Central to the science of hydrology is the localised nature of the medium through which water flows. This fact leads to a large amount of hydrological models, specifically made for a certain region (catchment). The fact that these models are made by different individuals, in different programming languages severely hinders re-use and reproducibility. This in turn is leading to a “crisis of reproducibility” in Hydrology. This science demonstrator seeks to create a fully FAIR Hydrological forecasting system, combining local and global models. With this we showcase and hope to demonstrate best practices on how data as well as software can be made FAIR in Hydrology.
Achievements
Using a combination of the CWL standard for workflows, Cylc, and Docker software containers, we were able to create a fully reproducible (low resolution) version of the eWaterCycle forecast. Output data is stored via OneData, and available for analysis in a notebook environment, as well as visualization in a web application. The forecast is running daily now, without any manual intervention, and runs on supercomputer systems with no changes needed to the software. We plan to integrate the work done in this demonstrator into our larger eWaterCycle II project, with a focus on FAIR Hydrological modelling https://www.ewatercycle.org/
Niels works on the water management project as well as general eScience infrastructure. Niels is also part-time guest researcher at the Leiden Observatory, where he applies distributed computing techniques to the AMUSE computational astronomy simulation framework.
Profile pageMaarten is a core team eScience Engineer, where he focuses on visualization.
Profile pageStefan is specialized in software development for life science projects.
Profile pageVincent is specialized in analytical methods for life science projects.
Profile page