In urbanizing areas, increasing imperviousness is responsible for rising runoff volumes and erosion, leading to greater nutrient and pollutant loads into downstream water bodies. This leads to eutrophication and proliferations of cyanobacteria, harmful for human and animal health. In order to protect surface water resources, water management stakeholders need tools to forecast the impact of urban stormwater on downstream ecosystems taking into account different mitigation scenarios, such as the improvement of sanitation or the reduction of runoff. Our objective was to set up such a modelling tool, coupling a hydrological model of the catchment and a physical-ecological model of the lake. The study site is Lake Pampulha, Brazil. The lake and its catchment were intensively monitored from October 2011 to October 2014. This data set was used for the parameter calibration and validation of both models. The hydrological model provided good results for the runoff quantity and an acceptable agreement with field measurements for the runoff quality. The lake ecological model successfully represented the cyanobacteria dynamics. A sensitivity analysis of the coupled catchment-lake model was used to investigate its ability to simulate scenarios of changes in the catchment, namely the reduction of phosphorus loading corresponding to an improvement of sanitation efficiency and the increase of impervious cover due to catchment urbanization. Impervious cover appeared as the more sensitive driver of cyanobacteria biomass. One option to mitigate the impact of impervious cover, without limiting future urbanization, is the introduction of nature-based stormwater control in the catchment. Green Blue Infrastructure (GBI), an interconnected network of natural and artificial green spaces and water bodies within urban areas, can play an important role for the protection of water resources. The analysis of land use and land characteristics like, for instance, topography, water table depth and pedology, makes it possible to identify types of GBI that can be implemented in each land use patch. This analysis was carried out for a pilot sub-catchment and the results of the coupled catchment-lake model were compared between the current “reference” situation and a scenario including GBI implementation. Extending this methodology to the whole catchment will allow us to identify the best management strategies for land use in the catchment.