Open, Shareable Delft-FEWS Configuration on GitHub: Follow Up on Community Call

Open, Shareable Delft-FEWS Configuration on GitHub: Follow Up on Community Call

On January 17, 2024, the North American FEWS User Community had a call to present and discuss an open and shareable Delft-FEWS configuration (CanOpenFEWS). The goal of this open system is to provide a basis for shared operational workflows, and to bring recent research in nowcasting and snow modelling closer to application.


The beautiful Bow Watershed in Alberta, Canada was selected for demonstration, but the functionality can be applied across North America. The shared configuration approach has promise particularly in Canada, where FEWS is very well adopted, however there are many different forecasting centers that operate independently. Although a variety of modelling frameworks and approaches are employed, the data access and processing needs are consistent. With CanOpenFEWS there is a platform where best practices, configuration and research can be tested and shared.


Some interesting functionalities integrated in the testbed and discussed during the call: 


1. Advances in Nowcasting shared in an Open Configuration: Ruben Imhoff demonstrated the latest advancements in nowcasting, in extending radar short-term precipitation forecasts. This aspect garnered significant interest, showcasing the configuration's potential in enhancing accuracy and reliability for predicting intense rainfall events.


2. Capacity for Community Engagement and Development: In a live survey, 30 respondents said they would have capacity to pull from the repository, and 20 that they could contribute back to it. By making the configuration available on GitHub, there’s an open invitation for users and researchers to contribute.


3. Snow Monitoring for Connection to Forecasting Models: The meeting highlighted the configuration's ability to link with operational data sources, especially focusing on snow data. This feature was well-received, as it opens avenues for more bringing novel technologies and methods to the practice of forecasters. Planned research is to advance leading snow models and data with hydrological forecasting models.


Low, Medium and High development routes for the repository were discussed. With the lowest development route being that it remains static as an example of current capability, medium that the forecasting community contributes and high that a community testbed results. In any case, the plan is to carry the system forward to peer-reviewed publication and improvement of the forecasting workflows to demonstrate systematic improvements. Also training is planned on how to access, use and contribute back to the repository.


We were encouraged by positive feedback during and after the call and are happy to share the current system. This approach can help to bring in the next generation of modelling tools and methods for the FEWS User Community. Check out the GitHub page for a more detailed description, and the complete system configuration: