An Infrastructure Ecosystem for Global Applications at NCEP

The global-workflow is the workhorse of the Global Forecast System (GFS) development at the United States National Weather Service (NWS). In operations, it runs both the Global Forecast System (GFS) as well as the Global Ensemble Forecast System (GEFS). For development of these applications, the workflow supports numerous different configurations. This affords both the researchers and developers a platform and the flexibility to provide innovations in the areas of model development, data assimilation techniques, post-processing, verification, and validation while maintaining the standards of the operational forecasting system. This system also supports multiple HPC platforms, including several NOAA Research and Development High Performance Computing Systems (RDHPCS) and the NWS operational supercomputer; WCOSS2. With cooperation from NOAA EPIC, the system is being ported to several cloud service providers, as well as additional R&D supercomputers for use by the wider community.

In addition, the global-workflow is expected to extend support to the development (and implementation) of the upcoming Seasonal Forecast System (SFS) including reanalysis and reforecasts, forecast model development (ranging from data-only to coupled models), and coupled data assimilation applications. In order to support the above-mentioned modes of operability, a continuous testing framework (i.e., CI) has been developed for applications ranging from global-workflow forecast applications to more granular job-by-job functional tests. Unit tests for the respective infrastructure components are also being developed as part of regular CI.

This presentation will cover these areas of development as well as improvements and modifications to the workflow infrastructure to support global applications at NCEP