Synthesizing hydrologic process knowledge to determine global drivers of dominant processes

A fundamental challenge in hydrology is to explain where and when different hydrologic processes occur, and how they are controlled by climate and landscape. Hydrologic processes describe how water moves through and is stored within the landscape on its path from rainfall or snowfall to river flow. Large-scale process knowledge is valuable for a wide range of applications such as designing realistic and accurate flow forecasting models, and choosing effective interventions to improve water quality. Many scientific organizations maintain intensively monitored watersheds that provide deep understanding of hydrologic processes at specific locations, but the knowledge is fragmented and difficult to integrate across regions and continents. This project will unify field hydrology studies into a global picture, by creating and exploring a searchable database of hydrologic process knowledge. The project will enable the hydrology community to apply modern, big data approaches to knowledge discovery, and to draw out emergent patterns that relate landscape organization and hydrologic function. The project team will work with undergraduate students to develop tutorials and hands-on activities to explore the database in the form of a Student Toolkit. To engage with the hydrology community and enhance the impact of the results, the project team will build a web app and online map discovery interface for the database. The project will create a relational database to represent process knowledge, and populate the database with extensive knowledge from hundreds of experimental watersheds, using standardized workflows to maintain quality. The database will use a taxonomy of hydrologic processes to enable labelling, organizing and hierarchical searching for process information. The investigators will explore the database to synthesize global patterns of hydrologic function, to analyze how dominant processes can be explained by physical watershed features, and to reveal transitions in processes at events and seasonal timescales. They will apply the database to evaluate new generation continental-domain hydrologic models, which currently lack the process data needed to select model structure. They will test whether flow predictions improve when models accurately represent processes, and whether model sensitivity analysis identifies the same dominant processes as those found experimentally. The long-term vision for the database includes a role as “ground truth” for future far-reaching machine learning efforts such as automatic literature and data analysis for process prediction. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.