A high-resolution grid-based water budget model for Mediterranean river basin districts: sensitivity to soil heterogeneity and evapotranspiration parameterization

Hydrological dynamics in Mediterranean regions are strongly controlled by temperature-driven evapotranspiration processes and marked spatial heterogeneity in soil properties and land use. In such settings, both the choice of evapotranspiration formulation and the representation of soil water storage play a decisive role in grid-based water budget modeling.

In this study, a recently developed high-resolution (1 km), monthly, grid-based water budget model (HYGRID-M) was applied to investigate the sensitivity of the simulated hydrological components to key methodological assumptions. The model was based on the temperature-driven Hargreaves equation for the estimation of reference evapotranspiration (ETo). The analysis focused on the effects of (i) soil water storage representation, comparing a uniform 1 m soil depth baseline with spatially heterogeneous soil depths; (ii) land use representation, considering both static and time-varying land use across the simulation period; and (iii) evapotranspiration parameterization, comparing the constant Hargreaves coefficient with a regionally calibrated coefficient. The analysis was conducted for the Southern Apennines District over the period 2000–2023. Land use dynamics were represented using static and time-varying configurations derived from the CORINE Land Cover database, while soil physical properties were derived from European Soil Data Centre (ESDAC) datasets and used to estimate soil hydraulic parameters through pedotransfer functions. Model outputs were evaluated by comparing simulated monthly actual evapotranspiration (AET) against independent satellite-based products (GLASS, ETMonitor, and MOD16), as well as against estimates derived from the BIGBANG model.

Results indicated that soil depth heterogeneity was the dominant factor influencing model performance. Compared to uniform soil depth assumptions, heterogeneous configurations improved agreement with reference AET datasets, reducing MAE and RMSE by ~2 and ~3 mm month⁻¹, respectively, and yielding higher KGE values. Although district-scale runoff metrics exhibited limited sensitivity, monthly runoff (Q) varies by up to ~30% in response to soil depth, particularly in winter. Land use dynamics further affected both AET and Q, with monthly Q variations reaching ~45%, whereas evapotranspiration parameterization had a comparatively minor impact, with differences of approximately 5%. Overall, these findings highlighted the critical importance of explicitly representing spatial heterogeneity in soil water storage and land use dynamics for improving large-scale water budget simulations in Mediterranean environments.