2014 Developers/Users Forum Minutes
CMAS Community Developers/Users Forum Minutes
CMAS Conference October 27-29, 2014
Developing a Vision for the Next Generation Air Quality Modeling Tools
Facilitator: Zac Adelman (UNC)
Panelists: Jon Pleim (USEPA ORD), Ravan Ahmadov (NOAA), Mike Moran (Environment Canada), Naresh Kumar (EPRI)
The objective of the meeting was for the CMAS community to have a conversation about the directions for future modeling and analysis systems. The forum began with series of brief presentations by each panelist and then the floor was opened for discussion. The Forum was presented as the beginning of an effort to develop the vision for the next generation modeling tools. A survey with the charge questions is available via the CMAS Center website for continuous feedback on this topic from the community.
Charge Questions: What questions need to be addressed/answered using the next-generation air quality modeling systems? What technical and diagnostic capabilities need to be included in these systems? What features should be included in the physics and chemistry parameterizations of the next-generation modeling systems? How should the software be designed/structured? What should the user interfaces look like?
Panelist Presentations:
Group Discussion:
- What is the timeline for developing new modeling systems? (Pleim: 3-10 years)
- Suggest leveraging other programs, see Earth System Model and next generation Global Modeling Systems (NOAA & NASA)
- Need to leverage other programs to distribute burden of funding on different agencies
- Concept of CMAQ as an operating systems presented; the CMAQ O/S would allow people to generate new "apps" as different science modules
- Need for data services to support new model features; input data are needed to support new features of modeling systems
- Encourage community development and EPA staff to collaborate with outside developers
- Parallelization: need to streamline the software for larger cores and larger clusters; system should be able to take advantage of expanding computational resources
- Stratospheric and marine chemistry need to be included, examples of coupled models exist
- Need for a global emissions model; SMOKE works well for the US and parts of Europe but need for tools to process global emissions data; need for global SMOKE/emissions results
- Analysis tools: need for basic stats analysis tools that can work with large data files
- Need for funding to bridge basic sciences and modeling, no resources for putting science innovations in a form for a model
- Need better support for chemistry mechanisms and science evaluation, there is no support for evaluating the literature and figuring out how/what goes into the model
- VERDI works well, need adaptable code structure in the models; relax the hard coding style and need for object oriented modularity
- EPA should consider engaging a software development house that specializes in code modularity and parallelization
- Interoperability: whatever is done now, make sure that it will still run in the future
- Need for more levels of parallelism; take advantage of the HPC revolution in hardware and compilers
- Would like to see a bigger suite of solvers in the chemistry
- Think of the models with a processed-based approach where each component is considered separately
- Need for adaptive grids
- Think of groups of processes in terms of their timesteps, i.e. group processes by timestep lengths
- Intelligent use of variable timesteps can be used to create efficiencies in the model
- Need to include adaptive chemistry and physics in the next generation of models
- This discussion is similar to the original Models-3 discussion: thinking of the broad evolutionary cycle of the software
- Need for a transition plan to move from the current tools to the next generation, what will this transition look like?
- How are emission inventories going to keep pace with the in-line modeling capabilities
