Flood risk : how can we estimate extreme flood levels ?

A rare event can nonetheless be a major risk. In order to be protected from such events, it is necessary to identify and estimate the levels of extreme or thousand-year floods, the details of which have been forgotten. The Extraflo project, led by Irstea, was completed in November 2013. It was given the task of developing recommendations on how to estimate the levels of extreme floods. Here are its main results.

The one-hundred year Seine flood of 1910 and that of the Rhône in 2003, though thought to be exceptionally severe [1], were nothing when compared to extreme floods. The recurrence of these events enables the development of survey data required for predictive models and pre-determining the risk of flooding. For extreme events, however, data is much scarcer. Consequently, the difficulty of estimating this major risk is evident. The Extraflo project, coordinated by Irstea [2], has enabled methods of estimating extreme floods to be re-evaluated. 

Focus on vulnerability

When faced with an extreme flood, the aim is not to stop the flooding. Building dams and flood barriers of sufficient size would be too expensive. Michel Lang, a hydrologist at Irstea, and the project’s coordinator explains : "We need to target vulnerable areas by working out contingency plans and a series of preventive measures.  On the Loire, for example, if companies moved storage so that it was one metre from the ground, this would considerably reduce any potential damage in the event of extreme events."

It was with this idea in mind that consultancy firms and various bodies responsible for the safety of network infrastructures, such as the SNCF, were invited to a project conclusion seminar on 15 November 2013. The seminar presented all the recommendations and the methods that had been assessed. 

Methodology and recommendations

The main goal of the project was to produce a set of recommendations for estimating risks. Very quickly it became apparent from the datasets of observations highlighted that there was a lack of robustness in the approaches, which relied on the statistical analysis of a series of observations collected over some tens of years. In order to strengthen them the Extraflo project involved a wide variety of competencies and disciplines: climatology, statistical modelling, hydrology and hydraulics, hydrogeomorphology, paleohydrology etc. In other words, these alternative approaches have enabled the analytical tools to be refined. In the final analysis, it seems that estimating extreme floods could be strengthened by using four different approaches : 

  • Use rainfall and flow records on a larger, regional scale ;
  • Collect historical data ;
  • Use simulation methods to convert rainfall into flow ;
  • Analyse morphological records left by earlier floods. 

Most important was a set of recommendations for use by industrial and local stakeholders. 

For further information

[1] A hundred-year flood is one whose probability of occurrence in a given year is 1/100 in terms of flow. Thus the likely recurrence interval between such floods is 100 years. The same logic applies to the thousand-year or extreme flood, with a probability of 1/1,000 and a recurrence interval of 1,000 years.

[2] Extraflo is an applied research project coordinated by Irstea with the help of Météo-France (Toulouse), CNRS (HydroSciences and GeoSciences Montpellier) and EDF/DTG (Grenoble) and exchanges with government agencies (Cete Méditerranée and Dréal Midi-Pyrénées) and consultancy firms (Artélia, Electralabel). The project was conducted over a period of 4 years (2009-2013) and benefited from ANR aid in the framework of the RiskNat programme.