Findable, Interoperable, and Reusable data and models in CUAHSI HydroShare

Presentation for AGU Frontiers in Hydrology Conference, June 23, 2022, Emphasizing F, I, and R in FAIR Hydrology: Bottlenecks and Solutions to Making Hydrologic Science More Reproducible Session https://agu.confex.com/agu/hydrology22/meetingapp.cgi/Session/142683

HydroShare is a web-based repository and hydrologic information system operated by the Consortium of Universities for the Advancement of Hydrologic Science, Inc. (CUAHSI) for users to share, collaborate around, and publish data, models, scripts, and applications associated with water related research. It serves as a repository for data and models to meet Findable, Accessible, Interoperable, and Reusable (FAIR) open data mandates. Beyond content storage, the HydroShare repository also links with connected computational systems providing immediate value to users through the ability to reduce the needs for software installation and configuration and to document workflows, enhancing reproducibility. Drawing upon experience with the development of HydroShare I will contribute to the discussion in this session through reflection on the contributions and challenges to each of Findable, Interoperable and Reusable FAIR elements. Findable is addressed through resource schemas (schema.org) that draw searchers to HydroShare and discover functionality within HydroShare that enables text, facet and geographic filtering. Interoperable is addressed through machine readability via an applications programming interface (API) and framework for linking computational systems to operate on HydroShare resources. Reusable is addressed through standards used to represent common data content types, and the overall Open Archives Initiative Object Exchange and Reuse based resource data model. HydroShare has capability to automatically recognize supported data types and harvest metadata from these data types, simplifying the process of metadata creation for users. Data types supported include multidimensional netcdf, time series, geographic rasters and features. For some of these, standard data services, such as OpenDAP services for netCDF or Open Geospatial Consortium web services for geographic data types are automatically established when data is made public, improving machine readability and system interoperability. Ongoing work is also advancing the use of containerization to encapsulate software and data dependencies needed to ensure persistent reproducibility of computational workflows, especially for data and computationally-intensive modeling. However, bottlenecks and challenges remain in the use of this functionality. In this session I will discuss some of these contributions from HydroShare and give thoughts on ongoing opportunities and challenges.