Minutes of the
CCSM Scientific Steering Committee Meeting
Wednesday, 7 April 2004
NCAR VisLab and Other Access Grids
Attendees: Bill Collins (Chair), Chris Bretherton and Cecilia Bitz via Access Grid at Boeing, Jim Carton via Access Grid at University of Maryland, Scott Doney via Access Grid at Woods Hole, Jim Hack, Bill Large, Danny McKenna, Gordon Bonan, Jay Fein, Dave Bader via LLNL Access Grid, Jeff Amthor via conference call, Phil Merilees, Tom Henderson, and Lydia Shiver
Members unavailable: Maurice Blackmon, Ben Santer, and Steve Reid
1. Welcome. Bill Collins welcomed Tom Henderson, new CCSM Software Engineering Working Group Co-chair and CCSM Software Engineering Manager replacing Tony Craig, and Jeff Amthor, DOE Climate Change Prediction Program (CCPP) Manager. He thanked Lydia Shiver and the Access Grid operators for their support of the meeting, especially Darin Oman, NCAR.
Collins discussed some action items from the last SSC meeting held in November 2003. The SSC can make recommendations to Anthes, Killeen, and Fein regarding who the new chair of the CAB will be. Shiver will send the CAB member's list to the SSC for their recommendations.
Collins updated the SSC members on the status of the Earth Simulator (ES) project. He reported that the MOU with CRIEPI/MEXT was signed and awaiting Killeen's signature.
2. Status of CCSM3. Collins outlined the schedule for the release of CCSM3 to the community, stating that model codes, data, and scripts would be frozen on 30 April 2004, and the code would be released to the community on 2 June 2004. The CCSM is designed to run on the IBM SP, HP Compaq, SGI Origin and Altix systems, numerous Linux clusters, and NEC and Cray vector systems. The range of platforms illustrates the flexibility of CCSM3. The model can be run at T85 with 1-degree ocean, T42 with 1‑degree ocean, finite-volume dycore 2x2.5 atmosphere and 1-degree ocean, and T31 with 3-degree ocean. Several biases have been reduced, and some challenges remain.
Due to NCAR's supercomputers being hacked several times, SCD has highly restricted access to CSL resources. CCSM has gotten special access so that the IPCC runs can continue. This down time could lead to a slippage in the release schedule. The SSC members requested that the details of exactly what was being released be posted on the SEWG web page, and Henderson agreed to that. Large suggested a sequential release if the schedule slips past 2 June 2004, and another suggestions was to use the ES runs for the control experiments and substitute the ES data for the control data. A serious "bug" was found in the code that increments CO2 in transient CO2 experiments. The volume mixing ratio and mass mixing ratio were not incremented in a consistent manner. This bug affects some results of CCSM2, which are used in papers published and in press.
3. Status of IPCC using CCSM3. Collins reported that there are 3 phases for the IPCC: pre-industrial, 20th century, and 21st and 22nd centuries. Collins made a motion that the SSC request an experimental design of forcings plan and implementation template on the IPCC runs from the Climate Change Working Group (CCWG) to use as a template for future IPCC experiments. After discussion, the SSC agreed that Collins should direct the CCWG to document the design now and discuss the implementation template in detail at their working group meeting in the summer and report back at the SSC/CAB/WG meeting. Fein reported that future assessments should come from more than the U.S. and that the expectation is that the assessment would foster international discussions.
4. Status of Biogeochemistry. Scott Doney reported that the Biogeochemistry Working Group (BGCWG) met on 30 and 31 March 2004, and the group finalized plans for IPCC runs using CCSM1, CCSM3 with carbon, and CCSM3 with carbon, dust, etc.
He stated that the group will need GAUs and programming support to integrate, test, and evaluate isotopes in CCSM3, and they will need several T31x3 runs for IPCC done very soon. Collins stated that he and Tom Henderson are working on programming support for the BGCWG. Doney said the computational burden will increase by a factor of 2, but that throughput is the issue.
Doney reported that a reactive atmospheric chemistry project is under way within CCSM, and ACD, CCSM, national laboratories, and universities. An offline chemical model is being developed to replace MATCH and MOZART, and an online chemical model is being developed, which is probably WACCM. The BGCWG is planning to develop a new ecosystem dynamics and biogeochemistry model for atmosphere and land in one year to one and a half years.
5. Status of Journal of Climate Special Issue. Collins reported that the focus of the special issue would be on the coupled system, one overview paper will be written with SSC members as co-authors, the submission date of all papers is November 2004 unless this slips due to computing down time, and the issue could double in size from the first one. The pruning of the number of papers will be discussed further, and a final list will be completed by the end of April, so we can begin to seek funding.
6. Future Directions and Agenda for SSC Meeting at Workshop. Collins stated that he thought that the near term (1 to 2 years) priorities for CCSM were to focus on 2 areas: 1) the first generation coupled chemistry-climate model, and 2) high-resolution process oriented climate modeling. He believed it important to make common cause with ESMF and SciDAC projects for software engineering support. He discussed a list of guests to invite to the SSC meeting in July to discuss future topics, such as tropospheric chemistry, isotopes, chemical transports, etc.
The SSC recommended that the focus of the next SSC meeting be rebalanced using CCSM3 as a driver for high-resolution experimentation and reduction of biases (tropical and physical). The other topics will be the success of the SGERs and release of CCSM4. Collins will send a revised agenda for SSC comment.
7. CCSM 9th Annual Workshop. Collins discussed the venue in Santa Fe, the use of webcasting and wireless connectivity, the agenda, and noted the SSC meeting before the workshop and the SSC/CAB/Working Group Co-chair meeting after the workshop.
Next year the Distinguished Achievement Award nomination request will be sent to the SSC and Working Group Co-chairs, will not allow SSC members to be nominated, and only the SSC will vote on the winner of the award.
8. Next CSL Proposal. Collins announced that the CCSM CSL allocation progress report is due at the end of April, and an email request was sent to the working group co‑chairs for their input. The next CSL proposal for the time period 1 August 2004 to 31 December 2005 is due on 28 June 2004. Jay Fein felt that the deadline should be delayed until after the workshop in order to have a final look at the proposal. He agreed to discuss a new mid-July deadline with Ginger Caldwell in SCD. It was agreed that every working group would have a development and a production allocation, and that a draft CSL proposal would be available at the workshop for all working groups to discuss and note any problems so they can be remedied before the proposal is sent to SCD. The total length of the development and production proposals should be less than or equal to 35 pages each. Collaboration with CPTs should also be considered, if appropriate.
Collins also noted recommendations from the panel were to better coordinate the working group components between each other, include a time request from the Ocean Model Working Group in the proposal, list integrations in order of priority, clearly distinguish between development and production runs, balance very high-resolution (T170) CAM tests and ensembles of high-resolution (T85) runs, use larger ensembles for Paleoclimate Working Group (PaleoWG) experiments
, include discussion of the role of ESMF, include discussion of performance and scalability, highlight code performance, and enable comparison to other application types.
9. Terms of Working Group Co-chairs. The working group co-chairs who have agreed to be reappointed for another two-year term are Leo Donner (AMWG), Gordon Bonan and Steve Running (LMWG), Jerry Meehl (CCWG), Scott Doney (BGCWG), and Cecelia DeLuca and Patrick Worley (SEWG). The working group co-chairs who will rotate off after this year's workshop are Dick Moritz (PCWG) and Inez Fung (BGCWG). A third co-chair, Steve Jayne, will be appointed to the OMWG, due to Rick Smith's accident. Lisa Sloan has stepped down as PaleoWG co-chair, and the SSC discussed her replacement. Jim Randerson will replace Inez Fung. Shiver will prepare letters of appointment and thank you letters.
Large suggested that working group co-chairs' terms should not end the same year so there is some overlap.
10. DOE Update. Jeff Amthor reported to the SSC that he had directed several DOE groups to make biogeochemistry a higher priority, and he is pushing Livermore staff to participate with CCSM on sulfur chemistry. The plan is to have a version of CCSM with sulfur chemistry running on the Cray X1 at Oak Ridge within one year. Amthor stressed that DOE wants to collaborate and not compete with CCSM activities, and that he is open to revisiting funding and priority issues for biogeochemistry with CCSM.
11. NSF Update. Jay Fein reported that the budget outlook beyond 2005 is up in the air, especially because it is an election year. He stated that generally budgets were bleak, so to make scientific progress, we will have to be very disciplined. The SSC stated that if NSF, DOE, and other agencies want a more complex next generation model, new and more funding will be needed.
12. CCSM Implementation Plan. Phil Merilees updated the SSC on the progress of the CCSM Implementation Plan. He stated that Phase 1 was completed with 3 new hires (George Carr, software engineer working on vectorization and portability; Zav Kothavala, associate scientist working as paleoclimate liaison; and the Climate Change Working Group liaison) and 1 reassignment (Sylvia Murphy, associate scientist working with the BGCWG). An offer will be made to a new Scientist I that correlates to the CCSM Business Plan.
In Phase 2 (FY05), we will entrain the new Scientist I and add two ASP postdocs whose interest is atmospheric chemistry, with all 3 being paid by NCAR. The Business Plan calls for the addition of one project scientist and one software engineer and the initiation of a CCSM Visitor Program. We will look for funds for Phase 2. Merilees noted that CCSM would need new funding to go to Phases 3 and 4.
Some SSC members stated that reallocation of support staff working on CCSM is needed based on the highest priorities of the project.
13. Status of CPT and GFDL Interactions. Jim Hack and Chris Bretherton reported that the atmosphere CPT has done comparative analysis of model simulations from NCAR, GFDL, and GMAO. There is funding in this CPT for a liaison position and interviews are being conducted. GFDL has hired a liaison. There are plans for meetings of the atmosphere CPT later this year for further collaboration between NCAR, GFDL, GMAO, and others.
Bill Large reported that there are 2 ocean CPTs and that the OMWG did some base runs for ocean only and ocean and sea ice, developed a data set, and sent it to GFDL for them to run and analyze both NCAR and GFDL's models. There will be an Ocean CPT meeting in June at GFDL, and another meeting is scheduled in September.
14. Update on SciDAC. Bill Collins stated that he was the NCAR PI for the SciDAC project now, and he plans to participate more. Jeff Amthor stated that DOE sees SciDAC as a quite positive collaborative effort, that the software engineering collaborations have been very successful, and that the needed focus for this project is toward biogeochemistry and atmospheric chemistry. Doney stated that Amthor has been in communication with him, Fung, John Drake, and others to make progress on biogeochemistry. Amthor encouraged everyone to find more effective ways of communication, because he sees this as a bottleneck because of the scattered locations of the project staff.
15. Update on ESMF. Cecelia DeLuca joined the meeting to update the SSC on the status of the ESMF project. She reported that in May 2003 the first prototype code was delivered and a community meeting was held. In July 2004 there will be a second release of the code, and another community meeting is planned at NCAR. She said that three experiments were completed in March 2003, in which different pieces of code from different institutions were coupled using ESMF. These experiments included CAM‑MITgcm ocean, CAM-NCEP SSI data analysis, and GFDL B-grid atmosphere‑MITgcm ocean. The experiments showed that the code could be turned into ESMF components fairly quickly. They also showed that the same model could be used in different contexts with no source code change, e.g., CAM coupled to both MITgcm ocean and SSI. The SSC requested criteria for CCSM and how ESMF's ease of use can be evaluated. Collins said the evaluation of CCSM with ESMF would take place in the next nine months. He also said that further discussion by the SSC is needed on how ESMF and CCSM would continue and what CCSM resources would be needed.
DeLuca suggested that CCSM start with CAM and evaluate that in the ESMF for added value, and then proceed to the other components. The ESMF milestones require that the CCSM submit a version of the model that uses ESMF for coupling to NASA by the end of the year.
DeLuca stated that NCAR is looking for future ESMF funding opportunities. There were comments that the ESMF-CCSM evaluation schedule does not fit the fiscal year budgeting process of NSF. To read more details about the ESMF project, please go to www.esmf.ucar.edu.
Grid Use. The SSC members liked the Access Grid approach to one
day meetings and said they would be willing to use this technology again.
Community Climate System Model