May 9, 2000
Dear CMIP2 participant,
At the last meeting of the CLIVAR Working Group on Coupled Models (WGCM) in Hamburg in September, 1999, it was decided to expand the CMIP2 database to include many more model fields, and daily data if possible. This additional data would greatly extend the scope of diagnostic subprojects. The issue then became one of how to go about collecting this additional model data. After much discussion over the past few months and upon reviewing current data processing and archival capabilities at PCMDI, in addition to looking at what is feasible with current technology, it has been decided to ask modeling groups to supply complete history tape data (atmosphere, ocean and sea ice, and daily data if available) for a control run of at least 80 years and a 1% per year CO2 increase integration of 80 years. We are calling this "CMIP2+" since it is taking the CMIP2 concept and expanding it to include more data.
This may seem a bit overwhelming at first, but recent experience with transferring large amounts of complete model history tape data to PCMDI from NCAR and GFDL has demonstrated this is feasible. The advantages for subprojects are numerous and obvious. By having a complete set of history tape data from a number of different global coupled models (and hopefully daily data from at least several models), subproject investigators can compare virtually any diagnostic that is possible to derive from the model data. The advantage to modeling groups is that they will not be required to strip out a lengthy list of selected fields, only to be asked repeatedly for different fields later.
After our experiences with the recent model data transfer exercises mentioned above, we have decided the best way to do this is for modeling groups to translate their history tapes into netCDF format before transferring the data to PCMDI. This has become fairly routine in the last year or so for many modeling groups, and hopefully will be relatively straightforward. If you are not sure how this can be done, please contact PCMDI for more information. When formatting their data in netCDF, groups are encouraged to follow a structure designed for climate model output, such as the GDT conventions described at http://www-pcmdi.llnl.gov/drach/netCDF.html. One way of enforcing these conventions is by invoking the COARDS option in the PCMDI-developed LATS software (which was used for AMIP2).
This request represents a significant advance in the area of model intercomparison. A traditional drawback of such exercises has always involved arguments over which fields to request and at what time resolution. Literally by definition this meant certain studies could not be done, and it was always extremely difficult to anticipate which fields would be most useful for subprojects. Clearly what the CMIP Panel has come to call the "send everything" option dissolves this major roadblock and will allow more comprehensive treatment of model data by subproject investigators than ever before. However, this will require a bit different way of doing things for modeling groups, though the netCDF translation process has become fairly routine at a number of modeling centers. Again, if you or your group is not familiar with netCDF, PCMDI can provide information to help you. A first point of contact there is Curt Covey (firstname.lastname@example.org) and he can direct you further.
Hopefully this data request can be fulfilled during the next month or two. When we have a sufficient representation of CMIP2+ data in the archive at PCMDI, we will send out a new subproject announcement soliciting analyses of the expanded database. Our target for sending out this announcement is early August, 2000. If you anticipate delays in fulfilling this request, please let me know as soon as possible.
On behalf of the CMIP Panel (Jerry Meehl, chair; George Boer, Curt Covey, Mojib Latif, Ron Stouffer) I thank you for your support of CMIP so far, and look forward to your participation in CMIP2+!
Jerry Meehl (for the CMIP Panel)
August 3, 2001
This is an invitation to submit Proposals for Diagnostic Subprojects to analyze model output collected in "CMIP2+", the Coupled Model Intercomparison Project Phase 2+, wherein complete history tape data (i.e. all fields, monthly means, control run and 1% per year CO2 increase run) from five global coupled models (NCAR CSM, DOE PCM, HadCM2, HadCM3, GFDL_R30) are being made available for analysis. In this initial announcement, atmospheric model history tape data (including SSTs on the atmospheric grid from a few of the models) are now available. In subsequent announcements, complete ocean and sea ice data will also be made available, in addition to full history tape data from several models of daily data for shorter periods from the simulations. This massive data collection and archival operation is being coordinated by PCMDI at Livermore, with Curt Covey as project lead there.
Recall that CMIP1 involved control runs from global coupled models, and CMIP2 is an intercomparison of climate change experiments from global coupled models where atmospheric CO2 is increased 1% per year compound. As in previous model intercomparison exercises, both CMIP1 and CMIP2 included only a subset of fields from the coupled models. This necessarily limited the range of analyses that could be performed.
In CMIP2+, we have collected and archived ALL available history tape data from five global coupled models for, at minimum, a 100 year control run and a 1% per year compound CO2 increase experiment. The collection of total history tape data from multiple global models for the purpose of model intercomparison is unprecedented. This presents a unique opportunity for subproject investigators, especially with regards to analyzing standard fields on the history tapes. Some non-standard fields whose formulations vary among the models (e.g. clouds, diabatic heating, etc.) may require more specialized analyses.
Though this dataset represents everything that was saved by the original model groups, the accompanying caveat is that some groups may not have archived data in the same way, or not saved some quantities.
The intent of the CMIP2+ subprojects is to encourage in-depth analysis and intercomparison of the CMIP2+ simulations by experts in the community. [See general description of CMIP1 and CMIP2 in the following two papers: Meehl, G.A., G.J. Boer, C. Covey, M. Latif, and R.J. Stouffer, 1997: Intercomparison makes for a better climate model. Eos, 78, 445--446, 451; Meehl, G.A., G.J. Boer, C. Covey, M. Latif, and R.J. Stouffer, 2000: The Coupled Model Intercomparison Project (CMIP). Bull. Amer. Meteorol. Soc., 81, 313--318.] As in previous CMIP phases, the CMIP Panel (Jerry Meehl, chair, Curt Covey, Mojib Latif, Ron Stouffer, and Bryant McAvaney) will facilitate this effort with the intent that subprojects: 1) have scientific merit, 2) have a high probability of being completed, and 3) are coordinated appropriately with one another and with the modeling community. The "participation protocol" and "collaboration protocol" are posted on the CMIP web site at diagsub.html.
If you have a previously approved CMIP1 or CMIP2 subproject, you need not formally re-apply for CMIP2+. Simply email Curt and me and let us know you will continue your subproject with the CMIP2+ data. Curt can then give you instructions on how to proceed.
If you are interested in initiating a new CMIP2+ subproject, please submit short proposals of no more than 1000 words (roughly 1-2 pages excluding references), consisting of:
- Short title.
- Name and address of proposed subproject leader(s).
- Background to the intended research.
- Objectives: A discussion of the intended diagnostic/intercomparison study and its importance.
- Methodology: e.g., specific computations and diagnoses to be made, sources of validation data to be used.
- References: Including a listing of relevant publications by the proposed subproject leader(s).
There is no deadline for proposal submission since this will be an ongoing process. As each proposal is received by the CMIP Panel, it will be reviewed and acted on. The goal is for rapid turn-around to facilitate access to the model data.
Proposals should be sent directly to Curt Covey at PCMDI (email@example.com).
Upon receipt of a proposal at PCMDI, it will be forwarded to the CMIP Panel members. Once the proposals have been evaluated by the CMIP Panel, subproject leaders of successful subproject proposals will be advised by the CMIP Panel Chair on how to proceed with the establishment of a subproject.
This announcement is being sent out to the original CMIP mailing lists but its distribution is by no means limited to people on those lists. Please forward this message to anyone who may be interested in initiating or participating in CMIP2+ diagnostic subprojects. Since we have several versions of CMIP email alias lists, we apologize if you receive multiple copies of this announcement.
Jerry Meehl (for the CMIP Panel)