[an error occurred while processing this directive] [an error occurred while processing this directive]

Report of the CSM Atmospheric Model Working Group

James Hack and David Randall, Co-Chairs

2 and 3 March 1998 and 23 June 1998


 March, 1998 CSM AMWG Meeting

This working group report summarizes the two most recent meetings of the CSM Atmospheric Model Working Group (AMWG), the first of which took place on 2 and 3 March 1998. The March meeting was the third assembly of this group and included a number of major themes (see attached agenda). The first objective of this meeting was to update the participants on overall CSM and Climate Simulation Laboratory (CSL) resource issues, the status of the CSM computer resource request, and the timing for working group input to a CSM planning document. The next objective was to provide the working group with status reports on a number of the more mature development projects underway. This phase of the meeting included presentations on the current plans for the Community Climate Model version 4 (CCM4), an NCAR-NASA collaboration on an alternative dynamical core, a proposal for an alternative longwave radiative transfer parameterization, sensitivity studies addressing shortwave radiative transfer, two alternative cloud overlap parameterizations, and an investigation into the CCM3 marine stratus parameterization. The next phase of the meeting included an extensive discussion of standard diagnostics for model evaluation. Finally, the various CSM AMWG focus groups (Tropical Convection, Planetary Boundary Layer, Middle Atmosphere, and Numerical Methods) met to discuss and prioritize their scientific and development objectives. This discussion included identification of development "deliverables" and the timescale on which they would likely become available.

The plenary discussion attempted to focus on the main problems with the CCM3 simulation to help in prioritizing development projects requesting CSM computational support. Key problems included the representation of marine stratocumulus in the eastern oceans, the lack of transience in deep convection (which was hypothesized to be responsible for things like the weak Madden-Julian Oscillation (MJO) and degraded tropical cyclone climatology), a pronounced lower-tropospheric tropical dry bias, systematic errors in the tropical precipitation distribution, deficiencies in the monsoon circulations, systematic biases in aspects of the simulated diurnal cycle, and polar circulation biases (attributed to inadequacies in parameterized physics, such as heat exchange over sea ice and marginal horizontal resolution).

Current simulation problems motivated a wide range of possible model improvements proposed on the timescale of CCM4 and on the longer timescale of CCM5. These included activities aimed at improving the large-scale dynamical core (higher accuracy and greater numerical efficiency), improved cloud overlap schemes (to provide a more physically realistic representation of cloud field), revisions to the marine stratus parameterization (to address eastern ocean surface and top-of-atmosphere (TOA) energy budgets), explorations of topographic filters and topographic form drag (e.g., to address coastal cloud and divergence anomalies), modifications to parameterized moist convection (to address the lack of transience in deep convective heating; errors in stationary structures), establishing minimal vertical resolution requirements, and the exploration of sensitivities to specified ozone distribution.

The need for a CSM planning document, something with a five-year vision that would be revised every year or two, was discussed. There was agreement that the CSM AMWG focus groups would contribute to such a document as soon as there was more general agreement on the timing for its preparation.

The CSM AMWG also discussed the importance of soliciting input from other CSM working groups on the most important scientific problems affecting the CCM simulation. It was generally recognized that this kind of input would be valuable and necessary for assembling a more comprehensive description of model simulation strengths and weaknesses. Input from other working groups would also help us allocate limited computational resources to those CSM AMWG projects aimed at addressing problems of broadest interest to CSM. There was a general sense that this interaction would naturally take place at the June, 1998 CSM workshop, with the recognition that the CSM AMWG might also need to actively take steps to solicit the participation of other working groups.

Finally, there was strong agreement that there was a need for an organized diagnostic activity to facilitate standard evaluations of proposed changes to the atmospheric model. This discussion prompted the creation of a Diagnostics Focus Group that would be charged with creating a standard evaluation procedure. Karl Taylor and Jim Boyle (Program for Climate Model Diagnosis and Intercomparison, PCMDI) agreed to participate as co-chairs in such a focus group. Unfortunately, we were unable to identify an NCAR co-chair for this activity. Several of the NCAR participants agreed to pursue this matter prior to the June, 1998 CSM workshop.

June, 1998 CSM AMWG Meeting

The fourth meeting of the CSM AMWG took place in conjunction with the annual CSM Workshop on 23 June 1998. The agenda was organized around proposed modifications to the model formulation on the timescale of CCM4, preliminary work on model modifications proposed on longer timescales (i.e., beyond CCM4), and identifying a process for deciding on proposed changes to the standard model formulation (see attached meeting agenda).

The session on proposed changes to be included in the CCM4 included eight presentations on a variety of topics including radiative transfer, generalized treatments of cloud overlap, explorations of alternative cloud parameterizations with an emphasis on marine stratocumulus, triggering mechanisms for deep convection, and a more efficient implementation of the proposed semi-Lagrangian dynamical core.


The session on development activities with a longer timescale included initial explorations of complete physical parameterization packages from Colorado State University, NASA Goddard Space Flight Center, and Center for Ocean-Land-Atmosphere Studies atmospheric general circulation models, as well as a new approach to representing marine stratocumulus convection that employs a prognostic inversion height formulation. The preliminary simulation properties of the NASA finite-volume dynamical core with CCM3 physics were presented, a collaborative project that was first discussed during the March, 1998 CSM AMWG meeting. The numerical advantages of a sigma-isentropic vertical coordinate for the CCM3 were also presented.

The plenary discussion, unfortunately limited to one hour, addressed a wide variety of topics of immediate importance to the atmospheric model development activity. Of greatest urgency is the need to establish a standardized process for evaluating and differentiating proposed changes to the atmospheric model. This was a subject that was extensively discussed during the March CSM AMWG meeting, culminating in the creation of a Diagnostics Focus Group under the CSM AMWG. Discussion following the creation of this focus group elevated its status to the working group level (the Data Management Working Group). This working group, still in the process of forming, met for the first time in early May, 1998. Major topics discussed during that meeting were on general data management, data access, data distribution, and data manipulation. The issues of greatest importance to the CSM AMWG remain to be resolved, namely the creation of an automated diagnostic procedure for illuminating the major simulation strengths and weaknesses of prototype configurations for the CCM. One aspect of such a process is achieving agreement on the set of diagnostics that will be produced. The CSM AMWG has proposed adoption of the Working Group on Numerical Experimentation (WGNE) Standard Diagnostics of Mean Climate as a start in this direction. Additional diagnostic procedures that characterize other aspects of the simulation (e.g., the MJO) can, and should, be added to a standard diagnostic suite by interested investigators as soon as is reasonably possible. In addition to proposing a minimum set of standard climate measures, the CSM AMWG agreed to a hierarchical experimental strategy for evaluating proposed changes to the atmospheric model, initially consisting of a single 5-year simulation forced by climatological sea surface temperatures. This control experiment would be followed by a 15-year AMIP II simulation, followed by an uncoupled ocean integration forced with the atmospheric history tapes, and finally followed by a fully coupled multi-decadal integration. Proceeding to subsequent stages in this evaluation hierarchy (i.e., beyond the initial 5-year simulation) would hinge on timely comments about the simulation strengths and weaknesses by members of the CSM AMWG, as well as the larger CSM community.

The urgent short-term impediments to implementing this kind of evaluation process are the establishment of a consensus on a set of comprehensive diagnostic metrics and finding a way to implement an infrastructure for automatically generating the agreed-upon set of diagnostics. This capability is essential to atmospheric model development progress and is needed immediately. It should be among the highest basic infrastructure priorities facing CSM. Additionally, the CSM AMWG is seeking guidance from the SSC on the matters of ownership and responsibility for changes that are incorporated to the CCM physical parameterizations and/or numerical methods. Experience has shown that ongoing support is required for all aspects of a model as complex as the CCM, such as dealing with coding or algorithmic problems. Therefore, it will be necessary to identify a source of contact/support for any algorithmic changes to the atmospheric model.

The working group recognizes that most of the current development activities on the timescale of the CCM4 will mostly provide more flexibility to the modeling framework and are not necessarily targeting deficiencies in the CCM3 simulation. The exceptions are the work on marine stratus, convective triggering functions, and the Rapid Radiative Transfer Model (RRTM) scheme. For the near-term, the group agreed that a reasonable goal was to assemble a prototype model containing the mature major components proposed for CCM4 by the end of the calendar year. These components currently include the two-time-level, 26 vertical level, semi-Lagrangian dynamical core (with an improved orographic filtering), prognostic cloud water, and Atmospheric and Environmental Research (AER) Inc.'s RRTM scheme. This configuration would be tuned for balanced TOA and surface energetics and would minimally provide a 5-year simulation forced with an annually repeating sea surface temperature climatology. If any of the process-oriented improvements (e.g., convective triggering mechanisms, improved marine stratus parameterization, etc.) become available prior to the completion of this prototype model, every effort should be made to incorporate them. In the meantime, CSM AMWG computing resources will continue to be allocated to those projects that are attempting to address the more severe CCM3 simulation deficiencies via improvements to major parameterization components. In this regard, the CSM AMWG will seek stronger links to other CSM working groups, in particular the ocean, land surface, natural variability, and polar working groups. As was pointed out last year, the CSM AMWG wishes to exploit the opportunities for interaction with other working groups, such as diagnostic projects that examine phenomenological behavior (e.g., MJO, tropical cyclone frequency, annual cycle, El Niño Southern Oscillation (ENSO) response, interannual variability) of the CCM that can be linked to deficiencies in specific physical processes.

A summary of the principal action items associated with this meeting includes:

The CSM AMWG anticipates meeting again in Boulder sometime during December, 1998 or January, 1999. The group hopes to be well along in a comprehensive evaluation of the CCM4 prototype and will look forward to updates on the status of process-oriented improvements to the basic model formulation.



Atmosphere Model Working Group Meeting


Tuesday, 23 June 1998

Forest Room, The Village at Breckenridge

Tuesday, 23 June 1998
8:30am - 3:00pm

8:30am - 10:30am

Candidate modifications for CCM4

-C. Zender

Absorption Effects Due to O2-O2 and O2-N2 Complexes

-M. Iacono

RRTM in the CCM3

-J. Bergman

Generalized Cloud Overlap

-X.-Z. Liang

Mosaic Treatment of Clouds

-B. Stevens

Marine Stratus Parameterization

-J. Norris

Simplified Large Scale Cloud Parameterization

-P. Rasch

Convection Trigger Functions

-D. Williamson

Two-time Level Semi-Lagrangian Configuration

10:30am - 10:45am


10:45am - 12:00 noon

Beyond CCM4

-C. Bretherton

A Realistic Stratocumulus Parameterization with a Prognostic Inversion Height

-W. Collins

Preliminary Evaluation of Enhanced Solar Absorption in CSM Integrations

-J. Shukla

COLA Physics in CCM3

-D. Randall

CSU Physics in CCM3

-Y. Sud

GSFC Physics in CCM3

12:00 - 1:30pm


1:30 - 2:00pm

Beyond CCM4 (continued)

-S.J. Lin

Climate Simulations Using the NASA Finite-volume Dynamical
Core with NCAR CCM3 Physics: Preliminary Results

-Z. Zhu

A Sigma-isentropic Coordinate for the CCM3

2:00 - 3:00pm

Atmospheric Model Evolution


Process for deciding on proposed changes, experimental configurations, climate measures, computational issues Ownership, responsibilities...



Atmospheric Model Working Group Meeting

The Broker Inn

Boulder, Colorado

2-3 March 1998







Check-in/buffet lunch

1:00 pm:


Resource/planning issues



  • Current CSL computer plans (O'Lear)



  • CSL computer resource request update (Randall)



  • CSM Planning Document (Blackmon)

-timing for working group input, etc.

1:30 pm:





  • CCM4 plans (Williamson)



  • Dynamical core collaboration and plans (Lin)



  • Alternative LW radiative transfer parameterization (Iacono)



  • SW radiative transfer investigations (Collins)

-19-band model

-sensitivities to enhanced cloud absorption



  • Cloud overlap activities (Bergman, Liang)



  • Marine stratus parameterization update (Stevens, Norris)

4:00 pm


  • Standard diagnostics for model evaluation (Boyle, Taylor)



-strawman and discussion

5:30 pm


Reception (cash bar)




Continental breakfast

8:30 - 11:30

Focus group break-out sessions


  • Identify focus group objectives and priorities


  • Identify deliverables and timescale



1:00 - 4:30



  • Working group reports (15-30 minutes)


  • Prioritize working group projects


  • Action items for June workshop (who, what, when?)



Be sure and register for the CSM Workshop on the CSM web page www.cesm.ucar.edu



2 and 3 March 1998 CSM AMWG Meeting Participants









University of Colorado



Climate and Global Dynamics









Lawrence Livermore National Lab



Climate and Global Dynamics


David G.


Center for Ocean-Land-Atmosphere Studies

Leo J.



Princeton University



Mesocale and Mircroscale Meteorology Division




Climate and Global Dynamics


Michael J.


Atmospheric & Environ. Research, Inc.



Climate and Global Dynamics




Colorado State University



Climate and Global Dynamics Division




Atmospheric Sciences Research Center




NASA/Goddard Space Flight Center




University of Colorado



Mesoscale and Microscale Meteorology




Mesocale and Mircroscale Meteorology Division












Climate and Global Dynamics




Climate and Global Dynamics




Dept. of Atmospheric Sciences

Colorado State University



Climate and Global Dynamics




Climate and Global Dynamics









Lawrence Livermore National Laboratory






Climate and Global Dynamics Division




Climate and Global Dynamics Division




Mesoscale and Microscale Meteorology

National Center for Atmospheric Research



Advanced Study Program




State University of New York