Morphological and Climatic Controls on the Frequency Distribution of Sub-Daily Rainfall Annual Maxima in the Po River District

An approach that relies on L-moments is used to examine the long-term annual series of precipitation maxima for two storm durations, namely 1 and 24 hours. Data are measured at a widespread network of rain gauges sited in the northern part of Italy. Based on 1951-2019 data on precipitation and several climatical information, the main focus of the study is to explore the statistical characteristics of extreme rainfall and to find statistically significant relationships between these characteristics and local mean annual precipitation (MAP), as well as other local morphological descriptors. First, this study uses a Weighted Moving Average (WMA) approach to assess the relationships between the statistics of observed rainfall annual maxima for 1 and 24 hours, namely the L-moments of order 1 (l1, arithmetic mean), 2 (l2 or L-Cv, analogous to the variance) and 3 (L-Cs, analogous to the skewness) and the mean annual precipitation (MAP), which is often used as a surrogate of geographical location in the literature. The l1 and l2 displayed the strongest dependence on MAP. Additionally, the coefficients of variation (L-Cv) showed a minor reduction when the mean annual precipitation was rising, which is more apparent in rainfall that lasts for an hour. Second, a stepwise regression procedure was applied to develop multi-regression models for predicting the considered L moments as a function of local morphological and climatic characteristics. Our study shows that MAP, and minimum distance to the Tyrrhenian coast (MinDistTir), are the most important descriptors, showing a strong correlation with the L-statistics of rainfall extremes.