When floods are predicted only on the basis of local data, there may be unpleasant surprises. A new method makes it possible to significantly improve predictions - using international data from hydrologically similar areas.
What can we expect in the worst-case scenario? In regions with a high risk of flooding, this is an important question: what extreme events should the protective measures be designed for? Often this is answered simply by looking at history: The worst flood events of the past decades or centuries are regarded as a realistic upper limit for what can be expected in the future.
However, this can be misleading, as so-called "mega-floods" have shown in recent years. Time and again, extreme flood events occur, extraordinary outliers that were not considered possible on the basis of local data. However, a major research project carried out under the leadership of TU Wien (Vienna) has now been able to demonstrate: If one considers the entire European continent, these local surprises are no longer surprising at all. If data from other regions with similar hydrological conditions is taken into account, the extent of these "mega-floods" suddenly becomes predictable. This has drastic consequences for the way in which flood protection must be dimensioned. The results have now been presented in the journal Nature Geoscience.
The simple extrapolation of probabilities
In 2021, there was a devastating flood disaster in Germany and Belgium in which over 220 people died. An event of this magnitude had not been expected. "Predicting the extent of such mega-floods is very difficult," says Prof. Günter Blöschl from the Institute of Hydraulic Engineering and Water Resources Management at TU Wien, who led the project. Until now, the usual strategy has been to statistically examine previous flood events in the region: There is a high probability of minor flooding, a lower probability of major flooding. From this, one can try to extrapolate the probability of even greater floods.
However, as has now been shown, there is a much better strategy: in an elaborate research project, data from more than 8,000 gauging stations across Europe, from the years 1810 to 2021, has now been analysed. "The decisive step was to anticipate mega-floods in one place by using data from similar river basins in other places on the continent" explains Dr. Miriam Bertola (Institute of Hydraulic Engineering and Water Resources Management, TU Wien), the lead author of the current publication. "In each river basin we can learn from other areas that have similar climatic and hydrological characteristics where mega-floods may have already occurred.”
Expecting the unexpected
If one looks at the historical flood data of all these areas simultaneously, a pattern emerges: locally surprising "mega-floods" are often below or close to the upper limit of previous floods observed in similar regions. By using a larger amount of data, a statistical outlier at the local scale becomes something expected at the continental scale.
The research team was able to show that flood disasters such as the one in the Rhine basin in 2021 could have been anticipated in this way - it was actually also well within the expected range if continental data is included.
An international perspective for better flood prevention
"It is important to consider not only geographically adjacent areas, but also areas with similar conditions - these may also be located further away," emphasizes Günter Blöschl. "It is therefore essential to move beyond national flood-risk assessment and share information on mega-floods across countries and continents, to reduce the surprise factor of their occurrence and save lives."