Inspiration-driven analysis sparked by science questions lead to industrial-strength research. The data analytics resources made available to the scientists are fundamental to facilitate inspiration-driven scientific exploration. In this presentation, we summarize ongoing workflow development at the Geophysical Fluid Dynamics Laboratory (GFDL), driven by ever-increasing data volume, increased opportunities (e.g Coupled Model Intercomparison Projects), and emerging technologies. At GFDL, we seek to make our workflow accessible to a broad, diverse community of researchers with varied computational resources. We constantly try to prototype solutions that enable big data analysis by focusing on aspects that reduce the overhead involved in finding, accessing and post-processing model output, that are resilient to technological changes. We will discuss a roadmap to: provenance tracking, unified data access and analysis drivers, scalable analytics, and repeatability. We will walk through examples from cloud-based technologies, such as Pangeo. Pangeo is a community promoting open, reproducible, and scalable science. Finally, we will describe how these ideas are implemented in the community-driven Model Diagnostics Task Force and its collection of open-source model diagnostics.
