Jane Toothill looks at the August 2002 flooding in central Europe.

The widespread river flooding that occurred in Central Europe during August 2002 has been hailed variously as the `event of the century' and the `flood of the millennium'. The accuracy of these terms is open to debate, but the disruption caused by the event, both on peoples' lives and the financial world, cannot be doubted. Whilst the lives of many householders in the affected area remain in chaos some four months later, however, there are a number of reasons why the insurance industry should be thankful.

First, simply put, it could have been worse. Compare the extent of flooding during this event (Figure 1) to the extent of the floods that occurred in Europe in 1345. This earlier event alone is enough to discount the `event of the millennium' tag given to the August 2002 event; the Elbe, Main, Donau, Moldau, Mosel, Oder, Rhine and Weser rivers all flooded, and the centres of cities as far apart as Cologne and Vienna were inundated.

Secondly, the industry was protected from the worst of the losses during the August event by the relatively low levels of flood insurance penetration in the countries affected: insured losses are likely to represent no more than 10%-20% of the total economic loss for the event.

The floods may, in fact, have served as a timely wake-up call for insurers and reinsurers. Governmental schemes such as CatNat in France, and the exclusion of flood from many householder insurance policies in many parts of Europe have allowed a certain amount of complacency towards flood hazard to develop during the last decade. There are surprises for the unwary, however. Many were caught off-guard by claims made on old East German insurance policies, and in other parts of Europe, flood remains covered by the private sector. This relatively easy situation for insurers may well change in the not too distant future as governments realise the potential for large economic loss from flooding in Europe. The August 2002 floods should serve as a reminder to the insurance world as to the scale of event possible in Europe. For this event, at this time, we escaped relatively lightly; given a similar event in the future or immediately in a different river basin, we may not.

Flood wisdom required

If companies are to take something positive from the August floods, an understanding of how such floods arise and appreciation for the various flood scenarios that may affect Europe in the future is required. Without this, any attempt to assess vulnerability and reduce exposure to flood will be in vain. The August 2002 event is only a single example, but is typical of the type of large flood event liable to affect Europe, in that a number of river basins and several countries were affected.

The floods were triggered by unusual, but not exceptional, meteorological conditions, which resulted in the passage of two rain-bearing depressions across Europe in close succession during the first half of August. The first of these events crossed the UK on 31 July, resulting in minor flooding, and by the end of the first week in August reached Germany and Austria, causing torrential rainfall. The system then moved eastwards, resulting in further heavy rain in Romania, the Czech Republic and the eastern coast of the Black Sea. Over 50 deaths resulted from flash flooding on the east coast of the Black Sea.

In Central Europe, this first event caused flash flooding in tributaries of the Elbe and Donau, notably on the Kamp and Krems. Water levels in many areas reached 10- to 20-, or more rarely, 100-year levels following the first event, but flooding was limited to smaller catchments. This event alone was not sufficient to cause widespread flooding in downstream areas. However, it raised both water levels in rivers and soil saturation levels in the catchments of the Elbe and the Donau. These levels did not have time to return to normal before the onset of the second storm.

The second rain-bearing depression, `Ilse', moved south-east across France from England and gave rain in northern Italy, before moving to the north-east and causing further heavy precipitation in Austria, the Czech Republic and southern parts of Germany on 10 and 11 August. Exceptional rainfall also occurred in Spain during this time. Over Central Europe, rainfall intensities were not significantly greater than during the first event, but rain fell on already saturated soils and was immediately converted to run-off. This fed into rivers already swollen by the first event, and many smaller catchments flooded immediately. As the rainfall continued, water levels in the larger rivers began to rise quickly, resulting in flooding on the Moldau, Mulde, Inn, Elbe and Donau. Rainfall from Ilse occurred over a widespread area, and large proportions of the catchments of the Elbe and Donau saw heavy precipitation. In addition, the front was slow moving, and rainfall continued until the 14 August. The cumulative rainfall for the two events was two to three times greater than normal for the time of year in many areas.

The area affected by the floods is illustrated in Figure 1. A more detailed picture is provided by satellite imagery of the flooded rivers, of which one example is given in Figure 2 for the Elbe upstream of Dessau in south-east Germany. In this part of the river, almost the entire floodplain was inundated, and the towns of Dessau and Wittemberg were badly affected. Pictures of residential properties in Dresden reflect the story along much of the flooded part of the Elbe and Donau.

The disruption caused by the flooding was widespread. In the Czech Republic, Germany and Austria, as many as 380,000 people were evacuated, and over 100 casualties were reported. Homeowners and small commercial properties were particularly badly hit by the flooding, and the social consequences of the event have been exacerbated by the low proportion of insurance coverage. Lifelines were also severely affected, with many areas suffering loss of electricity and water supplies. The transport network suffered damage to structures such as bridges, as well as direct stoppages caused by flooded roads and railways. The German railway authorities alone have reported losses of several hundred million euros.

Alleviate the damage

The chaos caused by the flooding has inevitably led to questions as to what can be done to alleviate flood damage in the future. Although it is easy to feel powerless in the face of natural disasters, there is in fact no such thing as a `natural' disaster. Risk is a function of hazard and vulnerability, and in nature phenomena such as floods or earthquakes are merely events. It is only when man gets in the way of these events that a catastrophe results. We may well be powerless to prevent the occurrence of a hazard event, but its consequences are a result of our own actions, and hence control of these consequences is an attainable goal.

For flooding, this control takes two primary forms: reducing the exposure in areas prone to flooding, and controlling the flow of flood water following heavy rainfall. While both these alleviation measures lie outside the direct control of insurance companies, an understanding and quantification of their effects is vital to a reliable assessment of potential flood losses likely to be faced in the future.

The obvious solution to reducing flood losses is to reduce the vulnerability to flood. In countries such as the UK, increased awareness as to the potential effects of flooding has led to legislation to limit development in floodplain areas. However, in heavily populated areas, it may not be plausible to entirely stop development in floodable areas. This method of controlling the effects of flooding also relies heavily on an adequate definition of what can and cannot be flooded - not always an easy task in light of changing weather patterns and levels of development in river basins. In addition, there is an existing vulnerability that cannot easily be reduced: houses cannot be moved or destroyed. These factors must be borne in mind by those underwriting flood risk.

Control of flood losses

The second method of controlling flood losses, flood management, must be considered for the river basin as a whole. This causes particular problems in Europe where the paths of many rivers cross several countries. The simplest approach to flood prevention in a local area may well be to build a river defence to protect that isolated location, but from a basin-wide perspective, this will in fact speed the passage of floodwaters to areas further downstream, where even greater problems may result. Flood management relies heavily on the retention of water, commonly through flooding of natural marsh or agricultural land. Development in floodplains not only directly increases vulnerability to flooding in that area, but also reduces the ability of the soil to absorb water and decreases the area of land available for water storage on the floodplain during flood events. A further means of flood management is the construction of water-retaining dams in the upper parts of a river basin. Such dams are used for flood management on many of Europe's largest rivers, including the Seine. Their ability to retain water is not guaranteed, however. In August, flood retention dams on the Vltava River (Czech Republic) were able to retain sufficient water to reduce the flood peak in Prague resulting from the first depression, but when the rainfall from Ilse arrived, the dams were already full and could not hold any more water. Even though emergency defences constructed at Prague immediately prior to the flooding were partially successful, Prague was perhaps the city worst affected by the floods.

In the absence of immediate and effective solutions to flood management and prevention, indirect methods of protection are sought, and cover of flood risk by the insurance market is a quick answer. For the August event, initial estimates for the economic loss ranged from ¤15bn to more than ¤20bn. The recent reduction in figures for the estimated economic produced by the German government from ¤15bn to ¤9.2bn is unlikely to be a comfort to the insurance sector, for which a total bill of around ¤3bn has been estimated. Of this, two-thirds is likely to reach reinsurers, with the majority split between Germany (about ¤1bn), the Czech Republic (around ¤700m) and Austria (about ¤200m).

Flood insurance in Europe varies in style and penetration from region to region, and the low insured proportion in the August floods has inevitably given rise to discussion as to whether changes to the insurance schemes in the countries affected area are required. Mandatory `all perils' coverage is one potential solution, and has the advantage of spreading the risk to a larger number of policyholders. One such scheme currently operates successfully in Norway via the Norwegian Natural Perils Pool, in which losses (not premiums) are pooled amongst participating companies in accordance with their market share. This scheme is seen to have advantages over the current system in Germany, where the government use tax income to reimburse losses following a catastrophic event. `Hand-outs' from the government in the wake of such events do not provide the insured with a definitive level of cover prior to an event, and do little to increase the responsibility taken for the consequences of flooding by the various parties whose decisions affect the overall level of risk. The devolvement of such responsibility is a means of lowering the overall level of risk faced by a country, and may be achieved by a variety of methods. One simple method is to increase the deductibles on individual insurance policies so that individuals cannot rely entirely on insurance or government hand-outs to cover their losses but must in part face the consequences of their own choices should flooding occur. A second approach is the financial involvement of authorities responsible for development in pooled cover so that their decisions must bear the risk resulting from their actions in mind.

What is the likelihood that similar losses will result from flooding in Europe in the future? A single weather depression alone is generally not sufficient to cause flooding in large river basins such as the Elbe or the Donau, and in August, it was only the occurrence of a second storm that caused large-scale flooding in downstream areas. However, as `pairs' of wind events such as Lothar and Martin have proven, the passage of two storms in quick succession across Europe is not unusual. Other factors may also increase the effect of a single storm on river levels, for example rainfall onto soils saturated by prolonged rainfall during the weeks prior to the event. Snow melt and frozen ground may also play a role in increasing run-off and were important factors leading to the flooding in Northern Europe in 1995.

More recent events

To put the August 2002 event in some sort of perspective, it is necessary to consider the other events that have occurred in Europe during recent times. Figure 4 lists some significant flood events that have occurred in Europe during recent years. Although the largest in terms of pure financial loss, events such as the 1910 flooding of the Seine, 1953 sea surge event in the UK and the Netherlands, and 1993/1995 river floods in Northern Europe are similar in scale, and could generate similar losses if they occurred today. This year's floods were not a `one-off', nor even `the event of the century'.

Will the frequency of large flood events increase in the future? The effects of global warming remains hotly debated in academic circles, and no two groups agree on exactly what will happen to global climate in even the next decade. Such studies largely concentrate on mean values of global temperature and precipitation, and the prediction of extremes is even more uncertain. However, evaluation of the climatic situation, and assessment of the various models available has led the Intergovernmental Panel on Climate Change (IPCC) to conclude, "river flood hazard will increase across much of Europe". The insurer of flood risks will do well to bear in mind that the situation will only be exacerbated by the effects of increasing risk value and development in river basins.

Faced with so many uncertainties, underwriters are in need of concrete information on which to base their decision-making. Precious little information is readily available regarding flood hazard, and in many areas, insurers are forced to rely on simplistic zonation models that do nothing more than define a best estimate of the n-year floodplain. Information relating to the scale of individual events and to the water depths they generate are not provided by such models, and information relating to the probability of defence failure, likely distribution of insured properties within an area and expected damage to buildings as a result of flooding are also not available. For an underwriter to make an informed decision, all these pieces of information, and more, are required. Gathering the relevant data alone is a time-consuming process, and combining them to get a full picture of the likely hazard is not a realistic goal for most underwriters.

Fortunately, the availability of probabilistic flood hazard models on the market is increasing. These are able to combine information relating to rainfall and flood hazard with event frequencies, resulting water depths and the likelihood of failure of flood defence schemes. A choice between single site analysis where the location of an insured property and analysis of aggregate data is typically available, with risk distribution models providing an assessment of the spread of risk with location and elevation in the latter case. In addition they provide vulnerability curves and the ability to model insurance conditions and reinsurance treaties so that specific portfolios/treaties can be analysed quickly and efficiently by the user. Such models may prove an ideal way forwards to resolve flood exposure management issues. EQECAT's `EUFlood' is one such model, allowing the user to analyse portfolios for river plain and sea surge flooding in Europe.

By Jane Toothill

Jane Toothill is a natural hazard analyst at EQECAT Europe.