Author(s): Carmelo Raspanti; Gabriele Freni
Linked Author(s): Gabriele Freni
Keywords: Flood risk Climate change Warning system Rainfall thresholds
Abstract: One of the most severe impacts of climate change is the increase in the frequency and magnitude of extreme precipitation events producing flooding (IPCC et al., 2022). According to the European Environment Agency, the overall losses for floods recorded in Europe in 1998–2009 summed to about EUR 52 billion (Agency, 2010). Flood risks and societal damages are projected to increase with global warming. In Europe, with 3°C global warming increase, the damage costs and people affected by precipitation and river flooding may double. (IPCC, 2022). The relationship between the growth of expanding cities, the number of catastrophes and the reduction of inhabitants in inland regions adds to the impacts of extreme events and the related damage. (Bertin, 2021) The measures taken to prevent or mitigate the impacts of floods can be identified as structural and non-structural measures. Standard structural measures are riverbank enlargements, urban drainage improvements, expansion or overflow ponds, and storage facilities. The latter are used to store floodwater, decrease the peak phase and retain floodwater to mitigate the impact of floods. Non-structural measures include flood early warning systems on which we will focus, emergency flood planning and response, and environmental education. Non-structural measures are applied to prevent and reduce harm through actions, legislation, standards and incentives. (Yang, 2020) It is essential to be able to forecast such events and prepare early warning systems to trigger emergency measures before flooding occurs. At the same time, an early warning system should be sufficiently robust to prevent false alarms that may be expensive and may reduce the system's reliability. This study aims to implement an early warning system considering climatic uncertainty. The analysis was applied to the San Bartolomeo basin in Sicily (Italy). The basin is characterised by large and highly populated urban areas near the river mouth and steep natural catchments in the upper part of the basin. The hydrological analysis of the natural catchment was made using state-of-the-art approaches based on regional extreme rainfall analysis and the SCS-CN method for rainfall-runoff transformation. To take climate change into account, we decided to follow the procedure proposed by Liuzzo and Freni (2015). As a first step, the Mann-Kendall test was used to verify the existence of trends; subsequently, it was possible to calculate the trend in rainfall annual maxima for fixed durations of interest considering the statistical hypothesis of scale invariance. Finally, using the Two Component Extreme Value Distribution (TCEV), rainfall depth and its intensity with projections to 2050 (short-term scenario) and 2100 (long-term scenario) were defined. Once critical flows from natural catchments were calculated, it was possible to proceed to the second phase, analysing the hydraulic risk in the urbanised downstream area. This second phase has allowed the simulation of flooding propagation in the urbanised area through a 2D model that applies the De-Saint Venant equations according to a bi-dimensional approach, obtaining the probability functions that relate flooding variables (water depth, velocities, specific energy) to river flows and by reverse approach to rainfall depths. The results have made it possible to identify the risks by combining hydraulic variables and considering different targets (humans, movable objects, buildings and urban furniture). Non-structural mitigation measures, such as early warning systems, are a good solution because they are quick, almost inexpensive and highly adaptable to climate change. on the other side, they suffer from intrinsic climate uncertainty primarily when their efficiency is related to the prediction of precipitations and the related impact on the ground. With the increasing level of exposure, the ability to quantify future risk is essential to develop suitable adaptation and mitigation measures. This study aimed to implement non-structural measures after a procedure to assess the risk of flooding at the local level. Sicily is sensitive to this risk due to the frequent proximity of inhabited centres to natural waterways. For this reason, it was decided to apply the procedure to a Sicilian catchment area, specifically the San Bartolomeo River basin, located in the western part of the island, which in the past caused victims due to flood events so suggesting to address the warning system the presence of the population in flood-prone areas. The study is divided into two main and successive phases. The first phase, defined as pre-processing, through the analysis of geographical data, made it possible to automatically identify, starting from a digital model of land elevations (DEM - Digital Elevation Model), the estimate of the morphological characteristics. The morphological characteristics include the hydrographic network, the watersheds and, therefore, the limits of the basin and sub-basins, each of which is accompanied by the main morphological parameters. Next step of the study consisted of the detection and quantification of trends in the annual maximum rainfall series of different durations (1,3, 6,12, and 24 h) observed at regional and local scale. The annual maximum rainfall series recorded using the rain gauges over the period 1931–2016 has been selected and analysed. For each duration, the moving averages were computed, and then the Mann-Kendall test was applied, considering different confidence levels. Statistical analyses have been carried out for each duration to estimate the quantiles for different return periods. The evaluation of the rainfall probability curves matched on the basin was analysed starting from the Two Component Extreme Value Distribution (TCEV). The results showed that, for all durations, increasing and decreasing trends occurred in the period examined. Once the trends were detected and quantified, the TCEV calculated extreme rainfall with return periods of 2,5, 10,20,50,100 and 300 years. The intensity-duration-frequency curves for all the raingauges in the region were estimated for current climate conditions and for the two future climate change scenarios. Subsequently, the hydrological loss was evaluated with the SCS-CN (Soil Conservation Service-Curve Number) model, with saturated soil conditions, obtaining results for the worst possible case. Subsequently, the convolution is applied to estimate the flood hydrograph for the given return period. Finally, flooding characteristics were evaluated using the FLO-2D model (FLO-2D Reference manual), which adopts the De-Saint Venant equations according to a two-dimensional approach, using as input data those obtained in the previous phase, the outflow, finally obtaining the values of the maximum flow rates at the height of the flood for the set return times in correspondence with the closure section of the main river branch. The hydrological study obtained a complete forecast of possible flooding and estimating the relative hydraulic risk. The second phase was aimed to the definition of rainfall thresholds to be used from the beginning of a rainfall events to trigger preventive actions and emergency management. Rainfall threshold values were defined on a geographical basis in order to differentiate actions in the flooded area depending on the characteristics of the triggering event. The designed warning system was able to guarantee a period of 75 minutes between the event triggering and the start of flooding. Such time is sufficient to allow the evacuation of people from the area but not enough to prevent physical damage to properties.
Year: 2025