Multi-scale impacts of climate change on flash floods in a heterogeneous, mixed land-use Mediterranean catchment

Flash floods are a major natural hazard in Mediterranean regions, causing significant damage to property, infrastructure, and loss of life. Climate change plays a crucial role in altering rainfall patterns, thereby directly affecting flash-flood behavior. The Mediterranean, a recognized climate change hotspot, is expected to experience more intense extreme rainfall events alongside decreasing total rainfall, both of which may influence flash-flood severity, with responses further modulated by land-use characteristics. Despite substantial research efforts, key gaps remain in understanding flash floods across scales, particularly regarding event-based assessments using high spatiotemporal resolution distributed models capable of capturing flash-flood dynamics in heterogeneous catchments and their sensitivity to climate-driven rainfall changes across catchment sizes, land-use types, and local rainfall characteristics.

This study addresses these gaps by investigating flash-flood behavior in the large Mediterranean Yarkon–Ayalon catchment, located in central Israel, covering 1,800 km². The catchment is characterized by pronounced spatial heterogeneity. The upper part is mountainous and dominated by natural and forested areas on highly permeable Terra Rossa soils, resulting in high infiltration rates. In contrast, the lower part of the catchment is flatter and characterized by lower infiltration rates due to heavy Grumusol soils underlying extensive agricultural land and widespread urban development, with built-up areas covering approximately 70% of the area, promoting rapid runoff generation during rainfall events. A unique streamflow network in the catchment includes 14 hydrometric stations spanning a wide range of spatial scales (7–953 km²) and dominant land use, enabling a multi-scale, multi-land-use evaluation of flash-flood response.

We employ the Grid-Based Hydrological Distributed Runoff (GB-HYDRA) model, an event-based, high-resolution (100 m, 5 min) hydrological model, developed to capture runoff and flash-flood dynamics. The model’s input includes high-resolution radar rainfall data, and it computes runoff at each grid cell and streamflow at any channel cell. To calibrate and evaluate model performance, 37 historical flash flood events with varying intensities and durations are simulated. Of these events, 24 were used for calibration and 13 for independent validation, and 5 hydrometric stations are excluded from calibration, allowing a fair evaluation of the model’s ability to simulate streamflow in ungauged locations. Calibration is performed using a multi-objective optimization approach, resulting in moderate overall model performance, with KGE values of approximately 0.75 for runoff volume and 0.70 for peak discharge across stations and spatial scales.

As a next step, we utilize high-resolution rainfall simulations for a set of storms, derived from the Weather Research & Forecasting (WRF) model under historical conditions and end-of-century projections (RCP8.5), as input to the calibrated hydrological model. The analysis focuses on comparative changes in flash-flood properties across different parts of the catchment and as a function of spatial scale and dominant land use. The results will provide insight into the processes linking changing rainfall patterns to flash-flood response, advancing understanding of flash-flood dynamics across scales in Mediterranean catchments and supporting improved flash-flood risk assessment under climate change.