Non-stationarities in extreme hourly precipitation over the Piave Basin, northern Italy

Dáire Healy 1∗ , Ilaria Prosdocimi 1 , and Isadora Antoniano-Villalobos 1 · 2025

Read Non-stationarities in extreme hourly precipitation over the Piave Basin, northern Italy by Dáire Healy 1∗ , Ilaria Prosdocimi 1 , and Isadora Antoniano-Villalobos 1 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-20069.txt.

Introduction

Healy, Prosdocimi, and Antoniano-Villalobos model how extreme downpours cluster over Italy's Piave Basin. Their 2025 paper studies hourly rainfall extremes in northeast Italy. Floods and landslides there trace back to intense rain bursts. Seasonal patterns shape both rainfall size and spatial spread. Dependence between sites weakens as events grow more extreme.

Hourly records came from sixty-five monitoring sites initially. Data spanned 1990 to 2023, averaging nearly twenty-eight years per site. Strict quality rules trimmed the network to forty-nine sites. Monthly maxima of hourly rain formed the analysis target. The team compared covariate-driven models for extremes. Climate variables entered at several space and time scales. A flexible max-id model handled dependence flexibly. Modeling only the margins missed real variability, they found. Seasonal change in dependence mattered just as much. Risk estimates improve when both layers vary.

Catalog record

CREATORDáire Healy 1∗ , Ilaria Prosdocimi 1 , and Isadora Antoniano-Villalobos 1
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE75 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-20069

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Non-stationarities in extreme hourly precipitation over the Piave Basin, northern Italy Dáire Healy 1∗ , Ilaria Prosdocimi 1 , and Isadora Antoniano-Villalobos 1 Abstract We study the spatio-temporal features of extremal sub-daily precipitation data over the Piave river basin in northeast Italy using a rich database of observed hourly rainfall. Empirical evidence suggests that both the marginal and dependence structures for extreme precipitation in the area exhibit seasonal patterns, and spatial dependence appears to weaken as events become more extreme. We investigate factors affecting the marginal distributions, the spatial dependence and the interplay between them. Capturing these features is essential to provide a realistic description of extreme precipitation processes in order to better estimate their associated risks. With this aim, we identify various climatic covariates at different spatio-temporal scales and explore their usefulness. We go beyond existing literature by investigating and comparing the performance of recently proposed covariate-dependent models for both the marginal and dependence structures of extremes. Furthermore, a flexible max-id model, which encompasses both asymptotic dependence and independence, is used to learn about the spatio-temporal variability of rainfall processes at extreme levels. We find that modelling non-stationarity only at the marginal level does not fully capture the variability of precipitation extremes, and that it is important to also capture the seasonal variation of extremal dependence. Keywords extreme sub-daily precipitation \cdot spatio-temporal extremes \cdot max-infinitely divisible process \cdot asymptotic dependence and independence \cdot complex multivariate extremes 1 Introduction Extreme precipitation events in northern Italy have become an increasing concern due to their devastating impacts, including significant flooding and landslides, which have major environmental and socioeconomic consequences ( 56 ) . Recent instances of such events underscore the urgency of this issue ( 10 ; 20 ; 21 ) . Detecting trends in extreme precipitation at a national scale remains challenging in Italy, likely due to the country’s diverse climate, complex terrain, and spatio-temporal dynamics of precipitation processes ( 39 ) . Inconsistent patterns across Italy ( 7 ) , suggest that national-scale analyses, based on coarse resolution datasets, may obscure local dynamics caused by the country’s high climatic and geographic heterogeneity. Extreme precipitation in northern Italy is particularly challenging to model due to its unique geographical and climatological setting, influenced by the Alps. Heavy precipitation trends in northern Italy show heterogeneous patterns, highlighting the necessity for localised studies within the region ( 34 ; 17 ; 39 ) . The literature calls for more nuanced local studies of extreme precipitation trends to effectively inform local authorities and address regional vulnerabilities. A localised understanding of precipitation extremes is critical for water management, infrastructure development, and over all risk mitigation ( 52 ) . To estimate the magnitude of rare precipitation events, extreme value theory (EVT) is routinely applied ( 38 ; 14 ; 48 ) . EVT provides a theoretically justified approach to identify the limiting distribution for extremes defined as either block maxima, i.e., the largest value within blocks of a pre-specified size, or peaks over threshold, i.e., all exceedances larger than a certain high threshold. Using probability theory, it is possible to derive the limiting distributions of these extremes: the Generalised Extreme Value (GEV) distribution for block maxima, and the Generalised Pareto (GP) distribution for threshold exceedances ( 12 , see) . EVT has furthermore been adapted to investigate changes in the distribution of extreme rainfall linked to climate change and other climatic drivers ( 35 ) . Several studies propose methods to reparametrise the GEV and the GP distributions to allow non-stationarity in the extremal models ( 23 ; 50 ) , enabling the practitioner to capture trends in the intensity and frequency of extreme events. In northern Italy, the use of EVT has significantly enriched our knowledge of the behaviour and the associated risks of extreme precipitation. For example, 17 identified long-term trends in the region, where some areas have seen increased intensities and frequencies of heavy rainfall, while others have remained relatively stable. 28 emphasises the inadequacy of stationary models of extreme precipitation given the increased variability of extreme rainfall over Italy in recent years. These findings highlight the utility of EVT models as a crucial tool for characterising changes in extreme precipitation patterns. Univariate models fail to capture the inherent spatial dependencies of extremal processes. A natural multivariate extension of the univariate block maxima approach is the max-stable process, which can be seen as an infinite-dimensional analogue of the GEV distribution, developed to describe the limiting behaviour of component-wise maxima in a spatial setting ( 19 ; 18 ) . Max-stable processes provide a popular and widely used framework for modelling the joint extremes of precipitation over space. A major concern with the use of max-stable processes is that they assume a very strong form of extremal dependence, known as asymptotic dependence. This assumption should not be made a priori, since it oversimplifies the dependence structure and may lead to the incorrect estimation of extreme events. It is often observed that environmental processes such as precipitation are not asymptotically dependent and become site-wise independent in the uppermost tail. To account for this, max-id models have been proposed as a flexible alternative to max-stable models. They are able to capture asymptotic independence, i.e., a weakening, eventually vanishing, extremal dependence at increasing quantiles ( 47 ; 5 ; 31 ; 64 ) . Our analysis entails a two-level modelling procedure. Firstly, we model the marginal distributions of the process using a covariate-dependent GEV distribution, accounting for seasonality, temporal non-stationarity and other geophysical influences acting on the extremal process. Secondly, we account for the extremal dependence of the process through a spatial max-id extreme value model. We enrich the original max-id process model by introducing covariate-dependent parameters to deal with the seasonal non-stationary behaviour of the extremal dependence. We find that this more complex model more accurately captures the spatial and temporal structure of the extreme precipitation data. The paper is organised as follows. Section 2 describes the observational data and study region. Section 3 provides an exploratory analysis to identify possible non-stationarities in the data, both marginally and spatially. Section 4 introduces the covariates used to account for the identified non-stationarities. Section 5 outlines the marginal and dependence models. In Section 6 , we present and discuss the results of our analysis. Conclusions and broader discussion are provided in Section 7 . 2 Data We model the spatio-temporal behaviour of monthly maxima of hourly precipitation over the Piave basin, in north-east Italy. The study domain is highlighted in Figure 1 . Monthly maxima are derived from a larger dataset of hourly precipitation recorded by the local environment monitoring agencies (Agenzia Regionale per la Prevenzione e Protezione Ambientale del Veneto, ARPAV 1 1 1 https://www.arpa.veneto.it/dati-ambientali/dati-storici ). The initial dataset consists of records from 65 observational sites between 1990 and 2023, with an average span of 27.8 years. In keeping with standard statistical modelling of precipitation processes, we define the day as starting from 09:00 AM. We remove stations with a substantial number of missing values; specifically, 4 sites with more than 20% missing data are excluded. To avoid deriving inaccurate or biased maxima due to gaps in the data, we omit any month which does not have

Read the complete text