Are Flood Warnings Futile?
Risk communication in emergencies
Flood warnings often don't work well and too frequently fail completely - and this despite great effort by the responsible authorities. Reasons for this may be inherent in the methodology and definitions used to assess warnings, for example, higher standards may be applied to warnings than to other forms of risk communication, and the definition of failure will often determine the outcome of an evaluation. Aside from these methodological issues, warnings may fail for a range of reasons associated with the meshing of the warning message with those at risk; as well as institutional factors such as cooperation between the organisations involved, and how they conceptualise the warning task. The task may be conceptualised narrowly leaving out important elements of the risk. These factors are examined in the context of recent European and Australian research and experience of warnings.
Despair is understandable; while resources devoted to warning systems are fairly static, the task is probably becoming more challenging by the day because of social evolution - and in turn this raises the issue of the validity of much earlier research. Although success with warnings may becoming more difficult to achieve, there are potential changes in the operating environment which may force higher performance. Other conclusions include: the importance of agreeing on a definition of "success", as a precursor to warning-system wide agreement on the task; a more negotiated approach to those at risk (rather than a monopolistic supplier approach); and targeting to ensure that no identifiable group is missed.
Are Flood Warnings Futile?
Risk communication in emergencies
Much of the risk communication literature can be categorised into the three themes set out in Table 1. The first and second of these themes are most familiar to those working in the field of natural hazards. They are about how to persuade people - formally or informally - to adopt self-protective behaviour. In the case of a predicted flood for example, the central concerns are to alert people that there is an immediate threat of water about to enter their homes, and to persuade them to take protective action. Similarly, longer term awareness-raising is about promoting self-protective behaviour, whether it concerns being aware of appropriate response to flood warnings, or the longer term aim of hazard mitigation and avoidance. The third area set out in Table 1 can be characterised euphemistically as building consensus over controversial issues. In many cases the last category can be better thought of as being more about persuading people to tolerate something that they would rather do without, or to legitimate unpopular activity. It can also cover the process of introducing controversial flood mitigation strategies such as restrictive land-use regulations for flood prone areas.
Table 1: Why undertake risk communication?
A flood warning turns a prediction or forecast into information in the form of an action statement. The purpose is to improve safety and reduce damages. They do this by communicating information to those at risk to take action to improve their safety and reduce damages: to enable "individuals and communities to respond appropriately to a threat in order to reduce the risk of death, injury, property loss and damage." (BoM and AEMI, 1993). Forecasting or prediction is valuable only in so far as it serves these purposes. There is an important difference between warning messages and individual capacity to understand and act on those messages, as discussed below and by Rohrmann (this volume).
In an attempt to spread good practice, a set of guidelines for flood warning system development and implementation has been developed in Australia. These are based around the integrating concept of a "total flood warning system" - in part to encourage a system wide view of the task. This approach is becoming widely accepted (Handmer et al, 1998). The Australian Guide has the following definition (EMA (Emergency Management Australia) (1995, Second Edition 1999):
"A total flood warning system integrates flood prediction, the assessment of likely flood effects, the dissemination of warning information, the response of agencies and the public in the threatened community, and review and improvement. These components must operate together for sound flood warning performance to be achieved." (p 5).
Achieving this requires integration, cooperation, shared responsibility, and thinking broadly about the problems - as well as involvement of the communities at risk. Self criticism and reflection are also required.
Despite this knowledge warning performance is frequently far from satisfactory. This paper examines why flood warnings fail, or appear to fail, and attempts to identify what can be done to improve the situation. The paper does this by setting out very briefly recent high profile warning failures in Australia and Europe, and then examining reasons for failure or perceived failure under three general categories. The material is illustrated with examples from Australia and Europe as appropriate.
2.1 The importance of warnings
Flood and storm warnings are issued at the end of an often complex chain of organisations or groups whose primary function is to deliver predictions to those at risk of flooding or whatever hazard is under consideration. The primary if not sole purpose of these organisational chains is to serve the needs of those at risk. Considerable effort and investment goes into the development and maintenance of warning systems, in particular for environmental monitoring and for the expertise and hardware required for computer modelling. The whole field has been supported by considerable research and many agencies have long struggled to improve their performance. Even though warning performance may be improving overall, the last decade is punctuated by high profile failures in Australia and Europe - as set out below.
Warnings may have shown their inadequacy, but it is almost certain that increased reliance is being placed on them. One fundamental reason for this is that warnings provide an approach to risk management that allows the risk to be taken. Development may occur in flood prone areas or in areas subject to other periodic and predictable hazards because warnings will (or rather should) trigger appropriate safety and damage reducing behaviour. This approach is particularly attractive in an era of apparent deregulation, but may mean that warnings could be serving administrative rather than practical needs. In addition, whatever the uncertainty over climate change and variability it seems likely that extreme events are and will continue to increase.
2.2 The impact of recent events
A signal event in Australian warning experience occurred in April 1990, when the three mainland states of eastern Australia experienced very severe flooding. Two country towns (Charleville, with 3,200 people, and Nyngan, with 2,500 people, in the states of Queensland and New South Wales respectively) had to be completely evacuated and there were substantial evacuations from small urban centres in the Gippsland area of the state of Victoria as well. Nyngan in particular was a major media and political event. Virtually the whole community was involved in placing over 200,000 sandbags to heighten the existing levee which created a dry "island" on the vast flooded western plains. Eventually the augmented levee was overpowered, the town was flooded and the population evacuated by helicopter.
The extensive and damaging flooding in the three states put warnings and emergency management under scrutiny. A national conference convened after the flooding concluded that warning practices in many areas were inadequate. Flood predictions had been the subject of strong criticism, and in places there was evidence that people at risk from flooding had either not been warned or had not understood the warnings provided. A review by the Bureau of Meteorology showed that flood prediction requirements had been fully defined at only 15 percent of the points for which the Bureau could provide predictions, indicating a lack of effort by emergency management and local agencies in planning their responses to flood forecasts. This experience stimulated the development of a set of national warning guidelines (EMA, 1999; Handmer et al, 1999).
Europe also experienced severe flooding during the 1990s, resulting in much damage, disruption and loss of life. In particular, the floods of 1993 and 1995 seemed to engulf much of the north and centre of the continent. These were preceded by devastating flash floods in southern France and northern Italy, and less serious but very damaging floods in Scotland and England. They have since been followed by further serious flooding in France, Spain and Italy, by the 1997 floods along the Oder affecting Poland, the Czech Republic and eastern Germany, by the inundation of parts of England during Easter 1998, and by flash floods which seriously affected southern France during November 1999.
In most cases warnings were an issue. In the 1998 British floods for example, the speed of onset of the flood caught people unawares and many flood victims were convinced that this was not a "natural flood", its rapid rise and fall being due to some negligent human action. This compounded their sense of frustration and anger about not being forewarned. There was public, media and political outcry. In most regions the primary reason for this outcry was the lack of warning and perceived inadequacies in emergency response. A Parliamentarian from one of the flooded areas (Evan Harris, MP for Oxford West and Adington), demanded an inquiry into why the responsible authority - the Environment Agency or EA - did not give people living in the path of the flood more warning. As with most of the European floods, an enquiry followed and made numerous recommendations concerned primarily with warnings (Bye and Horner, 1998).
Despite the picture of warning system failure - or at best under-performance - there was at least one outstanding success. The Netherlands learnt and built on the experience of the 1993 flooding, so that they were fairly well prepared in 1995. When the rising flood waters in 1995 meant that the security of river dikes could no longer be guaranteed, about 250,000 people were evacuated as a precautionary measure. This went very smoothly. The dikes held, but the evacuees were overwhelmingly satisfied with the decision to evacuate and with the conduct of the evacuation (Rosenthal et al 1998). There are many factors in this success, which are explored in Bezuyen et al (1997) and Rosenthal et al (1998), but one important hydrological factor was the relatively long warning time which gave the authorities time to think, liaise and develop their response. This was not a luxury available in the UK flooding for example.
There are many reasons why warnings often do not, or do not appear to, achieve their aim. Reasons can be loosely classified as methodological or substantive as shown in Table 2.
Before examining each of the categories set out in Table 2, a brief comment on the relevant research literature is in order. In assessing warnings the focus has not unnaturally been on the "official" warning system and those groups that have some formal responsibility for making the system work.
Even though the mass media is a key element in most warning systems, it is very rarely legally obligated or tied into a warning system and little research attention has been devoted to it. Although the relevant research literature is large the bulk of work is on the details of message design and factors important in individual response. This is not to say that other areas do not receive attention, only that research is skewed towards the technocratic side and highly skewed towards official networks - rather than the informal communication channels used everyday by everyone (Parker and Handmer, 1998). Examples of literature reviews include: Drabek (1986), Handmer and Ord (1987), Sorensen and Mileti (1989), Burkhart (1991), Janis and Mann (1997), Gruntfest (1997) and Mileti (1999). A comprehensive catalogue of published material can be found at http://www.colorado.edu/hazards.
Table 2: Reasons why warnings fail or appear to fail.
Key methodological issues include the definitions or criteria used for success or failure and the design of evaluative research. Attitudes and expectations held by different groups within the warning system may also be important. These issues have been canvassed by Penning-Rowsell and Handmer (1986). Comprehensive discussions of assessment are also found in Rohrmann (1999), and Mulilis (1998). Here, comments are made on criteria for success, as well as on the overlooked question of unrealistically high expectations for warning system performance. Both areas are important in viewing warnings as characterised by failure.
4.1 Pressure to perform and high expectations
Unlike much risk communication, flood warnings are often tested in operation against imaginary performance standards. The "tests" are in public. There is an identifiable constituency, often with outside support, who will complain loudly about any inadequacies. Those responsible for issuing warnings are usually clearly identifiable - although it is only the official agencies that are generally targeted. The media are usually too busy pointing to the failings of the government agencies to examine their own performance in passing on warning messages. As there is, or at least as there appears to be, a direct and unambiguous link between some agency or group with official responsibility and warning performance, they can be blamed.
Also, it is generally assumed by the media that the performance of warning systems can be readily and rapidly assessed. This does not appear to be the case in most areas of risk communication. How often is there outrage over a community or national awareness raising program - even though they have rarely achieved their objectives? What risk communication campaigns have minutes or hours to get their message across and are then subject to intense public evaluation? The result is that warning systems are put under more pressure to perform than many other comparable forms of risk communication. The formal high profile inquiries following flooding in the UK and Europe illustrate the extent and seriousness of this pressure (Bye and Horner, 1998; LAWA, 1995; Rosenthal and t'Hart, 1998), as does the activity in Australia during the 1990s.
Related to this is the high expectation in terms of warning reach. Although some agencies have targets, like the British Environment Agency (EA) with its target of reaching 80 percent of the population at risk, there is a popular expectation that none will be missed. Increasingly, people expect very high performance from public agencies - perhaps higher than can be achieved - and missing 20 or even 10 percent of the people flooded could result in much embarrassment, especially if that group had media or political support. From an agency perspective, it would be particularly unfortunate if those missed constituted an identifiable group such as non-english speakers or those living in an area for which no warning provision had been made.
The experience of the EA in the Easter 1998 floods stands in contrast to the official performance goal. Very few people received a warning (Bye and Horner, 1998), identifiable groups were not warned (Tapsell et al, 1999), and areas were excluded from the warning system. Areas excluded were protected by levees and therefore a low priority for warnings - however research orthodoxy has long been that such areas are at risk and need emergency planning including the provision of risk information (Ericksen, 1986; Tobin, 1995). Australian experience suggests that often a relatively small percentage of those at risk will receive or recall receiving official warnings (eg Handmer, 1988) - although many were warned through informal channels (Parker and Handmer, 1997). It is also clear that warnings which reach all those at risk may, according to one's point of view: be ignored, or mislead through optimistic interpretation or failure to fully explain the inherent uncertainty in any warning - as appears to have occurred at Nyngan.
4.2 Criteria for assessment
Any assessment needs to start with a clear statement of what constitutes success. Much past assessment by the responsible agencies has concentrated on the accuracy of the predictions. Usually, some indicator of timeliness is included as well. With reference to some point, a satisfactory warning may be defined as one that gives at least x hours notice of a flood within specified parameters such as plus or minus y cm. The actual wording of a public warning will usually be more precise than this. It might say: "The Muddy River is expected to reach 10 metres on the Rowing Club gauge by 6pm today," - thereby ignoring the considerable uncertainty surrounding this prediction. The details of acceptable accuracy and timeliness are generally kept by the responsible authority for its own assessments. This is useful to those developing prediction accuracy, but may be of limited value to those interested in the impact of the warning message. Even its purported accuracy may be misleading, as the predicted height is usually a point on a river as in the example above. To give this meaning it must be interpreted to relate to the water spread away from the river. Crucial advice on how flood water will actually affect people, and on appropriate action, is often minimalist or missing.
Many researchers - and some key officials within Australian and British warning systems - have argued that success should be seen in terms of the impact of the warning on reducing flood damages, as that is the reason for having warnings. For example, see the Australian and European cases set out in Handmer (1997; for earlier examples drawn from Australia see Smith and Handmer, 1987). For some of the officials consulted see the "Acknowledgements" section of this paper. This approach has been a basis of much of the assessment work performed at the UK's Flood Hazard Research Centre (eg, see Parker and Neal, 1990), and is implicit in the concept of a "total flood warning system" as set out in Section 1. Although there is by no means universal agreement on this approach within flood warning organisations, the approach is increasingly employed in warning assessments. It may be a very severe test. Taking action to reduce damages is a decision made by those at risk on receipt of a warning. A strong case can be made that the agency has responsibility to ensure that its message has meaning to the intended audience, but it is more difficult to assert that warning agencies should be responsible for people's decision making. It highlights the conflicts in risk management between individual rights and the obligations to protect individuals from danger. Nevertheless, taking this view helps ensure that warning agencies strive to understand the needs of their audiences, and very importantly helps them justify their budgets by showing the economic benefits of warnings.
A confounding factor is that warnings may be connected with household finances. It is difficult to admit to ignoring a warning and then to expect aid (in Australia) or insurance payments (in Britain). There is anecdotal evidence (from discussions with flood victims) that this factor has occasionally been significant in Australian and British floods.
5.1 Shared meaning is needed
If people at risk are to take action, then warning messages must mean something to them - but it appears that this is neither obvious nor easy to achieve. Post flood reviews, in for example Norway (Killingtveit, 1997) and Britain (Bye and Horner, 1998), make the point (see also Rohrmann, this volume). To have any chance of "success" warnings need to have meaning which is shared between those who draw them up and those for whom they are meant to inform. They must also appear relevant to the individual decision-maker. This is no easy task given the distinctions between scientific-technological organisations and the "public" highlighted in particular by the emerging discipline of "sociology of knowledge" (Jasanoff et al, 1995). Words and media are important, but so too are the diverse range of priorities and attitudes among those involved, the role of personal networks, and the characteristics of our dynamic fragmented societies.
A starting question for those designing and delivering warning messages is "what do those at risk need to know to reduce damages and improve their safety; and what is the best way of ensuring access to that information?" A detailed understanding of the community at risk and their warning needs as a prerequisite for implementation has long been a theme of the general risk communication literature (see eg Handmer and Penning-Rowsell, 1990; Vaughan, 1995). Unfortunately, flood prone populations are usually very diverse making such understanding and subsequent message and delivery tailoring problematic. This also highlights the difficulty of achieving 100 percent coverage. It is not like advertising or health promotion where a decision may be taken to concentrate on only certain identifiable segments of the population, leaving others to be attended to later via a different communication strategy.
It is difficult - if not impossible - to answer questions about local needs, priorities and access properly without consulting the people involved. The consultation should be a two-way process, more akin to negotiation, with the various stakeholders discussing their perspectives on the flood risk, and approaches to managing it. In many areas local people may be unaware of the risk - a perception that will influence their response to warnings. The development of shared meaning through a negotiative process is consistent with the evolution of practice in risk communication generally towards the development of partnerships (Fischhoff, 1995). Fully shared meaning cannot be achieved without a thorough understanding of the population at risk.
It is clear that such consultation has not taken place often enough. Warning messages are frequently written in jargon: in language that may mean a lot to those using the words but little to those for whom the message is intended. Some recent examples (of which there are many) include: the heavy use of jargon (eg in Belgium, van Hassel and van Lindt, 1998:75); and the Environment Agency used a colour coded warning system which the post flood independent inquiry found was generally not understood:
"Colour coded warnings appear to be misunderstood by nearly all who receive them... The interests of the public are not well served by warnings given on the colour coded basis." (Byre and Horner, 1998:57).
Flood warnings in Britain also used Greenwich Mean Time, however Britain was using summer time which resulted in needless confusion; and in the border regions of Belgium and the Netherlands, warnings were given according to different gauge zero levels, with the result that predicted flood levels differed substantially across the border (van Hassel and van Lindt, 1998:75).
Positive examples come from some areas of Australia where local emergency service groups identify key warning requirements which the warning agency tries to accommodate. Also, the Dutch experience during the 1990s was generally positive (Rosenthal et al, 1998; see Section 2.2 above).
5.2 Shared meaning is not enough
Shared meaning is a prerequisite for success, but it may be increasingly elusive as society becomes more atomistic. Worse, even if achieved there are factors which may nevertheless undermine the warning process. The main points are set out in Table 3.
Table 3: Reasons for failure whether shared meaning exists or not.
Shared meaning may exist but is of limited value:
Shared meaning difficult to achieve:
There is also the opposite issue of hypersensitivity to rain after flooding. This was seen following the 1986 floods in southwest Sydney (Handmer, 1988). The scale of the problem is unknown.
A question rarely asked in this context is whether modern communities are less interested in warnings or put another way: do the highly individualistic structure and priorities of modern society make it more difficult to achieve warning success - while increasing the likelihood of criticism and demands for compensation? I do not have an answer for this question, but feel that it is extremely unlikely that the rapid evolution of western society is making the warning task easier. It is more likely that the atomisation and hyper-mobility of society is making it increasingly difficult to design and deliver effective warning messages to all those occupying or using a flood prone area. The enormous improvements in information technology over the last decade or so have yet to have any significant impact on communicating warnings to those at risk.
6.1 Inter-organisation cooperation
As stated at the outset, warning systems are composed of many groups, some official and some not, each with its own agenda and priorities. To function effectively, the groups must have some shared meaning and agreement on the warning task - or a lot of luck. The evidence from failures suggests that there is often neither, and that warnings agencies may see that their task is to provide a prediction or to be involved in some specific way in the warning process, but not to see themselves as directly serving the aim of warning those at risk. As a result, many warning systems have no formal way of reaching those at risk. In some cases - including one inland Australian case in the late 1980s - local warning plans contained no procedures for actually issuing a warning to those most at risk (Handmer, 1993). Exclusion from warnings of some leveed areas in the UK has been commented on above.
There is often reliance for dissemination on the mass media, but there are rarely formal arrangements with media to ensure that warnings are dealt with in a timely manner. But, in some countries, for example Belgium, agreements with the media exist to cover this. In most jurisdictions faith must rest on discussion and informal agreement. In the very recent past in Britain, reliance for warning dissemination was placed on the Police based on an agreement now decades old and often ignored. The British Police are increasingly looking after their core business and for most forces this does not include flood warnings.
Ways are needed to improve this situation: that is the failure to conceptualise the warning task properly. There is a failure in practice to focus on what warnings are really for. Improvements may be sought through internal processes or through changes to the external operating environment. Some options are explored in the following paragraphs.
6.2 Potential pressures for change
Processes are needed to develop consensus over the aim of warning systems and to clarify the roles and responsibilities of all those involved - the flood warning consultative committees established in each Australian state/territory have gone a long way towards this, as has the preparation of national guidelines through a consultative process (see Handmer et al, 1999). In Europe, some groups have been established at the international level (where some of the more obvious failures of communication occurred), and at the interagency level, but the picture is very patchy. The British EA for example, has opted for an overarching "National Flood Warning Centre" which would develop the needed cooperation (Handmer et al, 2000).
Major changes are also occurring within the organisations operating warning systems. Managerialism or at least aspects of corporate practice have of course affected the organisations involved. But much larger changes are at least possible even if not probable. Part of the warning process - or almost the entire process - could be contracted out, in effect privatised (Handmer, 2000). One of the supposed advantages of this is that the tasks are specified very precisely and that performance measures related to these tasks are employed. In theory, such an approach would focus the minds of warning providers. The assumption here is that the specifications would stipulate that warnings were to enable "individuals and communities to respond appropriately..." as set out in Section 1 of this paper.
Another major change, independent of the role of government, would be development of appropriate warnings and associated emergency protection as a "human right" - as part of the gradual evolution of the concept of "safety" towards a "right". Health and safety issues, very broadly conceived, have long been part of the European legal scene as part of the Commission's efforts to reduce social exclusion, and to promote industrial safety. The "Seveso Directive" of the 1970s marked the start of these efforts in a pan-European context - see Schuetz in this volume. The European Union has steadily extended the scope of these requirements (for example through the Social Chapter of the Maastricht Treaty the early 1990s), and court decisions at national and European levels have given many aspects of safety the status of a legally enforceable right. A "right" removes many of the usual avenues for legal argument. A right to warnings could mean full disclosure of risk related information where agencies or enterprises have knowledge or where they should have knowledge of potential hazards. Those occupying flood prone areas would have to be fully informed of the risks.
Staff in the relevant agencies in Britain and Australia are concerned about these possibilities. Privatisation seems most unlikely. A "rights" approach would be antithetical to British traditions of governance more comfortable with discretionary approaches - although this is changing under the influence of the European Union. Agencies in common law jurisdictions like Australia and Britain are concerned about their potential legal liability under such an approach. However, in practice most Australian warning operators would be protected by their emergency services legislation. British authorities have no legislative protection - and it is clear that they are inhibited by the legal uncertainty (Handmer and Crook, 1999).
Flood warnings provide a margin of safety that is readily consumed by or traded for behaviour which benefits someone economically; for example a warning system connected to evacuation plans might allow the apparently safe development of a flood prone area. Warnings are likely to become an increasingly important tool for risk management as reliance on them allows more otherwise risky behaviour. However, this strategy assumes reliability. Unfortunately, recent experience in Australia and Europe shows that despite increased effort flood warnings are too often seen as failures.
Reasons for poor performance are found in the areas of flood detection failure, and inadequate message design and communication. But there are other important reasons for apparent failure inherent in the high profile testing in short-duration flood warning operations, often in situations where lives are at risk. Unlike most forms of risk communication a failure may result in immediate deaths. These situations invite critical scrutiny, blame allocation and expectations of high reliability performance. Often the definition of success employed is problematic and there are other methodological issues. In terms of actual performance, a lack of shared agreement about the task among all those involved is a common failing.
While resources devoted to warning systems are fairly static, the task is probably becoming more difficult by the day because of social evolution including: continually rising expectations, the growth of individualism and high levels of mobility. Changes that could potentially support improved warnings such as modern information technology are yet to have an impact in terms of communicating information to those at risk, and improvements in the general level of language skills are offset in many communities by the diversity of languages and significant levels of functional illiteracy. In turn the issue of an evolving society raises another question: is the good research done decades ago - and in another national context, primarily the US, still valid? or do the massive social changes of the last two decades render it at least partly invalid?
The important priorities in terms of policy and practice are:
Warnings are unlikely ever to achieve 100 percent coverage and response - but they are far from futile and much of the alleged poor performance is a feature of the time pressure and life and death nature of the task. Failures, enquiries and political interest provide opportunities for major improvements. The Australian Flood Warning Consultative Committees and the British National Flood Warning Centre, are examples of positive institutional changes which provide obvious vehicles for promoting continual improvement.
My thanks to Bernd Rohrmann, editor of this collection, and two anonymous referees for their helpful suggestions. I would also like to express my appreciation to the many officials in Australia and the UK who have shared their flood related experiences and insights with me, in particular Chas Keys of the NSW State Emergency Service, Jim Elliott of the Australian Bureau of Meteorology, and Chris Haggett of the UK Environment Agency. I have generally avoided naming these and other individual officials in the text of the paper because of organisational sensitivities. The UK work was part of the EC funded SIRCH project (Societal and Institutional Responses to Climate Change and Climatic Hazards: managing changing flood and drought risk, #ENV4-CT97-0447). The work of Emily Crook is acknowledged with gratitude.
Bezuyen M., van Duin M. and Leenders P. (1997) Flood management in the Netherlands. In Handmer, J W (1997) Flood warning issues and practice in total system design. London: Flood Hazard Research Centre, Middlesex University: 71-88.
BoM and AEMI (Bureau of Meteorology and Australian Emergency Management Institute or AEMI) (1993) Guidelines for effective warning. AEMI, Mt Macedon.
Burkhart, F.N. (1991) Media, emergency warnings and citizen response. Boulder: Westview Press
Bye P. and Horner M. (1998) Easter 1998 floods: Report by the independent review team to the board of the Environment Agency. Volume 1. Bristol: Environment Agency. 30 September.
Drabek, T.E. (1986) Human systems responses to disaster: an inventory of sociological findings. Springer Verlag, New York.
EMA (Emergency Management Australia (1999) Flood warning: an Australian guide. Canberra: Emergency Management Australia. Authored by Elliott J, Handmer J, Keys C, and Salter J.
Ericksen, N.J. (1986) Creating flood disasters? New Zealand's need for a new approach to urban flood hazard, Water and Soil Directorate, Ministry of Works and Development, Wellington.
Fischhoff, B. (1995) Risk perception and communication unplugged: twenty years of progress. Risk Analysis 15(2): 137-145.
Gruntfest, E. (1997) Twenty years later: what we have learned since the Big Thompson flood. Boulder: University of Colorado.
Handmer, J.W., (1988) The performance of the Sydney flood warning system, August 1986. Disasters 12 (1): 37-49.
Handmer, J.W., (1993) Cross-cultural emergency planning: the Alice Springs flood. In Y. Aysan and I. Davis (eds) Disasters and the Small Dwelling. London: James and James.
Handmer, J.W. (1997) Flood warning issues and practice in total system design. London: Flood Hazard Research Centre, Middlesex University.
Handmer, J.W. (2000) Are emergency services becoming private? Australian Journal of Emergency Management. (In press).
Handmer, J.W. and Ord, K.D. (1987) Flood warning and response. In Flood warning in Australia: policies, institutions and technology. (D.I. Smith, and J.W. Handmer, eds.), pp. 235-257. Canberra: CRES, Australian National University.
Handmer, J. and Penning-Rowsell, E.C. (1990) Hazards and the communication of risk. Aldershot: Gower.
Handmer J, Keys C, and Elliott J (1999) Achieving lasting change in multi-organisational tasks: the case of flood warnings in Australia. Applied Geography. 19: 179-197.
Handmer, J.W. and Crook, E. (1999) The future of flood warnings in Britain. Flood Hazard Research Centre, Middlesex University.
Handmer, J.W., Keys, K. and Elliot, J. (1999) Achieving lasting change in a multi-organisational task: flood warnings in Australia. Applied Geography. 19: 179-197
Handmer, J.W., Crook, E. and Haggett, C. (2000) Institutional shock and adaptation: the British floods of Easter 1998. Flooding: risks and reactions, conference. Terrace Dalton Publishers. London.
Janis, I.L. and Mann, L. (1997) Emergency decision making: a theoretical analysis of response to disaster warnings. Journal of Human Stress. 3:35-48.
Jasanoff, S., Markle, G. Petersen, J. and T. Pinch (eds), (1995) Handbook of Science and Technology Studies. London and New Dehli, Sage.
Killingtveit, A. (1997) The large spring flood in Norway in 1995. Presented at First RIBAMOD workshop: River basin modelling, management and flood mitigation, the challenge for European cooperation. Delft Hydraulics, 13-15 February.
LAWA (Landerarbeitsgemeinschaft Wasser) (1995) Guidelines for forward looking flood protection. Floods - causes and consequences. Stuttgart: Lander Working Party on Water.
Mileti, D.S. (1999) Disasters by design: A reassessment of natural hazards in the United States. Washington DC: Joseph Henry Press.
Mulilis, J-P. (1998) Persuasive communication issues in disaster management: a review of the hazards mitigation and preparedness literature and a look towards the future. Australian Journal of Emergency Management. Autumn 1998: 51-55.
Parker, D. and Handmer, J. (1997) The role of unofficial flood warning systems. J of Contingencies and Crisis Management. 6(1):45-60.
Parker, D J and Neal, J (1990) Evaluating the performance of flood warning systems, In Handmer, J.W. and Penning-Rowsell, E.C. (Eds), Hazards and the Communication of Risk. Aldershot: Gower Technical Press: 137-156.
Penning-Rowsell, E.C. and Handmer, J.W. (1986) Evaluating flood warning effectiveness: the conceptual, methodological and practical research problems. (CRES Working Paper 1986/6). Canberra: CRES, Australian National University.
Rohrmann, B. (1999) Assessing hazard information/communication programs. Australian Psychologist. 33(2):105-112.
Rosenthal, U. and t'Hart, P. (1998) Flood response and crisis management in Western Europe: a comparative analysis. Springer Verlag.
Rosenthal U, van Duin M, Bezuyen M, de Vreeze-Verhoef M-L, and t'Hart P (1998) Flood management in the Netherlands. In Rosenthal U and t'Hart P, Flood response and crisis management in Western Europe: a comparative analysis. Springer Verlag: 103-150.
Smith, D.I. and Handmer, J.W., (Eds), (1987) Flood warning in Australia. Canberra: CRES, ANU.
Sorensen, J. and Mileti, D. (1989) Warning and evacuation: answering some basic questions. Industrial Crisis Quarterly. 2(3-4): 195-210.
Tapsell, S., Tunstall, S., Penning-Rowsell, E.C. and Handmer, J.W. (1999) The health effects of the 1998 Easter flooding in Banbury and Kidlington. Prepared for the Thames Region of the Environment Agency. Flood Hazard Research Centre, Middlesex University.
Tobin, G.A. (1995) The levee love affair: a stormy relationship? Water Resources Bulletin 31(3):359-367
Hassel, H. and van Lindt, L. (1998) Flood management in Belgium. In: Rosenthal, U. and t'Hart, P. (eds) Flood response and crisis management in Western Europe: a comparative analysis. Springer Verlag: 57-101.
Vaughan, E. (1995) The significance of socioeconomic and ethnic diversity for the risk communication process. Risk Analysis. 15(2): 169-180.
Massey University, New Zealand
Last changed October