Report for Middlesex University Flood Hazard Research Centre

29th April 2003.

Flood incidents outside the indicative flood plain maps. - An insurance perspective.

By Professor David Crichton

Fellow of the Chartered Insurance Institute, Chartered Insurance Practitioner

Visiting Professor, Middlesex University Flood Hazard Research Centre

Contents Page

Executive Summary 3

Introduction and objectives 3

Methodology and conclusions 3

1. The Indicative Flood Plain Maps 4

2. Sources of discrepancies 4

3. Size of discrepancies 5

4. Social and human issues 6

5. Future outlook 6

6. Insurance implications 7

7. Differences in Scotland 7

8. Further research 8

Conclusions 8

Recommendations 8

Annex 1 Questionnaire used for this project 9

Annex 2. List of organisations contacted.10

Annex 3. Discrepancies from non river floods.11

Annex 4. Discrepancies from river and watercourse floods14

Annex 5. Discrepancies from other sources.15

Acknowledgements

The author wishes to gratefully acknowledge the assistance of the organisations listed in Annex 2. In particular those organisations which supplied data, and Gill Holland of the National Flood Forum, who gave a valuable insight into some of the problems of flood victims.

Executive Summary

It is well understood that the “Indicative Flood Maps” published by the Environment Agency and the Scottish Environment Protection Agency are not to be considered as more than an indication of areas at risk from non tidal main river floods. Indeed, “second generation” flood maps using better data and modelling tools are already in the pipeline in England, Wales, and Scotland.

In the meantime, it has been suggested that it would be useful to seek information from the insurance industry about the number of flood insurance claims arising outside the indicative flood map area during the autumn 2000 flood event. The author extended this to include the April 2000 flood event in Edinburgh.

Larger insurers supplement these indicative flood maps with other sources of data for underwriting or claim validation purposes. This means that data about whether a claim came from within or outside the indicative map area is not readily available from the industry, even if insurers were prepared to release it.

The author was asked to use his contacts within the insurance industry to elicit information and combined this with his own experiences in the industry. Due to the short timescale much of the information is anecdotal, but the message from various sources is consistent. The overall conclusion is that there are many factors at work and it is not possible to give a categorical statement about the size of the discrepancy. However it would not be unreasonable to conclude that between 30 and 40% on average of inland flood insurance claims are derived from outside the indicative flood map area.

Introduction and objectives

Around 11,000 properties were flooded in England and Wales in the autumn 2000 floods, and it has been estimated by Middlesex University that 15% of these properties were not in postcode districts that had any of their area in the indicative flood plain maps. There is an impression within the insurance industry that at the more detailed unit postcode level the percentage may be much higher. The objective of this exercise was to explore this aspect.

Methodology and conclusions

The author approached a number of relevant contacts with a questionnaire (see Annex 1 and 2). It was not possible in the timescale to obtain sufficient hard data to form conclusions but sufficient anecdotal information and impressions were provided in the responses to form a consensus for the following main conclusions.

  1. Up to 30 or 40% of flood insurance claims overall could have arisen from outside the indicative flood map area.
  2. The proportion of claims outside the indicative flood plain maps varies depending on local factors, season, and circumstances. An overall figure can therefore be misleading.
  3. Many floods arise from blockage or backup into small watercourses or culverts that have not been modelled for the maps.
  4. A significant number of properties are flooded from non river sources, or from tidal stretches of rivers, which the modelled data do not include.
  5. Insurance claims from outside the indicative flood map area tend to be less costly.

1.The “Indicative Flood Plain” Maps

Defra claim that the current maps utilise the best available data, and are only intended to show floodplain areas, not flood risk or definitive flood boundaries. Defra has stated that they support the principle of improvement, but caution that unrealistic accuracy should not be expected of such maps.

Defra also state[1] that

“The Indicative Flood Plain Maps are being improved as better data becomes available from Agency modelling, modelling carried out by developers, from flood events, and from information supplied by the public (and verified by the Agency). The development of the extreme flood outline, as required by PPG25, continues though this has been held back so as to use a more accurate digital elevation model that the Agency will acquire in May 2003.”

(The latter comment presumably refers to the “NEXTmap” product, produced by Intermap, a US company, with funding from the Norwich Union. There have been some problems with this product, and a delay of at least six months is now expected.)

Defra also predict that new extreme flood outline maps will be available in mid 2003.

In Scotland, the Scottish Environment Protection Agency (SEPA) has similar indicative flood plain maps from the same source. The Scottish Executive signed a licence with Intermap in late April 2003 to obtain airborne radar survey data to produce a second generation set of flood maps[2]. These are scheduled to be put into the public domain by October 2004.

2.Sources of discrepancies

This report identifies 20 reasons for discrepancies in the autumn 2000 floods (see annexes 3,4, and 5.)

The main ones are listed below. It should be noted that in England and Wales, the Environment Agency modify the modelled maps by including data from their flood records (known as “Section 24” data). This makes the maps more accurate, but only where historic records are available. Such records tend to be sparse for areas that were unpopulated during a historic event, and therefore may not be available for more modern housing developments.

  1. Non river sources. There are many non river sources of flood, such as sewage back up, surface water run off, rising groundwater etc (for a detailed list, see Annex 3). This could account for 15 to 20% of discrepancies.

2.Tidal Floods. The model on which the maps are based does not include tidal sections of main rivers (apart from the Thames). Backup into main rivers at high tide could account for more than 15% of the properties damaged in autumn 2000.

  1. Small Watercourses are not included in the model, yet insurers say these are one of the main sources of flood claims, although they tend to be smaller claims than those from properties within the flood map areas.
  2. Blockages. Blockages of bridges, culverts, etc., can result in flooding in unexpected places.

5.False positives. Not everywhere in the indicative map area is actually at risk.

  1. Topography. The elevation data used to create the modelled flood maps are not very accurate and in relatively flat areas this can be an important source of errors.
  2. Probabilities. The maps only claim to show the 1 in 100 year flood, yet in autumn 2000 the flood severity in some places was estimated as 1 in 200 or 1 in 300.
  3. Dynamics. Land use changes, climate change, and catchment processes can all change the flood hazard. Also data and modelling techniques have become much more advanced.
  4. Fraud. Insurers have a perception that some people who were not flooded may have made claims. This is in line with the “bandwagon effect” which is well known in the insurance industry.
  5. Reluctance to claim. On the other hand, those living in the flood hazard area of the maps may be reluctant to claim in case it results in increased insurance premium or problems with availability of cover.
  6. Seasonality. Any conclusions about the flood of autumn 2000 need not necessarily apply to other floods, as the time of year and previous rainfall levels have an important impact on the flood extent.
  7. Postcodes. Insurers use postcodes for locating claims, but these are not always captured correctly and in any case are not entirely suitable for this purpose.

For more details of all of these factors, see Annexes 3, 4, and 5.

3. Estimates of discrepancies

A. Edinburgh floods

The Edinburgh City Council hydrologist has carried out a very detailed analysis of the April 2000 floods in Edinburgh, and estimates that the number of properties flooded in areas not shown as flooding on the map, is very roughly, about 25%.

B. Association of British Insurers (ABI) research project

In 2002, the ABI commissioned research from HR Wallingford based on insurers’ computer records, but while the author has access to some of the results, he considers the findings are too inaccurate to use. He understands that they are unlikely to be published. (For example the ABI research states that only 600 out of 2,400 flood claims in Edinburgh in April 2000 were within the indicative flood map. This is at odds with the estimates of the Edinburgh City hydrologist and raises doubts about the results.) The ABI will use the information selectively to support the need for further action in appropriate areas.

TheABI indicated to Middlesex University at an industry seminar in the summer of 2002 that the percentage of flood claims that were not in the flood plain was much larger than had been expected. There is no statistically valid analysis of the causes of claims outside the flood plain included in the ABI research although the indications are that most claims do relate to flooding from smaller watercourses that feed into the river system. As the rivers flood, all watercourses (including culverts and small streams) leading into them, back-up and flooding then happens in areas not included in the indicative flood maps.

C. Author’s approaches to key individuals in the insurance industry

Due to the short timescale for this work, the results must be considered preliminary and dependent to a large extent on anecdotal evidence. Nevertheless there seems to be a view in the insurance industry that a main cause of claims outside the flood plain during the autumn 2000 event is backing up from smaller watercourses (including culverts and small streams). The picture emerging is something like the following:

  • Smaller watercourses, including backup: not estimated, but significant, perhaps dominant.
  • Non river related floods (as described in Annex 3) 15%
  • Tidal backup into rivers 15%

Overall the impression from insurers is that around 30 to 40% on average of flood insurance claims occur outside the indicative flood map areas.

All of these are very rough figures and it must be emphasised that the discrepancy will vary considerably depending on the location, time of year, and other circumstances.

  • Thus for location, it will depend, for example, on whether there are tidal rivers in the area.
  • For time of year, a big problem in October and November is the ploughing of fields for planting winter cereals, which can increase surface water run off, or the use of crops such as potatoes which leave the field bare after harvesting.
  • Other circumstances could include the increase in urbanisation since the maps were produced, and the extent to which historic flood records are available for the area.

An important point is that insurers have also found that those claims within the indicative flood map tend to be more expensive than those outside it.

This may mean that while the number of claims outside the map area may be as high as 40%, the percentage cost of claims outside the indicative map could be much lower.

4. Social and human issues

None of the above has taken into account the impacts on social and human issues which could apply well beyond the areas flooded, for example temporary relocation, business interruption, job losses, impacts on health etc. The “Learning to Live With Rivers” report[3] suggested that

“this human cost should be built in to future benefit/ cost assessments, so that the true worth of interventions is established” and stated that “preliminary research shows that the intangible costs of flooding are of the same order as the tangible costs.”

Defra has announced a new manual for assessing the impacts of flooding. "The Benefits of Flood and Coastal Defence – Techniques and Data for 2003". Defra also say that research on the intangible impacts of flooding is continuing and due to be completed in Spring 2003.

5. Future outlook

Largely due to the recent spate of flood events and the efforts of the Environment Agency, the public is becoming more aware of whether they live in flood hazard areas. The Environment Agency’s 4th annual public awareness campaign results show a significant increase (72% versus 47% in 2001) in awareness of those living in a flood risk area. A new £3M Multimedia Warning Dissemination Service Project has been started for England and Wales and should help to produce better warning dissemination from late 2003. In order for this to be properly targeted, better flood maps will be essential.

The National Flood Forum (NFF) started in January 2002 as an independent group representing flood risk communities. It is funded by the Environment Agency, and is providing an excellent service in England and Wales, providing closer contact at a grassroots level, and involving flood victims in decision making. Defra is also looking at the feasibility of providing a Flood Advisory Service and local advice on increasing the flood resistance of property.

The Environment Agency plans to develop community flood information boards and markers to raise flood awareness. A pilot flood information board has been installed in the main street in Lewes. Unfortunately there are some problems with this idea:

  • This information board only gives the broad code bands, it does not give any detail on the final level or the timescale.
  • It is a passive system; people have to walk into the middle of Lewes to look at it.
  • Condensation inside the plastic cover on the board means that no one can read it when it rains.

Perhaps a better solution would be sirens in areas at risk?

6. Insurance implications

The 1961 guarantee of availability of insurance expired at the end of 2002. Insurers are now free to refuse flood insurance cover where the hazard exceeds 1 in 75 per year, although most are prepared to maintain cover for existing customers even where the hazard is greater than this. Increasingly, owners of property, especially property within the indicative flood map area, may find it impossible to obtain insurance or may find insurance unaffordable. This could be a particular problem for new housing constructed in the floodplain, because if the buyer cannot obtain or afford the insurance, a mortgage is unlikely to be available. It is conceivable that speculative housing could remain unsold, resulting in blight.

7. Differences in Scotland

Scottish Councils have a duty to keep records, (not SEPA) so are the obvious people to consult, especially as they are obliged to publish updates to these records every two years. However this means consulting 32 different councils.

There are big issues over liability to be sorted out in Scotland where flood maps have not been designated as a Statutory Public Register as in England and Wales. This means that there is a reluctance to publish details of the indicative flood maps or any other flood maps.

The National Flood Forum does not operate in Scotland. However, unlike England and Wales, there is a network of “Flood Appraisal Groups” which provide expertise to most local councils and help them to deal with complaints from flood victims, planning issues, insurance problems and development of flood alleviation schemes. As a result of flood appraisal groups, new housing in flood hazard areas has virtually ceased in Scotland. Also, there has been a major programme of flood defence construction with more defences being completed in the four years since Devolution than in the previous 40 years. Since 1995, such defence designs have followed a policy of combining attenuation with defence walls, in order to keep the height of walls as low as possible, consistent with a minimum standard of service of 100 years plus climate change to 2050, plus freeboard.

8. Further research

The author has explored alternative sources of data and is satisfied that good quality insurance data do exist, but not in the public domain, and not so far available to the ABI or anyone else. The author has contacts with the holders of these data and given time and modest resources, may be able to tap into these more accurate sources of information on the location, cause, and cost of a large number of flood insurance claims.