Report on the CCSM Atmosphere Model Working Group Meeting
Breckenridge, June 25, 2003
The meeting covered:
The working group session ended with a discussion of near-term development issues for CAM, including selection of a dynamical core and primary spatial resolution for the next public release planned later in 2003. At the conclusion of this meeting, Phil Rasch replaced Bill Collins as the NCAR co-chair of the AMWG.
Report on the CCSM/CLIVAR Workshop:
GFDL hosted a Workshop on Reducing Biases in Coupled Model Simulations of the Tropical Oceans on Seasonal and Longer Timescales May 28-30, 2003, sponsored by CLIVAR and CCSM. Most of the major international centers with coupled models participated. Workshop participants discussed the status of biases in major models, related observational uncertainties, and the roles of idealized and reduced models. The workshop focused on biases related to the double ITCZ, mean temperature and humidity fields, MJO, ENSO, Atlantic-Pacific relationships, and the links between biases in double ITCZ and other aspects of the general circulation.
Recent development cycles at several model centers (NCAR, GFDL, Hadley Centre, and Max Planck Institute) have not resulted in large reductions of tropical biases. There is some variation in the extent to which an unrealistic double intertropical precipitation zone (ITPZ) develops in atmospheric GCMs, but a double ITPZ is quite general in coupled ocean-atmosphere GCMs. Recent atmosphere GCM parameterization experiments have produced mixed results. For example, the use of diffusive cumulus momentum transport reduced the double ITPZ in the GFDL atmosphere model but had less impact in the NCAR CAM. The use of non-local cumulus momentum transport in NASA NSIPP worsened its double ITPZ. The double ITPZ was found to be sensitive to evaporation of rain in NASA NSIPP, but other groups did not report this result. A more general result was a strong ITPZ dependence on low clouds in eastern oceans in coupled models. Resolution effects on tropical biases have also been inconsistent among models.
Experiments with the HadGEM1 model suggested that the ocean cold-tongue bias is related to excessive trade winds. Experiments at COLA also implicated surface heat fluxes in this problem.
Further development of GCM parameterizations of clouds and convection appears essential for further progress. Closer diagnosis of surface fluxes in coupled models is planned.
Status of CAM:
A new version of CAM was frozen in June 2003 and delivered to the CCSM Software Engineering Group. This version is proposed for use in the next public release of CCSM, pending approval by the AMWG and the CCSM SSC. It includes a number of major changes and improvements relative to CAM2, including:
The model includes several other improvements:
Several critical biases identified in CAM2 and CCSM2 have been reduced or mitigated:
However, several critical biases remain as outstanding issues in the fidelity of the climate simulated with CAM:
Near-term Development Issues for CAM:
The AMWG and SSC need to decide which dynamical core, the Eulerian or Finite Volume, will be the "official" dynamical framework for the next public release. This decision will also affect the integrations performed by NCAR for the next Intergovernmental Panel on Climate Change (IPCC) report. The designation of an "official" framework certainly does not preclude use or experimentation with the other dycores. CAM will continue to support all three dynamical formulations. However, the model will be tuned for optimum simulation fidelity for the official dynamical formulation, and future control integrations and development will be conducted primarily with the official dycore.
According to the CCWG, the IPCC integrations need to begin by mid-September 2003 to ensure timely completion by the first IPCC deadlines in July 2004. This schedule implies that the dycore decision needs to be made by early fall 2003. The plan discussed at the AMWG meeting, which was generally accepted, is to conduct parallel coupled and uncoupled integrations of the FV and EUL dycores. The complete results and diagnostics would be distributed to the AMWG for its recommendation to the CCSM SSC.
A second major issue is the "official" resolution for the next public release. The CCWG has expressed strong interest in conducting the majority of IPCC simulations at the equivalent of T85 in the atmosphere, and the CCWG has indicated that they would like to conduct a few test simulations at T170. At the moment, the AMWG and Climate Modeling Section at NCAR have much less experience with T85 than T42, and it is not clear if the model physics will scale to higher resolutions. The issue of whether to run at higher resolution was discussed at length during the CCSM Forum in Breckenridge, and a number of participants expressed concern that higher resolution versions could not be developed without the risks of inadequate testing and strain on the scientific and software engineering resources of the project. The NCAR staff is committed to attempting to build a high resolution version of the model for possible use in the IPCC project, but final decision regarding the development, configuration, and release of the higher-resolution CAM have not been made.
The final issue discussed in the AMWG session is the release date for the next version of CAM. The CCSM SSC has recommended that the release not be rushed in order to give the CCSM software engineers time to test and document the software with sufficient care. The current consensus of the SSC chair and
CSEG is to release the model sometime after December 1, 2003.
Bill Collins, co-chair
Leo Donner, co-chair
P. Rasch, incoming co-chair
L. Ruby Leung