Projects
BayWaT-AI upcoming
Process- and AI-based modelling of water temperature in Bavarian rivers, today and tomorrow — tools for Bavarian water management.
Funder: Bavarian State Office for Environment (LfU).
Rising water temperatures are one of the central climate risks for Bavarian rivers. Warmer water holds less oxygen, accelerates the turnover of nutrients and pollutants and reduces a river’s capacity to purify itself, and these effects are sharpest during low-flow phases, when temperature and sensitivity to additional stressors rise together. The consequences reach from aquatic ecosystems to energy supply, since many power plants depend on river water for cooling.
BayWaT-AI combines process-based and data-driven models to predict and project river temperatures across Bavaria. My part sits on the data-driven and uncertainty side: comparing the skill and spatial transferability of both approaches, developing regionalisation and transfer-learning strategies so that catchments with sparse measurement records can still be modelled credibly, and quantifying how much confidence the resulting projections warrant. Spatially explicit drivers such as land use, riparian shading and river morphology are accounted for, so that regional differences in vulnerability are resolved rather than averaged away. The project delivers a tool that the LfU can operate itself, combining long-term projections, short-term forecasts and uncertainty estimates.
ClimEx II
One of my largest projects to date. ClimEx II studies low-water extremes and drought under climate and land-use change across Bavaria (2020–2025), run by LMU Munich together with Québec partners, building on the earlier ClimEx project. Within its climate module, my contribution centred on using single-model large ensembles (SMILEs) to decompose climate model output into forced signal, bias and internal variability, with application to drought: a multivariate bias correction method based on zero-inflated vine copulas (VBC), the application of a European standardised drought index, and a deep-learning drought forecasting framework that propagates ensemble-derived uncertainty into probabilistic forecasts. For scientific contributions, see the publications.
Funders: Bavarian State Ministry for the Environment and Consumer Protection, and the Bavarian State Office for Environment (LfU).
Statistical Consulting (StaBLab)
As a member of StaBLab, LMU’s Statistical Consulting Unit, I advise Bachelor, Master and PhD students as well as researchers on the statistical side of their work, from study design and analysis planning through to modelling, interpretation and reporting.
Applied projects have ranged from eye-tracking studies for linguists and research projects in veterinary and human medicine to many other consultations across disciplines. For a while I also maintained the Blindgänger project, which maps risk zones for unexploded World War II bombs using spatial point processes.
Funder: Central University Administration (Zentrale Universitätsverwaltung), LMU Munich.