CFSv.3 Planning Meeting

August 25-26, 2011

ESSIC Building, College Park, Maryland

Agenda (PDF)

Climate Forecast System (CFS) Version 3 Planning Meeting

Background

The National Centers for Environmental Prediction (NCEP) implemented the Climate Forecast System version 2 (CFSv2) in March 2011. CFS is a coupled global climate model used for operational intraseasonal-to-interannual (ISI) prediction at NCEP. The current version of CFS, which took 7 years to develop and implement, includes many improvements in all components, including the latest developments in the global atmospheric model (GFS, also used for numerical weather forecast), coupling with a new version of the global ocean model (MOM4), a sea ice model, and an improved land surface model (Noah).

Now that CFSv2 has been implemented, NOAA is turning its attention to the next generation of CFS (CFSv3). In advance of the development cycle for CFSv3, this meeting aims to bring NCEP scientists and the external community together to develop a strategy for CFSv3. Although focused on CFSv3 development, the meeting will benefit from recent discussions being organized by the National Research Council on the US climate modeling strategy for the next 10-20 years. This meeting will also enhance NOAA-wide collaboration as well as partnerships between NOAA and the external community on climate model development, and strategies for improving ISI climate prediction and potentially decadal climate prediction.

Key questions to be addressed

1. Why is there a need for CFSv3? Discussions will focus on i) NOAA's strategic requirements for climate predictions and projections; ii) major strengths and shortcomings of the current operational versions of CFS, including model components, infrastructure, etc.; and iii) strategies to link modeling efforts for weather, ISI and longer-term climate predictions and projections within NOAA.
2. What does the future CFSv3 look like? Discussions will address i) model complexity (level of process representations); ii) potential benefits of higher resolution vs. larger ensemble size; and iii) model infrastructure (i.e., portability, scalability).
3. How to improve the CFS development process? Topics of discussions include i) a review of the CFSv2 development process, including how input was obtained from outside of NCEP; ii) how external community can contribute to the CFSv3 development effort and what the requirements and mechanisms are (e.g., documentation, accessibility, archive and access of model code and supporting data sets)? and iii) possible synergies between CFSv3 and climate modeling efforts at other modeling centers (e.g., GFDL, NCAR, NASA, COLA, DOE, ECMWF).

Meeting Format:

The two-day meeting will consist of overview presentations and breakout discussions on the key questions and related discussion topics. The meeting will be limited to around 40-50 invited participants. There will be two parallel breakout sessions on each of the three key questions.

Expected Outcome:

A white paper that summarizes the discussions and recommendations tied to the key questions and related topics above. The recommendations may include general advice on the CFSv3 development strategy and mechanisms for follow-up.

Key Questions for Breakout Discussions

Breakout Discussion 1: Why is there a need for CFSv3

• What are NOAA's requirements for climate modeling and prediction?
• What is NOAA's vision for operational climate prediction?
• What should be the strategy to link modeling efforts in weather, intraseasonal-to-interannual, and long-term climate predictions and projections within NOAA?
• What are the major model biases in CFSv2?
• What can be done to improve data archive and access for the research and user communities?

Breakout Discussion 2: What does the future CFSv3 look like?

• What level of process representations should be included in the CFSv3?
• How should the benefits and costs of higher resolution, ensemble size, hindcast period, etc be balanced?
• What are the requirements for atmospheric and oceanic data assimilations?
• What is the need to improve CFS model infrastructure (i.e., portability, scalability)?
• What are the benchmarks and metrics for model development?

Breakout Discussion 3: How to improve CFS development process?

• What are the experiences and lessons learned from the CFSv2 development cycle?
• What is the balance of model development at NCEP vs. other NOAA labs/centers vs. the external research community?
• What are the effective mechanisms/facilities to engage the external community to work on CFS diagnosis/evaluation/testing/development?
• How can we improve the model documentation, accessibility, code readability and usage by the external community?
• How to integrate the model evaluation/diagnosis process for CFSv2 into the CFSv3 development process?

List of Invited Participants

NOAA: Uccellini, Karl
CPO: Koblinsky, Mariotti, Barrie
CPC: Higgins; Huang, Kumar, van den Dool, Vintzileos, Halpert
EMC: Lapenta, Pan, Moorthi, White, Saha, Beringer, Ek, Howard, Wu, Derber, Kleist
GFDL: Rosati, Ramaswamy
PMEL: Harrison
OST: Lord, Zhou
Universities: Bretherton, Busalacchi, Dickinson, Kinter, Rood, XB Zeng, Janowiak, B.Huang, Marx, Barneston
NASA: Suarez
NCAR: Hurrel, Tribbia, Neale
International: Molteni (ECMWF), Navarra (CMCC/Italy), Lu (Taiwan), Rao (India)
Users: XZ Liang, Restrepo, Ray, Timofeyeva
WYLE: Yang