Economic losses from floods have increased dramatically in the past couples of decades. Jean-Paul Conoscente and Anja Sigl report.

This trend has been caused by increasing economic development in flood prone areas such as coastlines or flood plains, which has resulted in changes in land and vegetation, as well as in global climate.

Risk in EuropeThere are three types of floods, which can cause severe economic damage to a region: flash floods, river floods and storm surges.

Flash floods
Flash floods can be triggered either by torrential rain leading to sudden flooding, or by catastrophes, like dam failure after an earthquake.Although flash floods can occur almost anywhere in the world, mountainous areas are more prone to this type of phenomenon. In addition to mountains fostering rain formation, mountain slopes and small basins increase the probabilities of run-offs. Unfortunately, urban centres are not exempted from the ravages caused by flash floods. In cities, rainfall water cannot infiltrate easily into the ground, causing the sewage system to quickly saturate, leading to flooding. Although radar and satellite data are available they do not serve as good warning systems.

These events are known to cause substantial human and economic losses. Europe, and more particularly the Mediterranean region, is especially susceptible to these losses. In November 1994, Italy suffered one of the worst flash floods in its history, with economic losses of $12.5 billion and insured losses of $65 million. Furthermore, in November 1992 the South of France was hit by flash floods resulting in insured losses of over $30 million.

River floods
River floods are caused by continuous rain over a long period on a large geographical area. As ground capacity to absorb excess water diminishes and the river water level rises, the water overflows protection barriers, causing floods.

During the last decade, the Rhine and Danube have had “Centennial Floods” despite flood protection dikes. For instance, in December 1993, the Rhine, Mosel and Main rivers flooded and then 13 months later, in January 1995, there were new “Centennial Floods” in Germany, the Netherlands, Belgium and the north of France.In contrast to flash floods, river floods are easier to forecast and prepare for, as the flood districts can be modelled, and flood waters rise slower than flash flood waters.

Storm surges
Storm surges occur when the sea level rises by typically 2 to 3 metres, although it can rise up to 10 metres. This phenomenon can occur over a stretch of 10 to 100 kilometres and can last several hours.

The 1953 flood in the east coast of the United Kingdom was one of the worst storm surges ever, flooding over 800 km2. The flood caused more than 300 deaths in the UK and 1,800 deaths in the Netherlands. Gasworks, power stations, and water supply services were interrupted, and over 200 miles of railway track were made impassable. Up to 160,000 acres of flooded agricultural land suffered salt contamination, and most areas experienced deposition of vast quantities of sand and other debris.

Flood risk and insurance
The increasing economic losses caused by floods have led to an increase in demand for insurance coverage, especially from the recently affected populations. The simplistic model for assessing river flooding hazard requires the following elements:
• a model of the river bed at regular intervals for different return periods;
• a digital terrain model (DTM) of the surrounding topology;
• the possibility of simulating a single event along the river.
The first two are used to determine a flood height at a given point and the third is used to create realistic accumulation zones. Such work has been done in many parts of Europe and sophisticated models of the flood hazard are currently available in countries such as France and Germany. The development of realistic accumulation zones is a critical point for the insurance and reinsurance industry, and European model development efforts are still underway.

Modelling of flash floods is more difficult and requires the following elements:
• a torrential precipitation model which could be based on historical data;
• a model of the catchment area affected and the run off;
• superposition of these models with DTM of the surrounding topography.
Several companies, including EQECAT, are currently tackling this issue but the complexity of the physical model still remains to be solved.

Models of tidal height and storm surge have been carried out intensively in the United Kingdom and the Netherlands, and to a lesser extent in Germany. These models have been incorporated as part of the insurance risk assessment tools for the United Kingdom, and current development is underway to do the same for continental Europe.

The main barriers that prevent a rapid expansion of flood models are the lack of:
• economically available hazard data;
• a reliable and affordable DTM;
• information on the vulnerability of flood protection measures.
While in places such as the United States the information is publicly available, this is not the case in Europe, where data is difficult to get and the expenses required to get it are not justified by the current insurance market conditions.

Flood models
With increasing availability of high quality data, quality modelling of flood risk is improving dramatically. The following chart shows the three pillars needed for developing a hazard risk model.

Conclusion
Flood risk will continue to be one of the main challenges for the reinsurance industry into the next century, and although scientific research has made tremendous headway over the past 20 years, flood risk assessment tools are still not commonly used by insurers and reinsurers.

Jean-Paul Conoscente, director, Anja Sigl, engineer, EQECAT.