All-hands-on-deck data management: Building the team, tools, and workflows to forecast future water quality

Iterative, near-term ecological forecasts have the potential to advance the discipline of ecology but developing end-to-end forecasting workflows pose significant data management challenges. Synthesizing complex, heterogeneous, and sometimes messy datasets that are necessary for automated forecasting requires a flexible cyberinfrastructure that connects sensors to the cloud, runs ensemble models, and provides model output securely. Our interdisciplinary team of ecologists and computer scientists has developed a workflow to calibrate a reservoir ecosystem model forced by future weather predictions to create 16-day water quality forecasts for managers. Here, we share lessons learned on how to develop workflows for near-real time data assimilation.