Cyanobacteria blooms commonly occur in eutrophic lakes and reservoirs. They are expected to increase their frequency and intensity in response to climate change and expanding urbanization. This excessive growth of cyanobacteria in aquatic environments is responsible for decreasing the ecosystem biodiversity and disrupting water uses like drinking water supply and recreational activities. In addition, some species of cyanobacteria are potential producers of toxic substances which are dangerous to animal and human health. Frequently, attempts to restore the water quality in lakes by only reducing external input of nutrients have failed because the mechanisms involved in the control of cyanobacteria growth and toxin production are very complex and specific to each species. In this context, water quality surveys and warning systems are compulsory tools for the management of water resources. Continuous monitoring coupled with numerical modelling can provide a deeper understanding of cyanobacteria dynamics and can supply a powerful tool to forecast cyanobacteria growth within a short-term (1 week) warning system. Once validated, the model can also be run for predicting the cyanobacteria response to medium-term (10-20 years) scenarios of watershed development like changes in rainfall and air temperature patterns or land-use evolution. In order to study the impacts of watershed changes on cyanobacteria dynamics and help to define appropriate strategies to restore lake water quality, we propose a modelling approach in which an existing ecological lake model is connected to a hydrological watershed model. Two study sites have been selected in contrasted regions: Lake Enghien in a very densely urbanized watershed in France and Lake Pampulha in a fast-growing periurban watershed in Brazil. In 2009, the water quality (temperature, pH, conductivity dissolved oxygen), the cyanobacteria associated chlorophyll-a and the meteorological variables above the water surface, were continuously monitored in Lake Enghien. These high-frequency data (30 min) allowed us to catch the critical steps of cyanobacteria dynamic during short-duration events like the exponential stage of bloom initiation or the decrease of biomass due to adverse weather conditions. We will present the ecological model implementation and validation based on the very reliable information provided by this data set in Lake Enghien. In a second step, we will discuss the preliminary results of the ecological modelling of Lake Pampulha. We will introduce the methodology used to link the ecological lake model and the hydrological watershed model in the Brazilian study site. Finally, the next steps of the work, dealing with Lake Pampulha field survey and continuous monitoring, will be presented.