Near-term, iterative ecological forecasts with quantified uncertainty have great potential for improving lake and reservoir management. For example, if managers received a forecast indicating a high likelihood of impending impairment, they could make decisions today to prevent or mitigate poor water quality in the future. Increasing the number of automated, real-time freshwater forecasts used for management requires integrating interdisciplinary expertise to develop a framework that seamlessly links data, models, and cyberinfrastructure, as well as collaborations with managers to ensure that forecasts are embedded into decision-making workflows. Here, we provide an overview of FLARE (Forecasting Lake And Reservoir Ecosystems), an open-source forecasting framework that we applied to a drinking water reservoir to assist water quality management.
Results/Conclusions
The application of real-time, near-term, iterative forecasting for lake and reservoir management is in its nascent stages, providing an exciting opportunity for this research community to make great progress in developing and running freshwater forecasting systems. We share the major lessons learned from our experience developing and running FLARE over two years, which can hopefully inform other forecasting projects. Our goal is to break down the barriers to forecasting for freshwater researchers, with the aim of improving freshwater management in lakes and reservoirs globally. The take-homes from our work are: 1) Building and maintaining a forecasting system takes an interdisciplinary, highly-coordinated team; 2) Cyberinfrastructure is not trivial; 3) Let your forecasting goals guide your modeling approach; 4) Uncertainty partitioning informs forecast interpretation and forecast improvement; 5) Human-centered design improves the utility of forecasts for managers; 6) Forecasts should be reproducible and archived; and 7) Build a sustainability plan for your forecasting project. While our experience and others highlight that freshwater forecasting can be challenging, our goal in sharing our lessons learned is to assist new research teams as they begin this endeavor. Given the increased variability facing many freshwater ecosystems, ecological forecasting has much potential for improving preemptive management and minimizing water treatment costs. The iterative nature of near-term forecasting, in which data assimilation sequentially improves models and forecast performance over time, emphasizes that there is no better time than the present to start forecasting lake and reservoir water quality.