Actual evapotranspiration dynamics in Italy: trend detection and regional clustering analysis

Understanding the spatial and temporal dynamics of Actual Evapotranspiration (AET) is critical for water resource management, agricultural planning, and climate change adaptation. This study investigates AET variability across Italy from 1950 to 2024 using ERA5-Land reanalysis data. A physically-based approach was adopted to estimate AET by adjusting Potential Evapotranspiration (PET), computed via the FAO-56 Penman–Monteith (PM) equation, with soil moisture and vegetation stress factors. Statistical analyses reveal strong latitudinal contrasts: PET increases from north to south in response to climatic demand, while AET exhibits the opposite trend, limited in southern Italy by persistent water scarcity. Clustering analysis using K-means, Hierarchical, and Gaussian Mixture Models (GMM) consistently identified three primary hydroclimatic zones: humid northern, transitional central, and arid southern Italy, corresponding to distinct AET regimes. Among the methods, K-means provided the most robust spatial partitioning. The Seasonal Kendall (SK) test was applied to detect long-term trends in AET. Results show statistically significant positive trends over most of the peninsula, especially in the north and central regions, with Z-values exceeding 4.42 in Alpine zones. Southern areas and the major islands exhibit weaker or non-significant trends, reflecting limitations in moisture availability. This integrated approach highlights the contrasting controls of atmospheric demand and water availability on evapotranspiration. The observed trends suggest intensification of surface water losses in water-abundant areas and persistent constraints in arid zones. These findings emphasize the urgency of region-specific strategies for water and land-use management, particularly in southern and insular Italy where evapotranspiration remains tightly coupled to precipitation variability.