Being the most widely used generators of multifractal measures, multiplicative cascade models have been extensively applied in the field of geophysics, and particularly in hydrometeorology. As in any modeling effort, solving the "inverse problem" is essential, and in this case, it can be described as finding the appropriate cascade model that generates a given multifractal measure. Direct measurement of a generated field (e.g., a rainfall field, or a time series thereof) results in an immediate decomposition into breakdown coefficients, producing a microcanonical (strictly normalized) multiplicative cascade over a limited range of scales. Yet, the canonical (expectation-normalized) phenomenology at underlying scales may generate statistical properties that are non-trivial to reproduce. The present work analyzes such properties for the simplified case of a one-dimensional, beta-lognormal discrete multiplicative cascade.

Given the importance of breakdown coefficients (BDCs) for the parameterization of multifractal processes, whose distributions attract, under multiplication, all weights exhibiting lacunarity and finite-moment logarithms, this paper elaborates an analytical approximation for the probability distribution of BDCs in binary beta-lognormal processes. This case is not covered by exact results in the existing literature, notwithstanding its central role, which is due mainly to the attractive property of the lognormal distribution under multiplication of independent weights.

Global warming is significantly altering precipitation patterns in the Mediterranean Basin, a region already susceptible to water scarcity and extreme weather events. Changes in precipitation can exacerbate both drought conditions and soil degradation caused by increasingly erosive rainfall. Where wet season patterns mismatch peak irrigation demands, shifts in rainfall seasonality may pose a serious threat to agricultural sustainability. This study analyses century-long (1922-2022) trends in three regions of Southern Italy representative of Mediterranean semi-arid areas (i.e., Campania, Sardinia, and Sicily), using a wide spectrum of rainfall regime indicators. A novel methodological framework is applied to an extensive database of daily rainfall time series from over 700 rain gauges, using both point-based (at gauge level) and spatially distributed analyses. This approach ensures the robust detection and localization of trends in precipitation characteristics, with a particular focus on rainfall seasonality. Results indicate consistent temporal variations across the three regions for most indicators, including reduction in mean annual precipitation, frequency and length of dry and wet spells. Crucially, the study area exhibits reduced rainfall seasonality compared to previous historical periods, with earlier onset and longer duration of the wet season. These shifts have profound implications for water resource management, agricultural planning, and ecosystem resilience across the Mediterranean. Furthermore, spatial autocorrelation analysis highlights that some trends are highly localized, identifying priority areas for the implementation of targeted climate change adaptation strategies.

Multifractal cascades, as the generic constructors of multifractal measures, have been widely used in hydrometeorology and other areas of geophysics. In order to parametrize the cascade models, the inverse problem in this case consists in finding the generator parameters, given a realization of the multifractal measure. From the direct measurement of a field or time series (e.g. rainfall), its decomposition into breakdown coefficients is immediate, producing a microcanonical (deterministically normalized) multiplicative cascade over a certain range of scales. However, the canonical (stochastically normalized) construction at underlying scales may generate statistical properties that are not easy to reproduce. This work analyzes those properties for the case of a one-dimensional, beta–lognormal discrete multiplicative cascade.

Agricultural heritage is a cultural pillar of the Mediterranean region, where durum wheat plays a central role in traditional landscapes and food systems. Projected climate change is expected to alter crop productivity and place additional pressure on water resources. This study assesses future variability in durum wheat productivity and related implications for water resource management in Sardinia, Italy, where durum wheat is a major rainfed C3 crop. The AquaCrop-OpenSource model was calibrated to local conditions and applied to simulate historical (1950–2023) and near-future (2024–2050) scenarios using projections from seven climate models. Results indicate a modest increase in average yields under future conditions, accompanied by a higher frequency of crop failures. Elevated atmospheric CO2 concentrations emerge as the primary driver of yield increases, while changes in precipitation represent the main limiting factor. The role of aid irrigation as an adaptation strategy to stabilize yields and enhance productivity was evaluated. Scenario analysis shows that aid irrigation aimed at preventing crop failure remains sustainable in the near future, requiring approximately 14–17% of current agricultural water use in Sardinia. In contrast, irrigation used to maximize productivity would increase water demand by more than 40%, intensifying competition for water resources.

Predicting precipitation maps is a highly complex spatiotemporal modeling task, critical for mitigating the impacts of extreme weather events. Short-term precipitation forecasting, or nowcasting, requires models that are not only accurate but also computationally efficient for real-time applications. Current methods, such as token-based autoregressive models, often suffer from flawed inductive biases and slow inference, while diffusion models can be computationally intensive. To address these limitations, we introduce BlockGPT, a generative autoregressive transformer using batched tokenization (Block) method that predicts full two-dimensional fields (frames) at each time step. Conceived as a model-agnostic paradigm for video prediction, BlockGPT factorizes space-time by using self-attention within each frame and causal attention across frames; in this work, we instantiate it for precipitation nowcasting. We evaluate BlockGPT on two precipitation datasets, viz. KNMI (Netherlands) and SEVIR (U.S.), comparing it to state-of-the-art baselines including token-based (NowcastingGPT) and diffusion-based (DiffCast+Phydnet) models. The results show that BlockGPT achieves superior accuracy, event localization as measured by categorical metrics, and inference speeds up to 31x faster than comparable baselines.

This study quantifies the uncertainty in monthly streamflow predictions under future climate scenarios in two periods (near and far future) for the Riu Mannu di Narcao basin in Sardinia, Italy. The sources of uncertainty include the hydrological model structure, model parameters, and variability in climatic inputs derived from global and regional climate models (GCM-RCM coupling) and representative concentration pathways (RCPs). Three conceptual and lumped hydrological models (GR3M, ABCD, and IHACRES) were combined with four climate models and two RCPs (RCP 4.5 and RCP 8.5) to assess future streamflow. Monte Carlo simulations were performed to evaluate parameter uncertainty, and the analysis of variance (ANOVA) method was applied to quantify the different sources of uncertainty. The results reveal that, as a single source, GCM-RCM coupling is the largest contributor, accounting for 47.32% (54.64%) of total near (far) future monthly streamflow projection uncertainties, followed by the hydrological model structure at 16.02% (21.09%), RCP scenarios at 15.35% (8.54%), and parameter uncertainty at 0.79% (1.39%). A consistent decline in median monthly streamflow is projected, especially during winter months (December to February), raising a concern about water availability in the region. Our study quantified different sources of uncertainty in monthly streamflow predictions under climate change, disentangling the roles of the hydrological model, model parameters, climate model, and climate scenario for reliable future streamflow projections.

Many different nature-based solutions have been proposed in the literature to contribute to the sustainable development of the urban environment. Among them, multilayer blue-green roofs are becoming more and more popular, thanks to their multiple benefits. As traditional green roofs, the multilayer ones guarantee high retention capacity during rainfall events, contributing to the pluvial flood mitigation. Thanks to the additional storage layer, not only the mitigation capacity is increased, but there is the possibility to store the collected water, and reused it for some urban purposes, such as garden irrigation. Moreover, these nature-based solutions ensure thermal insultation for the underneath buildings and they help lowering the air temperature, contributing to the mitigation of the urban heat island effects. Finally, they improve the air quality, promote the biodiversity, and increase the aesthetic value of the overall city. In June 2019, a multilayer blue-green roof prototype has been installed at the university of Cagliari, and subsequently equipped with multiple sensors to monitor and evaluate the ecohydrological and thermal dynamics. The multilayer blue-green roof, with a surface of 16 m2, presents an 8 cm layer of soil, classified as sand, and a 10 cm additional storage layer. It is characterized by Cactaceae vegetation, which shows resistance to the high temperature and low water availability and does not require additional maintenance. The prototype has been equipped with a Smart Mill, that beside opening and closing of the valve to control the storage layer, enables to measure climatological variables, such as rainfall, air temperature and wind speed, and the water level in the additional layer. Four HOBO thermometers have been installed to measure the temperature in the soil, underneath the structure and on the lateral side. Two soil moisture sensors have been placed at opposite sides of the multilayer blue-green roof. Finally, a tank with a sensor to measure the water level have been collocated at the valve opening, to measure the outflow from the additional storage layer. The collected data have been used to model the ecohydrological and thermal dynamics, with the aim to quantify the potential benefits in terms of pluvial flood mitigation and thermal insulation. Results, collected during two full years of monitoring the prototype in Cagliari, are discussed, highlighting the potential benefits of a large-scale installation for the sustainable development of urban areas.

Durum wheat is a critical staple crop in arid and semi-arid regions worldwide, that plays a significant role in local food security. Providing essential nutrients and a high protein content, it is widely used for the production of pasta and couscous. Various constraints and drivers affect durum wheat productivity, including biotic and abiotic stressors, agronomic practices, and CO2 concentrations. Their influence varies based on duration and intensity of the stressor, as well as the durum wheat growth phase in which they occur. Drought and heat were shown to act as primary yield limiting factors. Furthermore, the water footprint, a comprehensive measure for the volume of water associated with crop production, helps to analyse durum wheat cultivation from a water-food nexus perspective. Given that climate change is affecting the main influencing factors of durum wheat’s productivity and of its water footprint, such as precipitation, temperature, and atmospheric CO2 levels, its cultivation is expected to undergo alterations as well. In this context, we explore the present state of durum wheat productivity and the potential influence of changing climatic conditions on its future cultivation worldwide. The current state of research on future durum wheat production is characterised by contradictory results, compromising projections of significant declines due to heat and drought stress as well as strong increases in productivity as a consequence of the CO2-fertilisation effect, for the same or nearby locations. Understanding the complex interactions between climate change, durum wheat productivity and the associated water footprint is of great importance to derive sustainable adaptation strategies and move one step closer into ensuring future food and water security.

Convection-permitting climate models have the potential to capture crucial processes in the climate system, presenting an opportunity to significantly enhance climate projections by providing more accurate representations of precipitation extremes. In this work, we conduct an evaluation of the accuracy of sub-daily precipitation extremes obtained from VHR-PRO_IT (Very High-Resolution PROjections for Italy, Raffa et al., 2023) over the Italian peninsula,. VHR-PRO_IT is generated through dynamic downscaling of the Italy 8km-CM climate projection at approximately 2.2 km resolution under the IPCC RCP4.5 and RCP8.5 scenarios, employing the Regional Climate Model COSMO-CLM.Gauged locations are used to assess the accuracy of VHR-PRO_IT in reproducing observed extremes. More specifically, the observed dataset used as ground truth for the comparison is I2-RED (Improved Italian – Rainfall Extreme Dataset; Mazzoglio et al., 2020). For this work, 742 rain gauges covering the entire country with a minimum of 30 years of short-duration (1, 3, 6, 12, 24 h) annual maximum rainfall depths recorded from 1980 to 2022 are used. Conversely, the dataset derived from the VHR-PRO_IT climate projections includes annual maxima from a 30-year time series, connecting the historical period (1981-2005) with 5 years of the RCP8.5 scenario (2006-2010) of the CPM. Return levels are obtained for both dataset by means of a GEV distribution and inform the assessment of the CPM simulations. Preliminary results outline the quality of the CPM simulations, especially at 24 hours duration, and show the impacts of return period, seasonality, elevation, latitude and proximity to the sea on the CPM model deviations. The results from this work are expected to have implications for both water resources management and adaptation measures.ReferencesMazzoglio P., Butera I., Claps P. (2020). I2-RED: a massive update and quality control of the Italian annual extreme rainfall dataset. Water, 12, 3308.Raffa M., Adinolfi M., Reder A., Marras G.F., Mancini M., Scipione G., Santini M., Mercogliano P. (2023). Very High Resolution Projections over Italy under different CMIP5 IPCC scenarios. Scientific Data, 10, 238.

Rainfall erosivity is a fundamental parameter in estimating soil erosion as it quantifies the potential of raindrops to detach soil particles and make them available for subsequent transport by surface runoff. Erosivity depends mainly on the intensity, duration, and energy of precipitation events, which directly affect the impact of raindrops on the soil surfaces and runoff. The most common methods for identifying erosive events emphasize short-duration, high-intensity rainfall events, while introducing critical thresholds for characterizing erosive events, such as the 30-minute maximum rainfall intensity (I30) and storm separation criteria (see e.g. Wischmeier and Smith, 1978, Foster et al., 1981 and Renard et al., 1997).Nevertheless, both historical and recently proposed frameworks occasionally consolidate rainfall events that should be disaggregated according to the established six-hour dry period threshold, leading to overestimation of rainfall event durations and erosivity factors. The present study aims at refining the identification and analysis of erosive rainfall events, a key component of soil erosion prediction, by introducing an alternative approach that strictly adheres to the original criteria introduced by Wischmeier and Smith (1978) and Renard et al. (1997), ensuring precise segmentation of rainfall events when rainfall accumulation is below the 1.27 mm threshold over a six-hour period.The proposed method classifies rainfall events as erosive when total rainfall exceeds 12.7 mm or meets intensity thresholds of 6.4 mm in 15 minutes or 12.7 mm in 30 minutes. Comparative analysis with existing approaches demonstrates improved alignment with fundamental criteria while addressing modern computational challenges, contributing to the advancement of soil erosion prediction by bridging historical methodologies with contemporary analytical precision.ReferencesWischmeier, W.H., Smith, D. D. (1978) Predicting rainfall erosion losses: A guide to conservation planning. Agric. Handb. 537. US Gov. Print. Office, Washington, DC.Foster, G.R., McCool, D.K., Renard, K.G., Moldenhauer, W.C. (1981) Conversion of the universal soil loss equation to SI metric units. J. Soil Water Conserv. 36, 355–359.Renard, K., Foster, G., Weesies, G., McCool, D. and Yoder, D. (1997) Predicting Soil Erosion by Water: A Guide to Conservation Planning with the Revised Universal Soil Loss Equation (RUSLE). US Department of Agriculture, Agriculture Handbook No.703USDA, USDA, Washington DC.

The assessment of rainfall erosivity is often hindered by the limited availability of high-resolution rainfall data. A large dataset, comprising 10-minute rainfall data collected over the last two decades from 335 rain gauges across three regions of southern Italy, was utilized in this study to estimate benchmark values of mean annual rainfall erosivity according to the Revised Universal Soil Loss Equation. A set of ten existing simplified models based on coarser resolution rainfall data (from daily to annual) were compared to two newly developed empirical models based on daily-resolution data. The first proposed model uses two compound meteorological predictors, namely the rainfall episodicity and intensity indices, derived from the mean annual rainfall, the Gini's coefficient, and the mean annual number of rainy days. The second model integrates the previous one with geographic and topographic covariates, including latitude, elevation, and minimum distance to the coastline. We evaluated and compared the performances of all models using various metrics, including the Root Mean Square Error (RMSE), the mean error, the adjusted coefficient of determination, the Kling-Gupta efficiency, and the Akaike information criterion. All the performance indices showed how the newly developed models outperformed the ten existing recalibrated equations, obtaining a reduction in absolute percentage error from 27 to 18 %. Our extensive dataset enabled robust calibration and validation of existing and new simplified models, paving the way for an ensemble modeling approach that enhances model transferability in data-scarce environments under similar precipitation regimes.

The magnitude and frequency of heavy precipitation are expected to increase under warming temperatures caused by climate change. These trends have emerged in observational records but with much larger evidence on a daily rather than a subdaily scale. Here, we quantify long‐term changes in heavy precipitation frequency in the United States using hourly observations in 1949–2020 from 332 gauges. We demonstrate that, when analyzed collectively, the frequencies of heavy precipitation at multiple durations from hourly to daily exhibit an increase that cannot be explained by natural climate variability. Upward trends are significant at ∼20%–40% of the gauges throughout the country except for the coastal western and southeastern regions, with higher percentages for longer durations. We also show that the frequency of hourly heavy precipitation has mainly grown after ∼2000, thus explaining the limited evidence of trends at the subdaily scale reported in past studies.

Climate change is changing the intensity and frequency of extreme precipitation. Understanding the impact of climate change on extreme precipitation quantiles is fundamental for managing flood risk and taking adaptation measures. Convection-Permitting Models (CPM), run at spatial resolutions for which deep convection is resolved (≤ 4 km), have been demonstrated to be more accurate than Regional Climate Models (RCM, ~10 km resolution) in describing the intensity of extremely short-duration events.This study uses the projections of a CPM to evaluate quantiles of precipitation extremes at the national scale (Italy) with a high spatiotemporal resolution. Indeed, VHR-PRO_IT, a recent downscale product of the CMCC model at a convection-permitting scale of 2.2 km, with 1h temporal resolution, is used as a dataset. So far, this is the only CPM projection that covers the entire Italy in both emission scenarios (RCP 4.5 and RCP 8.5) and for a temporal coverage of 90 years (1981-2070).A non-stationary implementation of the Simplified Metastatistical Extreme Value (SMEV) non-asymptotic approach is used to evaluate continuous changes in precipitation quantiles for different durations (1h, 3h, 6h, 12h and 24h) over the period 1981-2070 (1981-2005 historical + 2006-2070 emission scenarios). We adopt a two-parameter Weibull distribution to model the marginal distribution of the ordinary precipitation events. Three different models are compared: i) a stationary SMEV, with the two parameters constant over the entire time series; ii) a non-stationary model in which the higher-order parameter is kept constant; iii) a fully non-stationary model in which both parameters are allowed to change linearly in time.The results show a clear geographical organization of the projected changes, with both increases and decreases in precipitation quantiles depending on the zone, the emission scenario, the precipitation duration and the return period of interest. The non-asymptotic approach allows us to discuss the results in terms of dynamic and thermodynamic drivers. The research is carried out within the RETURN – multi-Risk sciEnce for resilienT comUnities undeR a changiNg climate Extended Partnership and received funding from the European Union Next-GenerationEU (National Recovery and Resilience Plan – NRRP, Mission 4, Component 2, Investment 1.3 – D.D. 1243 2/8/2022, PE0000005).

Study region: The Arno River basin (Tuscany, Italy), located across the Northern Apennine chain. Study focus: A parsimonious approach for calibration of a multiplicative random cascade model for multifractal rainfall disaggregation is tested and investigated. The performance evaluation in the case study area revealed a particular behavior of the applied procedure for canonical disaggregation of daily rainrates to sub-daily levels; consequently, we introduce an improvement to the suggested approach, and we evaluate its effectiveness. New hydrological insights for the region: The key advancement of the proposed approach is that it allows for parsimonious estimation of the canonical multiplicative random cascade framework parameters (specifically, Cβ and CLN) using only daily rainfall time series data and the power-law exponent n of commonly used intensity-duration-frequency curves. The case study analysis suggests that the proposed method can be effectively adopted in the Arno River basin, simplifying and improving the simulation of sub-daily rainfall time series while also enhancing its practical applicability, as the required inputs are generally available to most users.

Increasing empirical evidence has been showing that, over the last decades, the frequency of daily heavy precipitation has risen in some regions of the United States (U.S.); less evidence has instead been presented at subdaily resolutions. In this study, we describe the challenges and opportunities associated with the detection of trends in subdaily heavy P in the U.S. using Version 2 of the Hourly Precipitation Data (HPD) from the National Climatic Data Center (NCDC). This dataset comprises records from 1897 gages which we found to be affected by several issues preventing their use in trend studies, including long periods with missing observations, changes of instruments, and different signal resolutions (largely, 0.254 and 2.54 mm). Despite this, after proper checks, we were able to identify 370 gages with ≥40 years of statistically homogenous data in 1950-2010 that cover the U.S. with a good density. To improve the ability to detect trends, we designed a framework that quantifies the degree to which the observed over-threshold series above a given empirical q-quantile are consistent with stationary count time series with the same marginal distribution and serial correlation structure as the observations. We also applied the false discovery rate test to account for spatial dependence and multiplicity of the local tests. Analyses were performed for the signals aggregated at Δt = 1, 2, 3, 6, 12, and 24 h and for q = 0.95, 0.97, and 0.99, finding that most gages exhibit increasing trends across all Δt’s and that their statistical significance increases with Δt and decreases with q, but only for Δt ≥ 2 h. This might indicate that the physical generating mechanisms of precipitation have changed in a way that leads to larger accumulations over durations >1 h but similar intensities within 1 h. An alternative possible explanation for these outcomes is instead that the coarse signal resolution (2.54 mm) reduces the power of the test for trend detection as Δt decreases. Investigating these issues will be the subject of our immediate future work.

Global warming may induce significant alterations to the rainfall regimes, especially in the Mediterranean basin, which can be considered as a hot-spot for climate change. Several previous studies focused on the variations in annual rainfall and extreme values, while rainfall seasonal variations were less explored. Rainfall seasonality is a critical climate factor affecting the evolution of natural vegetation, water resource availability, and water security. Rainfall seasonality anomalies may have a high impact, especially in areas of the Mediterranean basin where water supplied during the wet season is used to offset rainfall shortages in the dry season. In southern Italy, the occurrence of long water deficit periods and extremely concentrated rainy seasons could limit water uses and cause serious effects on crop yield and, consequently, on food production. This study aims at exploring potential variations in rainfall seasonality over the last 100 years across three regions of southern Italy (Campania, Sardinia, and Sicily) through a dynamic approach proposed by Feng et al. (2013). The study area is characterized by a Mediterranean climate, where the hydrological year consists of a net alternation of two seasons: a cold-rainy period (wet season), usually including fall-winter months, and a hot-dry period (dry season), typically including spring-summer months. The analysis proposed involves the determination of time-variant values of rainfall magnitude and frequency of the two seasons (wet and dry). Daily rainfall values, recorded between 1916 and 2023, are gathered from hundreds of rain gauge stations distributed over the three regions. A pre-processing procedure was applied for data quality check, data reconstruction in years with less than 80% of missing data, and rain gauge selection; then, only rain gauge datasets with adequate data availability (i.e., more than 70 complete years, with at least 15 years in the last two decades, 15 years in the pre-World War II period, and without significant data interruptions) were retained and used for data analyses. Rainfall depth over each season is idealized as an exponentially distributed independent random variable with mean values h (mm), whereas the seasonal rainfall occurrence is modelled as a Poisson process with rate l (d-1). Rainfall seasonality at each rain gauge was defined annually, considering different indices: the Dimensionless Seasonality Index (DSI); the seasonal rainfall depth and the seasonal values of h and l; the wet season timing (i.e., centroid of the season) and duration. The reference period was divided into different equal-size and non-overlapping subperiods. Differences in the various rainfall seasonality indices and their distributions among the various gauges, regions, and subperiods were analyzed, also investigating the influence of some climatic and topographic factors (i.e., temperature, gauge distance from the sea and elevation). A trend analysis based on Mann-Kendall's and Sen's Slope Method with statistical significance at 95% level of confidence, was also carried out considering a limited subset of gauges with the largest data availability for each region.

Convection-permitting climate models (CPMs) represent a significant advancement compared to regional climate models, enabling more accurate simulations of extreme precipitation at fine spatial and temporal scales. Assessing the reliability of CPM projections for extreme short-duration precipitation requires understanding how well CPMs reproduce observed extremes—especially in Mediterranean regions, where such evaluations are rare. In this study, we assess the accuracy of simulations from a high-resolution CPM covering the entire Italy (VHR-PRO_IT), in reproducing sub-daily precipitation extremes. For this, we exploit observations from I2-RED, a comprehensive dataset of more than 5 000 quality-checked annual maximum time series from rain gauge observations. The comparison is performed by considering the median values of the annual maxima at 1, 3, 6, 12 and 24-h as a first step and rainfall quantiles up to 200-year return period as a second step. Our results show that model performance is influenced by both the distance from the coastline and elevation, highlighting an important role of orography and land-sea contrast in explaining CPM biases. Moreover, we find better performances when longer duration extremes are considered, while shorter durations are affected by strong underestimations, especially in coastal and low-elevation areas. These results hold significant implications for stakeholders and policymakers dealing with climate adaptation and flood risk management.

The need for long-term synthetic sub-daily rainfall time series is crucial in various hydrological applications, particularly in flood frequency analysis. Traditional sub-daily rainfall simulation models rely on high time-resolution data, typically spanning only 20–30 years, which is insufficient for generating the long synthetic time series required for high return period design value estimation. In contrast, longer datasets of daily rainfall records and annual maximum values are more widely available, often covering 50–80 years. These datasets underpin the derivation of Intensity-Duration-Frequency (IDF) curves, a cornerstone of current hydrological practice.This study introduces an innovative framework for simulating sub-daily rainfall time series using only daily rainfall records and IDF curves, thus eliminating the need for sub-daily observational data. The approach integrates a daily rainfall simulation model, Complete Stochastic Modelling Solution, calibrated with observed daily data, with a multifractal disaggregation scheme informed by IDF curves. The resulting framework offers a robust and parsimonious solution for generating sub-daily rainfall data.By leveraging readily available datasets, this method expands the applicability of sub-daily rainfall simulations to a broader range of hydrological and climate modeling contexts, providing a valuable tool for advancing flood frequency analysis and related applications.

Assessing the impact of climate change on extreme precipitation is a critical task for adapting flood mitigation strategies. Nowadays, Convection Permitting Models (CPM) are the state-of-the-art tool to simulate current and future extreme precipitation, as they explicitly represent convective processes, outperforming climate models with coarser resolutions. Nevertheless, CPM are computationally demanding and typically limited to decadal periods, making them potentially more affected by internal variability. Here, we exploit for the first time long-term CPM simulations over Central-Southern Europe to develop a nonstationary analysis of extreme precipitation. We use VHR-PRO_IT (Very High-Resolution PROjections over Italy), recent CPM projections with high spatiotemporal resolution (2.2 km and 1 h) and simulations spanning 90 years in two emission scenarios. We apply a non-asymptotic statistical approach (SMEV) with a non-stationary implementation using time as a covariate. This approach allows us to capture the transient change in precipitation extremes and in their distribution parameters, describing trends beyond the natural variability, which could affect short (decadal) simulations. We find a general increase of sub-daily extreme precipitation in a warmer climate, especially in the Mediterranean areas. Larger areas with statistically significant changes are found at shorter durations. Higher increases are expected with higher return periods (median value of about 3-4 % / decade at 100 yr return period), especially for longer durations. This can be explained by the increasing distribution tail heaviness found at those durations. Average expected changes in extreme precipitation are similar in the two scenarios, but larger nonstationary areas are expected in the RCP8.5.