The combined effects of climate warming and increased nutrient inputs from human activities have resulted in increasing frequency and severity of cyanobacteria blooms worldwide. However, in many lakes there is substantial variability in the strength of algal blooms within and between years which is not well understood, complicating efforts to predict and manage future blooms. Furthermore, it is likely that not all cyanobacteria blooms are qualitatively the same class of phenomenon. Developing robust forecasting methods therefore requires understanding the differences in the ecology of bloom events in different types of lakes, and selecting appropriate forecasting methods based on our theoretical understanding of system dynamics.
In this project, we are combining process-based models of lake physics with new machine-learning methods to understand the processes giving rise to the development of cyanobacteria blooms, and to develop hybrid modeling approaches to forecast cyanobacteria blooms in different types of lakes throughout Switzerland. Using these methods, we are building new understanding of the relative predictability of cyanobacteria blooms in different environments, constructing near-term forecasts of bloom occurrence in Swiss lakes, and building long-range forecasts to estimate the frequency of future cyanobacteria blooms in Swiss lakes in response to a variety of climate change and anthropogenic nutrient loading scenarios.