The first day and a half was spent listening to talks from various groups directly or indirectly involved in TransCom. Click here
to see the agenda and attendee contact list.
The second half of day 2 was spent discussing the next steps in the TransCom experiment. The following is a summary of this discussion:
We will consider a live access server in order to make access to the submitted model output available to all the participating modelers. Jorge Sarmiento will provide Kevin Gurney with contact information regarding this approach (got it, thanks Jorge).
At a minimum, we will immediately place the high-frequency station data on the TransCom 3 website. We will also provide the annual mean pre-subtracted and basis function fields at the surface. Finally, we may make available the G matrices (response functions) and a series of observational datasets.
We will create a discussion list for the T3 analysis. Groups can individually propose analysis so that there is not unnecessary redundancy and tabs can be kept on the various bits of work that are occurring on the T3 submissions. Such a list might include the group, the proposed project and deadlines.
We will password protect the portions of the T3 website that contain the model output.
Some of the basic descriptive plots that have already been made at CSU will be made available through the website.
Level II logistics
There was some discussion on adding some extra high-frequency stations to the existing list. Philippe Ciais suggested that he could provide these extra stations. For those modelers that have not performed the level II analysis yet, these stations could be formally included in their forward integrations.
Given the computational demands of the Level II forward integration, many modelers expressed a desire to shorten the forward simulation (currently set at three years in the T3 protocol). Rather than pick a particular amount of time, modelers can decide independently when to truncate their forward integration. However, the submitted output should follow the protocol which in most cases requires three years of output. Modelers can use an extrapolation scheme to complete the time series for all the required fields.
There was also some interest in having the Level II results broken into smaller, more manageable netCDF files. Kevin Gurney will work on rewriting the netCDF routines.
Set a new deadline on the level 2 results to August 31st.
The Princeton group volunteered to perform inversions across the T3 models using their own code.
The Ciais group proposed to run an interannual inversion across the T3 models.
The UCI group proposed to explore the implications of both CO oxidation and stratospheric transport to the CO2 inversion results.
There was interest in beginning the process on putting together a level 3 protocol. Martin Heimann was willing to begin this.
Given the large amount of analysis that can be done on the TransCom 3 model output, there was some discussion on maintaining the core goals of the project. The two that continue to stand out are formalization of the model to model differences and the ability of this project to contribute to observational network optimization. To this end, we may want to expend considerable energy ordinalizing the TransCom 3 results. Outcomes like the preliminary result in the southern oceans may be an example and could contribute beyond the inversion community to the oceanographic community; providing a challenge to carbon cycle work in that community. Concern was also voiced against presenting the TransCom experiment as 'solving the missing sink'. Individual modeling groups are quicker and can be more innovative in the near term - TransCom is a cumbersome process by it's very nature and perhaps cannot be on the 'cutting edge'. There is some obvious tension here, but clearly stated goals should help in this regard.
The 3rd TransCom 3 meeting will be held in Aspendale, Australia sometime the week of March 19th, 2001. This will be structured as a working meeting with more time devoted to analysis of T3 results and paper-writing. The October, 2001 meeting in Sendai, Japan will afford us an opportunity to propose a special TransCom session and present the level II/III results. This could be viewed as s deadline for having very mature Level II and Level II results ready.
Many TransCom participants will likely be attending the Fall AGU meeting in San Francisco. We will try to organize an informal meeting on TransCom at AGU.