Taking advantage of model diversity: benefits of ensemble modelling for managing algal blooms in polluted lakes

Climate change and increasing anthropogenic stress have intensified the occurrence of nuisance algal blooms worldwide. Toxic and highly adaptive blooms of cyanobacteria can threaten drinking water safety and break down ecosystem functions by suppressing aquatic macrophytes. To prevent the deterioration of aquatic ecosystems, ecological models play an important role to simulate possible scenarios and provide options for environmental management. However, the complexity of ecosystems makes it difficult to simulate all the physical, chemical and biological processes in one model. Instead of looking for a panacea, the urgent demand for such management tools has accelerated the development of a large and diverse number of ecological models for different contexts. Ensemble modelling is an approach inspired by weather forecasting. Model diversity is exploited to improve the robustness of algal bloom prediction. Ensemble modelling might also result in important insights how the differences in model structure contribute to the fit of the models to data. In this study, we selected two ecological models to examine their underlying causality. One widely applied model is PCLake, which is a dynamic model and includes food web interactions. PCLake is often used for so-called bifurcation analysis to define the critical loading that define lake regime shifts. Another widely applied model is BLOOM, whose phytoplankton module is built with linear programming and supported by an empirical database. To see how the models’ conceptual differences reflect on the simulated outcomes, we analyze their differences in model structure and thereafter run them for theoretical scenarios that vary in temperature, nutrient loading and light intensity，respectively. As a final step, we plan to apply both models in real-world scenarios and validate them with observed data, to see and explain how ensemble modelling works in practice. Our results show that ensemble modelling can be beneficial for managing algal blooms in polluted lakes