Preview

PDF, English Download (10MB) | Terms of use

Abstract

To improve understanding of how climate change affects forested catchment hydrology through interactions between climate, vegetation, and water balance processes, this dissertation investigates historical and projected hydrological dynamics in the Eyach catchment, a 52 km² forested watershed in Germany’s northern Black Forest. A comprehensive multi-model approach is applied using four hydrological models—TRAIN, HBV-Light, LWF-Brook90, and Raven-HMETS—to simulate key components of the water cycle including evapotranspiration, soil moisture, streamflow, and snow cover duration. The analysis spans the period 1975–2050, combining a historical climate and hydrology assessment (1975–2019) with future projections (2020–2050) under RCP 2.6, 4.5, and 8.5 climate scenarios. Historical climate data reveal statistically significant warming and rising solar radiation, while precipitation trends remain non-significant. These atmospheric shifts are reflected in the hydrological simulations: evapotranspiration and soil moisture show strong interannual variability, but clear long-term drying signals are found in soil moisture. A multi-model ensemble represents structural diversity across hydrological process models. TRAIN delivers high-resolution evapotranspiration and soil moisture simulations; HBV-Light provides robust discharge estimates; LWF-Brook90 allows species-specific representation of canopy fluxes and soil processes; and Raven-HMETS best captures interannual variability in snow cover dynamics, though it tends to underestimate total snow duration. Model evaluation against observational and satellite-based datasets—including GLEAM, ERA5 and snow cover records—demonstrates that each model performs differently depending on process and variable. TRAIN exhibits the most balanced performance in simulating snow cover duration, with the lowest root mean square error and minimal bias. Raven-HMETS, while underestimating absolute snow duration, achieves the highest correlation with observed interannual variability—highlighting how structural features such as refreezing algorithms and spatial discretization influence snow dynamics across models. Species-specific simulations using LWF-Brook90 reveal clear hydrological differences between Norway spruce (Picea abies), European beech (Fagus sylvatica), and oak (Quercus robur). Deciduous species consistently show lower evapotranspiration and higher soil moisture retention than evergreen conifers under identical conditions, highlighting the role of forest composition in drought resilience and water availability. Future scenario analyses indicate that under high-emission conditions (RCP 8.5), evapotranspiration increases modestly but begins to plateau, while soil moisture declines significantly due to enhanced atmospheric demand and warming-induced limitations on water availability. While Evapotranspiration projections from TRAIN and LWF-Brook90 converge, their soil moisture outputs diverge—emphasizing the importance of internal model structure in shaping subsurface hydrological sensitivity. Overall, this study fulfils six integrated objectives: from historical trend analysis to multi-model evaluation, species-specific simulation, and scenario-based forecasting. The findings underscore the dual importance of climatic drivers and vegetation composition in shaping catchment hydrology. The integrative approach—including model validation, species comparisons, and scenario simulations—provides a robust foundation for understanding and managing water dynamics in temperate forested regions under climate change.