Metodologie per l’aggiornamento delle curve segnalatrici di possibilità pluviometrica nel Veneto sulla base di modelli climatici ad alta risoluzione

Intensity–Duration–Frequency (IDF) curves are an essential tool for hydraulic design and flood risk management. However, their definition based solely on observational data does not adequately capture the impacts of climate change on extreme precipitation. This work proposes innovative methodologies for updating IDF curves in the Veneto region, using high-resolution climate simulations from Convection-Permitting Models (CPMs) for both historical and future climates, integrated with advanced statistical analyses of extreme events. As a preliminary step, a spatial correlation analysis of extreme precipitation was conducted, based on Moran’s Index and the estimation of the integral scale, for the study area and for its subdivision according to orographic features. This was done to characterize the spatial dependence of precipitation extremes and to define an appropriate use of model information. Subsequently, two approaches were developed and compared: (i) the application of CPM-based percentage changes to observed quantiles; (ii) the use of bias correction procedures derived from the comparison between observed and simulated data. Both methods were applied to regional rainfall records, adopting the Metastatistical Extreme Value Distribution (MEVD)—which better reduces the uncertainty associated with limited samples compared to the classical Generalized Extreme Value (GEV) distribution—and incorporating future climate projections under the RCP8.5 scenario for the periods 2040–2050 and 2090–2100. Results show that both approaches lead to updated IDF curves with higher quantiles compared to the present, consistent with projections of increasing extreme precipitation. The percentage-change method provides more consistent and stable results across stations, while the bias correction highlights the complexity of a systematic adjustment, influenced by orographic factors and the varying performance of the models. Overall, the analysis confirms the potential of CPMs in representing convective extreme events, while also emphasizing the need for more robust and generalizable correction procedures.

Il dimensionamento delle infrastrutture idrauliche richiede stime affidabili delle precipitazioni estreme, ma l’ipotesi di stazionarietà su cui ad oggi si basano le Curve Segnalatrici di Possibilità Pluviometrica (CSPP) non è più adeguata in un clima in rapido cambiamento. La presente tesi utilizza modelli climatici ad alta risoluzione (Convection-Permitting Models, CPMs) e serie storiche misurate per valutare l’evoluzione futura delle precipitazioni nel Veneto e propone e confronta tre metodologie per aggiornare le CSPP agli scenari futuri 2050 e 2100. Attraverso l’uso della Metastatistical Extreme Value Distribution (MEVD), della sua estensione spaziale e di un’analisi dell’autocorrelazione regionale, si superano le limitazioni dei dataset modellistici, caratterizzati da brevi serie temporali. Le metodologie sviluppate vengono confrontate mostrando incrementi significativi degli eventi estremi, con variazioni dipendenti da durata, tempo di ritorno e posizione geografica. I risultati confermano la necessità di revisionare le curve pluviometriche attualmente in uso e mostrano come i CPMs rappresentino oggi uno strumento indispensabile per progettare opere resilienti ai futuri scenari climatici.