Evaluating the D4Science virtual research environment platform for agro-climatic research

CONTEXT: From its early beginnings the Internet has been used by scientists to collaborate and share information about their research. Increasing connectivity and networking capabilities have resulted in improved collaboration functionalities ultimately combined in complete virtual research environments (VRE) as a type of virtual laboratories. These aim at providing collaborative online workplaces with access to all needed tools, data, and computing resources, and supporting data sharing. Since each research domain has its own characteristics, requirements, and preferred tooling, VRE providers must make trade-offs between the specificity of components and the functionality provided. The D4Science VRE adopts a modular approach based on open standards for constructing VREs for interested communities. The agro-climatic science domain develops diverse analytical tools that it connects to heterogenous data sources (i.e. climate data, experimental fields, satellite data, soil samples) originating from other domains, which is often poorly standardised and sparsely interlinked at best. OBJECTIVE: The aim of this paper is to test and evaluate the usefulness of the D4Science based VREs for this agro-climatic science domain, using crop growth simulation and crop phenology estimation as characteristic use cases, with specific attention to Open Science. METHODS: Based on the needs of the use cases a VRE has been composed and further developed in an iterative approach and evaluated at the end of each implementation cycle. RESULTS AND CONCLUSION: Both the development work and the evaluation results point at the foreseen potential benefits of adequate VREs and the current existence of sufficient opportunities and capabilities for constructing them. The focus when developing VREs should be on supporting research with proven and stable tools, instead of striving to include the latest and greatest. SIGNIFICANCE: The agro-climatic research domain has ambitious requirements concerning the availability and integration of data and models, which proved to be particularly challenging for incorporating in a VRE. Yet, a clear but gradual adoption of digital techniques to further the science itself is happening and VREs represent an ultimate possible end-state of Open Science. To conclude, this paper provides a few recommendations that we think can help this ongoing transition.