| WRF container limited to single node computing |
New I-WRF container will enable multi-node WRF |
Container-based modeling that scales |
| Limited visualization capabilities |
New tools will be integrated with the Analysis/Visualization Container such as ParaView |
Seamless data analysis experience with enhanced visualization capabilities |
| Community confidence in research results |
Containerized METplus tools will be optimized for I-WRF framework |
Faster access to verifcation tools and refinement of workflows |
| Difficulty compiling and configuring WRF |
Standard multi-node capable containers will be built, tested, and ready to use |
Rapid application deployment |
| Target systems with specialized hardware and/or limited container capabilities |
Easy to modify features. Extendable design. |
Platform portability to desktop, cloud/edge, and HPC for research flexibility |
| Consistent execution environment |
Reusable and repeatable framework |
Reproducible scientific results |
| Automated container deployment, scaling, and management |
Scripted Kubernetes orchestration via Terraform solution will be created for WRF |
Reliable container orchestration |
| Difficulty developing new containers from scratch |
I-WRF use case scripts, build files, etc. will be shared with sample data and storage configurations |
Creating I-WRF container framework will spur community development of new containers and science apps |
| ~50% students have difficulty configuring their machines to start WRF training exercises |
Turnkey container for easier education of students and early career scientists |
Increased student recruitment from all relevant domains; bigger pipeline to close the diversity gap |
| Lack of CI technical support skills, and/or time available |
I-WRF will reduce need for CI support; 24x7 virtual workshop training available. NSF NCAR committed to sustaining I-WRF. |
Researchers can get started with multi-node WRF without CI staff support, lowering the bar for less resource rich institutions |
