Co-variability and relative strength of hydroclimatic drivers in the Mediterranean region

The Mediterranean region is often described as a “climate change hotspot” in model projections due to pronounced warming signals. However, recent empirical analyses indicate that the hydrological response to regional warming is more nuanced and complex than represented by climate models. In this study, we examine the co-variability of several atmospheric and land–surface drivers that influence the behaviour of key hydroclimatic variables—precipitation, temperature, and evaporation—across the Mediterranean domain. The analysis is based on the ERA5 reanalysis dataset and explicitly distinguishes between land and sea domains to account for their differing dynamical and thermodynamic characteristics. To assess the strength and structure of associations, we employ complementary approaches including feature-importance metrics from machine-learning models and a revised formulation of the impulse response function based on the stochastic covariance structure, suitable for hydroclimatic dynamics. We investigate how different drivers relate to each other across space and scales, and we discuss methodological implications for developing more reliable hydroclimatic scenarios for water-resources and climate-impact studies.