Flood policy approaches in the UK and France
While the last two European summers have been characterised by an extreme lack of rainfall rather than a surfeit, major flood events continue to occur in other parts of the world, most recently (July 2005) in India and China, and research into flood hazard and risk continues apace. With regard to policy, the UK Government’s Department for Environment, Food and Rural Affairs (DEFRA)¹⁴, launched a major consultation exercise in July 2004, the results of which are published in a series of documents – under the heading Making space for water: taking forward a new government strategy for flood and coastal erosion risk in England and Wales, that can be accessed online. The consultation was initiated in response to the findings of an earlier government report that highlighted the huge potential cost of future flooding and is designed to develop a holistic approach to future flood risk, which addresses – in particular – the issue of integrating drainage planning to tackle flooding in urban areas.

Continuing the flood policy theme, Nathalie Pottier⁵¹ of the University of Versailles and colleagues, contrast different approaches to flood plain land use and flood protection in France with those of England and Wales. In a paper in Applied Geography, the authors compare a central government led ‘coercive’ approach in France with the more ‘cooperative’, locally-led approach that is currently adopted in England and Wales. Porter and her co-workers highlight pros and cons in relation to both approaches, but rule that one cannot be said to be better than the other. The French procedures, for example, appear to be successful at preventing urban development in high flood hazard areas, but not so good in the lower risk areas. In the UK, rather than be excluded from high hazard areas, urban development may be permitted provided that it is coupled with appropriate flood defences; hardly a sustainable way forward. The authors recognize that pressures on floodplains in France, England and Wales, continue to grow (for example the proposed Thames Gateway development in the south east of England), while at the same time, climate change threatens to raise the level of flood risk. Both approaches, they point out, have yet to prove themselves strong enough to stem the tide.

Uncertainties in flood frequency analysis
Uncertainty, whether due to climate change or other factors, remains a major issue in addressing flood frequency analysis. In this regard, Bruno Merz and Annegret Thieken⁴⁴ of GeoForschungsZentrum in Potsdam address, in the Journal of Hydrology, the issue of how natural and epistemic uncertainty can be separated when looking at flood frequency. In this context, natural uncertainty arises from the variability of the flood process, while epistemic uncertainty is a result of our incomplete knowledge of this process and is reflected, for example, in uncertainty relating to both sampling and modelling. The authors argue that in flood frequency analysis, the two types of uncertainty should be separated, as this can provide a more accurate picture. While natural uncertainty cannot be reduced, epistemic uncertainty can. A measure of the total uncertainty – in relation, for example, to the safety of a flood defence wall – will, inevitably, hide this distinction, giving the impression that more observations are all that are needed to improve the wall’s safety. Merz and Thieken conclude that an approach that separates out the uncertainties in flood frequency analysis can guide efforts for obtaining more information and enable these to be prioritized.

Flood damage to buildings
When a flood does occur, its effects on buildings – as revealed by extensive television coverage – might appear to be obvious. In fact, building damage and failure due to flooding arise from a range of other factors than simply the relatively slow rise of water levels (figure 11). In a paper in Engineering Geology, Ilan Kelman and Robin Spence³⁴ of the University of Cambridge, undertake a comprehensive review of the range of damaging flood actions on buildings. In addition to a slow increase in water depth, these include: lateral pressures arising from water velocity or from differential internal and external water levels; buoyancy effects; capillary rise; and the consequences of incorporated debris, turbulence, waves and erosion. Kelman and Spence conclude that the extent to which flood actions, other than the slow rise of water, affect direct flood damage has not been considered in sufficient detail to be fully understood. They call for more research in order to evaluate how flood damage arises so that it can be more effectively prevented.
Figure 11. In addition to the slow increase in water depth, floods can damage buildings in a range of other ways including; lateral pressures arising from water velocity or from differential internal and external water levels; buoyancy effects; capillary rise; and the consequences of incorporated debris, turbulence, waves and erosion. Courtesy: Alan Thomson, Symonds Group Limited.

Precipitation characteristics and flood estimation
Ultimately, flooding is directly related to the level and intensity of precipitation, a link examined by Daniela Rezacova⁵³ and co-researchers from the Institute of Atmospheric Physics in Prague. In a paper in Atmospheric Research, the authors take a look at how the probable maximum precipitation (PMP) can be estimated for river basins in the Czech Republic. The study was partly a response to major flood events in 1997 and 2002 (figure 12), and the resulting need to estimate design floods for reservoir outflow structures in the country. PMP values are critical in assessing the probable maximum flood (PMF) related to the catchment of a dam, which in turn is employed in designing hydrological structures, such as slipways, that minimize risk of dam overtopping. PMP is defined as ‘the greatest depth of precipitation for a given duration that is physically possible over a given storm size area at a particular geographical location at a particular time of year’. Comparison with rainfall during the 1997 and 2002 floods, showed that the maximum area rainfalls over small catchments on these occasions did not exceed 63 percent of corresponding PMP values.
Figure 12. Cumulative precipitation over central Europe in a 72 hour period leading up to 13 August 2002. The resulting floods, affecting the river basins of the Elbe (Labe), Danube, Moldau (Vltava), Inn, Salzach and Kamp, were some of the worst ever recorded. Despite the torrential and persistent precipitation, maximum area rainfalls over small Czech river catchments on these occasions did not exceed 63 percent of corresponding (probable maximum precipitation) PMP values. Courtesy: Olly Willetts.

Regional flood hazard and risk studies
While precipitation is the main driver of flooding, many other factors come into play, not least land use and its change over time. This is a relationship examined by Min Tu⁶³ of the UNESCO-IHE Institute for Water Education in Delft (Netherlands) and co-workers, with respect to extreme floods in the Meuse river over the past century. In a paper in the Physics and Chemistry of the Earth, Tu and colleagues highlight the fact that five out of the last seven largest floods recorded on the river between 1911 and 2003, have occurred in the last decade. This, it has often been assumed, is a consequence of an increase in the magnitude and frequency of extreme floods due to rapid land-use changes in the river basin since the 1950s. The study by Tu and co-authors show that both the flood peaks in the Meuse river, and the antecedent precipitation depths in the basin, have increased since the early 1980s. Since there have been few land-use changes since this time, however, the authors of the study conclude that climate variability can provide the only explanation for changes in frequency and magnitude of the Meuse floods over the past two decades.

In another regional study, published in Advances in Water Resources, G. Calenda¹⁰ and colleagues at the University of ‘Roma Tre’ have undertaken an evaluation of the distribution of extreme peak floods on the Tiber river that have affected Rome over the last 500 years. The authors used a large and varied database, drawing – for example – on a long record of daily river stage measurements (up to the 18th century) and contemporary descriptions of several extreme floods, as well as on more recent records of rainfall depth, river flow measurements and bed surveys. Twenty-two flood events have been recorded in Rome since 1000 AD, the last in December 1870. The distribution of these events leads Calenda and colleagues to the same conclusion as Tu and co-workers, with respect to the Meuse. Both the long-term variability of the annual peaks of the Tiber river since 1782, and the highly variable frequency of recorded flood events going back many centuries, suggest that these are not related to land-use or other changes related to human activities, but point to a link with variations in climate.

Following the deaths of over 1,800 people and the destruction of more than 3,000 homes in the North Sea flood of 1953, great importance has been attached in the Netherlands to quantitative flood risk assessment for the polders (areas of land reclaimed from the sea and protected by dykes). This issue is addressed by Sipke van Manen and Martine Brinkhuis⁴² of the Dutch Ministry of Transport, in a paper in Reliability Engineering and System Safety. The paper describes the application of new tools to calculate the actual flood risk of a polder that is entirely surrounded by protecting dykes – known as dyke rings. Van Manen and Brinkhuis model a breach in a dyke ring in an area of central Netherlands that is enclosed by two arms of the river Rhine. Sixteen hours after breaching the local water depth reaches a maximum of about 6m, by which time it can be expected to have claimed 153 victims (assuming no evacuation) and caused Euro 16 billion worth of damage. While presenting just this single example, the authors call for the collection of more data in order to improve risk analysis with respect to polder vulnerability. In particular, they highlight the need to examine the probability of failure of structural artifacts – such as sluices – in dykes, something that has not, so far, been adequately considered.