Using FloodRisk GIS freeware for uncertainty analysis of direct economic flood damages in Italy

The considerable increase in flood damages in Europe in recent decades has shifted attention from flood protection to flood risk management. Assessments of expected damage provide critical information for flood risk management efforts. The evaluation of potential damages under different flood scenarios through quantification of their ability to provide relative short-, medium- and long-term risk reduction, supports decision-makers in discriminating among several alternative mitigation actions. End-users should be aware of, and knowledgeable about, the limitations and uncertainties of such analyses, as well-informed decisions regarding efficient and sustainable flood risk management will become increasingly relevant under future climate and socio-economic changes. In this context, a method was developed to identify and quantify the role of the input parameters in the uncertainty of the potential flood economic damage assessment in urban areas with low sloping/flat terrain and complex topography using a GIS-based, free and open-source software called Floodrisk. Sets of plausible input parameters for the model's two flood loss modelling subroutines (hydraulic modelling and damage estimation) were dynamically combined to quantify the contribution of their inner parameters to the total damage assessment uncertainty. To estimate the contributions of each input to overall model uncertainty, the combination of input parameters that minimized the error in the spatial distribution assessment of the extensive damages affecting (downtown) Albenga (Italy), enumerated after the historical Centa River flood of November 5, 1994, was taken as a reference. In this specific case, a high epistemic uncertainty for the damage estimation module was noted for the specific type and form of the damage functions used. In the absence of region-specific depth-damage functions, the vulnerability curves were adapted from a range of geographic and socio-economic studies. Given the strong dependence of model uncertainty and sensitivity to local characteristics, the epistemic uncertainty associated with the risk estimate was reduced by introducing additional information into the risk analysis. Implementing newly developed site-specific curves and a more detailed classification of the construction typology of the buildings at risk, led to a substantial decrease in modelling uncertainty, along with a decrease in the sensitivity of the flood loss estimation to the uncertainty in the depth-damage function input parameter. These findings indicated the need to produce and openly disseminate data in order to develop micro-scale risk analysis through site-specific vulnerability curves. Moreover, this study highlighted the urgent need for research on the development and implementation of methods and models for the assimilation of uncertainties in decision-making processes.