Uncertainty in flood frequency analysis and its implications for infrastructure design

Working with environmental data means dealing with complex processes, limited data (in space and/or time) and the impossibility of setting up controlled experiments, leading to uncertain predictions of system behaviour. In the field of statistical hydrology, many efforts have been made during the last decades to provide methods to quantify uncertainty, but the common practice of infrastructure design has not yet incorporated them. This may be due to several reasons, including the complexity of the methods, which are often difficult to apply in most everyday cases, and regulations that "favour" well-established requirements. Here we present the "uncertainty compliant design flood estimator" (UNCODE) method, which accounts for aleatory uncertainty in the estimation of the design flood value. The method provides a corrected design value and is easy to use for practical purposes as simplified formulae are provided to quantify the correction factor. However, in addition to its practical application, it can also be used to compare different models with different levels of uncertainty and to highlight the "cost" of uncertainty. Finally, its mathematical formulation allows a direct link to be made between the classical approach to hydrological design, based on a fixed hazard level (or return period), and a risk-based design approach, which is widely recognised as a more flexible method but is not usually included in regulations.