Systematic investigation of the numerical methods and parameters in the SurEau model Betreuung: Gregor Rickert The SurEau model contains a variety of options to specify the numerical solution: time integrators (i.e. explicit, semi-implicit, implicit), time steppers (i.e. fixed, variable, adaptive), stability criteria, as well as (de-)activators of specific terms in the equations. The student will conduct a systematic investigation on these options and their role in the numerical solution as well as the output of the model.
Judging Solver Performance - Metrics, Benchmarks and a Hands-On Example Betreuung: Vera Brandtner Investigate how solver performance should be assessed by examining the strengths and limitations of common metrics and benchmark choices. Showcase your analysis by evaluating a provided solver for the 1D Richards equation with benchmarks from the established HYDRUS software. The focus and details of this seminar project can be discussed with your supervisor.
Modelling the effect of surface sealing on ecohydrological processes at Schlosspark/Schloss-Arkaden, Braunschweig Betreuung: Ilhan Özgen In 2005, the city of Braunschweig let its 3.5-hectare Schlosspark to be turned into the Schloss-Arkaden shopping centre. How did this change in land-use affect (eco)hydrological signatures, such as evapotranspiration, infiltration, and runoff, at the site? In this project, we will set up a parsimonious bucket hydrological model to investigate the site's precipitation partitioning across its successive land-cover states.
Identifying ecosystem-level traits across different stands Betreuung: Ilhan Özgen Do ecosystems have their own functional traits? And if so, can we identify them? In this modelling study, use a plant hydraulic model and inverse modelling techniques to investigate this issue. We will generate in silico stands that follow a certain trait distribution and try to inversely estimate them. We will then see whether these ecosystem-level traits can capture the "real" evapotranspiration of the whole stand.