The collaborative Climate Science for a Sustainable Energy Future (CSSEF) project started in July 2011 with the goal of accelerating the development of climate model components (i.e., atmosphere, ocean and sea ice, and land surface) and enhancing their predictive capabilities while incorporating uncertainty quantification (UQ). This effort required accessing and converting observational data sets into specialized model testing and verification data sets and building a model development test bed, where model components and sub-models can be rapidly evaluated. CSSEF’s prototype test bed demonstrated, how an integrated testbed could eliminate tedious activities associated with model development and evaluation, by providing the capability to constantly compare model output—where scientists store, acquire, reformat, regrid, and analyze data sets one-by-one—to observational measurements in a controlled test bed.
The CSSEF-DIT team worked closely with climate scientists on the tools for evaluating model components and with user facilities to deploy the prototype test bed. In addition, team members collaborated extensively with national and international institutions and universities that specialize in data-intensive science and exascale computing. The goal was for the new climate-model test bed to incorporate:
• A calibration platform where UQ techniques are used to calibrate a model against regional observational data sets; and
• A validation platform where simulation quality is quantified against global observational data sets.
The team designed the prototype test bed infrastructure so that it could be easily customized to the specific requirements of each science component area, and demonstrated its capabilities on one of them. The DIT team integrated closely with the modeling activities and UQ methods, and provided interactive access to the (derived) data products to support the goals of CSSEF science community.