Department of Computational Hydrosystems
The overarching questions/hypotheses that guide the research of the department are firstly that improved predictability of hydrologic models cannot be achieved by hyper-resolution alone. Secondly, improved model predictability can only be achieved with smart models. This means models that are complex enough with effective parameters that take into account the subgrid variability of the land surface and subsurface characteristics. Thirdly, "predictions everywhere" on the spatial scales required for decision-making can only be achieved with smart models with parameters which are properly regionalized so that they exhibit quasi-scale invariance.
The Department works as a computational laboratory devoted to the development, validation and integration of hydrologic models on multiple scales of the water cycle and their interrelationships with terrestrial ecosystems. Spatial heterogeneity of land surface and subsurface characteristics (like soil properties, topography and vegetation) are of central importance for our research. The main challenge is to develop scale-independent hydrologic models able to estimate matter and energy fluxes everywhere on regional scales. Therefore, our challenge is to develop efficient models with less complex parameters so that they are transferable across scales and locations.
At the core of our approach are process-based, distributed models which portray the behaviour of the different natural subsystems. This requires (1) the identification and description of the relevant physical, chemical and biological processes within the interactive system, (2) the ability to translate these processes into conceptual and computational models, taking into account inherent uncertainty, and (3) the development of robust and efficient computational algorithms to solve them (e.g. by complexity reduction, multi-scale codes or hybrid models).