information.



50

� Flood extents should be complemented with depth and depth-velocity data to give an indication

of the hazard threat – whether this is based on risk to life thresholds or depth-damage

thresholds. These are arguably mislabelled indications of risk rather than hazard (as referred to in

this study; the purpose of placing these within the hazard face of the tool was to emphasise that

the calculation was based on the flood parameters only, rather than calibrated to actual

household data (as would be required for this to be a true calculation of risk). Nonetheless, the

household presentation of flood data serves as a useful means of clarifying where flooding may

have a tangible impact.

� Depth-damage hazard thresholds should be calibrated to the EA’s property database and FHRC’s

multi-coloured manual to give a more accurate representation of impact.

� ‘Cleansed’ map images that show only the impact on the road network or properties, are useful

but not essential. According to the responders interviewed here, these layers would be

supplemented with the flood extent and are not therefore as useful when considered as stand-

alone layers.

� Re-colouring flood extent to clearly indicate and enable the user to interpret the varying degrees

of the hazard posed, was a useful but those interviewed on the whole, would not want to

manipulate these hazard thresholds; expert-declared, pre-set thresholds will suffice.

� Interactive animation is in invaluable means of depicting the spatial-temporal patterning of the

flood (as it accumulates, recedes and ponds): Where and when are key questions in planning for

and responding to flood incidents. Providing the animation is a slick operation i.e. can be sped-

up/slowed-down/paused etc. Summary tables should be used to support the visualisation and

give a rapid summary of what is being displayed on screen (e.g. flood start, end, peak).

� Uncertainty in flood modelling needs to be communicated. One concern for visualisation is that it

portrays an image of accuracy and although a caveat of uncertainty may accompany this it is not

likely that it will be understood unless it can be equally visualised in some form. Whether and

how certain or uncertain information is visualised is a matter for further research; this study has

shown that a key requirement is that this information shouldn’t add confusion and risk

complicating the key message of the map. Many of those interviewed did however express an

interest in seeing visualisations of uncertainty.

� There is a mismatch between the hazard and vulnerability data, both in terms of spatial and

temporal resolution. Whilst the hazard information can be calculated and mapped at the

household scale, the vulnerability data for the same household is based on an aggregated score

of the census-defined output area (ca. 150-200 households). Flood modelling captures the

dynamics of the hazard in both in space and time; yet vulnerability assessments provide only a

static snapshot, limited by the decadal nature of the census. Consequently vulnerability

information cannot support operational or tactical response, given its lack of spatial and

temporal clarity; however, it is conceivable that it may have a role to play in broad scale events

exceeding the nature of 2007 summer floods in the UK.

� Vulnerability information is best utilised in meeting the sustainability goals of FIM i.e. targeting

and tailoring mitigation strategies.

� Vulnerability is integrated into operational and tactical response via information sharing (e.g.

location of critical patients). This information cannot be centrally stored unless the said-

household has formally signed permission for authorities to do this; this is in accordance to the

Data Protection Act 1998. Future tools could facilitate the integration of this information with

hazard data when there is sufficient time to do so.

� A novel feature of this tool is the interactive, user-control construction of a vulnerability index. It

is acknowledge that this feature is not necessarily suitable in the context emergency response,



51

and its associated time constraints. However, given the number of FIM activities beyond the

response phases, it is argued here that this level of interactivity and the ability for the users to

integrate their informed subjectivities is an asset. This recognises the limitations of the objective

scientific expert when it comes to informing professional’s decision making – i.e. the subjectivity

of the professional becomes the expert voice in the tool. This feature provides an adjustable and

malleable vulnerability index and has the ability to meet various end-user needs and to support a

variety of FIM activities.

� The limitations of vulnerability assessments, means that a meaningful calculation of risk at the

household scale is not appropriate. Furthermore, this feature seemed to create some confusion

amongst professionals, who would have preferred the standard layering of hazard and

vulnerability information. Some professionals expressed an interest in weighting hazard and

vulnerability information to view how the property and people estimates changed between

different combinations. There may be a context for weighting hazard and vulnerability differently

to support different tasks within FIM, although the limited number and nature of professionals

interviewed in this study cannot shed further light on this and requires further research to

ascertain its potential value to decision makers.

� The requisite for simplistic tools is tied to a contentious debate concerning the dualistic meaning

of simplicity: Do practitioners require simplistic-user-friendly tools or simplistic-information tools.

It is apparent from the professional interviews reported on in this report, that simplistic-user-

friendly tools are essential. This is due mainly to i) the varied demands placed on professionals

which mean flood-related matters are but one, small component of the day-job; and ii)

inexperience, and a lack of confidence and self-efficacy in using new software. There is mixed

support for the interpretation that the desire for simplicity reflects a desire for simplistic-

information. For instance, most responders acknowledged the difficulties of defining and

mapping vulnerability and appreciated the nuances in this concept; on the other hand, while

acknowledging uncertainties in flood modelling, some responders seemed reluctant to engage

with it – this may be because they found it difficult to visualise how uncertainty might be

integrated in mapping, or because uncertainty is common place in the day-job and therefore not

regarded as an issue. There is a need for further research on this matter which implies a greater

tension involving the translation of science to practitioners and the professional context.

� ArcGIS was selected as the building-and-launch platform for this tool for the purpose of

facilitating discussion with professional stakeholders and addressing a number of research

questions. It is acknowledged that this software has a number of limitations which would prevent

this tool functioning in real-life situations. Aside from expensive licensing, the desktop

application is not suitable for creating and sharing a ‘common picture’ of the situation as

required for multi-agency working. The software is also unfamiliar to many practitioners and

would require training. The platform for launching a tool of this nature would need to be

something that is not reliant upon licensing and open to all potential users. An open source GIS

via a secure website would probably be the closest to meeting these requirements.



52

Figure 14: The ‘ingredients’ for developing future decision support tools in flood incident management

7 Reflections This research sought to understand the utility of social vulnerability assessment within the context of

flood incident management. The previous section of this report concluded that different aspects of

vulnerability (indicators and scales of mapping) are required to meet the diverse needs of emergency

responders and the different demands placed of responders through different phases of FIM. These

included;

1. Specific locations of critical infrastructure (e.g. hospitals, electrical substations): To support strategic-

scale decision making

2. Location of critical households (e.g. dialysis patients): To support operational and tactical response



53

3. Area-wide vulnerability indicators, adjustable to different contexts: To support strategic-scale decision

making, emergency planning, targeting mitigation initiatives and supporting training and learning amongst

professionals

The flood risk assessment tool developed within this research has principally engaged with the third type

of vulnerability assessment and demonstrated a novel way of allowing the user to manipulate a classical

approach for measuring vulnerability. This approach is arguably more appropriate than the traditional

method which imposes an objective, scientific view on what constitutes vulnerability; indeed, even within

academic literature indicator selection and aggregation are hotly debated and there is a tendency to

reproduce equal-weighted additive models – which in itself imposes a bold statement about the

relationship between vulnerability indicators. This research has argued a case for not only integrating the

end-users views on the matter, but allowing the user to actively construct their own vulnerability index,

acknowledging that the subjectivities of the end-user, informed by professional roles, responsibilities and

context, in fact represent the expert voice (rather than the objective scientist).

Although this research illustrates that there are possibilities for extending and utilising vulnerability

indicators within pragmatic decision support tools, the response from professional stakeholders

demonstrate that there are some fundamental limitations of this approach. In particular there is a

mismatch between the scales available for assessing flood hazard and social vulnerability, which

inherently constrain local risk assessments. On one hand, flooding can be modelled and depicted through

space and time and thus easily transformed into useable visualisations (e.g. animation), capturing the

dynamism of the hazard; conversely, vulnerability assessment remains constrained by a static-snapshot-

layering approach.

The Social Flood Vulnerability Index (SFVI, after Tapsell et al., 2002) is adopted as an example here and

represents a classical approach in vulnerability assessment. This method may be adapted so that

vulnerability is standardised according to different spatial scales and relative metrics of vulnerability are

assigned accordingly. Furthermore, vulnerability metrics can be simply intersected in ArcGIS with smaller

units of interest. Both these methods have been integrated into the flood risk assessment tool. However,

the fact still remains that the underlying data structure originates from a broader spatial scale. This is

commonly dictated by available spatial boundaries, such as that used by the census. The smallest spatial

unit of the census in the UK is the output area (OA), which includes ca. 200 households. Households

within this output area are then assigned a category of vulnerability – that is to say, they are all assigned

the same category. The abrupt and meaningless shifts in vulnerability that are created at the boundaries

of administrative districts is particularly problematic. Figure 15 illustrates this. Indeed, what is the

rationale for a category jump of 2 to 5 between neighbours A and B?

Figure 15: Abrupt shifts in vulnerability. Household A is a

category 5 (very high vulnerability); Household B is a

category 2 (low vulnerability); Household C is a category 4

(high vulnerability); and Household D is a category 3

(average vulnerability).

This form of vulnerability assessment is subject to the principles of ecological fallacy (Lloyd, 2010).

Essentially, the aggregation and mapping of data within such census-defined boundaries creates an



54

impression of area homogeneity and masks the intrinsic variability that exists within. In terms of FIM, this

assumption is acceptable in the context of high-level, strategic decision making, but becomes problematic

when seeking to adopt such methods to inform local-based (operational and tactical) decision making; in

this context, knowledge concerning the local variability (i.e. ‘hotspots’ of vulnerability and critical

households) is essential. Furthermore, area-wide assessments of vulnerability are also subject to what has

been termed by Openshaw and Taylor (1979) as the Modifiable Areal Unit Problem (MAUP). Essentially,

the scale at which vulnerability is calculated will directly influence the resulting score and interpretations

of vulnerability. Assessments of vulnerability are subject to these principles. Professionals interviewed in

this study heavily discussed these limitations in adopting census-derived indicators to support decision

making; indeed this received proportionally more discussion-time than the uncertainties surrounding

flood modelling. The acute awareness of these limitations (e.g. the decadal timescale of the census and

issues of accuracy) means that this form of vulnerability assessment is considered useful only as a ‘broad

brush’ approach to painting an areas social make-up. It cannot therefore be bolted-onto local flood

modelling and applied to a risk calculation. The two approaches in hazard and vulnerability appraisal are

divorced in terms of scale (spatial and temporal) and must be equally resolved in order to inform a

meaningful risk assessment.

It has been argued that more meaningful assessments of vulnerability could derive for instance, from the

use of existing social data regarding people’s attitudes and responses to flood risk (Twigger-Ross, 2010).

One might question who has the authority to impart this information and define what is meaningful; is it

the professional stakeholder, the academic researcher or the role of the community under scrutiny? This

report highlights potential methods for adapting existing vulnerability approaches (using the SFVI as an

example) and the potential for integrating this flexibility and end-user control within a decision support

tool. Limitations in the area-wide approach, resulting from the dependency upon existing census data,

could be resolved with the inclusion of locally-informed information; whether this arises from exiting

social surveys regarding flood experience and responses specifically or from more generic discussions with

the public in at-risk locations. Social science could facilitate this process of seeking more meaningful

assessments of social vulnerability.

Although there is scope to continue research which seeks more meaningful indicators for vulnerability,

and its partner resilience, issues remain with the static nature of the assessment. Future technological

advancements in agent-based modelling may serve to capture the dynamics of vulnerability in time. This

approach has already been applied for instance in simulating city-scale rescues (Jennings, 2011: Ramchurn

et al., 2009), evacuation (EA, 2009) and the effects of catastrophic flood events upon behavioural

dynamics and resulting loss to life (Dawson et al., 2011: Lumbroso et al., 2005; Johnson et al., 2005).

There is a clear application-potential for this type of research for informing operational planning and flood

incident management. While there are a number of constraints and uncertainties governed by the

underlying algorithms which dictate agent behaviour, this approach is the best means available for

capturing the dynamic nature of vulnerability that results from human decision making (collective and

individual) and institutional responses (e.g. emergency management); as well as the effect of the hazard

itself upon predisposing vulnerability characteristics (i.e. demographic details). Furthermore, agent-based

methodologies illustrate the dynamic partnership between the hazard and vulnerability in governing risk.

However, whether agent-based modelling can become a practical reality remains to be seen and is

proposed here as a potential addition to the FIM toolkit.

Conclusions

Flood risk is set to escalate over the coming century and therefore, there is a necessity to develop and

refine tools to support decision making to alleviate the increasing burden on practitioners and the future

costs from flooding. This research aimed to utilise 1D-2D urban flood modelling developed in FRMRC1

(Exeter University and Richard Allitt Associates) and integrate these outputs with social vulnerability

metrics in two socially contrasting locations; Keighley, West Yorkshire and Cowes, Isle of Wight. Both



55

hazard and vulnerability assessments are centralised in a GIS-based tool to facilitate local-scale risk

profiling.

The tool was designed on the basis of preliminary interviews with emergency professionals (e.g. blue light

services, emergency planning and Environment Agency) and the completed product has since been

demonstrated to a sample of these stakeholders for evaluation. The tool was not designed to roll-out in

practice but rather to inspire some practical recommendations for improving the assessment and

visualisation of hazard, vulnerability and risk for future real-life applications. Furthermore, the tool has

facilitated researcher-professional engagement and furthered understanding into the utility of social

vulnerability assessment within flood incident management.

It is argued here that future vulnerability assessments which rely on census-derived indicators should be

presented to FIM professionals in an interactive and malleable form to suit the flexibility required in

meeting the diverse needs of different professional groups; as well as the varied requirements of different

phases within FIM. This approach to vulnerability is considered appropriate for purposes of emergency

planning, steering recovery efforts, targeting and tailoring mitigation strategies and for professional

exercising and training. There is a limited role for this form of assessment in informing emergency

response, which instead requires accurate, point-scale information on critical facilities and households. It

was suggested that strategic decision making required for broad-scale flood events could employ this

form of vulnerability assessment to help target resources when stretched; however, as this scale of event

is yet to be seen in the UK it is difficult to infer whether vulnerability could practically play a role.

Limitations in this approach and the mismatch between vulnerability and hazard scales (temporal and

spatial) mean that the calculation of local risk as presented in this tool is problematic. While the weighting

criterion seems to have some scope, it is arguably inappropriate to calculate risk at this scale (unless

household level vulnerabilities are available).

Interactive assessments and map-making are seen as a powerful tool for not only communicating science

at the professional interface, but also integrating professional knowledge; i.e. mixing objectivity with

informed subjectivities of professional end-users. Arguably, pragmatic flood research requires the

stakeholder to become an active participant in the research process. True end-to-end research (as

described by Morss et al., 2005) could facilitate two-way knowledge exchange between science and

practitioners, the uptake of new ideas and tools in practice and prompt new thinking; otherwise stifled

behind traditional communication barriers. This type of pragmatic research constitutes a broader effort to

enhance the translation of science at the practitioner interface – and one of the main goals sort by the

Flood Risk Management Research Consortium.

Acknowledgement This research was performed as part of a multi-disciplinary programme undertaken by the Flood Risk

Management Research Consortium. The Consortium is funded by the UK Engineering and Physical

Sciences Research Council under grant GR/S76304/01, with co-funders including the Environment Agency,

Rivers Agency Northern Ireland and Office of Public Works, Ireland. We would like to further acknowledge

the work of Richard Allitt Associates Ltd and their data providers, Southern Water, as well as the team at

Exeter University (Albert Chen and Slobodan Djordjević in particular) for supplying the relevant inundation

outputs for Cowes and Keighley, respectively.



56

References

Alexander, M., Viavattene, C., Faulkner, H. and Priest, S. (2011) Methods for creating a flood risk assessment tool.

Guideline report submitted to the Flood Risk Management Research Consortium (FRMRC2). July 2011.

Allitt, A., Blanksby, J., Djordjevic, S., Maksimovic, C. and Stewart, D. (2009) Investigations into 1D-1D and 1D-2D urban

flood modelling. WaPUG Autumn Conference 2009.

Alphen, J.V., Martini, F., Loat, R., Slomp, R. and Passchier, R. (2009) Flood risk mapping in Europe, experiences and

best practices. Journal of Flood Risk Management. 1-8

Brown, J.D. and Damery, S.L. (2002) Managing flood risk in the UK: Towards an integration of social and technical

perspectives. Transactions of the institute of British Geographers. 27 (4). 412-426

Cabinet Office (2010) Emergency response and recovery: Non statutory guidance accompanying The Civil

Contingencies Act 2004. Available from http://interim.cabinetoffice.gov.uk/ukresilience.aspx (accessed 14/04/2011)

Cabinet Office CCS (2010) National Resilience Extranet Newsletter. Vol 1 (7). November 2010.

Cashman, A.C. (2009) Alternative manifestations of actor responses to urban flooding: Case studies from Bradford and

Glasgow. Water Science and Technology. 60 (1).

Chen, A.S., Viavattene, C., Faulkner, H., Blanksby, J., Djordjevic, S., Saul, A. and Savic, D.A. (2009) Urban flood risk

communication using modelling results. COST C22 Final conference, ‘Road map towards a flood resilient urban

environment’, Paris 25-27th

November

City of Bradford Metropolitan District Council, Leeds City Council, Yorkshire Water, Environment Agency and Pennine

Water Group. (2008) River Aire Strategic Studies Report. TR344. April 2008

City of Bradford Metropolitan District Council (2005) Review to consider the future of water management and the

associated problems of flooding in the Bradford district. City of Bradford Metropolitan District Council.

Communities and Local Government (2008) The English Indices of Deprivation 2007. March 2008.

Dawson, R., Peppe, R. and Wang, M. (2011) An agent based model for risk-based flood incident management.

Natural Hazards.

Department for Environment Food and Rural Affairs (2005a) Making space for water: Taking forward a new

Government strategy for flood and coastal erosion risk management in England. Defra Publications: London

Department for Environment Food and Rural Affairs (2005) Flood Risk Assessment Guidance for New Developments:

Phase 2, R & D Technical Report FD2320/TR2

DiBiase, D.A.M., Maceachren, A.M., Krygier, J.B., and Reeves, C. (1992) Animation and the role of map design in

scientific visualisation. Cartography and Geographic Information Systems. 19. 201-214

Djordjević, S., Prodanović, D., Maksimović, C., Ivetić, M. and Savić, D., (2005). SIPSON - Simulation of interaction

between pipe flow and surface overland flow in networks. Wat. Sci. & Tech., 52(5): 275-283

Environment Agency (2010) Preliminary flood risk assessments: “Living draft” guidance for lead local flood authorities.

May 2010. EA: Bristol

Environment Agency (2009a) Flood in England: A national assessment of flood risk. EA: Bristol

Environment Agency (2009b) Reliability in flood incident management planning. Final Report – Part B: Technical

report. SC060063/SR2. EA: Bristol

Environment Agency (2008) River Aire Catchment Flood Management Plan. Draft main stage summary document.

Environment Agency: Leeds



57

European Commission (2007) EU Floods Directive 2007. 2007/60/EC

Experian (2009) Improve outcomes through applied customer insight: Experian’s Mosaic Public Sector citizen

classification for the United Kingdom. http://publicsector.experian.co.uk/Products/Mosaic%20Public%20Sector.aspx

Faulkner, H., Parker, D., Green, C. and Beven, K. (2007) Developing a translational discourse to communicate

uncertainty in flood risk between science and the practitioner. Ambio. 36 (7). 692-703

Faulkner, H., Alexander, M. and Leedal, D. (2012) Translating uncertainty in flood risk science. [In] Hall, J. and Bevan,

K. (eds) Applied uncertainty analysis for flood risk management. (in press)

Fereday, J. and Muir-Cochrane, E. (2006) Demonstrating rigor using thematic analysis: A hybrid approach of inductive

and deductive coding and theme development. International Journal of Qualitative Methods. 5 (1)

Flood Hazard Research Centre, Middlesex University (2010) The benefits of flood and coastal risk management: A

handbook of assessment techniques 2010.

Gaist (2010) Emergency services: Commanders Resource Platform.

http://www.gaist.co.uk/BusinessLines/EmergencyServices.aspx (accessed 25/07/2011)

Gouldby, B. and Samuels, P. et al., (2005) Language of risk – project definitions. T32-04-01. March 2005, FLOODsite

Halcrow (2011) Instant insight into flood risk: FloodViewer. Available from

http://www.halcrow.com/Documents/flood_alert/FloodViewer.pdf (accessed 20/07/2011)

HM Government (2010)The Flood and Water Management Act 2010.

HM Government (2009) The Flood Risk Regulations 2009.

HM Government (2008) Identifying people who are vulnerable in a crisis – Guidance for emergency planners and

responders. Civil Contingencies Secretariat. February 2008

HM Government, Environment Agency and Defra (2008) Preliminary guidance: Developing a Multi-Agency Flood Plan

(MAFP). Guidance for local resilience forums and emergency planners. Civil Contingencies Secretariat. February 2008.

HM Government (2004) Civil Contingencies Act 2004. Available from

http://www.legislation.gov.uk/ukpga/2004/36/introduction (accessed 16/06/2011)

HR Wallingford, Flood Hazard Research Centre and Risk and Policy Analysts Ltd (2006) Flood Risks to people. Phase 2.

FD2321/TR2. Guidance document. Defra/Environment Agency Flood and Coastal Defence R&D Programme.

Johnstone, W.M., Sakamoto, D., Assaf, H. and Bourban, S. (2005) Architecture modelling framework and validation of

the BC Hydro’s virtual reality life safety model. [In] Vrijling et al. (eds). Proceedings of the international symposium on

stochastic hydraulics. May 23rd

– 24th

2005, Nijmegen, Netherlands

Kot, B., Wuensche, B., Grundy, J. and Hosking, J. (2005) Information visualisation using 3D computer game engines

case study: A source code comprehension tool. CHINZ 2005 Proceedings of the 6th

ACM SIGCHI New Zealand

Chanpter’s Internation Conference on Computer-Human Interaction: Making CHI natural.

Lane, S.N., Odoni, N., Landstrom, C., Whatmore, S.J., Ward, N. And Bradley, S. (2010) Doing flood risk science

differently: an experiment in radical scientific method. Transactions of the Institute of British Geographers. 36 (1). 15-

36

Leedal, D., Neal, J., Bevan, K., Young, P. and Bates, P. (2010) Visualisation approaches for communicating real-time

flood forecasting level and inundation information. Journal of Flood Risk Management. 3. 140-150

Lloyd, C.D. (2010) Spatial data analysis. An introduction for GIS users. Oxford University Press: New York

Lumbroso, D., Di Mauro, M. and Ramsbottom, D. (2008) Recent developments in loss of life modelling for flood

defence and dambreak risk assessments. 43rd

Defra Flood and Coastal Management Conference. 1-3 July, 2008.

Manchester University



58

McCarthy, S., Tunstall, S., Parker, D., Faulkner, H. And Howe, J. (2007) Risk communication in emergency response to

extreme floods. Environmental Hazards.7 (3). 179-192

MacEachren, A.M., Robinson, A., Hopper, S., Gardner, S., Murray, R., Gahegan, M. and Hetzler, E. (2005) Visualising

geospatial information uncertainty: What we know and what we need to know. Cartography and Geographic

Information Science. 32 (3). 139-161

Maceachren, A.M. and Ganter, J.H. (1990) A pattern identification approach to cartographic visualisation.

Cartographica. 27 (2), 64-81

Morss, R.E., Wilhelmi, O.A., Downton, M.W. and Gruntfest, .E. (2005) Flood risk, uncertainty and scientific information

for decision making: Lessons from an interdisciplinary project. American Meteorological Society. BAMS-86-11-1593

Met Office (2011) Hazard Manager. Available from http://www.metoffice.gov.uk/publicsector/hazardmanager

(accessed 16/06/2011)

Openshaw, S. and Taylor, P.J. (1979) A million or so correlation coefficients: three experiments on the modifiable area

unit problem. [In] Wrigley, H. (ed) Statistical applications in the spatial sciences. Pion: London. Pp127-144

Pitt M Sir (2007) "The Pitt review, interim report". HM Government Cabinet Office, London

Priest, S., Tapsell, S., Penning-Orwsell, E., Viavattene, C., and Wilson, T. (2008) Task 10: Building models to estimate

loss of life for flood events. Produced for FLOODsite, Report number T10-08-10

Priest, S., Wilson,, T., Tapsell, S., Penning-Rowsell, E., Viavattene, C. and Fernandez-Bilbao, A. (2007) Building a model

to estimate risk to life for European flood events – Final report. T10-07-10 FLOODsite.

Ramchurn, A.D., Farinelli, A., Macarthyr, K.S. and Jennings, N.R. (20010) Decentralised coordination in RoboCup

Rescue. The Computer Journal. 53 (9). 1447-1461

Tapsell, S., Penning-Rowsell, E., Tunstall, S.And Wilson, T. (2002) Vulnerability to flooding: Health and social

dimensions. Flood risk in a changing climate. Philosophical Transactions of the Royal Society of Mathematical, Physical

and Engineering Sciences.360 (1796). 1511-1525

Tapsell, S., McCarthy, S., Faulkner, H. and Alexander, M. (2010) Social Vulnerability and Natural Hazards. WP4 Report

for CapHaz-Net, Social Capacity Buildings for Natural Hazards. Available from http://caphaz-net.org/ (accessed

01/04/12)

Twigger-Ross (2010) Developing and using measures of social vulnerability and resilience in FRM. [In] Living with

Environmental Change (2010) Social and economic theme topic review index.

Wilkinson, D. (2007) Flooding at Stockbridge, Keighley October 30th, 2000: A case study of response and recovery.

(from an EA report ‘joining-up: Stockbridge pathfinder, science report E2-9057/SR4

Wilson, T. (2008) Defining and Mapping Societal Vulnerability and Resilience: A Literature Review. Deliverable 3.7a for

FRMRC2, from FHRC at Middlesex University.

A gis based flood risk assessment tool supporting flood incident management at the local scale

Environment

flood incident management

risk mapping

frmrc funders

sole risk

frmrc partners

urban scale flood modelling

local scale meghan alexander

frmrc research report