U.S. JGOFS Implementation Plan for
Synthesis and Modeling
The Role of Oceanic Processes in
the Global Carbon Cycle

Jorge L. Sarmiento
Robert A. Armstrong
AOS Program
Princeton University
P.O. Box CN710
Princeton, NJ 08544-0710

Printed copies are available from the U.S. JGOFS Planning Office. Contact Mary Zawoysky at mzawoysky@whoi.edu or (508) 289-2834 if you would like a copy.


  • A. Historical Overview
  • B. The U.S. JGOFS Program
  • 1. Global CO2 Survey
  • 2. Process Studies
  • 3. The U.S. JGOFS Time-series Studies
  • C. SMP Charges
  • D. SMP Goals
  • A. Introduction
  • B. The Contemporary Carbon Cycle
  • C. Future Changes
  • D. The Challenge
  • A. Introduction
  • B. Creating a Mass Balance
  • C. Global and Regional Mass Balances
  • 1. Estimating global carbon inventories
  • 2. Estimating the global balance of DIC fluxes
  • 3. Extrapolation
  • D. Implementation
  • A. Introduction
  • B. Qualitative Differences Among Oceanic Regimes
  • C. Mechanistic Controls Within Oceanographic Regions
  • 1. Ultimate limits to the draw-down of DIC
  • a) Nutrients
  • b) Light
  • 2. Partitioning of productivity by ecosystem processes
  • a) Size structure
  • b) Biogeochemical diversity
  • 3. Remineralization and export
  • 4. Sediment diagenesis and burial
  • D. Local and Regional Mass Balances
  • 1. Production and export rates of DOC, PIC and POC
  • 2. DIC fluxes
  • 3. Remineralization
  • 4. Time rates of change
  • E. Implementation
  • A. Introduction
  • B. Approaches to Extrapolation and Prediction
  • 1. Parameterizing of unresolved processes
  • 2. Evaluating the uncertainties
  • 3. Tools and techniques
  • a) Models
  • b) Data
  • c) Data assimilation and inverse modeling
  • C. Implementation
  • 1. Extrapolating from local to global scales
  • 2. Estimating, simulating, and predicting current conditions
  • 3. Predicting the ocean's response to climate change


    The goals of the Joint Global Ocean Flux Study (JGOFS) are (1) to determine the processes that control the partitioning of carbon and related biologically active substances within the ocean and between the ocean and atmosphere and (2) to develop a capability to predict how changes in the environment will influence this partitioning (JGOFS Report No. 5, SCOR, 1990). JGOFS has gathered an unprecedented data set on ocean biogeochemistry, providing an historic opportunity to address these goals.

    The central objective of the U.S. JGOFS Synthesis and Modeling Project (SMP) is to synthesize knowledge gained from U.S. JGOFS and related studies into a set of models that reflect our current understanding of the ocean carbon cycle and its associated uncertainties. Emphasis will be given to processes that control partitioning of carbon among oceanic reservoirs and the implications of this partitioning for exchange between the ocean and atmosphere. To this end, the following specific SMP goals were adopted.

    To synthesize our knowledge of inorganic and organic carbon fluxes and inventories, both natural and anthropogenic.

    To identify and quantify the principal processes that control the partitioning of carbon among oceanic reservoirs and between the ocean and atmosphere on local and regional scales, with a view towards synthesis and prediction on a global scale.

    To determine the mechanisms responsible for spatial and temporal variability in biogeochemical processes that control partitioning of carbon among oceanic reservoirs and between the ocean and atmosphere.

    To assess and implement strategies for scaling data and models to seasonal, annual, and interannual time scales and to regional and global spatial scales.

    To improve our ability to monitor and predict the role of oceanic processes in determining current and future partitionings of carbon between the ocean and atmosphere, and to evaluate uncertainties and identify gaps in our knowledge of these processes.

    These goals will be addressed by three major program elements:

    1. Global and regional mass balances: synthesis of improved estimates of natural and anthropogenic carbon inventories and of fluxes of carbon and related biologically active chemical substances.

    2. Mechanistic controls of local carbon balances: identification and modeling of the principal processes that control within-ocean and ocean-atmosphere partitioning of carbon and related biologically active chemical substances, with a view towards developing regional and global syntheses and models.

    3. Extrapolation, monitoring, and prediction: development and application of methods that will allow knowledge gained on small spatial and temporal scales to be scaled to seasonal, annual, and interannual time scales and to regional and global spatial scales; and development and application of methods that will improve our ability to monitor and predict the role of oceanic processes in determining the partitioning of carbon between the ocean and atmosphere and the resulting feedback to the climate system.

    Implicit in this effort is the quantitative evaluation and estimation of associated uncertainties, as well as the identification of gaps in our knowledge that may significantly compromise monitoring and prediction of carbon partitioning.


    A. Historical Overview

    In the period since 1988, the U.S. Joint Global Ocean Flux Study (U.S. JGOFS) has gathered an oceanic biogeochemical data set of unprecedented scope and detail. These data provide the oceanographic community with an historic opportunity to enhance our understanding of ocean biogeochemistry. The field component of U.S. JGOFS will end in 1998 following completion of the Southern Ocean Study. In anticipation of this, the U.S. JGOFS Steering Committee asked at its October 1995 meeting for the preparation of a plan for culminating the U.S. JGOFS program with a program of synthesis and modeling. A Synthesis and Modeling Project (SMP) steering committee was appointed to accomplish this task and to lead implementation of the plan. This planning document was written by the SMP steering committee with wide participation from U.S. JGOFS investigators and other oceanographers at a meeting held at Durham, New Hampshire, in August 1996. (A full list of participants is given in Appendix A, and the agenda of the meeting is given in Appendix B.)

    The two major JGOFS goals, as embodied in the International Science Plan (JGOFS Report No. 5, SCOR, 1990), are:

    1. To determine and understand on a global scale the processes controlling the time-varying fluxes of carbon and associated biogenic elements in the ocean, and to evaluate the related exchanges with the atmosphere, sea floor, and continental boundaries.

    2. To develop a capability to predict on a global scale the response of oceanic biogeochemical processes to anthropogenic perturbations, in particular those related to climate change.

    These goals define the central aims of the JGOFS program as the attainment of an improved understanding of ocean fluxes of carbon and associated biogenic elements and the embodiment of this understanding in models. The discussion of modeling in the U.S. JGOFS Long Range Plan (U.S. JGOFS Steering Committee, 1990) contains a brief summary of the distribution of biologically active chemical compounds in the ocean, concluding that,

    "these observations challenge us to develop models which couple the physical forcing, biological agents of transformation, and chemical substrates so that theories about the ocean's production system can be formulated and tested. Through large-scale models the theories found to work in areas of study can be applied far afield in regions remote to study sites. The legacy of U.S. JGOFS will be the understanding of the system, as encoded in algorithms and models, which will enable us to monitor the state of the ocean in real-time and to predict its future course in an era of climate change."

    Achieving U.S. JGOFS SMP goals will require a sustained effort to gain a comprehensive understanding of ocean biogeochemical processes and to embody this understanding in models capable of monitoring and prediction. Indeed, incremental funds for components of the U.S. Global Change Research Program, like U.S. JGOFS, have been justified by the argument that the collective understanding gained from the ensemble will be greater than the sum of the individual elements. Investigators play a critical role in synthesis when they analyze their individually-funded efforts and submit these data to the U.S. JGOFS data base. In addition, many process study grants were funded with the specific expectation that groups of investigators would be working together to interpret multiple, linked data sets.

    However, the more complete synthesis and modeling called for in U.S. JGOFS requires specific planning; it is too important to be left to serendipity. While the empirical studies already undertaken as part of U.S. JGOFS and related studies have produced a wealth of new data on process rates and a number of new hypotheses concerning their mechanistic controls, the quantitative consistency and large-scale implications of this new knowledge can only be evaluated through quantitative modeling; a dedicated synthesis and modeling effort is therefore necessary for realizing the full fruits of U.S. JGOFS empirical studies. Achieving this quantitative synthesis will require an unprecedented and systematic commitment to new and creative analyses of multiple data sets by those who made the observations, with the dual aims of producing new knowledge and supporting development of models, and by modelers working closely with those who made observations to capture in mathematical form the key insights gained from these analyses.

    In this implementation plan, we discuss data synthesis and modeling activities that will be required to meet the overall goals of JGOFS as summarized in the International Science Plan and in the U.S. JGOFS Long Range Plan. The following sub-sections provide a brief overview of the U.S. JGOFS program, a discussion of the charges of the U.S. JGOFS Steering Committee, and a statement of U.S. JGOFS SMP goals that were adopted at the August 1996 meeting.

    B. The U.S. JGOFS Program

    Until now the primary U.S. JGOFS activity in support of JGOFS goals has been a set of major field programs organized in three different ways: as a global survey of ocean properties; as time series studies; and as detailed process studies of regions important to the ocean's carbon cycle. These programs were complemented by related efforts from other agencies as well as independent projects with related goals. A body of historical data also exists that is of direct relevance to our understanding of the carbon cycle. Collectively, the programs and historical data represent the starting point for the SMP. A short summary of the U.S. JGOFS field programs is presented below.

    1. Global CO2 Survey

    The U.S. JGOFS Global CO2 Survey, co-sponsored by the U.S. Department of Energy and the National Oceanic and Atmospheric Administration, was designed to collect high-quality CO2, nutrient, tracer, and hydrographic measurements with the goals of obtaining more realistic estimates of the inventory of anthropogenic carbon dioxide in the oceans and of providing a better understanding of the fluxes of CO2 across the air-sea interface. The CO2 survey, conducted collaboratively with the WOCE Hydrographic Program, was initiated in 1990 and will be completed in 1997. Since the beginning of the U.S. component of the survey, 13 oceanographic cruises have been carried out in the Pacific Ocean, 7 cruises in the Indian Ocean, and 9 in the Atlantic Ocean. The number of cruises in the Atlantic will be increased to 12 by the end of the program. To date, over 100,000 samples have been collected and analyzed for CO2 system parameters. In addition to U.S. participation there have been additional contributions to the Survey by France, the U.K., Japan, Germany, Taiwan and the Nordic countries. The accuracy of today's measurements is dramatically better than that of previous studies. This unique and extensive set of calibrated measurements provides a basis for estimating global carbon inventories and the amount of fossil-fuel-derived CO2 in the oceans. Data from the U.S. Survey can be accessed from Tom Boden or Alex Kozyr at CDIAC:



    2. Process Studies

    One of the primary U.S. JGOFS field components is a set of regional process studies designed to measure a variety of inventories, process rates, and fluxes that are believed to control the cycling of carbon in the ocean. The sites for the four process studies, the North Atlantic (NABE), the equatorial Pacific (EqPac), the Arabian Sea, and the Southern Ocean (AESOPS), were selected to provide insight into phenomena that link carbon dioxide in the atmosphere to the processing of carbon in the water column and sediments. The process studies were designed as nested observational programs in which measurements of carbon inventories and rates made from ships were expanded in the temporal domain by moorings, and in which both ship-board and mooring data were further extrapolated in space and time using satellite remote sensing, sediment deposition patterns, and models.

    Process study sites were chosen to reveal local details of biogeochemical processes that control partitioning of carbon among oceanic reservoirs and between the ocean and atmosphere. Key issues involved (1) balances of carbon between the atmosphere and upper ocean, (2) cycling of carbon within the upper ocean, (3) export of carbon downward from the upper ocean, (4) remineralization within the water column, and (5) remineralization and deposition of carbon at the sea floor. A key uncertainty addressed in each process study was whether the region in question was a net source or sink for atmospheric carbon dioxide.

    These four sites were selected for intensive study because of characteristic differences in physical forcing and biological response. The North Atlantic was chosen to represent the development and northward propagation of the spring bloom and the associated drawdown of CO2 in a subtropical gyre. The equatorial Pacific provided an opportunity to study High-Nutrient Low Chlorophyll (HNLC) conditions in a region characterized by strong variability in seasonal and interannual (ENSO) physical forcing. The Arabian Sea was chosen because strong and regularly alternating monsoonal forcing results in the greatest seasonal variability in carbon cycling observed in any ocean basin. The Southern Ocean process study was conducted because the fluxes of carbon are large and play an important role in the global carbon cycle.

    Table 1 summarizes the status of the U.S. JGOFS process studies and the availability of data. Data can be accessed either through the U.S. JGOFS Data Management Office through Christine Hammond at:


    or from the National Oceanographic Data Center. NABE and EqPac results have been widely reported in the literature; many NABE results were published in Deep-Sea Research II, Volume 40, Nos. 1/2 (1993), and most EqPac results were published in Deep-Sea Research II, Volume 42, Nos. 2/3 (1995) and Volume 43, Nos. 4/5/6 (1996), with a third volume nearing completion.

    Table 1. Status of U.S. JGOFS Process Studies

                  Process Study                  Field     Expected Time of    
                                                 Work      Data Availability   
    North Atlantic Bloom Experiment (NABE)       1989      Already available   
    Equatorial Pacific Study (EqPac)            1992-3     Already available   
    Arabian Sea Expedition                      1994-6           1998          
    Southern Ocean Study                        1996-8           2000          

    U.S. JGOFS process studies were conducted in the context of international JGOFS studies. The resulting collaborations have greatly enhanced the total knowledge gained. The cooperation that has been achieved nationally and world-wide is a significant accomplishment of U.S. JGOFS and should continue to be fostered. U.S. agencies that have supported related research, as well as other countries that were involved in each of the process studies, are summarized in Table 2.

    Table 2. U.S. agencies and other countries involved in U.S. JGOFS Process Studies
         Process Study          Related US      US Agencies   Other Countries   
    North Atlantic Bloom        ML-ML (ONR)         NSF            Canada       
    Experiment (NABE)              WOCE             NASA          Germany       
                                                    ONR       The Netherlands   
                                                               United Kingdom   
    Equatorial Pacific           IGAC-MAGE          NSF          Australia      
    Study (EqPac)                 IronEx            DOE            Canada       
                               OACES (NOAA)         NASA           France       
                                   WOCE             ONR           Germany       
    Arabian Sea Expedition      FUODE (ONR)         NSF           Germany       
                                  GLOBEC            DOE            India        
                               OACES (NOAA)         NASA      The Netherlands   
                                   WOCE             ONR             Oman        
                                                               United Kingdom   
    Southern Ocean Study           WOCE             NSF          Australia      
                                                    NASA          Belgium       
                                                    NOAA           France       
                                                                New Zealand     
                                                               United Kingdom   

    3. The U.S. JGOFS Time-series Studies

    Two time-series stations were established as part of the efforts of U.S. JGOFS to understand the ocean-atmosphere system on annual to decadal time-scales. In the North Atlantic, the Bermuda Atlantic Time-series Station (BATS) is operated by scientists at the Bermuda Biological Station for Research, Inc; and in the North Pacific, the Hawaii Ocean Time-series (HOT) program is operated by scientists at the University of Hawaii. Both programs have WOCE as well as U.S. JGOFS components with support from NSF and NOAA. HOT has additional support from the State of Hawaii, and BATS has additional support from NASA as well as ancillary programs from Canada, Germany, Switzerland, and the United Kingdom. Mainly based on ship-board observations, the time-series programs continue to conduct biweekly-to-monthly measurements of a suite of hydrographic and biogeochemical parameters at their respective deep-water stations. These measurements include discrete and continuous measurements through the water column and a variety of underway measurements. Rates of particle production and export are also determined experimentally. A more complete description of these time-series programs can be found in a special volume of Deep-Sea Research II, Volume 43, No. 2/3 (1996).

    U.S. JGOFS ocean time-series data are available as hard copies from the U.S. JGOFS planning office and from the National Oceanographic Data Center. Data can also be accessed electronically through the internet system using the anonymous file transfer protocol (ftp) or through the world-wide-web. The corresponding addresses for BATS are:


    http://www.bbsr.edu/ and follow links to various data sets

    and for HOT:

    ftp access - mana.soest.hawaii.edu


    Direct inquiries can also be made to the respective principal investigators.

    As of 1995, there were at least three other JGOFS time-series studies in progress: a French effort at Kergulen Island in the southern Indian Ocean, a joint German-Spanish measurement program in the eastern North Atlantic Ocean near the Canary Islands, and a joint German-Chilean station in the eastern South Pacific Ocean off the coast of Chile. Several additional ocean time-series research efforts conducted outside the context of the IGBP-JGOFS programs are also underway, but few have the physical-biogeochemical focus of the Hawaiian Ocean Time-series and Bermuda Atlantic Time-series Study. Collectively, these oceanic time-series research efforts have provided an unprecedented view of oceanic variability on a variety of time scales from days to decades.

    Both time-series programs build upon historical data sets within their respective regions. In the North Atlantic, a hydrographic data record has been maintained at Hydrostation S since 1954, and data from moored sediment traps near Bermuda have been collected since 1976. In the North Pacific, several short-lived and irregular time-series studies were conducted within the vicinity of the Hawaiian Islands. The time-series programs have provided opportunities for ancillary research programs to conduct open-ocean time-series studies at both sites. Institutions and federal sources have provided funding for these projects. The time-series programs maintain a listing of recently-completed and on-going projects at their study sites.

    C. SMP Charges

    Anticipating the completion of field measurements in 1998 (except for the time-series stations, which are expected to continue), the U.S. JGOFS Steering Committee appointed a Synthesis and Modeling Project (SMP) steering committee at its October 1995 meeting (Table 3) and gave it the charges shown in Figure 1. This planning document has been prepared in response to the first two SMP charges.

    Table 3. Members of U.S. JGOFS Synthesis and Modeling Project Steering Committee appointed by the U.S. JGOFS Steering Committee in October 1995.

    Jorge Sarmiento, Chair
    Rob Armstrong, SMP Project Scientist
    Dick Barber
    Eileen Hofmann
    Marlon Lewis
    Dennis McGillicuddy
    Tony Michaels
    Jim Murray

    Figure 1. Charges to the U.S. JGOFS Synthesis and Modeling Project approved by the Steering Committee in October 1995.

    U.S. JGOFS Synthesis and Modeling Project (SMP) Charges

    (1) To plan and direct U.S. JGOFS synthesis and modeling activities that apply the knowledge gained by the JGOFS time series stations, process studies, and global CO2 survey, as well as satellite observations and the WOCE and GLOBEC Programs, in support of the two major JGOFS goals:

    JGOFS GOAL 1. To determine and understand on a global scale the processes controlling the time-varying fluxes of carbon and associated biogenic elements in the ocean, and to evaluate the related exchanges with the atmosphere, sea floor, and continental boundaries.

    JGOFS GOAL 2. To develop a capability to predict on a global scale the response of oceanic biogeochemical processes to anthropogenic perturbations, in particular those related to climate change.

    (2) To provide advice to the Scientific Steering Committee on field studies that would explore creative new ideas or fill important gaps in our knowledge in support of meeting the JGOFS goals.

    (3) To work with the SSC to encourage cross-fertilization of ideas within JGOFS, and dissemination of the knowledge gained by synthesis and modeling activities, through the planning of workshops and encouragement of publications.

    (4) To serve as a counterpart and direct liaision to the Global Synthesis and Modelling Task Team of the International JGOFS SSC, and to the IGBP-GAIM and other relevant bodies.

    As a first step towards planning the SMP, it was agreed that the major focus would be on the partitioning of carbon and related elements among oceanic reservoirs, and the implications of this partitioning for exchange between the ocean and atmosphere. The primary factors that led to this decision are (1) the fact that the oceanic carbon cycle provides a focal point for addressing virtually all major oceanic biogeochemical processes, (2) the major role of the ocean-atmosphere CO2 balance in determining climate, and (3) the importance of the ocean as a sink for anthropogenic CO2.

    The ocean-atmosphere balance of carbon is affected by a wide range of oceanic biogeochemical processes that were examined as part of the U.S. JGOFS field program. In addition, all components of U.S. JGOFS made direct measurements of carbon distributions that will make it possible to obtain improved constraints on air-sea fluxes of CO2. Indeed, understanding the cycling of carbon was at all times a central goal of U.S. JGOFS field programs. There is thus a large body of data and a considerable improvement in our understanding of ocean biogeochemical processes that can be applied towards advancing our knowledge of the ocean carbon cycle. It is our expectation that when the increment in understanding that U.S. JGOFS data make possible is combined with improved models of ocean circulation, the predictive power of the resulting models will be clearly superior to that attainable by current models.

    The second step in articulating a specific set of activities was to determine what is required to satisfy JGOFS goals. JGOFS GOAL 2 asks us to develop "a capability to predict." Prediction requires solving some form of the tracer continuity equation:


    where c is the tracer concentration (e.g., in mol m-3), is the velocity vector, is a diffusivity tensor that accounts for the combined effect of molecular and eddy processes, and (e.g., in mol m-3 s-1) is the combined net effect of all biogeochemical sources and sinks such as photosynthesis and remineralization, as well as sinking of particulate forms and direct biological transport. This equation applies to ecosystem components such as phytoplankton and zooplankton as well as to dissolved or suspended matter. For some of these properties, the ocean transport terms may be negligible compared to the movements of the organisms themselves. By way of illustration, Figure 2 shows the seven component Fasham et al. (1990) model for the biogeochemical source minus sink terms. The Fasham et al. model includes four forms of nitrogen (nitrate, ammonia, labile dissolved organic nitrogen, and detritus) and three biotic components (phytoplankton, zooplankton, and bacteria). Fasham et al. solved their ecosystem model in a simplified mixed layer model of ocean physics. The Fasham et al. ecosystem model has also been added to a three-dimensional ocean general circulation model that solves the full form of equation (1) Sarmiento et al. (1993).

    Figure 2. Fasham et al. (1990) ecosystem model in the form that was solved in the Sarmiento et al. (1993) ocean general circulation model.

    To satisfy JGOFS GOAL 2 we must determine the structure of the terms for carbon and related substances, as well as the magnitudes of relevant parameters, their variability in space and time, and the uncertainties in our knowledge. In effect, we must develop improved biogeochemical cycling/ecosystem models. Much effort in the U.S. JGOFS process and time-series studies was devoted specifically to attempting to characterize the potentially complex structures of many of the terms. A crucial challenge in U.S. JGOFS SMP will be to develop these characterizations more completely. We must also account for physical transport associated with fluid motion. Satisfying JGOFS GOAL 2 (Figure 1) therefore requires improvements in our understanding both of mechanistic controls of biogeochemical cycling and of the influences of physical processes on these controls. The gaining of this improved understanding constitutes the first part of JGOFS GOAL 1, which challenges SMP investigators to quantify fluxes associated with processes of interest.

    The second part of JGOFS GOAL 1 is to evaluate the exchanges of carbon and associated biogenic elements with the atmosphere, sea-floor, and continental boundaries. The quantitative expression of this goal is the box model, which is defined as the integral of (1) over the volume V of a box surrounding a given region of interest, for example, the world ocean, the North Atlantic, or the BATS station region. This gives:


    where n is a unit vector perpendicular to the surface. Note that the volume integral of the transport and advection terms are converted into integrals around the surface S of the box, since the behavior of these terms within the box does not affect the total tracer within the box. Equation (2) is usually expressed in a form such as:


    where C is the volume integral of the tracer within the box (e.g. in mol), the F's are the total fluxes across the various faces of the box (e.g. in mol s-1), and FC is the volume integral of all biogeochemical source-minus-sink terms in the interior of the box (e.g., photosynthetic uptake or remineralization in mol s-1).

    Solving equation (3) for the air-sea flux, a major focus of U.S. JGOFS SMP, reveals the dependence of this flux on physical and biogeochemical processes occurring within the interior of the ocean, including the integrated effects both of the terms and of physical transport and sinking of organic material into or out of the box. Similar box model equations may be derived for other interfaces.

    D. SMP Goals

    Based on the foregoing considerations, the set of implementation goals given in the EXECUTIVE SUMMARY were adopted at the New Hampshire SMP planning meeting. The central goal of U.S. JGOFS SMP is to synthesize our knowledge into a set of models that can be used for prediction. Achieving this goal will require that synthesis and modeling be conducted in tight coordination and with an awareness of their mutual dependencies. In particular, model development must be driven by data, with model structures, functional forms, and parameter values chosen in a way that allows them to reproduce large-scale spatial and temporal patterns in inventories and fluxes. Conversely, data synthesis efforts should be driven in large measure by their usefulness to model development. Models based on mechanistic controls should prove particularly useful in this concurrent pursuit of synthesis and modeling goals. The needed interplay between synthesis and modeling will be achieved most easily through close working arrangements between empiricists and modelers; such close collaborations will be encouraged in SMP proposals.

    Three major program elements are proposed for attaining these goals. These elements should not be considered in isolation, but instead as functional parts of an overall strategy for attaining U.S. JGOFS SMP goals. The pursuit of one element should therefore always be considered in the context of how its attainment will affect other program elements. The proposed program elements and a list of specific objectives for each element are:

    1. Global and regional mass balances: synthesis of natural and anthropogenic inventories and fluxes of carbon and related biologically active chemical substances. The flux balance is given by box model equations such as (3).

    Specific objectives:

    To synthesize global and regional carbon inventories and flux balances, including changes in inventories caused by anthropogenic and other processes.

    To extrapolate existing data to regional and global scales to create a more complete understanding of the global ocean carbon system.

    2. Mechanistic controls of local carbon balances: identification and modeling of principal processes that control within-ocean and ocean-atmosphere partitioning of carbon and related biologically active chemical substances, with a view towards developing regional and global syntheses and models. Essentially we require knowledge of the in (1) and of the FC in (3).

    Specific objectives:

    To construct mass balances for all appropriate biogenic elements at the U.S. JGOFS time-series and process study sites, including explicit comparisons among U.S. JGOFS sites.

    To describe and quantify the principal mechanisms controlling the partitioning of carbon and related elements.

    To determine how this collection of mechanisms is expressed regionally and temporally.

    To develop models and parameterizations of these mechanisms and their interactions and associated fluxes to facilitate regional and global synthesis and modeling.

    3. Extrapolation, monitoring, and prediction: development and application of methods that will allow knowledge gained on small spatial and temporal scales to be scaled to seasonal, annual, and interannual time scales and to regional and global spatial scales; and development and application of methods that will improve our ability to monitor and predict the role of oceanic processes and feedbacks in determining the partitioning of carbon between the ocean and atmosphere. This element also has a component in which local models are tested by determining whether or not they successfully account for large-scale spatial and temporal patterns.

    Specific objectives:

    To estimate the contribution of ocean biogeochemical processes in determining the partitioning of carbon among oceanic reservoirs and between the atmosphere and ocean under current climate conditions.

    To develop the ability to monitor and predict the role of oceanic processes and feedbacks in determining the partitioning of carbon among oceanic reservoirs and between the atmosphere and ocean in response to large-scale changes in climate.

    Recognizing the strong dependencies among program elements is essential for attaining U.S. JGOFS SMP goals. First, carbon balance studies define qualitatively and quantitatively the spatial and temporal patterns that predictive models must reproduce; they also provide constraints to assess whether the models are working as required. Second, results produced locally must be scaled regionally and globally before they can be used for carbon cycle predictions at a global scale. Scaling will be required for complete analysis of the carbon survey data, for direct extrapolation of correlations to changed climate scenarios, and for scaling mechanistic models to the global domain. Finally, the most powerful models will be based on a knowledge of mechanistic controls. These models must automatically produce correct spatial and temporal patterns in response to appropriate physical and geochemical forcing; the evaluation and modeling of mechanistic controls is therefore an essential basis for constructing models needed for extrapolation and prediction.

    The following sections begin with a discussion of the rationale for the SMP (Section II); this discussion is followed by more detailed accounts of the three program elements and their relationships to one another. In section III ("Global and regional balances of carbon and their mechanistic controls") we discuss carbon and related nutrient balances on both global and regional scales. The U.S. JGOFS/WOCE carbon survey data are discussed as the major basis for constructing a synthesis of global inventories of both natural and anthropogenic carbon. In this section we also discuss the need to develop methods for extrapolating relationships uncovered in the evaluation of carbon balances to regions with sparse coverage. Section IV ("Local carbon balances and their mechanistic controls") presents a discussion of processes that must be understood to enable construction of the powerful mechanism-based models that will be needed for predicting the partitioning of carbon both among oceanic reservoirs and between the ocean and atmosphere. Regional mass balances are discussed as a mechanism for synthesizing and understanding results from U.S. JGOFS time series studies and process studies, and consideration is given to empirical and modeling studies that could reduce or help characterize uncertainties in these balances. Section V ("Extrapolation and prediction") discusses the need for methods for scaling mechanism-based models globally to allow prediction of the carbon system under current forcing and in response to altered climate. Scaling requirements will determine to a large extent the structure of mechanism-based models, most notably by determining the mixture of parameterization and explicit representation of important processes. Finally, Section VI ("Implementation") discusses how the SMP will be implemented.


    A. Introduction

    The major practical justification for JGOFS has been the need to understand more fully the important role of the ocean as a sink for anthropogenic CO2. Mankind has emitted an estimated 341 Pg (1 Pg = 1 Gt = 1015 g) of carbon to the atmosphere since 1850, 219 Pg by fossil fuel burning and cement production and 122 Pg by changes in land use (Figure 3). The only fraction of this total that has been measured precisely is the 42% that has remained in the atmosphere, where it can affect climate through its impact on the earth's radiative balance (Figures 3 and 4). Ocean circulation models calibrated with radiocarbon observations predict an oceanic uptake of 30% of the total (Figure 3). Direct observations of the oceanic increase are extremely difficult, but the greatly improved measurement accuracy of recent observations obtained by U.S. JGOFS on the WOCE hydrographic cruises (1 mmol kg-1 today, versus ~20 mmol kg-1 during the GEOSECS era two decades ago; the total anthropogenic signal is ~40 mmol kg-1 in surface waters) should allow significantly better estimates of oceanic inventories and fluxes to be produced as part of the SMP. The remaining 28% of the emissions is thought to be taken up by the terrestrial biosphere, but the evidence for this is as yet weak and the proposed mechanisms are highly controversial.

    Figure 3. Anthropogenic carbon sources and sinks. The oceanic uptake is estimated using the model of Sarmiento et al. (1992) forced with atmospheric data (Figure 4) linearly interpolated between measurements. Fossil fuel production is from Marland et al. (1989) and land use is from Houghton (1992). Y-axis should be labeled Pg C/yr. Figure 4. CO2 concentrations over the past 1000 years from recent ice core records and (since 1958) from Mauna Loa. The inset shows the period from 1850 to the present in more detail, including CO2 emissions from fossil fuel. Data sources: D47 and D57 (Barnola et al., in press); Siple (Neftel et al., 1985 and Friedli et al. 1986); South Pole (Siegenthaler et al., 1988). The smooth curve is based on a 100-year running mean. All ice core measurements were take in Antarctica. The uptake of anthropogenic CO2 by the ocean is insensitive to biological processes so long as these processes remain unchanged (Broecker, 1991). In such a case, CO2 would enter the ocean as a passive tracer that dissolves in the surface mixed layer and would be advected into the interior. Most scenarios of future atmospheric CO2 content have been based on the assumption that ocean circulation and ocean biological processes will remain constant. The assumption of a steady-state ocean carbon cycle is supported by the remarkably constant atmospheric CO2 content in the millennium prior to the industrial revolution (Figure 4), during which CO2 concentrations varied by less than ±10 ppm. If the oceanic carbon cycle had changed significantly throughout this period, it would have been reflected in atmospheric CO2 concentrations.

    However, ocean biogeochemical processes will become important to the air-sea CO2 flux if circulation or ocean biology change, or if we consider the spatial and temporal variability of the air-sea flux. The latter theme is presented in section B, where we discuss patterns of spatial and temporal variability of air-sea CO2 fluxes that have emerged from detailed studies of the contemporary carbon budget; these results have been made possible by improved atmospheric and oceanic measurements and models. In section C we discuss issues that arise when the possibility of changes in ocean circulation and biology are considered. Section D outlines the challenges for U.S. JGOFS SMP in characterizing and predicting oceanic responses to climate change.

    B. The Contemporary Carbon Cycle

    The study of the global carbon cycle has moved away from simply cataloguing the total strengths of various sources and sinks of each component of the carbon budget and towards a sophisticated treatment of spatial distributions and how these change in time. Analyses of the spatial distribution of sources and sinks of CO2, such as those by Keeling et al., 1989b; Tans et al., 1990; and Enting and Mansbridge, 1991, have provided a powerful new tool for studying the global carbon balance and its oceanic component. These particular studies focus on the generation and maintenance of the 3 ppm inter-hemispheric gradient in atmospheric pCO2 during the period 1981-1987. This gradient is consistent with the north-to-south atmospheric CO2 transport expected from major anthropogenic fossil carbon sources in the northern hemisphere. However, the magnitude of the transport implied by this gradient is very small. Atmospheric models constrained by the observed gradient can transport only 0.9 to 1.2 Pg C/yr. to the south, not enough even to feed the southern hemisphere atmospheric increase of 1.6 Pg C/yr. Southern hemisphere fossil emissions of 0.4 Pg C/yr can close the atmospheric budget, but leave nothing to feed a southern hemisphere oceanic sink. If correct, these results imply that the entire oceanic uptake of anthropogenic carbon must occur in the northern hemisphere.

    However, a number of observational studies and models suggest a major oceanic sink for anthropogenic CO2 sink in the southern hemisphere. One proposed solution to the contradiction between atmospheric transport estimates and ocean anthropogenic CO2 flux estimates posits a large northern hemisphere oceanic uptake of 2.3 Pg C yr-1, which would feed the southern hemisphere oceanic sink by a southward transport of CO2 within the ocean. This transport would need to have been present pre-industrially (Keeling et al., 1989b; cf. also Broecker, 1992) An alternative scenario posits a small northern hemisphere oceanic uptake of about 0.6 Pg C yr-1, as estimated from observations of the air-sea CO2 gradient multiplied by a gas exchange coefficient. This latter solution rejects the existence of a large southern hemisphere oceanic sink and argues that the total oceanic sink is only 0.3 to 0.8 Pg C yr-1, thus requiring that the major sink for CO2 be in the large terrestrial regions of the northern hemisphere (Tans, et al., 1990).

    The three pre-industrial air-sea flux estimates of Figure 5 summarize our lack of quantitative understanding of the steady-state oceanic carbon cycle, let alone our ability to predict changes that may have occurred to that cycle due to anthropogenic forcing. The Keeling et al. (1989a) scenario requires a large pre-industrial southward inter-hemispheric transport. As a consequence, the large equatorial efflux is balanced by a northern hemisphere uptake, and the Southern Ocean actually has a loss of CO2 to the atmosphere rather than the gain that is typical of high latitudes. The estimate based on observational data presented by Tans, et al. (1990) shows a large equatorial efflux balanced primarily by uptake in the southern hemisphere. The atmospheric model study of Tans et al. rejects the notion of a large southern hemisphere oceanic uptake, which they argue is poorly constrained by the observations. The Princeton biogeochemistry model described by Sarmiento et al. (1995) gives an equatorial efflux balanced by equal uptake in both hemispheres. A further difference among the three results is that the equatorial efflux of the ocean model is about half the magnitude of the other studies. It is apparent that improved estimates of the spatial distributions of air-sea CO2 fluxes would add greatly to our understanding of this major anthropogenic CO2 sink.

    Figure 5. Pre-industrial sea-air CO2 fluxes (Pg C/yr).

    Pre-Industrial Sea-Air CO2 Fluxes

    (Pg C/yr)


    Observations Ocean GCM Atmospheric


    (Pre-Ind. Based (Princeton Carbon Model) (Keeling et al., 1989)

    on Takahashi)


    15 N-90 N -0.16 -0.59 -1.68

    15 S-15 N 2.16 1.20 1.83

    90 S-15 S -1.66 -0.61 -0.14


    Total 0.34 0.00 0.01

    In addition to spatial variations, the ocean exhibits large temporal variability on a wide range of time scales. The variations that are easiest to document and study are those associated with eddies, seasonal fluctuations in temperature, and interannual variability, of which the El Niño/Southern Oscillation (ENSO) phenomenon is the most dramatic example.

    Monthly observations of atmospheric CO2 and O2 display regular patterns of seasonal oscillations (Keeling et al., 1989; Keeling and Shertz, 1992) where the oscillations in CO2 are strongly correlated with seasonal changes in terrestrial productivity. Oceanic productivity is also highly seasonal and is comparable in magnitude to terrestrial productivity. The seasonal changes in the partial pressure of CO2 in the oceanic mixed layer induced by the seasonal cycle of productivity can be quite large. However, very little of this signal reaches the atmosphere due to the long time-scale for CO2 equilibration between the atmosphere and ocean (about 8 months for a 50 meter mixed layer). In contrast, the time scale for equilibration of O2 produced by photosynthesis and consumed by remineralization is about two weeks. Atmospheric O2 therefore exhibits a far larger ocean-driven seasonal signal than the oceanic signal in CO2. The oxygen signal also shows that Southern Hemisphere seasonality is of comparable magnitude to that in the Northern Hemisphere. Such atmospheric data therefore provide strong constraints on seasonal patterns in oceanic cycling of carbon.

    On interannual time scales, the major signals in atmospheric observations are fluctuations associated with El Niño and with additional as-yet-unexplained fluctuations, such as the recent reduction in atmospheric CO2 that began about the time of the Pinatubo eruption. The oceanic role in the El Niño phenomena has been explored (e.g., Bacastow et al., 1980; Siegenthaler and Wenk, 1989; and Winguth et al., 1994), but is still controversial, given recent contradictory atmospheric carbon-13 measurements (Ciais et al., 1995; Francey et al., 1995; Matear and Holloway, 1995). U.S. JGOFS global survey and equatorial Pacific studies should provide strong constraints on the mechanisms determining interannual variability.

    C. Future Changes

    That a slowing or even a collapse of the ocean thermohaline circulation might occur in response to global warming was demonstrated using a coupled general circulation model of the ocean and atmosphere (Manabe and Stouffer, 1993). A recent simulation (Sarmiento and LeQuere, 1996) of ocean CO2 uptake in this model shows that the potential long-term (350 year) impact of such changes on CO2 uptake is very large: a reduction of almost 40% in the cumulative uptake in a CO2-doubling scenario and almost 50% in a CO2-quadrupling scenario in a "solubility" model that leaves out the effects of biological processes (Jain et al. 1995). Only a small fraction of the reduced uptake is due to temperature effects on solubility and carbon speciation. Adding a parameterization of biological processes to these simulations has a dramatic effect: by exporting organic carbon to the abyss, the biological response is able to cancel much of the reduction in uptake caused by the change in ocean circulation. However, the uncertainty associated with the biological response is very high.

    Another important result from the coupled model simulations is that large changes are produced in the oceanic environment. After 350 years, the simulations predict (i) an increase in surface ocean temperature, which averaged 2.4°C in the CO2 doubling scenario and a staggering 4.9°C in the quadrupling scenario; (ii) a large reduction in oceanic pH by 0.25 units in the CO2 doubling scenario and 0.5 units in the quadrupling case; (iii) a reduction in ocean buffering capacity by a factor of about 2 in the CO2 doubling case and 4 in the quadrupling case, with accompanying comparable reductions in carbonate ion concentration; (iv) a possible impact on photosynthesis from increased in surface ocean CO2 concentrations; and (v) additional changes such as might occur in the supply of micro-nutrients such as Fe and Zn due to changes in atmospheric dust transport. The Southern Ocean is affected more strongly than other regions by all changes (Table 4 from Jain et al., 1995). The extent to which biological processes will alter these predictions depends on the way organisms respond to changed environmental conditions; little is known about how organisms will respond to such changes.

    Table 4. Latitudinal Breakdown of 350 year cumulative atmosphere-ocean flux of anthropogenic CO2 in the 4XCO2 scenario in Pg C.

                       Baseline     Reduction due   Increase due    Super biota   
                      Solubility     to Climate      to Biology    minus abiotic  
                        Model          Change                                     
     North of 30°N  382            -257            61              159            
       30°S-30°N    493            -41             -76             273            
     South of 30°S  1265           -744            193             683            

    The role of biological processes in regulating the historical atmosphere-ocean balance of CO2 also provides indications of changes that might be expected in the future. Two basic explanations have been proposed to explain the glacial-interglacial difference in atmospheric CO2 concentrations (see summary in Sarmiento and Bender, 1994). The first explanation is that biological productivity was enhanced during glacial times, so that organisms near the sea surface removed more CO2 from the atmosphere and, by sinking, carried it to the deep ocean. Knox and McElroy (1984), Sarmiento and Toggweiler (1984) and Siegenthaler and Wenk (1984) have emphasized the importance of the Southern Ocean in this regard, because this region of the ocean contains a very large nutrient pool that would allow enhanced production if other necessary ingredients (specifically iron) were supplied (Sarmiento and Orr, 1991; Sarmiento and LeQuere, 1996). The second explanation is that the oceans might have been more alkaline, thereby drawing down atmospheric CO2 as required by the physical chemistry of the carbonate system. Boyle (1988), for example, has presented evidence that the deep glacial oceans contained a higher concentration of metabolic CO2 than the modern ocean. He argued that this would have led to increased dissolution of CaCO3 from sea floor sediments and, therefore a higher sea water alkalinity. The higher alkalinity would lower the partial pressure of CO2 in the surface ocean and, hence, in the atmosphere. Many variations on these hypotheses have been proposed (e.g., Archer and Maier-Reimer, 1994), along with intriguing discussions of the detailed behavior of CO2 in the oceans and atmosphere during the transitions between glacial and interglacial climates (e.g., Broecker and Denton, 1990).

    These considerations suggest two major conclusions. The first is that the overall magnitudes of oceanic ecosystem processes such as production and export may change in response to climate change, particularly if ocean circulation changes. Such changes may have far-reaching implications for ocean-atmosphere partitioning of carbon.

    The second conclusion is that climate-induced changes in ocean chemistry, such as increased rates of iron input, will almost certainly affect particular aspects of ocean ecology and biogeochemical cycling; the possible effects of these changes must be understood regardless of their consequences for the ocean-atmosphere balance of CO2. For example, changes in the rates of nitrogen fixation and denitrification, the rate of calcification, and the relative dominance of organisms such as diatoms that are efficient exporters of organic matter may all have significant influnces on the production, remineralization, and partitioning of carbon and inorganic nutrients among oceanic compartments.

    D. The Challenge

    One of the major challenges of U.S. JGOFS SMP is to provide a summary description of the present state of the ocean carbon cycle. The recent development of techniques for determining the anthropogenic carbon inventory from temporal changes in total carbon or from reconstruction of the pre-industrial total carbon distribution (e.g., Gruber et al., 1996) suggest that this goal can be attained on a global scale. However, attempts to balance simultaneously the biogeochemical cycles of several elements on a local scale (e.g., nitrate and total carbon) and attempts to reconcile measurements made by different techniques (e.g., sediment traps versus reductions in surface concentrations or radiocarbon versus oxygen-18 estimates of primary production) show that many inconsistencies remain to be reconciled.

    A second major challenge is to provide improved models of the biogeochemical cycling of carbon; such models will be needed both for prediction and for monitoring. During the development of the JGOFS program, models with multiple compartments were constructed to represent then-current paradigms of the controls of production and export. One of the most influential models was that of Fasham et al. (1990) (Figure 2), whose seven nitrogen-based state variables (phytoplankton, zooplankton, detritus, bacteria, dissolved organic nitrogen, nitrate, and ammonium) reflect and are reflected in the suite of JGOFS core measurements. The dominant paradigm at the time, on which this model was based, was one of production (both new production based on nitrate and regenerated production based on ammonium) dominated by large phytoplankton species (notably diatoms) and of export dominated by the sinking of fecal pellets produced by large zooplankton (Dugdale and Goering, 1967; Eppley and Peterson, 1979; Evans and Parslow, 1985; see also Frost, 1987). By now there exist many examples of coupled biological-physical models based on principles similar to those embodied in Fasham et al. (1990). While further developments along these lines will be a necessary part of U.S. JGOFS SMP, development of a reliable predictive capability will also require extensive development of methods for representing processes whose importance was not appreciated at the time the JGOFS program was created. These include:

    (1) The existence and importance of small-celled phytoplankton species (notably of the genera Prochlorococcus and Synechococcus) in oceanic environments was not appreciated, nor was the fact that a large percentage of global productivity is performed by these species (Chisholm, 1992). Similarly, we have gained a greater appreciation for the role of micro-zooplankton and the bacterial loop in surface biogeochemical processes.

    (2) While in many places and at many times both biomass and productivity may be dominated by very small phytoplankton species such as Prochlorococcus, export is often dominated by discrete blooms of large species such as diatoms. This realization has led some workers to propose that the distinction between steady-state and bloom-dominated systems is of fundamental importance. If this proves true, we will need to include explicit representations of export mediated by mechanisms (e.g., the formation of diatom aggregations; see Jackson, 1990) that have not traditionally been included in ecosystem models.

    (3) Export by long-lived ( 1 yr half-life) Dissolved Organic Matter (DOM) has been found necessary for accurately predicting surface ocean export of organic matter (Michaels et al., 1994) as well as deep-water concentrations of dissolved inorganic carbon (Anderson and Sarmiento, 1995; Bacastow and Maier-Reimer, 1991; Najjar et al., 1992; Sarmiento et al., 1988). A more detailed understanding of the production and export of long-lived DOM is beginning to emerge; this picture shows wide variability in contrasting regions such as the Sargasso Sea, where it appears to be high (Ducklow et al., 1995), and the Ross Sea, where it appears to be low (Carlson et al., in review). Long-lived DOM is rarely included in ecosystem models, where DOM, if represented at all, is assumed to decay rapidly.

    (4) The importance of iron limitation in the genesis of the High Nutrient Low Chlorophyll (HNLC) condition in the equatorial Pacific is now beyond dispute (Coale et al., 1996b), but the applicability of this mechanism to other HNLC areas (most notably the Southern Ocean) is still controversial. Even in the equatorial Pacific, much work remains to be done to characterize the sources of iron (e.g., aeolian versus upwelling; Coale et al., 1996a), the responses of phytoplankton growth to iron concentration, and the recycling of iron within the mixed layer.

    (5) Nitrogen fixation (Carpenter and Romans, 1991) and mesoscale eddies (McGillicuddy and Robinson, 1997; McGillicuddy et al., 1995; McGillicuddy et al., 1995) may deliver much more nitrogen to the ocean's mixed layer than previously thought, increasing significantly our estimates of both new production and of the ocean's ability to reduce the partial pressure of CO2 in the mixed layer. In addition, phosphorus may be delivered to the mixed layer by migrating nitrogen fixers such as Trichodesmium (Karl et al., 1992) resulting in the need to represent both phosphorus and nitrogen as limiting macro-nutrients. The apparently large iron requirement for nitrogen fixation is an additional complication that must also be addressed.

    (6) Diel vertical migration of zooplankton may represent an important mechanism for moving carbon to depth. As with the migration of nitrogen fixers, vertical migration of organisms is not easily captured in traditional simulations, which solve ecosystem equations layer-by-layer (e.g., Fasham et al., 1993; Sarmiento et al., 1993). Redistribution of carbon and nutrients by the active movement of organisms must either be resolved using GCM's capable of representing diel cycles of mixed layer depth or must be suitably parameterized.

    The critical problem of adapting nitrogen-based ecosystem models for use in predicting the partitioning of carbon has only begun to be addressed (e.g., Anderson, 1993); similarly, the processes that determine when calcareous and siliceous organisms will be present are still poorly understood. Published data on stoichiometric ratios suggests that uptake in the surface ocean may be highly variable (Sambrotto et al., 1993), but this variability does not appear to be reflected in the relatively uniform remineralization ratios estimated by analysis of nutrient and carbon data below 400 m (e.g., Anderson and Sarmiento, 1994). The extremely high C:N ratios of nitrogen fixers such as Trichodesmium (Karl et al., 1992) may also have an important influence on overall stoichiometric ratios. Observations bearing on these issues must be synthesized and incorporated into quantitative models.

    The role of physical processes in biology has received a great deal of attention during JGOFS, and models must reflect this new knowledge. The role of re-stratification in driving the North Atlantic spring bloom has been elucidated, and the start-and-stop nature of the bloom has become more widely appreciated. The importance of warm and cold states in the equatorial Pacific has received considerable attention, as has the broad reach of El Niño forcing. The role of monsoonal forcing in the Arabian Sea and major frontal features in the Southern Ocean have been or are being studied. The ubiquity of small-scale features such as mesoscale eddies and equatorial instability waves and their impact on biology have been documented.

    Our ability to capture the importance of these processes in predictive models is in its infancy, and a central challenge of U.S. JGOFS SMP will be to support the development of this capability. Broadly speaking, the essential tasks are

    (a) to create model representations of alternative hypotheses for the controls of production and stoichiometric ratios (grazing control; iron or zinc limitation; phosphorus, nitrogen, or silicon limitation; light limitation in areas with deep mixed layers; ocean circulation);

    (b) to predict when and where each of these controls will dominate, and how they will determine the composition of organic matter exported from the photic zone; and

    (c) to predict the consequences of differences in export processes for rates of water column remineralization and sedimentary burial.

    For example, where most export is generated by diatom blooms, depth scales for remineralization will be much longer than in regions where export by DOM dominates, and our models must automatically reflect these differences. That different mechanisms dominate at different times and places is strongly suggested in Figure 6, where the relationship between export production and total production varies markedly among oceanic regions; the controls of these differences and their implications for the depth-distribution of remineralization must be understood and modeled. Also, whether silica forming diatoms, CaCO3 forming coccolithophorids, nitrogen fixing Trichodesmium, or other organisms dominate in a given region may determine the composition of organic matter exported from the surface.

    Figure 6. Comparison of relationship between "ThE" ratio ("ThE" ratio = POC flux derived from 234Th/primary production; y-axis) and primary production (x-axis) for a variety of oceanic provinces. Local range of POC/234Th ratio of particulates is shown in parenthesis. Data ranges (not error bars) are shown for both export ratios and production estimates. Data from K. Buesseler, unless otherwise indicated. Figure adapted from Buesseler, submitted.

    These considerations suggest that the goal of developing an improved model of biogeochemical cycling cannot be met simply by deriving from JGOFS observations better estimates of parameter values for use in standard models (e.g., the model of Fasham et al., 1990; see Figure 2). Instead, achieving U.S. JGOFS SMP goals will likely require major structural changes in the modeling of a range of critical ocean processes (e.g., to include better parameterizations of directed movements by organisms). The development of such models will require close collaboration among those who make observations, modelers of detailed processes, and modelers who will attempt to capture the essence of carbon cycle processes compactly enough to be usable in large-scale models.

    A third major challenge for the SMP is to develop methods that will enable us to incorporate data from ongoing observational programs, particularly from satellites, to monitor ocean biogeochemical cycling. For example, in the open ocean the most abundant calcifying phytoplankton are coccolithophorids. In these organisms, calcification leads to the formation of microscopic plates that can detach from the cell. Light scattered from these plates can be detected by satellite sensors, allowing the spatial distributions of these organisms to be explored using remotely sensed data (Holligan and Balch, 1991; Aiken et al., 1992). Long term monitoring will also require development of measurement devices that can be deployed on floats and ships of opportunity. While the SMP does not have as its goal the development of such devices, the SMP will provide a stronger and more focused rationale for them.

    Finally, the recent IRONEX study in the Equatorial Pacific (Coale et al., 1996a, 1996b) is a shining example of the type of study that could serve as a model for future open ocean experiments that could provide insight into specific changes. One of the important accomplishments of U.S. JGOFS SMP will be to provide background and rationales for such studies in the future.

    The challenges for U.S. JGOFS SMP will only be achieved through quantitative syntheses based on modeling; such quantitative syntheses can only be achieved through the concerted efforts of U.S. JGOFS investigators who gathered data during time-series and process studies, and who therefore have intimate knowledge of the implications and limitations of these data, working with modelers who can capture in mathematical form the key insights gained from these studies.


    A. Introduction

    Conservation of mass provides a fundamental constraint on our understanding of the ocean carbon cycle on all time and space scales. The balance of fluxes of an element into any defined region over any defined period must equal the changes in the inventory of that element over the same region and period. Likewise, changes in the concentration of a constituent (e.g., nitrate or dissolved organic carbon) must equal the sum of the rates of creation, destruction, and exchange. An imbalance in a budget implies incorrect parameter values and/or gaps in our knowledge of the processes involved. This fundamental constraint can be incorporated into biogeochemical studies of the ocean carbon cycle to improve both our understanding of the mechanisms controlling the distribution of carbon and other bioactive elements and our ability to model these mechanisms.

    The use of mass balances to test hypotheses in ocean biogeochemistry is well-established. Historically, however, our ability to constrain balances has been hampered by a lack of data on inventories or fluxes of one or more constituents. This is particularly true for the ocean carbon cycle. For example, controversies over the measurement of dissolved organic carbon once led to large uncertainties in carbon balances. Through JGOFS and other studies over the past decade, we have made important advances in our understanding of biogeochemical flux processes and in our ability to estimate total inventories of dominant constituents. Through U.S. JGOFS SMP we will use these increases in knowledge to create regional carbon and nutrient budgets, to understand the uncertainties in these budgets, and to use the results of these exercises to focus research and synthesis on the inventories and fluxes with the largest remaining uncertainties.

    U.S. JGOFS studies have obtained data that describe features of the ocean over different scales of time and space. The Global Survey is inherently global but not synoptic. The Process Studies have been local-to-regional with near-seasonal temporal resolution. The Time-series Studies are local with a seasonal- to near-decadal temporal resolution. Other studies complement the U.S. JGOFS efforts. In some cases, these are independent efforts in the same geographical areas; in other cases, they are unique efforts elsewhere. One challenge facing the SMP is to synthesize the data and understanding from these studies to create elemental balances for specific regions and for the world ocean.

    U.S. JGOFS SMP is focused strongly on carbon; this focus is determined in large part by the explicit goal that we understand and be able to predict the air-sea partitioning of CO2. However, the carbon budget, both locally and globally, is controlled by a variety of processes. In particular, nutrient availability and utilization regulate carbon cycling in most areas of the world. U.S. JGOFS has spent considerable effort improving our understanding of carbon and nitrogen cycling, and to a lesser extent the cycling of phosphate, silica, and iron. For each of these important nutrient elements and for other biogenic tracers, the constraints imposed by the construction of mass balances can improve our understanding of the carbon system. The simultaneous construction of mass balances for carbon and nutrient elements is a powerful approach for understanding the basic features of ocean ecosystems. We are particularly interested in studying the extent to which cycling of carbon and nutrients is de-coupled by physical forcing (e.g., phosphorus input from mixing versus iron input from aerosols) and biogeochemical processes (e.g., preferential degradation of nutrient-rich biogenic chemicals in sinking particles).

    In this section we present background material on mass balance efforts within the SMP. We begin with a brief description of issues involved in creating mass balances. We then describe a range of research questions relevant to U.S. JGOFS SMP goals where mass balance calculations and constraints may improve our understanding of the oceanic carbon cycle. This list of projects is extensive but not exhaustive, and should be seen as a series of examples of the kinds of synthesis projects that may be appropriate in the SMP context. One issue that occurs throughout these examples is that errors and uncertainties in each approach must be accurately determined on appropriate time and space scales.

    B. Creating a Mass Balance

    Mass balances can be useful on different spatial and temporal scales. For global budgets it is often useful to follow processes along isopycnal surfaces in regions where diapycnal processes can to a first approximation be neglected. Each term in equations (2) and (3) can represent a range of exchange mechanisms, which may vary depending on the chemical constituent in question.

    Optimally, the rate of change of concentration within a reservoir should be measured on the same time and space scales as the fluxes into and out of the reservoir. For many constituents, it may also be possible to assume that the rate of change is zero, particularly over long time periods, simplifying the mass balance significantly. For other constituents, such as the input of anthropogenic carbon, the rate of change represents the signal to be determined and cannot be assumed to be zero.

    On shorter time scales inventory changes can be significant. Air-sea exchange processes include the transfers of heat, momentum, and gases, and wet and dry deposition of various compounds. Physical exchange processes include diffusive exchange along concentration gradients, advective exchange, upwelling and downwelling, convective overturning, and Ekman pumping. Physical exchange processes operate on all dissolved and suspended constituents simultaneously, so that, for example, the upward mixing of nitrate is usually associated with a downward mixing of organic nitrogen. Biological exchanges, a particular focus of U.S. JGOFS, are frequently the most difficult to quantify; they include zooplankton migration, gravitational export of particles, upward transport by positively buoyant organisms, and other processes where the actions of organisms supplement or dominate physical processes in determining transport of material.

    A powerful application of the mass balance approach is to consider simultaneously the mass balances of several constituents. If the biogeochemical cycling of constituents is connected by elemental or isotopic ratios (e.g. C:N; d13C; C:Th), one can use these relationships as independent constraints. One advantage of this approach is that a valid model must produce balance for all elements using the same set of transport parameters. For example, invoking an increased diapycnal mixing rate to balance a carbon budget will simultaneously increase the transport rates for nitrate, phosphate, oxygen, and silicate.

    Simultaneous comparisons of the mass balance of multiple elements can lead to the identification of important but previously unrecognized processes. For example, the importance of nitrogen fixation as a source of nitrogen in subtropical oceans has been inferred. At Bermuda, nitrogen fixation shows up as an imbalance in the nitrogen and carbon concentrations relative to phosphate concentration (Michaels et al., in press). Mass balance equations for 15N, as well as for total nitrogen, provide two independent equations for this element. Nitrogen mass balance calculations at HOT suggest that 20 to 40% of the nitrogen input to the euphotic zone is due to nitrogen fixation. The mass balance for 15N in particulate matter from sediment traps suggests that about half the settling particulate organic nitrogen derives from nitrogen fixation (Karl et al., submitted).

    Mass balances for silicate may also be valuable. For example, in the equatorial Pacific, variability in diatom abundance may be an important response to episodic inputs of iron. In NABE, silicate was depleted before nitrate and appeared to cause a shift in community structure from diatoms to other phytoplankton. Biogenic silica is also interesting because it is remineralized with a longer length scale than are particulate organic C, N, and P.

    Mass balance equations can also be written for particle-reactive tracers such as thorium (234Th, 230Th, and 228Th), polonium (210Po), and lead (210Pb). Mass balance approaches have been used to estimate the sinking flux of particulate carbon. The range of half-lives of these tracers provides an opportunity to calculate carbon fluxes on different time and space scales. For example, EqPac investigators used mass balance equations, including vertical and horizontal transport of 234Th, to calculate vertical flux. Mass balances with age tracers (14C and CFC's) are useful for calculating net respiration rates.

    C. Global and Regional Mass Balances

    1. Estimating global carbon inventories

    A fundamental goal of the U.S. JGOFS global survey is to improve our estimates of the distributions of total dissolved inorganic carbon (DIC) and anthropogenic carbon in the ocean. One of the major reasons for this effort is to provide a baseline for future changes. However, we can also use these data to estimate directly the inventory of anthropogenic carbon. There is a need to assess the errors and uncertainties associated with the extrapolation of data to the global scale.

    Several recent investigations have shown how the U.S. JGOFS Global CO2 Survey can be used to refine our estimates of oceanic inventories of CO2. Gruber et al. (1996) have used DIC, alkalinity, tracer, and hydrographic data to obtain a new estimate of the inventory of anthropogenic CO2 in the North Atlantic (20 ± 4 Pg C); this result agrees well with recent results from the Princeton GCM (Gruber et al., 1996). Wallace (1995) and Slansky et al. (submitted) have developed multi-parameter linear regression analysis procedures for estimating the fossil component of ocean CO2 inventories from data obtained on overlapping cruises separated by periods of a few years, with an estimated uncertainty of 3-4 mmol kg-1. These methods allow survey data to be used to estimate anthropogenic CO2 inventories throughout the oceans. These estimates provide powerful constraints on coupled atmosphere-ocean GCM's as applied to carbon budgets.

    The same global data set will allow a closer examination of regional variations in anthropogenic carbon inventories. This information, in turn, reflects the spatial distribution of fossil CO2 uptake. Small volumes of ocean with large exchange signals may be easier to constrain than the entire global system. Regional analyses will also provide a context for interpreting local process studies and time series cruises.

    2. Estimating the global balance of DIC fluxes

    Estimating the distribution of DIC in the ocean is but one step in understanding the dynamics of its control. Our current understanding of the ocean carbon cycle is that the net oceanic uptake of carbon from the atmosphere is approximately 2 Gt C/y. Our goals are to assess the accuracy of this flux estimate using measurements, to estimate its uncertainty, to understand mechanistic controls on this flux on time-scales of years to centuries, and to determine its spatial and temporal variability.

    The global survey data and a range of additional measurements will allow us to make improved assessments of the balance of fluxes into and out of the ocean and to determine whether these fluxes lead to changes in the oceanic inventory of carbon. This effort requires a careful assessment of all boundary exchanges on a global scale. U.S. JGOFS SMP activities can play a particularly important role in determining the air-sea flux of CO2. As with the inventory assessments, this will involve an extrapolation of existing survey data to global scales and their coupling to global estimates of gas exchange. The uncertainties associated with this effort must also be determined.

    As part of the quantification of fluxes in individual sub-regions, we must assess the balance of processes that determine these fluxes. Sub-regions of the ocean can have large systematic air-sea pCO2 differences with large net fluxes. River inputs of carbon and carbon burial on the sea floor also have strong regional differences. Regional ocean circulation features transport large amounts of carbon, and the survey data on DIC and DOC concentrations, when combined with extensive WOCE hydrographic data collected simultaneously, should allow vastly improved estimates of these physical transports (Holfort et al., 1995)

    For example, Takahashi et al. (in press) have provided a new estimate of regional sources and sinks of CO2 in the oceans. These estimates are based on seasonal pCO2 data combined with surface flows obtained from a general circulation model. Their findings indicate that the northern oceanic sink for atmospheric CO2 is about a factor of two higher than the previous estimate of Tans et al. (1990), the equatorial source is slightly smaller, and the southern oceanic sink is nearly a factor of two smaller than the previous estimate. Although the uncertainties are large, these results provide a significant new constraint on the spatial distribution of surface-ocean CO2 fluxes in GCM's. These regional balances will be substantially improved by local data from the process studies and time-series studies because these studies allow an improved understanding of processes controlling the carbon balance in each region.

    3. Extrapolation.

    A principal goal of JGOFS is to reduce uncertainties in the distributions of inventories and fluxes of carbon in the ocean. In general, logistical constraints required most JGOFS data to be collected on local or regional scales. An important exception is the U.S. JGOFS Global CO2 Survey. There are also additional U.S. JGOFS-related data sets (e.g., ocean color) that have global coverage. To extrapolate local and regional JGOFS data globally, they must be scaled using these global data sets. Comparison of property fields using observations and models will provide valuable constraints on fluxes of carbon and related chemical elements among the atmosphere, the ocean, and the terrestrial biosphere. Within U.S. JGOFS SMP, implementation will focus on both generation and joint analysis of global fields.

    D. Implementation

    JGOFS data have been collected on disparate scales of time and space. These include a global view of the distribution of oceanic DIC from the Global CO2 Survey, the long-term time-series data collected at Hawaii and Bermuda, and the seasonally-resolved research conducted by the process studies. The SMP will support activities aimed at extrapolating these results to the global scale. The following activities will be supported.

    (a) Production of regional and global fields of carbon and related substances. Examples of such activities include:

    (i) A standardized Global CO2 Survey data set. The Global Survey data set forms the basis for direct estimation of natural and anthropogenic inventories of CO2. It will also serve as an important constraint, validation tool, and initialization tool for ocean carbon cycle models. Related measurements of 13C, oxygen, and nutrients will also be required for such analyses. These data have been collected over the course of the past decade in various seasons, and are aliased in the upper water column by temporal variability. In order to be made most useful for global-scale analyses, the data must be interpolated globally with explicit recognition of seasonal and known interannual (e.g., ENSO) variability. Interpolation schemes should preserve observed relationships between carbon-cycle parameters (e.g., stoichiometric relationships), since these will provide important diagnostics of model performance.

    The extrapolations should provide measures of the uncertainty and aim at reducing uncertainties in such estimates as already exist. Given that the anthropogenic CO2 invasion has increased surface total carbon concentration by about 40 mmol kg-1 and the total carbon inventory by about 100 Pg C yr-1, the goal should be to produce concentration fields to better than about 4 mmol kg-1 at the surface, and global inventories to better than about 10 Pg C. Local inventories would have to achieve smaller uncertainties to be useful. For example, the North Atlantic has an anthropogenic carbon inventory of about 20 Pg C (Gaffin et al., 1995) which suggests that an uncertainty of order 2 Pg C should be aimed for.

    (ii) A surface pCO2 climatology. The spatial and temporal distribution of air-sea CO2 fluxes provides a key constraint on global sources and sinks of CO2 (Tans et al., 1990). A recent estimate of the global air-sea flux of CO2 has been developed by Takahashi et al. (1996), who used a model of the surface ocean to extrapolate sparse information on the seasonal variability of surface water pCO2 to much larger spatial scales. As a result of JGOFS, a global data set much larger than that available to Takahashi is being compiled under the auspices of the IOC-SCOR CO2 Panel. A re-analysis of the combined data set may be appropriate as part of U.S. JGOFS SMP.

    It is unlikely that the global pCO2 climatology can be known to sufficient accuracy to determine the total oceanic uptake of anthropogenic carbon. Variability in time, the difficulty in establishing correlations with wind speed and the continuing difficulty in establishing the gas exchange coefficient, all make this extremely difficult. The value of these measurements lies in establishing signs and regional differences in fluxes. For example, the extremely large air-sea CO2 difference implied by the proposed large northern hemisphere carbon sink of Keeling et al. (1989b) should be detectable if the oceanic pCO2 can be estimated to within approximately 5 ppm (10% of the North Atlantic air-sea CO2 difference they propose).

    (b) Production of regional and global fields of fluxes of carbon and related substances. Examples of such activities include:

    (i) Air-sea flux fields based on the pCO2 climatology and models of gas exchange. The value of such fields in estimating regional flux differences has been discussed above. Uncertainty regarding the gas exchange coefficient, e.g., Liss and Merlivat (1986) vs. Wanninkhof (1992), limits the confidence that one can place in flux predictions to a factor of roughly 2, but the regional flux distribution can still provide strong constraints. The total oceanic uptake is about 2 Pg C yr-1, with an estimated northern hemisphere uptake of order 0.5 Pg C yr-1. Regional fluxes would thus be most useful if they could be estimated to an accuracy of about 0.1 Pg C yr-1.

    (ii) Fluxes of carbon and related substances within the ocean, such as export fluxes of dissolved and particulate organic carbon from the surface to the abyss. Oxygen observations in both the atmosphere and ocean should prove very useful in estimating seasonal effects. If we take the air-sea flux of anthropogenic carbon as a measure, it would be most useful to have global estimates of these fluxes to better than 1 Pg C yr-1, and regional fluxes to about 0.1 Pg C yr-1. It seems highly unlikely that this will be achievable, even using satellite observations for spatial and temporal extrapolations. However, any significant improvement over present estimates would be extremely useful.

    (iii) A benthic flux field. Several regional and global maps of benthic fluxes have proven to be quite robust. These compilations could be updated with recent results from JGOFS and other benthic studies.

    (c) Production of joint analyses of flux and tracer fields.

    Extrapolation activities will include analyzing and comparing fields of properties and fluxes relevant to the carbon cycle. Studies that involve the analysis and comparison of these fields should provide insights into the ocean carbon cycle. One example of such studies would be a comparison of interannual variation in wind fields with interannual variation in ocean color, which might allow a test of whether atmospheric iron deposition at the sea surface influences production (or biomass) on a large scale and evaluation of the influence of wintertime winds (and the depth of wintertime mixing) on spring and summer productivity. Another example would be a comparison of spatial variation in total primary production estimated from ocean color data with variation in annual new production estimated from climatology-driven ocean GCM's; one outcome of this comparison would be a rough estimate of f-ratios on a global scale. A third example would be a comparison of benthic fluxes to upper ocean new and total production; this comparison could reveal spatial differences in the extent of organic matter recycling between the euphotic zone and the ocean floor.

    (d) Production of global fields of biogeochemical parameters.

    Most biogeochemical data from JGOFS have been collected at fixed locations, often with seasonal and interannual resolution. Certain rates, pool sizes, and mechanistic information gained from the time-series and process studies and related studies may require global extrapolation to assess their global significance and to allow their use in models. Satellite observations will play a central role in this activity. Extrapolation will surely not be justified for all JGOFS measurements, however, and any proposed activity in this area will require justification of its potential for increasing understanding of the ocean carbon cycle on basin and global scales.


    A. Introduction

    The ultimate goal of U.S. JGOFS SMP is to develop a capability to predict both the response of the oceanic carbon system to climate change and the resulting feedback to the climate system. Modeling the mechanistic bases of these responses is a key element of this program, since only by accumulating confidence in our understanding of mechanisms can we become confident in our predictive capabilities.

    The mutually supportive tasks of synthesizing and modeling mechanistic controls on the flows of carbon and associated biogenic elements is a critical link to the overall U.S. JGOFS SMP goals of understanding and prediction. Although processes that transfer large amounts of carbon are readily identifiable, the mechanistic controls of these processes are often not well understood. To help alleviate these uncertainties, we must develop techniques that will improve our understanding of these processes and their controls. And despite the uncertainties, we must move forward to improve the predictive capabilities of models. This challenge can only be met through a coordinated effort by empiricists and modelers to synthesize the knowledge gained from JGOFS and related studies in a manner that will support the development of better predictive models.

    Figure 7 summarizes the key features of the oceanic carbon cycle. The left-hand column in this figure refers to processes that take place in the deep-water formations regions of the North Atlantic and Southern Ocean, while the right-hand column refers to other locations.

    Figure 7. Key features of the oceanic carbon cycle.

    Within the euphotic zone, photosynthetic productivity removes carbon from the dissolved inorganic carbon (DIC) pool, converting this carbon to an organic form that is not exchanged with the atmosphere. The maximum potential draw-down in DIC is determined by the availability of mineral nutrients. In most oceanic regions, the availability of the macro-nutrients nitrate and/or phosphate limit productivity, while zooplankton grazing and the availability of other macro-nutrients (e.g., silica for diatoms) and micro-nutrients (e.g., iron, zinc) are among the factors that determine how this production is partitioned by algal taxa (diatoms; coccolithophorids; prochlorophytes; dinoflagellates) and by size. In so-called High Nutrient Low Chlorophyll (HNLC) regions, by contrast, not all phosphate and nitrate are used during the growing season, so that the draw-down of inorganic carbon is much less than in macronutrient-rich non-HNLC areas.

    Figure 7 also indicates that part of the carbon fixed by photosynthesis is remineralized within the mixed layer, while part is exported to depth and remineralized there. The partial pressure of carbon dioxide in the mixed layer depends only on the amount of photosynthetically fixed carbon that cannot exchange with the atmosphere, not on whether that carbon has been exported or simply sequestered in an organic form while remaining in the mixed layer. The export of organic carbon from the mixed layer is critically important, however, in determining the partitioning of carbon among shallow and deep oceanic reservoirs on longer time scales, so that the ability to predict the depth distribution of remineralization is of fundamental importance (Figure 7).

    Finally, remineralized nutrients and carbon are brought up to the photic zone from depth to fuel the next round of production and export. Depending on the depth of remineralization and the details of atmospheric circulation, this resurfacing of nutrients and carbon could occur locally after a short time interval, or in the case of the thermohaline circulation could occur on the other side of the globe hundreds of years later; the division of Figure 7 into deep-water formation regions versus other regions serves to emphasize the importance of transport by ocean currents.

    In the following sections we describe the functional processes that must be included in a predictive model and discuss what is known about these processes from empirical studies; these discussions define a range of synthesis and modeling studies that are most likely to be useful in achieving U.S. JGOFS SMP goals. While reflecting our current best guesses as to the key processes that must be represented, this list should not be view as exhaustive, since other processes may emerge as being of equal or greater importance relative to those discussed below. In Section B we discuss major qualitative differences among oceanic regimes; the genesis and maintenance of HNLC versus non-HNLC areas is the primary focus of this section. In Section C we discuss in more detail the processes depicted in Figure 7. In Section D we discuss the use of local and regional mass balances in the evaluation of mechanistic controls. Section E contains a summary of important studies that must be undertaken on mechanistic controls to enable the development of powerful mechanism-based predictive models.

    B. Qualitative Differences Among Oceanic Regimes

    In most oceanic regimes, new production is controlled by the abundance of the macro-nutrients nitrate and/or phosphate. If carbon in these regimes is fixed in a constant stoichiometric ratio to nutrient availability (see below), then when these nutrients are not fully utilized, the conversion of inorganic carbon to organic matter will be less than its maximum potential value, with consequences for the atmosphere-ocean exchange of carbon.

    The most important regions controlling the ocean-atmosphere partitioning of carbon are the deep-water formation regions of the North Atlantic and Southern Ocean, where the atmosphere communicates with the deep ocean. The level of biological activity in these regions determines the gradient of dissolved inorganic carbon between the surface ocean and the deep ocean, thereby determining to a large extent the ocean-atmosphere partitioning of carbon. These regions are currently High Nutrient Low Chlorophyll (HNLC) regions, where the biota do not totally consume available nitrate and phosphate at any time in the yearly cycle, so that surface concentrations of CO2 are not reduced by their full potential amount. A change in the level of biological activity in these regions would have important consequences for the partitioning of carbon between the ocean and atmosphere (Joos et al., 1991; Sarmiento and Orr, 1991).

    The mechanistic control of the difference between HNLC and non-HNLC regions appears to devolve from an interaction between micro-nutrient (iron) limitation of algal growth and grazing (Coale et al., 1996b; Landry et al., in press). In particular, smaller phytoplankton size classes are thought to be controlled by micro-grazers, while larger phytoplankton size classes are thought to be limited in their maximum growth rates by their smaller surface-volume ratios, which limit their ability to take up iron. The result is an ecosystem dominated by small-celled phytoplankton, but which can produce blooms of large phytoplankton in response to iron enrichment (Coale et al., 1996b). It is critical that appropriate mechanistic controls of large-scale ecosystem responses to altered forcing, such as alterations to the aeolian deposition of micro-nutrients, be incorporated into models, so that the locations and spatial extents of HNLC regions will respond appropriately under alternative climatic regimes.

    C. Mechanistic Controls Within Oceanographic Regions

    In this section we discuss in more detail the processes indicated in Figure 7. We first examine the central role of nutrient supply in setting the maximum conversion of DIC to organic carbon. We then examine the role of food web processes in determining the fraction of this production that is remineralized within the euphotic zone versus the fraction that is exported from the euphotic zone for remineralization at depth or burial in the sediments. Finally, the roles of deep-water remineralization and sedimentary diagenesis are considered within the context of carbon partitioning among oceanic reservoirs.

    1. Ultimate limits to the draw-down of DIC

    a) Nutrients

    The maximum amount of carbon that can be converted from DIC to organic forms is ultimately set by the supply of nutrients. Because organisms are constructed from a limited set of constituent building blocks, each with a characteristic ratio of C, N, O, P, and other nutrients, living organisms tend to have elemental ratios that are close to those proposed by Redfield. If the oceans were populated only by autotrophic organisms, then when the chemical element in shortest supply (relative to its stoichiometric ratio) was exhausted, no new organic material could be synthesized. However, non-living organic pools such as Dissolved Organic Matter (DOM) may have much larger C:N and C:P ratios than living phytoplankton, so that if a substantial amount of fixed organic carbon is converted to such compounds, the amount of carbon removed from the DIC pool in the face of complete utilization of a mineral nutrient such as nitrate may be much larger than predicted from Redfield stoichiometry (Sambrotto et al., 1993). Understanding and predicting variations in stoichiometric ratios is therefore fundamental to predicting the air-sea partition of carbon.

    Note particularly that control of the partial pressure of carbon dioxide, the important variable for exchange of carbon with the atmosphere, may have little relationship to the control of total primary productivity, a large part of which may be fueled by nutrients regenerated within the euphotic zone, resulting in no further net removal of DIC (Dugdale and Goering, 1967; Eppley and Peterson, 1979).

    Nutrients are transported from depth into the euphotic zone by deep wintertime convection, upwelling, mixing along isopycnal surfaces in the thermocline, surface ocean currents, large scale thermohaline circulation processes, mesoscale advection, and eddy diffusive processes. Consequently, in the absence of external nutrient sources, the rate of biologically driven export is ultimately controlled by the combined rate of these nutrient return mechanisms. Regional and seasonal variations in nutrient supply impose substantial variability on the rates of biological export. The length scale of remineralization will determine the time that is required to complete one circuit in this coupled physical-biological cycle of biogenic elements.

    In certain oceanic habitats, nutrient elements can enter the system by alternative processes. For example, in coastal regions land runoff may represent a major flux of biogenic elements. In other regions, atmospheric deposition can supply both major and trace nutrients to the ecosystem. In the case of iron, atmospheric deposition may be the most important source of "new" iron, especially in the northern hemisphere. Another potential mechanism for de-coupling the input-export nutrient balance is nitrogen fixation. Nitrogen fixation may occur in response to nitrogen stress in nitrogen-poor situations and may often be accompanied by the active transport of phosphate from below the mixed layer into the photic zone by motile nitrogen-fixing phytoplankton (Karl et al., 1996). The aeolian deposition of iron may be further linked to nitrogen fixation by the large iron requirement of nitrogen-fixing cyanobacteria.

    Ocean chemistry and biology are interconnected, and both are affected by ocean physics. This influence of physics on biogeochemistry is manifested particularly strongly at the oceanic mesoscale. Highly energetic currents, fronts, and eddies with spatial scales on the order of 100 km and temporal scales on the order of weeks to months are ubiquitous in the world ocean. Such phenomena have been present during each of the U.S. JGOFS process studies as well as at the time-series sites. The implications are twofold. First, mesoscale variability is aliased into observations, often making it difficult to distinguish between spatial and temporal variability. Second, mesoscale dynamical processes contribute to local and regional biogeochemical balances to varying degrees. Their role in mechanistic control can be broadly categorized into those processes that impact biological rates (e.g. eddy-induced nutrient transport [BATS, HOT, NABE]) and those that simply redistribute material (e.g. convergence at fronts [EqPac]).

    These considerations imply that modeling nutrient delivery mechanisms, both physical (upwelling; currents; mesoscale eddies; atmospheric deposition) and biological (nitrogen fixation; biological phosphate transport into the photic zone) will be an important task in U.S. JGOFS SMP.

    b) Light

    In special circumstances, such as the Southern Ocean, the combination of low light and deep convective mixing has been hypothesized to limit the maximum draw-down of DIC (Mitchell and Holm-Hansen, 1991; Mitchell et al., 1991; Nelson and Smith, 1991; Sverdrup, 1953). In high latitudes, light limitation dictates that draw-down of DIC will occur only during a limited part of the year, exerting strong control over the air-sea partitioning of carbon. Outside these regions, however, the effect of light on algal productivity has little effect on the potential draw-down of DIC. Even so, parameterizations of light-limited growth based on variable chlorophyll:nitrogen ratios have been found necessary for simultaneously fitting chlorophyll, nitrate, and productivity time-series data at BATS (Doney et al., 1996; Hurtt and Armstrong, 1996), so that further research on this topic may be warranted to allow fuller use of ocean color data sets in interpreting and extrapolating other data.

    2. Partitioning of productivity by ecosystem processes

    The food web structure of marine ecosystems has potentially profound implications for the flows and fates of carbon and biologically active elements; the partitioning of carbon among size classes and among taxa with distinct biogeochemistry are two areas of special importance.

    a) Size structure

    One of the guiding paradigms of the JGOFS Program has been the notion that the size structure of pelagic communities strongly influences the partitioning of primary production into particulate export flux, dissolved organic matter, and remineralized inorganic constituents. Where the base of the food web is dominated by small primary producers, as is the case for some macro-nutrient or trace-element-limited systems (subtropical gyres; HNLC ecosystems), small consumers (protistan micro-zooplankton) dominate the grazing process, production is dissipated in longer trophic pathways, and the fraction of organic matter that is exported is low. In contrast, where the system is dominated by larger primary producers (areas of sustained upwelling or convection during Arabian Sea monsoons; seasonal blooms in the North Atlantic and coastal ecosystems), the direct grazing of primary producers by larger consumers (meso-zooplankton) may exceed that of micro-zooplankton, the mean trophic pathway will be shorter and more efficient in transforming production into rapidly sinking fecal material, and the fraction of organic matter that is exported will be higher.

    The JGOFS data set will provide a basis for characterizing food web relationships within and across pelagic ecosystems. U.S. JGOFS SMP will support research directed towards determining and representing important aspects of size structure, including the selection of consumer diet by size, the direct role of grazing by larger consumers in determining particulate export fluxes, the transformations of food resources into particles and into dissolved organic and inorganic matter, and the sources and fates of dissolved organic matter.

    b) Biogeochemical diversity

    Several functional groups of the upper water column deserve special note. One group includes organisms associated with the precipitation and export of inorganic carbonate (coccolithophorids, foraminifera, and pteropods) and silica (principally diatoms, but also radiolarians and silicoflagellates). Another group includes nitrogen fixers (Trichodesmium; Rhizosolenia with its bacterial endosymbionts). A third includes producers of climate-relevant gases (e.g., DMS production by coccolithophorids and Phaeocystis). Finally, some organisms have mechanisms for the particularly rapid and efficient production of particles and their transport to the deep sea (e.g., Phaeocystis; diatom aggregations; pelagic tunicates). As in the case of community size structure, the relative abundance and biogeochemical importance of functional groups varies seasonally and spatially within regions, as well as from one region to another.

    At the oligotrophic HOT and BATS time-series stations, for instance, the importance of Trichodesmium spp. and nitrogen fixation may reflect the relative importance of nitrogen and phosphorus limitation, assuming that these organisms have the ability to mine phosphate at depth by regulating buoyancy (Karl et al., 1996). Diatoms appear to be competitively inefficient relative to smaller forms under macro-nutrient and trace element limitation (e.g., subtropical gyres and HNLC regions like the central equatorial Pacific and the Antarctic Circumpolar Current) but can dominate when growth substrates are available in excess (coastal ecosystems; Arabian Sea monsoon seasons; the Ross Sea). The depletion of dissolved silica when other nutrients are still available in excess may lead to successional changes in the phytoplankton community (e.g., coccolithophorid or Phaeocyctis blooms in the North Atlantic, Arabian Sea, and Southern Ocean).

    It is essential that the relative roles of physical forcing, grazing, and macro-nutrient and trace-element limitation in regulating the relative abundance of taxa of distinct biogeochemistry in space and time be determined and modeled; U.S. JGOFS SMP will therefore support synthesis and modeling efforts directed towards developing a capability to predict the relative abundance of taxa of diverse biogeochemistry and the implications of this diversity for the oceanic carbon cycle.

    3. Remineralization and export

    Food web structure determines the partitioning between remineralization in surface layers and export to deeper waters. Organic carbon can be exported as sinking particles, by animal migration, and by advection of DOM to depth by Ekman pumping, horizontal induction into the thermocline, mesoscale flows, or turbulent mixing. U.S. JGOFS SMP will support efforts to resolve these export mechanisms to create better representations of remineralization and export.

    The mechanisms, and hence the rates, of carbon export, are fundamentally controlled by ecosystem dynamics in the surface ocean. A distinction can be drawn between steady-state ecosystems without distinct phytoplankton blooms (e.g., the equatorial Pacific) and transient-controlled ecosystems whose dynamics are dominated by phytoplankton blooms (e.g., the North Atlantic, coastal upwelling regions, and marginal ice zones). This distinction may be especially relevant to the relative and absolute rates of production and export of particulate and dissolved organic matter. Steady-state regions tend to be dominated by small-celled phytoplankton, held in check by small grazers. This system is typified by suspended or slowly sinking particles, by the importance of ammonium for primary production, and by the abundance of dissolved organic material (DOM) DOM may accumulate or may be remineralized by bacteria, which in turn are remineralized by micro-grazers. Bloom-dominated regions tend to sustain larger size classes of phytoplankton, which contribute disproportionally to sinking POC, and to stimulate the growth of large grazers, which also contribute to POC export

    The export of biogenic carbon is responsible for maintaining the gradient of dissolved inorganic carbon. Traditionally, the export of particulate organic carbon has been thought to dominate this process. Recently, direct observations of DOC dynamics in the Mediterranean Sea (Copin-Montegut and Avril, 1993) and in the Sargasso Sea (Carlson et al., 1994) have demonstrated that a significant portion of the annual DOC production from the upper ocean can be exported vertically as a result of deep convective mixing. These observations indicate that DOC export can be an important part of the biological pump in some oceanic environments (Ducklow et al., 1995).

    The decline of the flux F of biogenic material with depth below a reference depth of z* is often formulated as

    A commonly used version of this equation was derived by Martin et al. (1987) based on a few stations in the eastern low-latitude Pacific. The exponent b reflects relative rates of degradation and sinking. This equation predicts the carbon flux well at the U.S. JGOFS HOT station but does not explain the shape of the decline of F with z in the Arabian Sea or the South Atlantic (Banse, 1994).

    An alternative approach that has provided considerable insight is the use of trace metal tracers to determine particle cycling rates (Bacon and Anderson, 1982; Murnane et al., 1996). This research reveals a much more complex view of particle cycling dynamics than can be represented by a simple power law relationship. Since the depth-distribution of remineralization is critical to predicting the partitioning of carbon among oceanic reservoirs, U.S. JGOFS SMP will support efforts to develop a better mechanistic understanding and quantitative description of remineralization and export processes.

    4. Sediment diagenesis and burial

    Marine sediments play important roles both as a buffer within the oceanic carbon cycle and as a long term repository for carbon and other biogenic matter. Diagenetic reactions in the deep ocean are important on time scales of ocean mixing (100's of years), and burial terms are relevant on longer time scales (1000's of years). Deep-sea sediment diagenesis measurements may also provide a test of the accuracy of deep sediment trap measurements and of euphotic zone export models and measurements.

    Burial of organic carbon and CaCO3 in marine sediments represents removal of carbon and alkalinity from the global cycle. The ratio at which these materials reach the sediments may play a major role in determining the diagenesis of CaCO3 (Archer et al., 1989; Emerson and Archer, 1990; Emerson and Bender, 1981) and could have a significant impact on atmospheric CO2 on longer time scales (Archer and Maier-Reimer, 1995).

    U.S. JGOFS process and time-series studies have produced new insights into the coupling between benthic and surface ocean processes. Photographic evidence from the sea floor collected during NABE identified rapid transport events that carry chlorophyll-rich material to the sea floor; such discrete export events should produce different profiles of remineralization with depth than do steady rains of material, so that understanding the differences between these mechanisms may be crucial to predicting the partition of carbon among oceanic reservoirs. Pulsed delivery is the natural state of the Arabian Sea, and even the equatorial Pacific sea floor shows evidence of rapid delivery events (Smith et al.). The sea floor beneath the Antarctic Polar Front Zone provides evidence of high export rates of organic matter from the euphotic zone, whereas surface ocean and satellite measurements would not predict such large fluxes. Organic carbon oxidation on the sea floor from the BATS region appears de-coupled from the rain of carbon, whereas within the northeast Pacific (Smith et al.) and within the EqPac region (Hammond et al., 1996) there appears to be a closer coupling.

    Of particular interest to U.S. JGOFS SMP would be benthic studies that could be used to constrain rates and locations of water column remineralization, with implications for the partitioning of material among water column and benthic reservoirs.

    D. Local and Regional Mass Balances

    Carbon balances are constructed using continuity equations that describe the time rate of change of pools of dissolved inorganic carbon (DIC), dissolved organic carbon (DOC), and particulate inorganic carbon (PIC), and particulate organic carbon (POC). One of the most important characteristics of the JGOFS program is that independent estimates of many of these terms have been made. If independent estimates agree in a certain region and over a particular time period, then our confidence in the carbon flux and production rate measurements are enhanced. In contrast, if these independent estimates disagree, then it becomes likely that certain flux measurements are in error and/or that important processes have been excluded.

    If the important carbon fluxes could be measured accurately, then a carbon balance would be insured. Typically, however, only one or two carbon fluxes are measured at any given site, so that a conclusive test of measurement accuracy is usually not possible. During JGOFS, a major effort was made to obtain independent estimates of critical carbon fluxes, so that in some circumstances, at least, tests of the accuracy of specific carbon fluxes are possible.

    1. Production and export rates of DOC, PIC and POC

    During the U.S. JGOFS field studies, primary production was measured using 14C uptake and O2 changes (e.g., Bender et al., 1992). New production was separated from recycled production using 15N (NO3- and NH4+ ) measurements (e.g. McCarthy et al., submitted). While the precision of these rate measurements is typically about ±20%, their accuracy is less well known. Net POC export was measured both directly (using sediment traps) and indirectly (by measuring the disequilibrium of 234Th in the water; the latter is then converted into an organic carbon export rate by assigning a C:Th ratio to the POC (see, e.g., Buesseler et al., 1994). A major question arising from the U.S. JGOFS studies involves the accuracy of the sediment trap and 234Th methods. Sediment traps are attractive because they measure POC flux directly and have been used with apparent success in the deep ocean. In the upper ocean, however, it is still unclear whether traps underestimate or overestimate the flux of POC. A definitive experiment to test the accuracy of trap collections has not yet been performed. Evaluating the accuracy of the sediment traps is further complicated by the likely dependence of trap performance on specific oceanic conditions (e.g., depth and current shear). The 234Th method provides an important independent, though indirect, estimate of POC loss. A major uncertainty, however, results from the correct choice of a C:Th ratio for the POC being exported and, to a lesser degree, the need to account for 234Th transported by currents and mixing (Murray et al., 1996 [2nd EqPac volume]).

    Rates of DOC production and export have typically not been measured directly and have been estimated by combining measured DOC gradients with rates of advection and mixing supplied independently from circulation models (e.g., Feely et al., 1995).

    2. DIC fluxes

    Two physical processes, air-sea CO2 exchange and DIC transport due to circulation and mixing, have a substantial impact on DIC budgets. Net CO2 gas exchange rates are typically determined by combining gas transfer rates (estimated from empirical relationships to wind speed) with measurements of pCO2 (e.g. Feely et al., 1995). Most of the uncertainty in the resulting CO2 flux estimate lies in the uncertainty of the gas transfer rate (e.g., Wanninkhof, 1992). In situations where pCO2 varies significantly over time, insufficiently frequent pCO2 measurements can contribute substantial uncertainty to the net CO2 gas flux estimate. The overall uncertainty in the gas transfer rate is about 30-40% at typical ocean wind speeds; however, the discrepancy between rates based on laboratory measurements and rates based on studies of ocean tracers increases substantially at higher wind speeds.

    Rates of supply of DIC by advection and mixing are difficult to quantify. Typically, DIC and nutrient supply rates are determined by applying estimates of the advection and diffusion terms from a circulation model or other independent source to measured gradients of DIC (e.g., Toggweiler and Carson, 1995). This approach has the inherent difficulty that the advection and diffusion terms are typically either climatological averages or represent specific experimental conditions, whereas the DIC gradients represent instantaneous conditions measured at the study site. The accuracy of the advection speeds and mixing rates predicted by the models is not well known. For these reasons, there is often large uncertainty in the DIC (and nutrient) supply rate. The rate of physical supply of DIC (or nutrients) can be estimated from the relationship of gradients in DIC to gradients in tracers of mixing (e.g., 3He, 228Ra, bomb 14C); however, these studies have had only limited application (e.g., Jenkins, 1988). The physical supply rate of DIC into the euphotic zone has been calculated using a combination of DIC and DI13C budgets (Zhang and Quay, submitted); this method relies on an estimate of the CO2 gas exchange rate.

    3. Remineralization

    In situ remineralization affects mass balances by determining vertical gradients of many chemical constituents, and hence their transport by advection, diffusion, and sinking particles and migrating organisms. Biological activity returns organic materials to an inorganic form and reprocesses and repackages organic materials from one form to another. Even when the total mass of an element is conserved, biological transformation among organic and inorganic forms creates the possibility of differential transport caused by differential interactions with sinking particles and with organisms. Accurate assessment of remineralization processes and the interaction of remineralization products with organisms and particles is therefore crucial to understanding overall mass balances.

    4. Time rates of change

    Rate-of-change terms in carbon budgets must also be estimated. Time series measurements may be used to estimate these terms; alternatively, steady-state conditions are often assumed. Accurately determining the rates of change can be difficult. For example, if the total inventory of DIC over a 100m deep layer is determined with a precision of 0.5%, then the error in two sets of DIC measurements separated by one month would translate to a DIC flux of 30 mmol m-2 d-1, which is greater than typical organic carbon export rates. Even time-series measurements do not guarantee a useful time-rate-of-change determination. If short-term (daily or weekly) variations are large, one is faced with separating spatial from temporal variability. Generally the longer time interval over which one tries to construct a carbon budget, the smaller the magnitude and error of the rates of change. Investigators have often had little or no data to evaluate these terms and a steady-state assumption was the only choice. During U.S. JGOFS time series studies, monthly sampling over many years has provided data to evaluate seasonal time rates of change.

    E. Implementation

    The foregoing discussion suggests several key issues that must be addressed during the SMP phase of U.S. JGOFS. Research topics that will be supported by U.S. JGOFS SMP include the following.

    (i) Reassessing the accuracy of key rates and inventories estimated from data collected during the field programs, including new production estimated from incubations, sediment trap fluxes, and 234Th-derived POC fluxes. Previous attempts to develop local flux balances at U.S. JGOFS time series and process study sites have met with mixed success, due in part to large uncertainties in the flux estimates themselves. The SMP will support well-focused experimental and modeling efforts aimed at improving our knowledge of the accuracy of key U.S. JGOFS rate and inventory determinations. It is important to distinguish clearly accuracy and precision, since errors accompanying measurements often represent precision rather than accuracy.

    (ii) Identifying and developing models of the mechanisms that introduce nutrients into the euphotic zone, both biological processes such as nitrogen fixation and the import of phosphate from depth by motile plankton, and physical processes such as mesoscale eddies.

    (iii) Developing models of primary and new production and the relationship of these to satellite observations.

    (iv) Identifying and developing models for the genesis and maintenance of HNLC and non-HNLC conditions.

    (v) Identifying and developing models of the mechanisms that control the C:N and C:P stoichiometries in different ecosystem processes and compartments .

    (vi) Identifying and developing models of the role of food web structure in determining the partitioning of productivity between particulate and dissolved organic matter and the consequences of this partitioning for export and remineralization.

    (vii) Identifying and developing models of the mechanistic controls on the occurrence and abundance of taxa with distinct biogeochemical properties and for predicting their influence on biogeochemistry (e.g., the importance for alkalinity of the ability to predict the distributions and abundances of organisms that precipitate calcium carbonate).

    (viii) Identifying and developing models of the transport of dissolved and particulate organic matter and the resulting implications for remineralization in the water column. An important task will be to determine the scales of time integration by different tracers of carbon budgets (e.g., POC flux, 234Th, DIC, 13C/12C of DIC, O2, NO3). Each of these tracers of carbon fluxes represent different time scales of integration so that the fluxes obtained by different measurements do not necessarily agree. How can we best make use of the multi-tracer approach to determining carbon balances?

    (ix) Identifying and developing models of the diagenetic mechanisms that alter organic carbon and related substances in the sediments by remineralization, dissolution, and sediment preservation.

    (x) Evaluating the impact of spatial and temporal heterogeneity in all processes at all scales. How can we improve our understanding of the impact that small-scale biological and physical processes have on the JGOFS field programs, and in particular at the time series stations?


    A. Introduction

    One of the key goals of the JGOFS program is to enable the extrapolation of knowledge gained from process studies and related efforts to regional and global spatial scales and to seasonal, interannual, and decadal time scales. This problem is important but difficult. We must be able to extrapolate existing data to large scales, predict oceanic physical and biogeochemical patterns in atmosphere-ocean partitioning of carbon under current climatological conditions, and then predict how this partitioning might be altered in response to climate variations that have occurred in the past and to those that are expected to occur over the coming decades.

    The basic tools for extrapolation and prediction are models. Models may range in complexity from simple empirical relationships, to highly parameterized models, to complex, highly resolved mechanistic models. These models require observations and data for initialization, parameter estimation, and assimilation. The required data include routine in situ measurements and remotely-sensed observations of the surface ocean. Use of data assimilation techniques to couple models to data should also prove useful.

    B. Approaches to Extrapolation and Prediction

    Achieving U.S. JGOFS SMP objectives for extrapolation and prediction will require a wide range of tools and techniques, and a variety of models and modeling approaches will be employed. Common to all these approaches will be the parameterization of unresolved processes, estimation of mean fields and their associated uncertainties, and utilization of in situ and satellite observations to constrain model structures, parameters, and boundary conditions.

    In this section we identify outstanding issues that must be addressed in the course of providing extrapolated fields and predictions; U.S. JGOFS SMP will support research activities in the following and related areas.

    1. Parameterization of unresolved processes.

    JGOFS studies have identified a large number of previously unknown or uncertain terms influencing the partitioning of carbon within the ocean and between the atmosphere and ocean. In many cases, the time and space scales for these processes are such that they cannot be included directly in the schemes for extrapolation and prediction; these processes must therefore be parameterized.

    Three issues arise in the development of parameterizations for use in regional and global extrapolation and prediction:

    creating simplified representations of critical carbon partitioning processes;

    scaling point observations and models to large space and long time scales; and

    understanding whether these parameterizations will be robust when applied to different regions and to large-amplitude changes in forcing.

    2. Evaluating the uncertainties.

    Throughout the development of parameterizations, uncertainties in the data on which a particular parameterization is based and in the ability of a given parameterization to reproduce the data must both be quantified. There are several sources of error in both extrapolated fields and predictions. The parameterization (or model) used for extrapolation is itself imperfect, and introduces some error. Estimates of parameter values, many of which cannot be measured directly, will also be uncertain. There is always the possibility that a model with incorrect mechanisms may adequately explain local observations, so that extrapolation based on that model can lead to error. Finally, the data used to initialize and constrain models are limited in their precision and accuracy and in their spatial and temporal coverage. For the extrapolations and predictions to be credible, it is important that these error sources be evaluated explicitly and their effects propagated through the models to give estimates of overall confidence.

    3. Tools and techniques.

    a) Models

    The development of quantitative models is critical to the success of the synthesis and modeling phase of U.S. JGOFS, because models are needed for extrapolating data, for predicting future trends, and for testing hypotheses related to the marine carbon cycle. Models used for extrapolation can be as simple as the regression of an important but infrequently measured variable (such as the rate of carbon export from the photic zone) on a frequently measured variable (such as remotely-sensed ocean color). More elaborate extrapolation techniques might use optimal interpolation or advection-diffusion models to fill data-void areas. It is envisioned that a wide variety of models will prove to be useful for these tasks, ranging from simple abiological box models to comprehensive ecosystem models embedded in ocean GCM's.

    b) Data

    Two classes of data sets are considered essential for meeting U.S. JGOFS SMP goals: intensive data sets and extensive data sets. Intensive data sets, such as those from U.S. JGOFS Process and Time-series Studies, are essential for developing mechanistic understanding and parameterizations for models. Extensive data sets cover broader space and/or time scales. These will be used to initialize and constrain models and to quantify transient events that cannot be resolved using climatological data. Table 5 provides examples of extensive data sets that are planned and the development of which should be encouraged by U.S. JGOFS SMP.

    Table 5. Extensive Data Sets for U.S. JGOFS SMP

         Measured Variable           JGOFS Relevance         Sensor Example     
    I. Oceanic and atmospheric   Contraint ocean carbon  Chemical               
    tracers of ocean carbon      fluxes                  instrumentation        
    fluxes CO2/N2, and d13C of                                                  
    CO2 in air; TCO2,                                                           
    Talkalinity, pCO2,                                                          
    nutrients and O2 in                                                         
    II.   Satellites (extensive                                                 
    in time and  space)                                                         
       Wind Stress               Piston Velocity, ocean  NSCAT on ADEOS         
                                 mixing and circulation                         
       SST                       Heat Flux, Circulation  NOAA Metsats           
       Sea surface topography    Circulation             TOPEX/Poseidon, ERS    
       Ocean color               Biological              OCTS, SeaWiFS,MODIS    
       Surface Irradiance        Productivity, Heat      Varied                 
    III.  Moorings extensive in                                                 
      Meteorological             Time series of heat,    TOGA TAO mooring       
    variations, currents,        carbon, nutrient        systems, BATS.         
    thermal structure, optics,   fluxes                                         
    pCO2, nutrients.             (Eulerian)                                     
    IV.  Drifters extensive in                                                  
      Meteorological             Time series of heat,    Upgraded WOCE buoys    
    variations, optics, thermal  carbon, nutrient                               
    structure, pCO2              fluxes                                         
    V.  Time series stations                                                    
    extensive in time.                                                          
    Varied, adaptable            All relevant fluxes at  Hydrostations, e.g.    
                                 local scale.            HOT and BATS           
                                                         Weather stations,      
                                                         e.g. S, P, November,   
                                                         Global tide gauge      
                                                         Coastal SST, nutrient  
                                                         Coastal surveys,       
                                                         process studies, e.g.  
                                                         CalCOFI, LOICZ??,      
    V.  "Analysis" products      Heat, momentum, mass    (ECMWF, NWS, FN),      
    extensive in time and space  fluxes at air-sea       winds, heat flux,      
                                 boundary.               currents               

    c) Data assimilation and inverse modeling

    "Inverse modeling" techniques provide powerful methods for confronting models with data, and we expect that their use will make important contributions to U.S. JGOFS SMP. In modeling efforts, it is often unclear whether a lack of fit to data is caused by a defect in model structure or by an unwise choice of parameters. Use of an automated method for finding optimal parameter values for a specified model structure eliminates poor parameter values from the list of possible culprits, forcing attention back to model structure. Inverse methods have been recently applied to such JGOFS-related applications as the problem of recovering parameter values from simulated BATS and HOT time series data (Lawson et al., 1996) and the fitting of a highly parameterized ecosystem model to BATS data (Hurtt and Armstrong, 1996).

    C. Implementation.

    1. Extrapolating from local to global scales

    Extrapolation will require the use of parameterizations of biogeochemistry on local scales to generate fluxes and distributions of carbon and other tracers at basin and global scales. The goal of this effort will be to improve both local process models and large-scale maps of flux and tracer distributions by repeatedly comparing model fields to observations.

    The objectives of this work are (1) to achieve deeper insights into biogeochemical processes that are important on basin and global scales (e. g., iron limitation; grazing; species composition; physical forcing); (2) to achieve an improved understanding of carbon fluxes and other bio-active elements on these scales; and (3) to reach a better understanding of the distribution of CO2 and other tracers in the oceans, so that the future evolution of the anthropogenic CO2 inventory of the oceans can be more accurately predicted.

    A key aspect of this exercise will be to define the range of uncertainties in properties predicted by different biogeochemical models. This analysis will determine the extent to which large-scale constraints may be used to infer the regional importance of various biogeochemical processes. If local models invoking different ecological dynamics can each successfully explain large-scale chemical and flux fields, then our mechanistic understanding of the underlying processes will have been shown to be incomplete.

    Heroic efforts in recent years have led to a tremendous expansion of data on the distributions of total CO2, pCO2, and alkalinity in the oceans. Nevertheless, the ocean remains severely under-sampled in time and space. There are a number of approaches for estimating global CO2 fields by interpolation and extrapolation of existing data. Several alternative approaches to the problem should be tried. We particularly emphasize the opportunity to derive global CO2 fields by modeling the interaction between ocean physics, biology, and chemistry, successively modifying models to fit observed fields of CO2 concentration and related properties.

    2. Estimating, simulating, and predicting current conditions

    An important test of our ability to predict the response of the oceanic carbon system to climate change is the ability to predict the present-day carbon cycle. Comparison of observations to results from analytical and numerical models should provide information on gaps in conceptualization, observation, and mathematical representation of relevant processes. Closing these gaps will require close collaboration between research groups focused on observations and those focused on modeling.

    3. Predicting the ocean's response to climate change

    The overall success of U.S. JGOFS SMP will be judged not only by the attainment of an improved understanding of the current state of the oceanic carbon system, but also by the development of a capability to predict changes in response to natural and anthropogenic climate change. As shown by the recent IPCC Climate update (Houghton et al., 1995), the potential for alterations of the Earth's climate over the next several centuries is large. An important aspect of U.S. JGOFS SMP will be to evaluate the response of the oceanic carbon cycle under various climate change scenarios, with particular emphasis on possible climate feedbacks caused by altered partitioning of carbon among compartments of the ocean-atmosphere system.

    Efforts to predict the future behavior of the ocean carbon system will clearly build upon the synthesis and modeling of the present ocean carbon system. However, the potential for nonlinear climate transitions and new physical regimes, such as a collapse of the North Atlantic overturning cell (Manabe and Stouffer, 1993), pose an additional set of problems that can only be addressed through a combination of dynamical ocean biogeochemistry models, sensitivity studies, and studies of similarly large climate changes in the paleo-record.

    The altered net air-sea fluxes of heat, fresh water, and wind stress will change the location and strength of ocean surface turbulence and, on longer time scales, wind- and buoyancy-driven circulation patterns. Specific circulation features of a higher-CO2 world may include reduced formation of North Atlantic deep water, weaker subtropical gyres, and a transition to more permanent El Niño-like conditions in the equatorial Pacific. Large-scale perturbations in atmospheric cloud and aerosol distributions may modify surface solar irradiance, while atmospheric dust deposition and corresponding input of trace metals such as Fe and Zn will likely vary with changing soil wetness, land use patterns, and atmospheric transport. Additional regional perturbations due to anthropogenic nitrogen deposition are also possible.

    Generation of future climate scenarios, although beyond the scope of JGOFS, is an active area of research in the climate community, and U.S. JGOFS SMP should strive to forecast the redistribution of carbon within the ocean-atmosphere system given projected changes in atmospheric CO2 levels, surface forcing, and ocean circulation. Furthermore, U.S. JGOFS SMP should provide ocean biogeochemistry models for use in fully coupled, global ocean-atmosphere-land carbon cycle models. The current generation of ocean transport and biogeochemical models show some skill at reproducing the current ocean carbon cycle, but the parameterizations are not always adequate for predicting responses to large-amplitude physical or chemical perturbations. The challenge is to build on the understanding developed through JGOFS and related programs to provide increased confidence in our ability to forecast future ocean boundary conditions for atmospheric carbon dioxide.


    Achieving U.S. JGOFS SMP goals will require coordinating the efforts of many teams of investigators while avoiding the imposition of an excessively rigid hierarchical structure. We propose a structure that will foster cooperation among modelers and those who make observations at two levels. First, submission of proposals by teams that include both modelers and empiricists is to be encouraged as the most straightforward method for insuring that synthesis is undertaken with a view towards its usefulness in model development. In these teams, it is the responsibility of empiricists to insure that the data upon which parameterizations and dynamical descriptions are based are sound and that mechanistic controls are accurately depicted by modelers. We envision that those writing proposals will concentrate on one or more areas of special expertise (e.g., relationships between total production and new production; water column remineralization of macro-nutrients and/or micro-nutrients; export processes) for intensive exploration of new model structures and parameterizations. Three major areas have been identified as critical for success in achieving SMP goals; they are discussed below.

    A second key point is to view the SMP process as an evolutionary cycle wherein periods in which variant model structures are created and tested alternate with periods in which the most successful model structures are identified and selected as starting points for the next round of innovation. Innovation will arise naturally during the course of research within each funded group. But this alone will not supply the necessary overall direction to the SMP; to provide this direction, there must be a regular and continuing series of workshops for Principal Investigators and other interested individuals at which successes, failures, and gaps are identified. These meetings should be held at two levels of organization. At the first level, all principal investigators should meet once per year for a week-long plenary workshop to exchange information on progress in the various component areas of U.S. JGOFS SMP and to consider the extent to which existing components could be assembled into predictive models of the ocean's role in the global carbon cycle. The latter exercise especially will aid in identifying misfits among results from different component areas and gaps in knowledge, and will focus research for the next cycle. In these meetings investigators should also agree on data sets to be used in demonstrating that new algorithms do indeed advance the central U.S. JGOFS SMP goal of reducing uncertainty in carbon cycle predictions. At the second level, principal investigators who are investigating related topics in one of the three major component areas identified below would meet every six months (perhaps for a day or two at the beginning of the plenary meeting and at one other time) to assess progress in their sub-specialties.

    These yearly attempts to evaluate the extent to which current knowledge could support a predictive capability are essential to the evolutionary approach. The need to define progress each year in terms of new synthesis results should also encourage the full participation in U.S. JGOFS SMP of those who made the observations upon which syntheses and models are based.

    The organization of U.S. JGOFS process studies may provide a useful model for the organization of the SMP. In the process studies, a set of investigators pool their expertise to allow the concurrent measurement of quantities that are thought to control biogeochemical cycling in one region of the ocean. The measurements are then shared and interpreted in coordination with modelers. By analogy, we have identified the following five major components as being essential to achieving U.S. JGOFS SMP goals. These are categorized within the three major elements of the SMP.

    (1) Global and regional mass balances

    This component will build primarily on global survey data and will use various techniques to extrapolate data to regions and times when measurements were not made. All major forms of carbon and related substances are included in this category, e.g., dissolved organic carbon as well as inorganic carbon.

    (2) Mechanistic controls of local carbon balances

    This component will build primarily on the process and time series studies. Comparisons of mass balances among sites will provide a powerful basis for extracting generalities from the data and will be strongly encouraged.

    (a) Euphotic zone production and export of carbon and related biologically active substances.

    This component will challenge empiricists and modelers to develop improved representations of biogeochemical processes that determine production and export in the surface ocean, including both ecosystem interactions and the role of physical processes. An important part of this effort is to determine how such models can be made to apply on regional and global scales.

    (b) Transport and remineralization of carbon and related biologically active substances.

    This component will challenge empiricists and modelers to develop improved mechanistic representations of biogeochemical processes that determine the transport and remineralization of organic matter within the deep ocean. As above, an important part of this effort is to determine how such models can be made to apply on regional and global scales.

    (c) Sediment diagenesis of carbon and related biologically active substances.

    This component will challenge empiricists and modelers to develop improved mechanistic representations of biogeochemical processes that control sediment diagenesis.

    (3) Extrapolation and prediction

    In this component, basin and global scale models will be used to apply knowledge gained in other components of the SMP to predict the cycle of carbon and related biologically active substances for the present ocean, the past ocean (particularly during time periods such as the ice age and Younger Dryas/Allerod events, when atmospheric CO2 changed, most likely in response to oceanic processes), and the future ocean, for which some coupled ocean-atmosphere models predict dramatic changes in ocean circulation.


    Aiken, J., G. G. Moore, and P. M. Holligan (1992). Remote sensing of ocean biology in relation to global climate change. J. Phycol., 28: 579-590.

    Anderson, L. A. and J. L. Sarmiento (1995). Global ocean phosphate and oxygen simulations. Global Biogeochemical Cycles, 9: 621-636.

    Anderson, L. A. and J. L. Sarmiento (1994). Redfield ratios of remineralization determined by nutrient data analysis. Global Biogeochemical Cycles, 8: 65-80.

    Anderson, T. R. (1993). Modelling the influence of food C:N ratio, respiration on growth and nitrogen excretion in marine zooplankton and bacteria. Journal of Plankton Research, 14: 1645-1671.

    Archer, D. and E. Maier-Reimer (1994). Effect of deep-sea sedimentary calcite preservation on atmospheric CO2 concentration. Nature, 367: 260-263.

    Archer, D, S. Emerson, and C. Reimers (1989). Dissolution of calcite in deep-sea sediments: pH and O2 microelectrode results. Geochimica et Cosmochimica Acta, 53: 2831-2845.

    Bacastow, R. and E. Maier-Reimer (1991). Dissolved organic carbon in modeling oceanic new production. Global Biogeochemical Cycles, 5: 71-86.

    Bacastow, R. B., J. A. Adams, C. D. Keeling, D. J. Moss, T. P. Whorf and C. S. Wong (1980). Atmospheric carbon dioxide, the Southern Oscillation, and the weak 1975 El Niño. Science, 210: 66-68.

    Bacon, M. P. and R. F. Anderson (1982). Distribution of thorium isotopes between dissolved and particulate forms in the deep sea. Journal of Geophysical Research, 87: 2045.

    Banse, K. (1994). On the coupling of hydrography, phytoplankton, zooplankton, and settling organic particles offshore in the Arabian Sea. Proc. Indian Acad. Sci., Earth Planet. Sci., 103:125-161.

    Barnola, J. M., M. Anklin, J. Procheron, D. Raynaud, J. Schwander and B. Stauffer, (submitted). CO2 evolution during the last millenium as recorded by Antarctic and Greenland ice, Tellus.

    Bender, M., H. Ducklow, J. Kiddon, J. Marra and J. Martin (1992). The carbon balance during the 1989 spring bloom in the North Atlantic ocean, 47 N, 20 W. Deep-Sea Res., 39: 1707-1725.

    Boyle, E. A. (1988). Vertical oceanic nutrient fractionation and glacial/interglacial CO2 cycles., Nature, 331.

    Broecker, W. A. and G. H. Denton (1990). What drives glacial cycles? Scientific American, 262: 49-56.

    Broecker, W. S. (1991). Keeping global change honest. Global Biogeochemical Cycles, 5: 191-192.

    Buesseler, K. O. (submitted). The de-coupling of production and particulate export in the surface ocean. Global Biogeochemical Cycles.

    Buesseler, K. O., J. A. Andrews, M. C. Hartman, R. Belastock and F. Chai (1995). Regional estimates of the export flux of particulate organic carbon from thorium-234 during the JGOFS EqPac program. Deep-Sea Res., 42: 777-804.

    Buesseler, K. O., A. F. Michaels, D. A. Siegel and A. H. Knap, (1994). A three-dimensional time-dependent approach to calibrating sediment trap fluxes. Global Biogeochemical Cycles, 8: 179-194.

    Carlson, C. A., H. W. Ducklow, W. O. Smith and D. A. Hansell (in review). Carbon dynamics during spring blooms in the Ross Sea polynya and the Sargasso Sea: Contrasts in dissolved and particulate organic carbon partitioning. Limnol. Oceanogr.

    Carlson, C. A., H. W. Ducklow and A. F. Michaels (1994). Annual flux of dissolved organic carbon from the euphotic zone in the northwestern Sargasso Sea. Nature, 371: 405-408.

    Carpenter, E. J. and K. Romans (1991). The diazotrophic cyanobacterium Trichodesmium: importance in C and N cycling in the tropical North Atlantic Ocean. Science, 254: 1356-1358.

    Chai, F., S. T. Lindley and R. T. Barber (submitted). Origin and maintenance of high nitrate condition in the equatorial Pacific. Deep-Sea Research.

    Chisholm, S. W. (1992). Phytoplankton size, in Primary productivity and biogeochemical cycles in the sea, vol. edited by P. G. F. and A. D. Woodhead, 213-237, Plenum Press, New York.

    Ciais, P., P. P. Tans, J. W. C. White, M. Trolier, R. J. Francey, J. A. Berry, D. R. Randall, P. J. Sellers, J. G. Collatz and D. S. Schimel (1995). Partitioning of ocean and land uptake of CO2 as inferred by d13C measurements from the NOAA Climate Monitoring and Diagnostics Laboratory Global Air Sampling. Journal of Geophysical Research, 100: 5051-5070.

    Coale, K. H., S. E. Fitzwater, R. M. Gordon, K. S. Johnson, and R. T. Barber (1996a). Control of community growth and export production by upwelled iron in the equatorial Pacific. Nature, 379: 621-624.

    Coale, K. H., K. S. Johnson, S. E. Fitzwater, R. M. Gordon, S. Tanner, F. P. Chavez, L. Ferioli, C. Sakamoto, P. Rogers, F. Millero, P. Steinberg, P. Nightingale, D. Cooper, W. P. Cochlan, M. R. Landry, J. Constantinou, G. Rollwagen, A. Trasvina, and R. Kudela (1996b). A massive phytoplankton bloom induced by an ecosystem scale iron fertilization experiment in the equatorial Pacific Ocean. Nature, 383: 495-501.

    Codispoti, L. A. and J. P. Christensen (1985). Nitrification, denitrification and nitrous oxide cycling in the Eastern Tropical South Pacific Ocean. Marine Chemistry, 16: 277-300.

    Copin-Montegut, G. and B. Avril, (1993). Vertical distribution and temporal variation of dissolved ogranic carbon in the North-Western Mediterranean Sea. Deep-Sea Research, 40: 1963-1972.

    Doney, S. C., D. M. Glover, and R. G. Najjar (1996). A new coupled, one-dimensional biological-physical model for the upper ocean: Applications to the JGOFS Bermuda Atlantic Time Series (BATS) site. Deep-Sea Research II, 43: 591-624.

    Duce, R. A. and N. W. Tindale (1991). Atmospheric transport of iron and its deposition in the ocean. Limnol. Oceanogr., 36: 1715-1726.

    Ducklow, H. W., C. A. Carlson, N. R. Bates, A. H. Knap and A. F. Michaels (1995). Dissolved organic carbon as a component of the biological pump in the North Atlantic Ocean. Philosphical Transactions of the Royal Society, 348: 161-166.

    Dugdale, R. C. and J. J. Goering (1967). Uptake of new and regenerated forms of nitrogen in primary productivity. Limnology and Oceanography, 12: 196-206.

    Emerson, S. and D. Archer (1990). Calcium carbonate preservation in the ocean. Phil. Trans. R. Soc Lond., A 331: 29-40.

    Emerson, S. and M. Bender (1981). Carbon fluxes at the sediment-water interface of the deep sea: calcium carbonate preservation. J. Mar. Res., 39: 139-162.

    Emerson, S., P. D. Quay, C. Stump, D. Wilbur and R. Schudlich (1995). Chemical tracers of productivity and respiration in the subtropical Pacific Ocean. J. Geophys. Res., 100: 15873-15887.

    Enting, I. G. and J. V. Mansbridge (1991). Latitudinal distribution of sources and sinks of CO2: results of an inversion study. Tellus, 43B: 156-170.

    Eppley, R. W. (1972). Temperature and phytoplankton growth in the sea. Fish. Bull., 70: 1063-1085.

    Eppley, R. W. and B. J. Peterson (1979). Particulate organic matter flux and planktonic new production in the deep ocean. Nature, 282: 677-680.

    Evans, G. T. and J. S. Parslow (1985). A model of annual plankton cycles. Biological Oceanography, 3: 327-347.

    Fasham, M. J. R., H. W. Ducklow and S. M. McKelvie (1990). A nitrogen-based model of plankton dynamics in the oceanic mixed layer. Journal of Marine Research, 48: 591-639.

    Fasham, M. J. R., J. L. Sarmiento, R. D. Slater, H. W. Ducklow and R. Williams (1993). A seasonal three-dimensional ecosystem model of nitrogen cycling in the North Atlantic euphotic zone: A comparison of the model results with observation from Bermuda Station "S" and OWS "India". Global Biogeochemical Cycles, 7: 379-415.

    Feely, R. A., R. Wanninkhof, C. E. Cosca, P. P. Murphy, M. F. Lamb, and M. D. Seckley (1995). CO2 distributions in the equatorial Pacific during the 1991-1992 ENSO event. Deep-Sea Research, 42: 777-804.

    Francey, R. J., P. P. Tans, C. E. Allison, I. G. Enting, J. W. C. White and M. Troller (1995). Changes in oceanic and terrestrial carbon uptake since 1982. Nature, 373: 326-330.

    Friedli, H., H. Lötscher, H. Oeschger, U. Siegenthaler and B. Stauffer (1986). Ice core record of the 13C/12C ratio of atmospheric carbon dioxide in the past two centuries. Nature, 324: 237-238.

    Frost, B. W. (1987). Grazing control of phytoplankton stock in the open subarctic Pacific Ocean: a model assessing the role of mesozooplankton, particularly the large calanoid copepods Neocalanus spp. Marine Ecology Progress Series, 39: 49-68.

    Gaffin, S. R., B. C. O'Neill and M. Oppenheimer (1995). Comment of "The lifetime of excess atmospheric carbon dioxide" by Berrien Moore III and B.H, Braswell. Global Biogeochemical Cycles, 9: 167-169.

    Gruber, N., J. L. Sarmiento and T. F. Stocker, (1996). An improved method for detecting anthropogenic CO2 in the oceans. Global Biogeochemical Cycles, 10: 809-837.

    Hammond, D. E., J. McManus, W. M. Berelson, T. E. Kilgore and R. H. Pope, (1996). Early diagenesis of organic material in equatorial Pacific sediments: stoichometry and kinetics. Deep Sea Research II, 43: 1365-1412.

    Hansell, D. A., C. A. Carlson, N. Bates and A. Poisson (submitted). A re-assessment of carbon export in the equatorial Pacific Ocean. Deep-Sea Research.

    Holfert, J., D. W. R. Wallace, K. M. Johnson and G. Siedler (1995). Transport and convergence of CO2 in the South Atlantic Ocean, U.S. WOCE Implementaion Report Number 7, Texas A&M University.

    Holligan, P. M. and W. M. Balch (1991). From the ocean to cells: Coccolithophore optics and biogeochemistry. In: S. Demers (ed) Partical Analysis in Oceanography, Springer-Verlag, Berlin pp. 301-324.

    Houghton, R. A., (1992). Effects of land-use change, surface temperature, and CO2 concentration on terrestrial stores of carbon.

    Houghton, J. T., L. G. M. Filho, J. Burce, H. Lee, B. A. Callander, E. Haites, N. Harris and K. Maskell (Ed.), (1995). Climate Change 1994, Radiative Forcing of Climate Change, and an Evaluation of the IPCC IS92 Emission Scenarios, Cambridge University Press, New York.

    Hurtt, G. C. and R. A. Armstrong (1996). A pelagic ecosystem model calibrated with BATS data. Deep-Sea Research II, 43: 653-683.

    Jackson, G. A. (1990). A model for the formation of marine algal flocs by physical coagulation processes. Deep-Sea Research, 37: 1197-1211.

    Jenkins, W. J. (1988). Nitrate flux into the euphotic zone near Bermuda. Nature, 331: 521- 533.

    Jain, A. K., H. S. Kheshgi, M. I. Hoffert and D. J. Wuebbles (1995). Distribution of radiocarbon as a test of global carbon cycle models. Global Biogeochemical Cycles, 9: 153-166.

    JGOFS Science Plan (1990). JGOFS Report No. 5, SCOR.

    Joos, F., J. L. Sarmiento and U. Siegenthaler, (1991). Estimates of the effect of Southern Ocean iron fertilization on atmospheric CO2 concentrations. Nature, 349: 772-774.

    Karl, D. M. and A. F. Michaels, guest editors (1996). Ocean Time-series: Results from the Hawaii and Bermuda Research Programs. Deep-Sea Research, Part II, 43 (2-3): 127-686.

    Karl et al., Dinitrogen fixation fuels substantial new and export production in the subtropical North Pacific Ocean, Nature, submitted.

    Karl, D. M., R. Letelier, D. Hebel, L. Tupas, J. Dore, J. Christian, and C. Winn (1995). Ecosystem changes in the North Pacific subtropical gure attributed to the 1991-1992 El Nino. Nature, 373: 230-234.

    Karl, D. M., R. Letelier, D. V. Hebel, D. F. Bird and C. D. Winn (1992). Trichodesmium blooms and new nitrogen in the North Pacific gyre, in Marine Pelagic Cyanobacteria: Trichodesmium and other diazotrophs, vol. edited by E. J. C. et al., 219-237, Kluwer Academic Publishers.

    Keeling, R. F. and S. R. Shertz (1992). Seasonal and interannual variations in atmospheric oxygen and implications for the global carbon cycle. Nature, 358: 723-727.

    Keeling, C. D. and T. P. Whorf (1991). Mauna Loa, in Trends '91: A compendium of data on global change, vol. ORNL/CDIAC-46, edited by T. A. Boden, R. J. Sepanski and F. W. Stoss, 12-14, Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, Oak Ridge, Tennessee.

    Keeling, C. D., R. B. Bacastow, A. F. Carter, S. C. Piper, T. P. Whorf, M. Heimann, W. G. Mook and H. Roeloffzen, (1989a). A three-dimensional model of atmospheric CO2 transport based on observed winds: 1 Analysis of observational data, in Aspects of Climate Variability in the Pacific and the Western Americas AGU Monograph 55, vol. edited by D. H. Peterson, 165-236, AGU, Washington D.C.

    Keeling, C. D., S. C. Piper and M. Heimann (1989b). A three-dimensional model of atmospheric CO2 transport based on observed winds: 4 Mean annual gradients and interannual variations, in Aspects of Climate Variability in the Pacific and the Western Americas, AGU Monograph 55, vol. edited by D. H. Peterson, 305-363, AGU, Washington D.C.

    Knox, F. and M. McElroy (1984). Changes in atmospheric CO2, influence of marine biota at high latitudes. Journal of Geophysical Research, 89: 4629-4637.

    Landry, M. R., R. T. Barber, R. R. Bigidare, F. Chai, K. H. Coale, H. G. Dam, M. R. Lewis, S. T. Lindley, J. J. McCarthy, M. R. Roman, D. K. Stoecker, P. G. Verity, and J. R. White (in press). Iron and grazing constraints on primary production in the central equatorial Pacific: an EqPac synthesis. Limnol. Oceanogr.

    Lawson, L., E. Hofmann and Y. Spitz (1996). Time series sampling and data assimilation in a simple marine ecosystem model. Deep-Sea Research 43: 625-651.

    Liss, P. S. and L. Merlivat (1986). Air-sea gas exchange rates: introduction and synthesis, in The Role of Air-Sea Exchange in Geochemical Cycling, vol. edited by P. Buat-Menard, 113-127, D. Reidel Publ. Co.

    Manabe, S. and R. J. Stouffer (1993). Century-scale effects of increased atmospheric CO2 on the ocean-atmosphere system. Nature, 364: 215-218.

    Marland, G., T. A. Boden, R. C. Griffin, S. F. Huang, P. Kanciruk and T. R. Nelson (1989). Estimates of CO2 emissions from fossil fuel burning and cement manufacturing, based on the U.S. Bureau of Mines cement manufacturing data, ORNL/CDIAC-25, NDP-030, Carbon Dioxide Info. Anal. Cent., Oak Ridge National Lab., Oak Ridge, Tenn.

    Martin, J. H. (1990). Glacial-interglacial CO2 change: the iron hypothesis. Paleoceanography, 5: 1-13.

    Martin, J. H., 1987, Deep-Sea Res. 34: 267-285.

    Matear, R. J. and G. Holloway (1995). Modeling the inorganic phosphorus cycle of the North Pacific using an adjoint data assimilation model to assess the role of dissolved organic phosphorus. Global Biogeochemical Cycles, 9: 101-119.

    McCarthy, J. J., C. Garside, J. L. Nevins and R. T. Barber (submitted). New production along 140 W in the equatorial Pacific during and following the 1992 El Nino event. Deep-Sea Research.

    McElroy, M. F. (1983). Marine biological controls on atmospheric CO2 climate, Nature, 302: 328.

    McGillicuddy, D. J. and A. R. Robinson, (1977). Eddy induced nutrient supply and new production in the Sargasso Sea. Deep Sea Research, in press.

    McGillicuddy, D. J., McCarthy, J. J., and A. R. Robinson, (1995). Coupled physical and biological modeling of the spring bloom in the North Atlantic (I) Model formulation and one dimensional bloom processes. Deep Sea Research, 42(8): 1313-1357.

    McGillicuddy, D. J., Robinson, A. R. and J. J. McCarthy (1995). Coupled physical and biological modeling of the spring bloom in the North Atlantic (II) Three dimensional bloom and post-bloom processes. Deep Sea Researc, 42(8): 1359-1398.

    Michaels, A. F., D. Olson, J. Sarmiento, J. Ammerman, K. Fanning, R. Jahnke, A. H. Knap, F. Lipschultz and J. Prospero (in press). Inputs, losses and transformations of nitrogen and phosphorus in the pelagic North Atlantic Ocean. Biogeochemistry.

    Michaels, A. F., N. R. Bates, K. O. Buesseler, C. A. Carlson and A. H. Knap (1994). Carbon-cycle imbalances in the Sargasso Sea. Nature, 372: 537-540.

    Mitchell, B. G. and O. Holm-Hansen (1991). Observations and modeling of the Antarctic phytoplankton crop in relation to mixing depth. Deep-Sea Res., 38: 981-1007.

    Mitchell, B. G., E. A. Brody, O. Holm-Hansen, C. McClain and J. Bishop (1991). Light limitation of phytoplankton biomass and macronutrient utilization in the Southern Ocean. Limnol. Oceanogr., 36: 1662-1677.

    Murnane, R. J., J. K. Cockran, K. O. Buesseler and M. P. Bacon (in press). Least squares estimates of thorium, particle, and nutrient cycling rate constants from the JGOFS North Atlantic Bloom Experiment.

    Murray, J.W., guest editor (1996). A U.S. JGOFS Process Study in the Equatorial Pacific, Part II. Deep-Sea Research, Part II, 43(3-6): 687-1436.

    Murray, J.W., guest editor (1995). A U.S. JGOFS Process Study in the Equatorial Pacific. Deep-Sea Research, Part II, 42(2-3): 275-904.

    Murray. J. W., R. T. Barber, M. R. Roman, M. P. Bacon and R. A. Feely (1994). Physical and biological controls on carbon cycling in the equatorial Pacific. Science, 266: 58-65.

    Najjar, R. G., J. L. Sarmiento and J. R. Toggweiler (1992). Downward transport and fate of organic matter in the Oceans: Simulations with a general circulation model. Global Biogeochemical Cycles, 6: 45-76.

    Neftel, A., E. Moor, H. Oeschager and B. Stauffer, (1985). Evidence from polar ice cores for the increase in atmospheric CO2 in the past two centuries. Nature, 315: 45-47, 1985.

    Neftel, A., H. Oeschger, J. Schwander, B. Stauffer and R. Zumbrunn (1982). Ice core measurements give atmospheric pCO2 content during the past 40,000 years. Nature, 295: 220-223.

    Nelson, D. M. and W. O. Smith, Jr. (1991). Sverdrup revisited: critical depths, maximum chlorophyll levels, and the control of Southern Ocean productivity by the irradiance-mixing regime. Limnol. Oceanogr., 36: 1650-1661.

    Quay, P. D. and H. Anderson (1996). Organic carbon export rates in the subtropical N. Pacific. EOS-Transactions AGU, 76: 85.

    Quay, P. D., B. Tilbrook and C. S. Wong (1992). Oceanic uptake of fossil fuel CO2: Carbon-13 evidence. Science, 256: 74-79.

    Sambrotto, R. N., G. Savidge, C. Robinson, P. Boyd, T. Takahashi, D. M. Karl, C. Langdon, D. Chipman, J. Marra and L. Codispoti (1993). Elevated consumption of carbon relative to nitrogen in the surface ocean. Nature, 363: 248-250.

    Sarmiento, J. L. and M. Bender (1994). Carbon biogeochemistry and climate change. Photosynthesis Research, 39: 209-234.

    Sarmiento, J. L. and C. LeQuere (1996). Oceanic CO2 uptake in a model of century-scale global warming. Science, 274: 1346-1350.

    Sarmiento, J. L. and J. C. Orr (1991). Three-dimensional simulations of the impact of Southern Ocean nutrient depletion on atmospheric CO2 and ocean chemistry. Limnol. Oceanogr., 36:1928-50.

    Sarmiento, J. L. and J. R. Toggweiler (1984). A new model for the role of the oceans in determining atmospheric pCO2. Nature, 308: 621-624.

    Sarmiento, J. L., R. Murnane and C. L. Quéré, (1995). Air-Sea CO2 transfer and the carbon budget of the North Atlantic. Philosphical Transactions of the Royal Society, 348: 211-218.

    Sarmiento, J. L., J. C. Orr and U. Siegenthaler (1992). A perturbation simulation of CO2 uptake in an ocean general circulation model. Journal of Geophysical Research, 97: 3621-3646.

    Sarmiento, J. L., J. R. Toggweiler and R. Najjar (1988). Ocean carbon cycle dynamics and atmospheric pCO2. Phil. Trans. R. Soc. Lond., 325: 3-21.

    Sarmiento, J. L., R. D. Slater, M. J. R. Fasham, H. W. Ducklow, J. R. Toggweiler and G. T. Evans (1993). A seasonal three-dimensional ecosystem model of nitrogen cycling in the North Atlantic euphotic zone. Global Biogeochemical Cycles, 7: 417-450.

    Schimel, D., I. G. Enting, M. Heimann, T. M. L. Wigley, D. Raynaud, D. Alves and U. Siegenthaler (1995). CO2 and the Carbon Cycle, in Climate Change 1994, vol. edited by J. T. Houghton, L. G. M. Filho, J. Bruce, H. Lee, B. A. Callander, E. Haites, N. Harris and K. Maskell, 35-71, Cambridge University Press, Cambridge.

    Siegenthaler, U. and T. Wenk (1989). Modeling the ocean's role for El Niño related CO2 variations, in Third International Converence on Analysis and Evaluation of Atmospheric CO2 Data Present and Past, vol. 59, edited by 189-194, WMO Environmental Pollution Monitoring and Res. Prog.

    Siegenthaler, U., H. Friedli, H. Loetscher, E. Moor, A. Neftel, H. Oeschger and B. Stauffer (1988). Stable-isotope ratios and concentration of CO2 in air from polar ice cores. Ann. Glaciol., 10: 1-6.

    Slansky, R.A. Feely and R. Wanninkhof (submitted). The stepwise linear regression approach to calculating anthropogenic CO2 invasion into the North Pacific Ocean. J. Geophys. Res.

    Sverdrup, H.U. (1953). On conditions for the vernal blooming of phytoplankton. Journal du Conseil, Conseil International pour l'Exploration de la Mer, 18: 287-295.

    Takahashi, T., R. A. Feely, R. Weiss, R. H. Wanninkhof, D. W. Chipman, S. C. Sutherland, and T. T. Takahashi, (in press). Global air-sea flux of CO2: an estimate based on measurements of sea-air pCO2 difference. Proceeding of the National Academy of Sciences, Roger Revelle Memorial Colloquium.

    Tans, P. P., I. Y. Fung and T. Takahashi (1990). Observational constraints on the global atmospheric CO2 budget. Science, 247: 1431-1438.

    Toggweiler, J. R. (1990). Modeling workshop offers first look at new simulation of Equatorial Pacific, U.S. JGOFS News, 2 (2):1.

    Toggweiler, J.R. and S. Carson (1995). What are upwelling systems contributing to the ocean's carbon and nutrient budgets? In Upwelling in the ocean: modern processes and ancient records. Summerhayes et al. (eds.)

    Wakeham, S. G., J. I. Hedges, C. Lee, M. L. Peterson and P. J. Hernes (submitted). Compositions and fluxes of lipids through the water column and surficial sediments of the equatorial Pacific ocean. Deep-Sea Research.

    Wallace, D. W. R., (1995). Monitoring global ocean carbon inventories, OOSDP Background Rep., 5, Ocean Obs. Syst. Dev. Panel, Texas A&M Univ., College Station, Texas.

    Wanninkhof, R. (1992). Relationship between wind speed and gas exchange over the ocean. Journal of Geophysical Research, 97: 7373-7383.

    Wanninkhof, R., R. A. Feely, D. K. Atwood, G. Berberian, D. Wilson, P. P. Murphy and M. F. Lamb (1995). Seasonal and lateral variations in carbon chemistry of surface water in the eastern equatorial Pacific during 1992. Deep-Sea Res., 42: 387-409.

    Winguth, A. M. E., M. Heimann, K. D. Kurz, E. Maier-Reimer, U. Mikolajewicz and J. Segschneider (1994). El Nino- Southern oscillation related fluctuations of the marine carbon cycle. Global Biogeochemical Cycles, 8: 39-64.

    Winn, C. D. F. T. Mackenzie, C. J. Carillo, and D. M. Karl (1994). Air-sea CO2 exchange in the N. Pacific subtropical gyre: implications for the global carbon budget. Glob. Biogeochem. Cycles, 8: 157-163.

    Woods, J. (1988). Mesoscale upwelling and primary production. In: Towards a theory on biological-physical interactions in the world ocean. B. Rothschild, ed.

    Zhang, J. and P. D. Quay (submitted). Organic carbon export rate based on 13C and 12C of DIC budgets in the equatorial Pacific region. Deep-Sea Research.

    Appendix A

    U.S. JGOFS Synthesis and Modelling Project Workshop

    August 13-19, 1996

    Participants List

    Chair: Jorge L. Sarmiento
    Bob Anderson
    David Archer
    Rob Armstrong
    Karl Banse
    Richard Barber
    Nicholas R. Bates
    Michael L. Bender
    Will Berelson
    Paul Bissett
    Peter G. Brewer
    Kenneth Caldeira
    Craig A. Carlson
    Fei Chai
    Scott Doney
    Hugh W. Ducklow
    Steven Emerson
    Richard A. Feely
    Dave Foley
    Mick J. Follows
    Marjy Friedrichs
    Catherine Goyet
    Dennis A. Hansell
    Raleigh R. Hood
    George A. Jackson
    David Karl
    Ralph Keeling
    David Kirchman
    Michael R. Landry
    Cindy Lee
    Ricardo Letelier
    Marlon Lewis
    Harilaos Loukos
    John Marra
    James J. McCarthy
    Julian P. McCreary
    Dennis McGillicuddy
    Anthony F. Michaels
    B. Greg Mitchell
    James W. Murray
    Raymond G. Najjar
    Warren L. Prell
    Paul Quay
    Michael Roman
    David A. Siegel
    Sharon Smith
    Walker O. Smith
    Taro Takahashi
    Luis M. Tupas
    Douglas W.R. Wallace
    Richard Wanninkhof
    Agency Participants
    Donald L. Rice
    James F. Todd

    Appendix B

    (This was the original agenda. In practice, there was significant diversion.)


    August 13-19, 1996

    Durham, New Hampshire

    The objectives of this workshop are

    (1) to present and discuss critical knowledge gained from JGOFS process and time series studies, and the JGOFS CO2 survey program, with a focus on how this and related knowledge will allow development of regional and global syntheses and models;

    (2) to determine the nature of the syntheses that must be achieved within each of the above study areas to enable results from these areas to be scaled globally;

    (3) to develop and endorse a set of specific objectives that can be attained within the time frame of the SMP; and

    (4) to write an implementation plan that will guide future SMP activities.

    In support of these purposes, the following sections contain the SMP steering committee proposals for SMP goals, meeting agenda, and meeting participants.

    I. Proposed SMP goals

    The central legacy of JGOFS SMP will be the synthesis of knowledge gained from the various JGOFS studies into a set of models that reflect our current understanding of how oceanic processes affect the global carbon cycle. To this end, the following specific SMP goals are proposed.

    A. To synthesize our knowledge of inorganic and organic carbon inventories, both natural and anthropogenic.

    B. To identify the first-order processes that control the partitioning of carbon between the ocean and atmosphere, with an ultimate view towards synthesis and prediction. JGOFS' unique contribution to this problem will be an improved understanding of the partitioning of carbon among oceanic reservoirs, and the implications of this partitioning for exchange between the ocean and atmosphere.

    C. Since biogeochemical and physical processes vary on a range of time and space scales, we must: (1) determine the mechanisms responsible for spatial and temporal variations, and (2) develop methods to extrapolate the cumulative effects of biogeochemical and physical processes to seasonal, annual, and interannual time scales and to regional and global spatial scales.

    D. To improve our ability to predict the role of oceanic processes in determining the future partitioning of carbon between the ocean and atmosphere, and to evaluate uncertainties and identify gaps in our knowledge.

    August 12-19, 1996 New Hampshire SMP workshop.

    ARRIVAL, Monday, August 12

    August 13-19

    I. SMP Goals Review: J. Sarmiento

    II. JGOFS Vision: Moderator: J. Sarmiento

    (JGOFS potential contribution to improving understanding of how the partitioning of carbon within oceanic reservoirs affects the balance of carbon between the ocean and atmosphere.)

    P. Brewer

    M. Lewis

    J. Sarmiento

    III. Process, Time Series, and Survey Team Overviews: Moderator: J. Sarmiento

    (Presentations will discuss the original goals of each study and reviewknowledge gained from each study that will enable conceptual models of the carbon distributions to be constructed and their mechanistic controls understood)

    1.- NABE J. Marra

    2.- BATS A. Michaels

    3.- EQPAC J. Murray

    4.- HOT D. Karl

    5.- Arabian Sea S. Smith

    6.- Southern Ocean W. Smith

    7.- CO2 Survey D. Wallace

    IV. Preliminary carbon budgets and and their mechanistic controls (Each process study and time-series study group will be responsible to summarize progress on the following three topics. One day will be dedicated to each topic to provide plenty of time for full discussion and for group meetings and writing sessions, as required. Each day will begin with plenary presentations by the named individuals not to exceed about 15 minutes each).

    A.- Preliminary carbon budgets for each study area and critical knowledge gaps

    Moderator: J. Murray

    (Synthesize natural and anthropogenic inventories)


    1.- NABE Ducklow

    2.- BATS Michaels et al.

    3.- EQPAC Quay

    4.- HOT Emerson

    5.- Arabian Sea McCarthy et al.

    6.- Southern Ocean Anderson

    7.- CO2 Survey Wanninkhof/Feely

    B.- Mechanistic controls on these budgets and critical knowledge gaps Moderators: R. Barber, D. McGillicuddy

    (Discuss the first-order processes that control within-ocean and ocean-atmosphere partitionings of carbon, with a focus on how this and related knowledge will allow development of regional and global syntheses and models)


    1.- NABE McGillicuddy & Doney

    2.- BATS Bates et al.

    3.- EQPAC Landry

    4.- HOT Karl

    5.- Arabian Sea Barber et al.

    6.- Southern Ocean W. Smith, Anderson & Banse

    7.- CO2 Survey Keeling/Caldeira

    V. Extrapolation Moderator: Marlon Lewis

    Several presentations given arranged at the meeting.

    (Cross-study working groups will be formed to debate methods that will allow knowledge gained on small spatial and temporal scales to be scaled up to seasonal, annual, and interannual time scales and to regional and global spatial scales)


    VI. Objectives

    (Develop and endorse a set of specific objectives that can be attained within the time frame of the SMP)


    LUNCH & End of Meeting