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

Minutes of the CCSM SSC Meeting

Held 11/10/06 at NCAR in Boulder, Colorado via Access Grid, Webcast, and Conference Call

Attendees: Peter Gent (Chair), Chris Bretherton, Steve Ghan, Scott Doney, Bill Collins, Bill Large, Jim Hurrell, Danny McKenna, Gordon Bonan, Mariana Vertenstein, Jay Fein, Dave Bader, Anjuli Bamzai, Lydia Shiver, and Emily Doremire

1. Update on C-LAMP and carbon cycle runs in T31x3. Doney stated that his presentation is from the Biogeochemistry Working Group perspective. He reported:

In summary he said that the impacts appear smaller than expected, and a preliminary look suggests including carbon may not cause the whole climate simulation to derail.

Doney moved on to give an update on carbon model spin up, stating that the land and ocean carbon are sensitive to the coupled biases, and the working group is trying to minimize the physical and biogeochemical coupling shock because of the long response times for the carbon reservoirs. He reported that there is a stable coupled carbon-climate run using CCSM3.1 T31x3. The working group is using an incremental coupling procedure, starting with the land and ocean biogeochemistry model to let the biogeochemistry adjust to the physics, then adjusting the biogeochemistry model to the atmospheric CO2, then coupled physics and radiative fully coupled carbon-climate. They are using a fixed atmosphere of carbon of 280 ppm, which is pre-industrial. In their initial attempt to run the model there were large internal drifts, so they are trying a new procedure. The new spin-up method uses a fully coupled physical climate with ocean carbon and runs to less than 0.01 PgC/year equilibrium with CO2 equal to or greater than 285 ppm. They will use 150 years of the atmosphere output and ocean CO2 to equilibrate the offline land model to less than 0.01 PgC/year.

Next Doney reported on the CCSM Carbon LAnd Model Intercomparison Project (C-LAMP), stating that they are developing observational data sets and metrics to evaluate the land carbon model and improve the CCSM Land-BGC model by comparing CLM-CN, CASA’, and LSX-IBIS against data; running simulations in CCSM3 at T42 at Oak Ridge using the Climate End Station; and archiving and making available the results at PCMDI. The simulations are out to 2500 years now, and the working group will begin 19th and 20th century runs in the next week. Doney thanked Washington, Hoffman, and Drake for their support of this project. The CCSM C-LAMP protocol is outlined below, and the protocol, output fields, and metrics for model evaluation are available at http://climate.ornl.gov/bgcmip/.

Collins and Gent asked Doney when the BGC Working Group would be ready to couple their model with the new CCSM model, and he said as soon as the new physics in the model are ready, they will be ready to couple. Doney also reported that because the scientists working on this new biogeochemistry model need time to publish their C LAMP results and spin-up experiments, that moving on to choose a CCSM model to fully couple to will bring tension. Bonan stated that there is a lot of development in the code base in all areas of the model and a common framework is needed. Bonan suggested that one model be chosen and another model continue to be further developed offline, noting that would further dilute resources. Large asked if the SSC needed to make the code base a priority for the CCSM project, and also if the SSC should steer groups to include “credible” nitrogen and transient vegetation. Doney responded that a small group of scientists is needed to lead the carbon project and that buy in to the model chosen is needed. Large said that if resources are needed, the SSC should be informed.

2. Update on coupled run with ocean changes. Large reported that immediate development is ongoing in the ocean component, and the Ocean Model Working Group’s hopes for significant improvements did not happen when the changes were coupled with the finite volume dynamical core model. The working group is also working on parameters for thickness distribution, moving from 40 to 60 levels, a new ocean grid, and incorporating results from the CLIVAR CPT teams. Large stated that the new ocean component should be ready to couple during the first part of 2007.

3. Update on coupled run with atmosphere changes. Collins reported on the atmospheric changes under consideration, which are boundary layer, convection, cloud fraction, sub-columns, radiation, microphysics, gravity waves, aerosols, and chemistry. The plans for these changes are on the Atmosphere Model Working Group (AMWG) Webpage.

There is a community convection intercomparison project ongoing with 3 groups (Zhang, Wu, Neale and Richter) working on the changes. A decision on the new parameterization will be made by the AMWG external co-chairs only. Large stated that Neale and Richter's work shows improvements in CCSM3’s cold bias along the equator. Hurrell asked when the new convection parameterization would be chosen, and Collins replied that a preliminary determination would be made in early 2007, and then it would be discussed at the AMWG meeting. Large suggested that the SSC update the implementation plan that was discussed at the CCSM annual workshop in Breckenridge with all the working groups. Hurrell asked if more resources are needed to improve the tropical biases. Collins responded that more resources are needed to analyze results, and Bretherton stated that help was available once the runs are done and posted to the Web. It was also stated that more resources are needed.

4. Update on land component. Bonan updated the SSC on the Community Hydrology Project, stating that the changes implemented help the soil water storage. A manuscript is being written now on the changes, and the software engineering tasks are being done and they should be completed in the next month. A new data set is also coming from this work, and the latest CAM has the new hydrology project included already. Some improvements are a dramatic improvement in the latent heat flux, changes in runoff, and improvement in surface energy balance. The Land Model Working Group (LMWG) needs to do some runs with the new CCSM development code as soon as possible. Large asked about the computational problems with the runoff scheme, and Bonan stated that the river routing model was developed years ago and no development has been done since then, so it needs to be revisited.

5. Update on sea ice component. Gent reported that the Polar Climate Working Group (PCWG) is moving the sea ice model base code to CICE4.0, and is also working on improving the ridging formulation, and the vertical thermodynamics. A melt pond parameterization and improved shortwave radiation treatment are also being tested. The influence of these modifications is being assessed now. The PCWG anticipates that the new sea ice model (CICE4.0) will include all of the new modifications. Collins asked about Lipscomb’s ice sheet model, and Gent stated that it would probably not be included in the interim version of the CCSM model. Gent said he would contact Lipscomb for a status report. Vertenstein reported that Lipscomb thinks the fully coupled ice sheet model will be ready by mid-2007, but that the sequential system must be in place. Collins stated that the ice sheet model is needed, and Vertenstein said that Lipscomb will be working with the LMWG.

6. What components are recommended for early 2007 version? Gent reported that the land and CICE components are in good shape and ready for an interim version of CCSM in early 2007, the ocean component is still having changes made to it and more work needs to be done, and the atmosphere component is deciding between new convection schemes with decisions coming in February after its working group meeting. Collins stated that Bretherton’s new convection scheme is ready to be input into the new model also, but he thinks the cloud microphysics scheme needs more analysis before including it. Gent asked how the indirect effect of aerosols could be handled without the microphysics scheme. Ghan said he is working on this, but will need several months, and Bretherton suggested putting in the microphysics and not turning on the aerosols in the early 2007 version. The SSC decided it was acceptable to not have the indirect effect in the early 2007 version.

Gent said that the target for the new model versions is the end of February 2007, and that on 1 March 2007, the new model components should begin to be coupled. This early version will be called CCSM3.5.

Bretherton wants some simulations without the biogeochemistry included. Large proposed that the model components begin to share information on the spin up right away and not wait until 1 March to figure out the procedure. It was decided that a group from all working groups is needed to work on the spin up of CCSM3.5. Large also said the issues of tuning and drift need to be explored. Collins said another issue is what CAM would be tuned to.

7. Discussion of using CCSM for short-term simulations. Gent updated the SSC on the background of this issue, stating that it has been discussed by Sundquist and Crowley at the last CAB meeting, at the SSC/CAB/WG co-chair meeting at last year’s CCSM workshop, at the Aspen WGCM/AIMES workshop held in August 2006, and on 6 November 2006 at a CGD meeting. The short-term simulations proposed are:

The advantages of short-term simulations are:

Large asked if CCSM would use other models to compare, or use CCSM with different physics as the multi-model ensemble. Collins said that several institutions/modeling groups will do short-term simulations, so those can be used for comparison.

The challenges of short-term simulations are:

McKenna stated that CCSM could do chemistry with a chemistry transport model (CTM) or regionally. Ghan said CCSM should have aerosols included even if they are done offline. Bretherton stated that we must do highest resolution for the impacts community. Bonan said that an interactive carbon cycle should be included so that wildfires and drought can be looked at. Bader said that if several things are added, then the simulations will not be that useful.

In summary, the next steps are to modify the CCSM Implementation Plan by adding the short-term simulations work and sending it out to the community by posting it on the Web. Gent also stated that the top priority for CCSM is to reduce the biases.

8. Computing. Gent asked Vertenstein to update the SSC on the use of computers at Oak Ridge National Lab. CCSM3 is up and running on Jaguar, and we can start running experiments with the development code changes right away. Vertenstein also discussed the data transfer problem, and the need for a special climate queue so that we can run for 24 hours a day instead of in 2 hour chunks of time. Bader said he would lobby for this for CCSM. Vertenstein then stated that there are many problems using Phoenix (Cray X1), and the performance is about one-half what was anticipated, and there are also IO problems. She recommended that we use Jaguar once the development version of CCSM is validated on it. The CCSM Climate Change Working Group wants to use the standard CCSM3 released version that runs very well on Phoenix.

9. Development code access policy. Vertenstein discussed with the SSC members how to clarify who can have access to the development code and what is the decision process. The SSC decided that it is the responsibility of the CCSM Working Group Co chairs to approve access to the development code, and Gent will send an email to them stating that is the process. The SSC requested Vertenstein add a question to the request form, so it is known what Working Group Co-chairs to contact about access.

10. 2007 Workshop. The SSC decided that the 12th Annual CCSM Workshop will be held at The Village at Breckenridge the week of 18-22 June 2007 with a SSC meeting on 18 June, and the workshop 19-21 June 2007.

11. Other Business. Collins recommended that the SSC discuss CCSM computing and storage needs for 2009 and 2010 with CISL. Large suggested that a discussion take place with the CCSM community about culling of CCSM data already stored. Shiver asked the SSC about their thoughts on having extra days of working group meetings at the workshop. Gent said that component model working group meetings must be held separate from, and well before the workshop, and that if a working group decides to have additional days of meetings at the workshop, CCSM would not pay those expenses. Gent again thanked Lydia Shiver for her years of work on the CCSM project, and introduced Emily Doremire as the new CCSM Administrator.

-notes by Lydia Shiver

For questions or comments on this page, please contact ccsm-admin@cgd.ucar.edu.

Last updated: 1/3/07 EAD