How well can we estimate flood risks?
Uncertainties associated with flood risk estimates are substantial and originate for a large part from components that are relatively scarcely researched.
In the Netherlands, located on the delta of the Rhine, Meuse and Scheldt rivers, the Dutch have had to live with the constant threat of flooding for many centuries. For more than a thousand years, communities have come together to build dikes and defensive structures to protect themselves from floods. In recent decades, however, flood management has started taking a broader approach than just building protective structures to cope with the danger of flooding. In such a so-called risk management approach, also the potential consequences are taken into consideration. This focus on a risk-based approach makes flood risk assessments increasingly more important. Such assessments aim to quantify the flood risk of an area and to assess the impact of certain management measures or future developments (e.g. climate change, population growth) on flood risk.
In such flood risk assessments, many different types of data, models and analyses are combined to estimate flood risk. All these analyses, models and data are surrounded by uncertainties, which propagate and accumulate through a risk assessment. In order to estimate the uncertainty surrounding the final flood risk estimate, it is necessary to make a large amount (hundreds) of calculations with input parameters that are constantly changed (known as Monte Carlo analyses). Such comprehensive uncertainty analyses are not performed often as the computational burden of doing so many calculations is large. This mostly results from the complexity of flood inundation models, a key model in flood risk assessments. To overcome this, a simplified inundation model has been developed for dike rings in the Netherlands. With this model, uncertainties in flood risk estimates have been investigated for the West (dike ring 14) and middle South (dike ring 36) of the Netherlands. Uncertainties that were considered were related to: the duration and magnitude of the flood event, the probability of the event, growth of the breach in the embankment, the damage calculation, and information on which assets are at risk (e.g. land use).
These cases illustrate that uncertainties surrounding flood risk estimates are substantial; estimates can easily be four times smaller or larger than the median estimate. The most influential parameters causing this uncertainty are related to the damage calculation, the duration of high water level conditions, and the probability of a flood event. Overall, it can be concluded that for flood damage of dike rings in the Netherlands, the order of importance of different parameters is: (1) the probability, (2) the damage calculation, (3) the duration of high water conditions, (4) the breach growth, (5) the estimation of the inundated area, and (6) the land-use information.
These results imply that, given its large contribution to the overall uncertainty, the damage calculation part deserves more attention than it generally receives in flood risk assessments, which often focus on the hazard. Furthermore, given the considerable uncertainty present in risk assessments, it is advised to perform multiple calculations using different combinations of input parameters in order to construct lower, middle, and upper estimates that reflect the uncertainty present in the risk estimate. This uncertainty can then be communicated more transparently and used in follow-up actions such as cost-benefit analyses. Moreover, the implicit cost of uncertainty in flood risk should also be considered in flood management. From an insurer's perspective, as well as from a societal perspective, high uncertainties are viewed as undesirable, and situations with high uncertainty should therefore be valued lower than situations with less uncertainty. Including such considerations in flood risk decision-making may result in a more resilient situation with respect to flooding.
Hans de Moel has performed research on this subject for his PhD thesis. His public defence is scheduled on September 20th at 13:45.
Publications on which thesis is based
De Moel, H., Asselman, N.E.M. & Aerts, J.C.J.H. (2012). Uncertainty and sensitivity analysis of coastal flood damage estimates in the west of the Netherlands. Natural Hazards and Earth System Sciences, 12, 1045-1058. doi:10.5194/nhess-12-1045-2012
De Moel, H., and Aerts, J. (2011). " Effect of uncertainty in land use, damage models and inundation depth on flood damage estimates." Natural Hazards, 58(1), 407-425. doi:10.1007/s11069-010-9675-6
De Moel, H., Aerts, J. C. J. H., and Koomen, E. (2011). "Development of flood exposure in the Netherlands during the 20th and 21st century." Global Environmental Change, 21(2), 620-627.doi:10.1016/j.gloenvcha.2010.12.005
De Moel, H., J. van Alphen, and J.C.J.H. Aerts (2009), “Flood maps in Europe – methods, availability and use”, Natural Hazards and Earth System Sciences, 9, 289-301, http://www.nat-hazards-earth-syst-sci.net/9/289/2009/
[eigen figuur uit proefschrift, overstromingssimulaties voor dijkring 36]
[foto uit Beeldbank van RWS – hoogwater Maas 1995 bij Itteren]