Modellazione statistica di eventi meteorologici rari presenti e futuri in aree scarsamente strumentate

This Thesis addresses the statistical modelling of rare hydrological and coastal extremes in settings where observations are short, heterogeneous, or spatially sparse, and where changing climatic and environmental conditions can alter the frequency and intensity of extreme events over time. It develops a coherent methodological framework to support the estimation of design-relevant quantiles for floods, extreme precipitation, and storm surges in poorly gauged and ungauged areas. The work builds on metastatistical extreme-value approaches, which estimate the distribution of block extremes by exploiting information from the broader population of ordinary events (and their occurrence rates), rather than relying only on annual maxima or a small set of threshold exceedances. Within this perspective, the Thesis advances methods that can better leverage limited samples, incorporate spatial information through regionalization, and interface with process understanding where beneficial. The proposed developments are organized across multiple scales and hazards, ranging from catchment-scale flood estimation to regional rainfall extremes and coastal storm-surge modelling, and extending to gridded applications relevant for large-scale risk assessment and climate-change impact studies. Throughout, the emphasis is on methodological generality, transparent uncertainty treatment, and operational applicability in engineering design and planning contexts.