Minutes of the CCSM Scientific Steering Committee Meeting
Monday, 25 June 2001
The Village at Breckenridge
Attendees: Kiehl (Chair), Blackmon, Chang, Boville, Dickinson, Gent, Moritz, McKenna, Fein, Bamzai, Bader, and Harper
1. Kiehl New Chair of SSC. Blackmon discussed his stepping down as Chair of the CCSM Scientific Steering Committee (SSC) to take the lead in planning the Earth System Model project and other NCAR strategic plans as requested by NCAR Director Tim Killeen.
2. Kiehl's Report on Organization of CCSM. Kiehl outlined his plans for the CCSM project by discussing organizational issues, model development issues, diagnostic capabilities, and future issues.
Regarding organizational issues, Kiehl reported:
- the coordinator position would be advertised in Eos and BAMS as soon as possible after the workshop, and further details of the coordinator position would be discussed in the SSC Executive Session;
- the formation of the CCSM Software Engineering Group within NCAR/CGD and the hiring of Tony Craig (with partial support by DOE) as the manager of that group; and
- the CCSM organization falls in the Program Office, with him as Chair, a new coordinator position to help him with day-to-day activities, Harper as administrator, and Craig as the Software Engineering Group Manager. Kiehl hopes that Jim Hack, head of CGD's Climate Modeling Section and co-chair of the Atmosphere Model Working Group (AMWG), and Peter Gent, head of CGD's Oceanography Section and co-chair of the Ocean Model Working Group (OMWG) will work together with him, the coordinator, and Craig to guide this new Software Engineering Group.
Kiehl stated his personal goals for the CCSM project are:
- to fill the coordinator position;
- to produce a plan of development for CCSM;
- to improve communications;
- to explore boundary issues (What is CCSM? How large should it be? What national efforts should it be involved in? What international efforts should it be involved in?);
- to create a "Living CCSM Plan" on the web that provides very clear and extensive information about what is in each component model, continually shows updates to the CCSM Plan, etc.; and
- to provide outreach, such as increasing visibility in the media and providing more information to NSF and Congress.
Kiehl reported that activities in the model development area are:
-currently testing the fully coupled version of the model;
-modifications to the cloud properties over sea ice;
-that preliminary results indicate improved tropical sea surface temperature variability;
- that there is improved summer sea-ice distribution;
- currently testing a river runoff scheme; and
- computing performance of the model must be addressed.
Kiehl reported that activities in the diagnostics capability area are:
- AMWG designed diagnostic packages that are currently on their website;
- other component model working groups need to add similar diagnostic packages to their web sites very quickly; and
- further ways to interact with the greater community on diagnostics need to be found.
Future issues reported on are:
- in the organizational area, we need to define the extent of collaborations, expand visibility of the CCSM project; and move toward a more coherent support group at NCAR;
- in the model development area, we need to include glacial processes, atmospheric chemistry, etc.;
- in the diagnostics area, we need to generalize the diagnostics packages used to all component models and open the diagnostics evaluation process to the whole community; and
- in the software engineering area, we need code efficiency and portability, coding standards, and code security. There are several collaborative projects (DOE's Avant Garde and NASA's CAN) that involve CCSM in software engineering initiatives.
3. CCSM and DOE Collaborations. Dave Bader reported that the DOE Climate Change Prediction Program (CCPP) core activities prior to 2001 consisted of ocean and sea ice model development at Los Alamos National Laboratory (LANL), model intercomparison and diagnostics work at the Program for Climate Model Diagnosis and Intercomparison (PCMDI), long-term climate change prediction research through the NCAR Cooperative Agreement, and computational support at LANL and Oak Ridge National Laboratory. In addition, university grants and single PI DOE lab projects were funded.
He then discussed the changes to the CCPP in 2001, which are a refocus at PCMDI with increased emphasis on diagnostics; closer interactions with modeling development such as participation in the analysis of CCSM atmosphere model candidates; linking with the DOE Atmospheric Radiation Measurement (ARM) project by using a Lin-Rood based atmospheric general circulation model (AGCM) as a parameterization testbed; Climate Model Intercomparison Program (CMIP) runs; diagnostic analysis tools and software; and community outreach and support. Further changes are the incorporation of ACPI (Accelerated Climate Prediction Initiative) and SciDAC (Scientific Discovery through Advanced Computing) with a new funding announcement.
The CCPP/SciDac RFP includes grant applications for predictability and climate dynamics, parameterization development, numerical methods and algorithms, and diagnostics; cooperative agreement applications for the climate model of the future that includes integrated, multi-disciplinary research of climate, applied math, and computational science; and emphasizes multi-disciplinary graduate training.
Bader invited all SSC members to attend the next DOE CCPP Science Team meeting in San Diego from 1 to 3 October 2001.
4. CCSM and NCEP Collaborations. Fein reported that an initial meeting was held at NCEP with Steve Lord and Louis Uccellini of NCEP, Maurice Blackmon and David Randall of the SSC, and himself to begin talks about collaborative projects between NCEP and using the CCSM. There are three initial fronts that are being proposed for interactions: (1) running the NCEP, NASA DAO, and CCSM atmospheric models from the NCEP GDAS and DAO FVDAS analyses to analyze the growth of the systematic errors as forecasts proceed; (2) exploring the utility of an isentropic coordinate in the CCSM atmospheric model; and (3) exploring ways to optimize for both NCEP and CCSM, the sea ice models newly developed by both groups. Fein stated that strong links between research groups and operational groups in the U.S. are highly desirable and that we must find ways to strengthen those links. The isentropic coordinate project has interested leadership and will go forward. Regarding the forecast initial error analysis and intercomparisons, the SSC concluded that a scientist needs to come forward, take the leadership role, and define what science they want to do within this framework between NCEP, DAO, and CCSM. Also, the lead scientist does not have to be an NCAR scientist. Boville stated that NCAR and DAO are already working together on initial errors analysis. He suggested that NCEP bring their resources and appoint people to work with NCAR and DAO on both issues. Randall had reported that DOE is organizing a group to study the issue of initial error tendencies in the context of the highly instrument ARM sites and Randall and Steve Lord, Bob Atlas, and Dave Williamson are involved. Fein will set up another meeting with NCEP, NCAR, NASA DAO, and others (e.g., CDC) that are interested in discussing integrating all these factors in developing "next steps."
5. CCSM Tutorial Workshops. A two-week summer workshop hosted by NCAR to help graduate students learn how to run the CCSM and analyze the output of computer runs was suggested and discussed. It was further suggested that training be included on how to design an experiment and get something useful from it. Kiehl will email several university professors to find out what would be useful during this workshop for their students and to help to define the needs of the workshop. Kiehl will discuss this topic at the next SSC meeting.
The SSC recommended that a one-day training workshop at the beginning or end of the annual CCSM summer workshop for postdoctoral students and scientists to learn how to run the CCSM be started at next year's workshop.
The SSC adjourned for their Executive Session. The minutes of the Executive Session are separate from these minutes.
6. Blackmon Recognition. During the SSC working lunch, Fein read a letter to Blackmon from Margaret Leinen thanking him for his leadership and work on the CCSM project.
7. Global Chemical Modeling and CCSM. McKenna outlined ACD's existing capabilities, scientific needs, future directions, science questions, and next steps. ACD's existing capabilities include the Model for Ozone And Related chemical Tracers (MOZART) 1, 2, and 3, the Whole Atmosphere Community Climate Model (WACCM), and the GLObal Biosphere Emissions and Interaction System (GLOBEIS). ACD's scientific needs are field studies and experiments in collaboration with universities, regional simulations in collaboration with MMM, impact studies in collaboration with ESIG, remote sensing studies in collaboration with MOPITT and HIRDLES, middle atmosphere in collaboration with HAO and CGD, and chemistry/climate/ biosphere interactions in collaboration with CGD and ESIG.
McKenna discussed future directions for chemical modeling with CCSM to include MOZART Chemistry/Solver, MATCH transport and physics, CCSM land and ocean model components coupled to emissions modules, and a CCM-Chemistry model using CCSM dynamics and transport. He also discussed some potential chemistry/climate studies, such as the O3 budget of the troposphere, influence of oxidants on aerosol formulation, influence of climate on greenhouse and other gas emissions, and influence of O3 loss on composition, climate and transport. The next steps are to build upon the WACCM framework, simplify as much as possible the tropospheric chemistry, find a consistent and suitable transport scheme (Lin/Rood, MATCH, other?), find consistent gas to particle conversions (nucleation, condensation, uptake, processing), and emission models (extend to other species, e.g., CO, CH4, SO2, etc.).
The CCSM Working Group Co-chairs (Houser, Bonan, Washington, Meehl, Craig, Schneider, DeLuca, Smith, Otto-Bliesner, Hunke, Fung, Hack, Sloan, Hurrell, and Taylor) joined the SSC meeting for joint discussions.
8. Component and Coupled Model Update. Moritz reported on the improvements to the sea ice model component (CSIM 4.4.4): (1) new elastic-viscous-plastic (EVP) scheme, (2) ice thickness distribution, (3) thermodynamics, (4) Greenland Pole grids in the x1 and x3 versions of the model, and (5) active ice only set up with mixed layers and prescribed atmospheric data. This model has been running for some time, and the Polar Climate Working Group (PCWG) has been cleaning up the code, using F90 standards, debugging, and archiving at NCAR. The code is available via FTP, and there are a developer's page (to be posted soon) and a draft documentation page online at http://www.cgd.ucar.edu/ csm/models/ice-csim4/.
Bonan reported on the development of a version of the Community Land Model (CLM) that provides a framework for simulation of carbon assimilation by vegetation and dynamic vegetation modeling and is in other ways compatible with internal NCAR requirements. He also reported simulations with this code to provide results that are overall consistent with the results reported at last year's workshop by Zeng for an earlier version of CLM. Significant improvements include reduction of cold biases in the simulated surface temperatures, but some features become worse because of the introduction of warm biases as the model becomes overall warmer and drier. A 1/2 degree grid for river routing has been developed based on earlier code provided by Jay Famiglietti. Plans for the future include building on the Land Surface Model (LSM) code for carbon fluxes in collaboration with the CCSM Biogeochemistry Working Group.
Gent reported that the CCSM ocean model is based on POP version 1.4, Greenland Pole, an open Bering Strait, a new horizontal viscosity, new discretization of Gent-McWilliams, improved equation of state, addition of river runoff and marginal seas, x3 version has a equilibrium solution, and there are capabilities to add biogeochemical links. Future plans are to add partial bottom cells; Beckmann and Doscher bottom boundary layer scheme; Visbeck et al. isopycnal coefficients; spatially varying background vertical diffusivity and viscosity; move to use freshwater flux; monotonic advection scheme, and coupled ocean, sea ice, and runoff simulations.
Hack reviewed the status of the AMWG working group activities since the last SSC meeting in January 2001. Additional diagnosis of the simulations for the model configuration proposed at that time (two time level semi-Lagrangian dynamical (SLD) core, updated longwave radiative transfer scheme, CSU Arakawa-Schubert cumulus convection scheme) led the working group to reconsider the available configuration options. An alternative configuration, which replaced the SLD core and CSU cumulus convection with a three-time-level reduced-grid spectral dynamical core and a modified version of the Zhang-McFarlane cumulus convection to include a precipitation evaporation term, was selected as the preferred configuration. The main strengths included improvements to the distribution of precipitable water, to the surface solar radiation budget in the eastern oceans, and to the eastern Pacific surface stress. The working group feels this configuration provides an improved physical framework over the CCM3, along with some significant simulation improvements that will be beneficial to the coupled system model. The working group is also satisfied with the experimental and evaluation framework that has been developed over the last year, and was central to the selection process.
Kiehl reported that the fully coupled model runs have been started and problems are being found. Debugging and detailed analysis are underway. His goal is to have stable code by the end of the summer and to freeze the code by the end of December 2001. After an adjustment period and all the issues are addressed, the code efficiency will be a priority, and he proposed that long integration simulations would be started in early fall.
Kiehl also reported that the Lin-Rood dynamical core would be tested in the summer/fall timeframe in the fully coupled mode.
Fung requested that all simulations be run with SF6 as a standard. Fung reported that the CCSM Biogeochemical Working Group (BGCWG) has done a lot of work on model development for a new terrestrial carbon model. In the future this group will add water isotopes, C13, dust, ocean biogeochemistry, nitrogen, and human dimensions.
Kiehl encouraged all the working groups and especially the co-chairs to help CCSM gain more visibility nationally and internationally. He also discussed the SSC's decision to send letters of appreciation for individuals that have gone above and beyond in their CCSM contributions. He encouraged the working group co-chairs to send information to the SSC on anyone they think should receive a letter of appreciation. He also encouraged the working group co-chairs to think imaginatively about long-term ideas, such as physical processes, biological processes, new components, etc., and discuss them with him for future planning purposes.
9. General Discussion with Working Group Co-Chairs. DeLuca described the CCSM computational infrastructure development that included completing the Software Engineering Plan, hiring of a software engineering manager, creating a software engineering group, and writing a software developer's guide with standard templates. She recommended that a computer scientist be hired to work on the CCSM project.
10. Frequency of Release of New Versions of CCSM. Fung stated that changing naming conventions poses a problem for the BGCWG because they have to go through their code and change the naming conventions then. Washington stated that several versions would be needed, such as paleo low resolution, climate change high resolution, and different dynamical cores. Dickinson stated that the model should be frozen for 5 years with release of additions to model codes that effect the community when necessary. Bader stated that too many options would cause instability and different results, so he recommended keeping the number of versions to a minimum. Craig recommended one or two core releases per year. From a software engineering point of view, different types of computers, different dynamical cores, and different configurations will need to be addressed.
The other issues that arose during this discussion were: (1) who would have access to the model codes? (2) when would researchers have access to the model codes? (3) how could subgroups have access to the component model codes, so they could add to the development of the community model? (4) is it an NCAR code or a community code? Blackmon suggested these issues be discussed in the working group meetings and input be sent to the SSC.
Kiehl will check with the UCAR Intellectual Property staff regarding copyrighting the CCSM code. UCAR/NCAR will hold the copyright for the community.
Kiehl will form a committee including himself, Craig, etc., to draft a policy on frequency of release and send to the SSC and Working Group Co-chairs for comment and approval.
11. Approval of CCSM2 Component Models. Until a few fixes are input, code efficiency is addressed, and a long control run is completed, the component models cannot be approved.