3.0 Executive Summary

3.1 Summary of Findings

The North Atlantic is unique among the major ocean basins as a strong CO2 sink, characterized by marked seasonal drawdowns of DIC and pCO2, which are driven by a spatially- and temporally-varying balance of physical and biogeochemical processes. In spite of intensive, but temporally- or spatially-limited studies of these large signals, carbon budgets have not been closed or successfully balanced, even to a first order. We are still unable to reconcile different geochemical and biological estimates of key processes of carbon fixation, transformation and export in the upper ocean. The problem is mainly due to poor conceptual understanding of the fundamental biogeochemical processes and their linkages to the physical climate system, which lead to inadequate sampling strategies. Uncertainties in the measurements, which in some cases are close to or even larger than the expected signals, also contribute to uncertainties in carbon budgets.

A new generation of coupled observational and modeling studies is likely to be the key to significantly improving our understanding of the processes driving the carbon cycle, refining flux estimates, and balancing local- to basin-scale carbon budgets. Such studies are ideally broadly collaborative among various national JGOFS and other global change programs (GLOBEC, IGAC, WOCE, EOS, CLIVAR, etc.), and last more than one field year. They ideally include a mix of the elements of the original JGOFS components of process studies, time series and survey operations, and the use of state of the art technologies including remote sensing and autonomous chemical and optical sensors.

3.2 Scientific Presentations

The first 1.5 days of the meeting were devoted to a series of presentations which are summarized here. Reports of each presentation are included in the remainder of this report. Presentations were divided between observational and modeling perspectives. The scientific rationale for a control volume study of carbon fluxes was presented by Tony Michaels and Nick Bates of BBSR. The term Control Volume Experiment has several different connotations to different people, including a very exact sense understood primarily by physicists. In the discussions at Bermuda, and in this report, it should be taken to mean any one of a class of linked observational and modeling studies seeking to estimate fluxes of mass through some three-dimensional volume of water, and transformations between chemical states within it. Observations of the annual carbon cycle at the BATS site and budgets constructed from sediment traps and other flux estimates suggest that our conceptual understanding of the carbon cycle, and/or our technical capability for observing that cycle, are still inadequate to close the budget. Well-designed CVE's, including real-time modeling studies, were suggested as one path toward better closure of local carbon budgets.

Taro Takahashi and Catherine Goyet presented summaries of their observations of pCO2 and TCO2 variability in the North Atlantic and equatorial Pacific oceans. Although the variability cannot be predicted without a better understanding of the processes generating it, Takahashi and Goyet were proponents of the idea that the variability itself should be the objective in these studies.

Dennis Hansell contributed some ideas on the potential importance of the DOC pool as a short- or longer-term storage pool. Chris Garside emphasized the longer-term goal of JGOFS, a basin-scale understanding of carbon fluxes, and presented an approach to constructing algorithms with which nutrient distributions could be estimated from more finely resolved data like temperature and salinity. John Marra used observations made near 59°N, 20°W in the Marine Light-Mixed Layers program to demonstrate how well (or how poorly) incubation-based estimates of primary production compare with those inferred from bulk changes in water column properties. The latter kind of estimates might be better constrained in a CVE-type approach. Finally Jim Price discussed some of the physical oceanographic constraints on a CVE study.

Modelers Scott Doney, Dennis McGillicuddy, Don Olson and Allan Robinson discussed a variety of approaches to supporting CVE's with models. In particular, Robinson described the application of Observation System Simulation Experiments (OSSE's) which could be used to design field observation programs. As Robinson put it, "The ocean is a very complicated nonlinear system. It is hard to outguess it. Observed behavior must be used to construct the models." Doney and McGillicuddy described rather simple ecosystem models coupled to more sophisticated physical models. Olson addressed some ways in which the ecosystem models can be refined to treat complex plankton systems more realistically. A major problem with either approach is validation of the model behavior.

3.3 Interactions and Collaboration with Other Programs

The resource spectrum available for future work might be enhanced by collaboration with other programs. Alex Pszenny provided an overview of the International Global Atmospheric Chemistry (IGAC) Program. IGAC has a strong ocean component, with particular interest in better understanding of the sulfur cycle. The IGAC North Atlantic Regional Experiment will be carried out in 1997 in the region of the Azores with aircraft support. Tom Powell talked about GLOBEC, a program addressing the effects of global change on oceanic ecosystems at higher trophic levels (zooplankton and fish). Modeling is a strong common thread between JGOFS and GLOBEC. Joint efforts in the development of OSSE's need to be carried out at several sites, including the North Atlantic. In spite of many areas of common interest, the precise nature of interaction and collaboration among these 3 programs (and others) still need to be spelled out. Lack of space on ships, different time frames, and different priorities make close interaction difficult to achieve.

3.4 Discussions

The group did not, and has not reached a final conclusion regarding the priority that should be assigned to a control volume experiment in the North Atlantic. The reason for a lack of consensus is that the three program elements in JGOFS, time series, surveys and modeling, require rather different mixes of observational approaches, and proceed on different time and space scales. Time series observations provide local understanding of temporal variability. Wide areal surveys provide global (at least basin-wide) understanding of spatial variability. Process studies, including CVE's, are complementary of the other two modes of study. They combine some aspects of each and are typically performed on different temporal and spatial scales from the two other modes of study.

Elements of all three observational modes are required for improving our understanding of North Atlantic carbon fluxes. JGOFS time series observations have been limited to lower-latitude sites, and the temporal (seasonal) context in which to view processes and fluxes over much of the basin is still lacking. A seasonally- resolved survey with moderate spatial resolution could be carried out over some limited area. Together, these studies provide the spatial and temporal context in which a CVE aimed at better constrained carbon budgets could be most profitably conducted. Meeting participants articulated a varying priority of resources to assign to each of these study modes.