To understand the impact on our global environment of societal challenges such as climate change or pollution, scientists need to model and measure the environment on a large scale and understand the interactions between the atmosphere, oceans, geosphere and biosphere as part of a cohesive global ecosystem. Current research support environments include data and ICT infrastructures that provide services for collecting and managing data and executing computational experiments; however, many of these infrastructures are built for specific domains or communities, relying on different standards and procedures for creating meta-information for data and resources, providing data services, and managing the data lifecycle. Such heterogeneity not only results in barriers for researchers wishing to access data and software of different research support environments and integrate them together for cross-domain scientific experiments, but also makes running those experiments across different underlying computational platforms particularly difficult.

The objectives of this session focus on identifying the best practices in using environmental data from heterogeneous sources and domains to support system-level scientific research. Topics include semantic alignments, metadata mapping, data identification and citation across research support environments, data integration from heterogeneous sources, dynamic optimization of execution environments and provisioning for data-intensive applications, and data provenance.