A US Carbon Cycle Science Plan

A U.S. CARBON CYCLE SCIENCE PLAN

A Report of thethe Carbon and Climate Working Group Jorge L. Sarmiento and Steven C. Wofsy, Co-Chairs A U.S. CARBON CYCLE SCIENCE PLAN

A Report of the Carbon and Climate Working Group Jorge L. Sarmiento and Steven C.Wofsy, Co-Chairs

Prepared at the Request of the Agencies of the U. S. Global Change Research Program

1999 The Carbon Cycle Science Plan was written by the Carbon and Climate Working Group consisting of:

Jorge L. Sarmiento,Co-Chair Princeton University Steven C.Wofsy, Co-Chair Harvard University Eileen Shea,Coordinator Adjunct Fellow, East-West Center A.Scott Denning Colorado State University William Easterling Pennsylvania State University Chris Field Carnegie Institution Inez Fung University of California,Berkeley Ralph Keeling Scripps Institution of Oceanography, University of California, San Diego James McCarthy Harvard University Stephen Pacala Princeton University W. M. Post Oak Ridge National Laboratory David Schimel National Center for Atmospheric Research Eric Sundquist United States Geological Survey Pieter Tans NOAA,Climate Monitoring and Diagnostics Laboratory Ray Weiss Scripps Institution of Oceanography, University of California,San Diego James Yoder University of Rhode Island

Preparation of the report was sponsored by the following agencies of the United States Global Change Research Program: the Department of Energy, the National Aeronautics and Space Administration,the National Oceanic and Atmospheric Administration,the National Science Foundation,and the United States Geological Survey of the Department of Interior.

The Working Group began its work in early 1998,holding meetings in March and May 1998. A Carbon Cycle Science Workshop was held in August 1998, in Westminster, Colorado,to solicit input from the scientific community, and from interested Federal agencies. The revised draft Plan was made available for input by a broad segment of the scientific community.

Cover photo credits: M.Harmon,Oregon State University (forest) University Corporation for Atmospheric Research (clouds,ocean)

Copies of the report may be obtained from:

U.S.Global Change Research Program 400 Virginia Avenue,SW Suite 750 Washington, DC 20024 USA phone:(202) 488-8630 fax:(202) 488-8681

ii Contents

Executive Summary...... I

Chapter 1: Introduction...... 1 Rationale ...... 1 Outline of the Research Program ...... 1 Structure of the Plan...... 5

Chapter 2: The Past and Present Carbon Cycle ...... 7 The Terrestrial Carbon Sink ...... 9 The Ocean Carbon Sink...... 12 Land Use...... 16

Chapter 3: Predicting the Future Carbon Cycle...... 19

Projecting Future Atmospheric CO2 Concentrations...... 19 Management Strategies...... 24

Chapter 4: An Integrated Carbon Cycle Science Program ...... 27 Goal 1: Understanding the Northern Hemisphere Terrestrial Carbon Sink...... 28 Goal 2: Understanding the Ocean Carbon Sink...... 36 Goal 3: Assessing the Role of Land Use...... 42

Goal 4: Improving Projections of Future Atmospheric CO2...... 43 Goal 5: Evaluating Management Strategies ...... 47 Observing and Discovering ...... 48 Expected Results...... 50

Chapter 5: Summary of Program Elements and Resource Requirements ...... 53 Mapping of Program Elements to CCSP Goals...... 53

Chapter 6: Program Implementation and Critical Partnerships ...... 57 Implementation Principles ...... 57 Proposed Program Management Structure...... 58 Critical Partnerships...... 60

Bibliography ...... 63

Acronyms and Abbreviations ...... 69

iii Executive Summary

Rationale global sinks has been observed to change from year to year and decade to decade,due to a variety of mecha- Carbon on earth is stored primarily in rocks and sedi- nisms,only partly understood. ments. Only a tiny fraction resides in mobile reservoirs (the atmosphere,oceans,and soil and terrestrial bios- The understanding of carbon sources and sinks has phere) and is thus available to play a role in biological, ad v anced enormo u s l y in the last decade. Th e r e is now physical and chemical processes at the earth’s surface. clear evidence that global uptake of anthrop o g enic CO2 The small fraction of carbon present in the atmosphere as oc c u r s by both land plants and by the ocean. The magn i - tude of the oceanic sink, pr evi o u s l y infer red from models carbon dioxide (CO2) is especially important: it is essential for photosynthesis,and its abundance is a major regu- and observations of chemical trac e r s such as oceanic rad i o - lator of the climate of the planet. carbon and tritium distrib u t i o n s , has rec e n t l y been confir med by direct observation of the increase in dissolved The major gases,nitrogen, oxygen,argon,which com- in o r ganic carbon. An a l ysis of new trac e r s such as chl o ro fl u - prise over 98% of the atmosphere,are transparent to far- or ocarbons provide further ref inements to our underst a n d - infrared (heat) radiation. Trace gases such as carbon diox- ing of carbon uptake by the oceans. The importance of the ide, water vapor, nitrous oxide,and methane absorb heat sink due to the terres t r ial biosphere has emerged from radiation from the surface, warming the atmosphere and an a l ysis of the global carbon budge t ,i n cluding improved radiating heat back to the surface. This process,called the estimates of the ocean carbon uptake, as well as data on greenhouse effect, is a natural phenomenon without 13 CO /12 CO isotopic ratios and from cha n g es in the abu n - which the Earth’s surface would be 30°C cooler than it is 2 2 dance of O2 rel a t i v e to N2. Isotopes can giv e inform a t i o n at present. on the terres t r ial sink, for exa m p l e , since plants prefe re n t i a l - Throughout the climate extremes of the past 400,000 ly select certain isotopes during photosynthesis and leave a years,during which there were four major glacial cycles, global signature in the isotopic ratio of carbon dioxide in the atmosphere. For est inven t o r ies and remote sensing of atmospheric CO2 concentrations varied by no more than twenty percent from a mean of about 240 parts per mil- veg etation appear to confir m a significant land sink in the No rt h e r n Hemisphere and provide insight into the underly- lion (ppm). Concentrations of CO2 and methane are now higher by more than 30% and 250%, respectively, than the ing mecha n i s m s . Ho wever , we cannot yet quantitativel y previous maxima. This very rapid increase has occurred de f ine the global effects of human activities such as agric u l - since the industrial revolution, raising concerns that the tu r e and fore s t r y or the influence of climate var iations such temperature of the atmosphere may rise as a consequence as El Niño. Studies to determine these effects have of an increase in the greenhouse effect. em e r ged as critical for understanding long-term cha n g es in at m o s p h e r ic concentration in the past, and will help to dra- The rapid increase in atmospheric concentrations of CO2 ma t i c a l l y enhance understanding of how the earth ’ s cli m a t e over the past 150 yea rs , rea c hing current concentrations of will evol v e in the future. about 370 ppm,co r responds with combustion of fossil fuels since the beginning of the industrial age. Co n ver sion of The Carbon Cycle Science Plan (CCSP) presented in for ested land to agric u l t u r al use has also red i s t r ibuted car- this document has several fundamental motivations. First, bon from plants and soils to the atmosphere. Th e r e has it is clear that the oceans and land ecosystems have been growing concern in recent yea r s that these high level s responded in measurable ways to the atmospheric of carbon dioxide not only may lead to cha n g es in the increase in carbon dioxide,but the associated mecha- ea rt h ’ s climate system but may also alter ecological balances nisms are still not well quantified. Second,the land and th r ough phys i o l o g ical effects on vege t a t i o n . ocean sinks and sources appear to fluctuate naturally a great deal over time and space,and will likely continue to Only about half of the CO2 released into the atmo- vary in ways that are still unknown. Third,to predict the sphere by human activity (“anthropogenic”CO2 from behavior of Earth’s climate system in the future, we must combustion of fossil and biomass fuels and from land use be able to understand the functioning of the carbon sys- changes) currently resides in the atmosphere. Over the tem and predict the evolution of atmospheric CO 2. last 10-20 years,more than half of the CO2 released by Finally, scientific progress over the past decade has burning fossil fuels has been absorbed on land and in the enabled a new level of integrated understanding that is oceans. These uptake and storage processes are called directly relevant to critical societal questions associated “sinks” for CO2, although the period over which the car- with the economic and environmental effects of forestry, bon will be sequestered is unclear. The efficiency of agriculture,land use and energy use practices. I A U.S. Carbon Cycle Science Plan

The development of the CCSP has been strongly influ- • Strengthening the broad research agendas of the agen- enced by the view that carbon cycle science requires an cies through better coordination, focus,conceptual and unprecedented coordination among scientists and sup- strategic framework,and articulation of goals. porting government agencies. The nature of the problem demands it. Carbon dioxide is exchanged among three major active reservoirs,the ocean,land,and atmosphere, Scientific Questions and through a variety of physical, chemical,and biological mechanisms,including both living and inanimate compo- In the very broadest terms,the present plan addresses two fundamental scientific questions: nents. Research on atmospheric CO2 therefore encompasses the full Earth system and involves many different • What has happened to the carbon dioxide that has research disciplines and approaches. already been emitted by human activities (past anthro- Consequently, a large number of government agencies pogenic CO2)? and programs are involved in supporting research on the • What will be the future atmospheric CO2 con- carbon cycle,including data-gathering, field research, centration trajectory resulting from both past and analysis,and modeling. Thus it is clear that there is future emissions? extraordinary value to be gained by coordinating research and encouraging disciplinary and organizational cross- The first of these questions deals with the past and fertilization through effective program integration. In present behavior of the carbon cycle. Information about addition,a recent report evaluating research programs in this behavior provides the most powerful clues for under- global environmental change by the National Research standing the disposition of carbon released as a result of Council highlighted the importance of developing a coor- human activities,the underlying processes in this disposi- dinated, focused,scientific strategy for conducting carbon tion,and their sensitivity to perturbations. Improved cycle research (NRC 19981). understanding will require targeted historical studies,the development of a sustained and coordinated observational The Carbon Cycle Science Plan presents a strategy for a effort of the atmosphere,land and sea,and associated program to deliver credible predictions of future atmos- analysis,synthesis,and modeling. pheric carbon dioxide levels, given realistic emission and climate scenarios, by means of approaches that can incor- The second question focuses directly on the goal of porate relevant interactions and feedbacks of the carbon predicting future concentrations of atmospheric CO2. The cycle-climate system. The program will yield better under- study of essential processes in the carbon cycle will be integrated with a rigorous and comprehensive effort to standing of past changes in CO2 and will strengthen the scientific foundation for management decisions in numer- build and test models of carbon cycle change,to evaluate ous areas of great public interest. and communicate uncertainties in alternative model simulations,and to make these simulations available for public The intent of the CCSP is to develop a strat e g ic and scrutiny and application. The research program outlined optimal mix of essential components, wh i c h include sus- in the CCSP will depend on parallel initiatives in human tained observa t i o n s ,m o d e l i n g , and innovat i v e process stud- dimensions programs to fully achieve its goals,especially i e s ,c o o rdinated to make the whole greater than the sum of that of predicting future atmospheric CO2 concentrations. its parts . The design of the CCSP calls for coordi n a t e d , rig - o ro u s ,i n t e rd i s c i p l i n a ry scientific res e a r ch that is strat e g i- Current estimates of terrestrial sequestration and ca l l y prio r itized to address societal needs. The planned oceanic uptake of CO2 vary significantly depending on activities must not only enhance understanding of the car- the data and analytical approach used (e.g.,the Forest bon cycle , but also improve capabilities to anticipate future Inventory Analysis,direct flux measurements in major conditions and to make infor med management decisions. ecosystems,inverse model analysis of CO2 data from surface stations, changes over time of global CO2,O2, 13 12 CO2/ CO2, measurement based estimates of the air-sea Basic Strategy CO2 flux and dissolved inorganic carbon inventory, and ocean and terrestrial vegetation model simulations). The There are two components to the CCSP strategy: proposed program is designed to reconcile these esti- • Developing a small number of potent new research ini- mates to a precision adequate for policy decisions by tiatives that are feasible,cost-effective,and compelling, delivering new types of data;applying new, stringent tests to improve understanding of carbon dynamics in each for models and assessments;and providing quantitative carbon reservoir and of carbon interactions among understanding of the factors that control sequestration of the reservoirs. CO2 in the ocean and on land. A sound basis for policy debates and decisions will be provided by the program as fully implemented.

1 National Research Council,Global Environmental Change:Research pathways for the next decade,National Academy Press,Washington, D.C.,1998. II Executive Summary

Long-Term Goals Goal 1: Quantify and understand the The elements of the proposed program (see the sec- Northern Hemisphere terrestrial tions that follow) have the following long term objectives: carbon sink.

Goal 2: Quantify and understand the Scientific Knowledge and Understanding uptake of anthropogenic CO2 in • Develop an observational infrastructure capable of the ocean. accurately measuring net emissions (sources and sinks) Goal 3: Determine the impacts of past and of CO2 from major regions of the world. current land use on the carbon • Document the partitioning of carbon dioxide sources budget. and sinks among oceanic and terrestrial regions. Goal 4: Provide greatly improved pro- • Un d e r stand the processes that control the temporal jections of future atmospheric tr ends and spatial distribution of past and current CO2 concentrations of CO . so u r ces and sinks, and how these might cha n g e under 2 fu t u r e conditions of climate and atmospheric che m i s t r y. Goal 5: Develop the scientific basis for societal decisions about manage • Determine quantitatively the factors (long term and ment of CO and the carbon cycle. transient) that regulate net sequestration of anthro- 2 The potential for near-term progress is demonstrated pogenic CO2. by the identification of two general hypotheses that • Develop the ability to predict and interpret changes in are testable by integration of the program elements atmospheric CO2 in response to climate change and described below. inputs of CO2, and including changes in the functional aspects of the carbon cycle. Hypotheses:

1. There is a large terrestrial sink for anthropogenic CO2 in the Northern Hemisphere. Application of Scientific Knowledge to Societal Needs 2. The oceanic inventory of anthropogenic CO2 will continue to increase in response to rising atmospheric • De velop atmosphere-ocean-land models of the carbon CO2 concentrations,but the rate of increase will be cy c le to (1) predict the lifet i m e s ,s u s t a i n ab i l i t y , and inter- modulated by changes in ocean circulation,biology, an n ual/decadal vari a bility of terres t r ial and ocean sinks, and chemistry. and (2) provide a scientific basis for evaluating potential ma n a gement strat e g ies to enhance carbon sequestrat i o n . • Develop the ability to monitor the efficacy and stability Goal 1: Understanding the Northern of purposeful carbon sequestration activities. Hemisphere Land Sink • Develop the capability for early detection of major Goal 1 is intended to establish accurate estimates of the shifts in carbon cycle function that may lead to rapid magnitude of the hypothesized Northern Hemisphere ter- release of CO2 or other unanticipated phenomena. restrial carbon sink and the underlying mechanisms that regulate it. Program Goals for Fiscal Years 2000-2005 Major Program Elements and Activities and Associated Program Elements • Expand atmospheric monitoring: vertical concentration data,column CO2 inventories,continuous measure- Achieving these goals requires new mechanisms to ments,new tracers (oxygen,carbon and oxygen integrate information on the carbon cycle and coordinate isotopes, radon,CO, and other traces gases). interagency efforts as well as additional funding in critical areas as described below.The following goals will provide • Conduct field campaigns over North America,and even- high-value deliverables upon full implementation of the tually over the adjacent oceans,using aircraft linked to proposed program. enhanced flux tower networks and improved atmospheric transport models. • Improve inverse models and strengthen connections between atmospheric model inferences and direct terrestrial and oceanic observations.

III A U.S. Carbon Cycle Science Plan

• Expand the network of long-term flux measurements • Synthesize results of recent global ocean carbon sur- for representative ensembles of undisturbed,managed, veys to elucidate the processes for carbon storage in and disturbed land along major gradients of soils,land the sea,and conduct ongoing oceanic inventory and use history, and climate,using new technology to tracer measurements at a level that assures coverage of lower costs. all ocean areas every 10 to 15 years. • Conduct manipulative experiments and process studies • Develop and implement process studies,models,and on ecosystem,local,and regional scales,coordinated synthesis,including manipulation experiments,large- with flux measurements to quantify terrestrial scale tracer releases,and direct measurement of air-sea sources/sinks and changes over time. fluxes of CO2; focus should be on defining physical and biological factors regulating air-sea exchanges and • Obtain remote-sensing data defining changes in vegeta- export to long-term storage in deep ocean waters. tion cover and phenology. • Develop remote sensing capability to monitor physical and biological properties on larger scales. Deliverables • Define the existence and magnitude of past and pres- Deliverables ent terrestrial carbon sinks. • Understand ocean processes in critical regions such as • Elucidate the contributions of climate variations,such the Southern Ocean. as rainfall,length of growing season,soil moisture,and long-term temperature changes,of fertilization (CO2, • Determine the existence,magnitude,and interannual NOx, etc.),and of past and present land use changes variability of oceanic carbon sinks and sources on and land use management to the contemporary regional scales. CO sink. 2 • Attribute observed changes in the ocean carbon sink to • Document and constrain the uncertainties related to variations in circulation,biology, and chemistry. the potential Northern Hemisphere terrestrial • Incorporate improved oceanic CO flux estimates for carbon sink. 2 better constraints on inverse models in estimating terrestrial sinks. Goal 2: Understanding the Ocean Carbon Sink Goal 3: Assessing the Role of Land Use Goal 2 is intended to establish accurate estimates of the More fully stated,Goal 3 is to obtain accurate esti-- oceanic carbon sink,including interannual variability, spa- mates of the effects of historic and current land use pat- tial distribution,sensitivity to changes in climate,and terns on atmospheric CO2, emphasizing the tropics and underlying mechanisms. In the near-term, focus should be Northern Hemisphere. given to the North Atlantic and North Pacific to complement the focus on the Northern Hemisphere in Goal 1. In the long term, focus should be on important regions Major Program Elements and Activities with major gaps in current knowledge,such as the Southern Ocean. • Document the history of agricultural expansion and abandonment and its impact on the contemporary carbon balance of North America by conducting intensive Major Program Elements and Activities historical land use and present-day land management analyses across environmental gradients in coordination • Develop new technology for measuring oceanic CO2 with the proposed North American network of eddy and related quantities on various observational plat- flux measurement sites. forms,including moorings,drifters,and new shipboard sampling techniques. • Improve observational capabilities in tropical regions through cooperation with tropical nations,including • Acquire air-sea carbon flux measurements,including international expeditions and exchanges and training of data from long-term stations and from field campaigns scientists and managers. combining aircraft,shipboard,and shore-based measurements;use improved ocean and atmosphere transport •Apply the observational programs discussed in Goal 1 models to refine estimates of the magnitude of oceanic in the tropics,including long-term observations and modeling. sources and sinks of CO2 based on pCO2 (partial pressure of carbon dioxide) data. IV Executive Summary

• Couple historical and socioeconomic analysis of land Goal 5: Evaluating Management Strategies management strategies with biogeochemical models to evaluate the integrated effects of human activities that Goal 5 is to develop a scientific basis for evaluating sequester carbon (see Goal 5). potential management strategies to enhance carbon sequestration in the environment and capture/ disposal strategies. Deliverables • Document the existence,location,and magnitude of Major program elements and activities carbon sources and sinks resulting from historical and current land use change and land management. • Synthesize the results of global terrestrial carbon surveys, flux measurements,and related information; • Provide the scientific basis for evaluating short-term strengthen the scientific basis for management strate- and long-term impacts of deliberate modification of the gies to enhance carbon sequestration. terrestrial biosphere • Determine the feasibility, environmental impacts,stabili- • Provide better assessments of carbon balance uncer- ty, and effective time scale for capture and disposal of tainties related to land use,particularly agriculture in industrial CO2 in the ocean and geological reservoirs. North America and the tropics. • Define potential strategies for maximizing carbon storage that simultaneously enhance economic and Goal 4: Improving Projections of resource values in forests,soils, agriculture,and water resources. Future Atmospheric CO2 • Identify criteria to evaluate the vulnerability and More fully stated,Goal 4 is to improve projections of stability of sequestration sites to climate change and future atmospheric concentrations of CO through a com- 2 other perturbations. bination of manipulative experiments and model development that incorporates appropriate biophysical and eco- • Document the potential sustainability of different logical mechanisms and carbon cycle-climate feedback sequestration strategies and the storage time for into global climate and carbon cycle models. sequestered CO2. • Develop the monitoring techniques and strategy to Major Program Elements and Activities verify the efficacy and sustainability of carbon sequestration programs. • Improve representation of physical and biological • Develop scientific uncertainty estimates that are processes in carbon and climate models. required for policy discussions and decisions related to • Provide a framework for rigorous analysis of observa- managed carbon sequestration. tions and for independent comparisons and evaluations • Create a consistent database of environmental,econom- of climate and carbon models using comprehensive ic,and other performance measures for the production, data and model assessment activities. exploitation,and fate of wood and soil carbon “from • Develop new generations of terrestrial biosphere and cradle to grave”(for wood,from stand establishment ocean carbon exchange models that include the roles through final disposal; for soil carbon,from formation of both natural and anthropogenic disturbances,succes- through burial). sion,and the feedback through changes in climate and atmospheric composition. • Develop coupled Earth system models incorporating Deliverables (Goals 4 and 5) terrestrial and oceanic biogeochemical processes in • Present a new generation of models, rigorously tested climate models. using time-dependent 3-dimensional data sets,suitable for predicting future changes in atmospheric CO2. • Provide projections of the future evolution of CO2 concentrations to evaluate the consequences of different • Support the development of management/mitigation emission scenarios and to provide a scientific basis for strategies that optimize carbon sequestration opportu- societal decisions. nities using terrestrial ecosystems,primarily forest and agricultural systems,as well as evaluating the efficiency (See under Goal 5, for deliverables.) of sequestration in the ocean • Use these new models to identify critical gaps in our understanding and help design future observational strategies. V VI Executive Summary

Mapping of Program Elements to Goal 2: Program Elements 2, 3, 6, 9-11 CCSP Goals The elements (9) global ocean measurements (surveys, The following paragraphs map the program elements time series, remote sensing),(10) a quantitative under- shown in the table above to the goals and objectives of standing of air-sea exchange processes,(2) airborne CO2 the Carbon Cycle Science Plan. observation program,and (3) global CO2 monitoring network together address the first part of Goal 2: to quantify the oceanic uptake of CO2. These elements provide direct Goal 1: Program Elements 1-4, 6-9, 11 long-term measurements defining sources and sinks on the scale of major ocean regions. A critically important The prog ram elements (1) expanded terres t r ial flux net- task is to successfully integrate expanded observations of work , (2) airborne CO2 ob s e r vation prog ram , (3) global time series at key locations,observations of atmosphere- CO 2 mo n i t o r ing network , and (4) land use inven t o ri e s , ocean exchange,periodic global ocean surveys, remote ta k en toget h e r , ad d r ess the fir st part of the CCSP Goal 1: sensing of the oceans,large-scale airborne measurements to quantify the North e r n Hemisphere terres t r ial sink and over the oceans,and atmospheric data from island sta- mo r e gen e ra l l y,to quantify the global terres t r ial sink for tions. The conceptual framework for interpreting these CO 2. Pro g ram elements (1), (2 ) , and (3) together provi d e observations will be developed and tested in element (6), di r ect long-term measurements defining sources and sinks strategically planned regional observation experiments. on reg ional scales. The proposed airborne observations (2) ar e capable of providing integrated measures of regi o n a l Ocean process studies and manipulations (10) tell us net exch a n g e sever al times per month; the conceptual why the carbon is going (or coming from) major ocean fra m e wor k for interpreting these observations needs to be regions and provide the basis for predicting long-term st r engthened and tested through a series of strat e gi c a l l y trends. Program elements (10) and (6) thus contribute planned reg ional observation exp e r iments (6).The exp a n d - significantly to the second part of Goal 2:to understand ed flux networ k (1) complements the other elements by the mechanisms of oceanic uptake of CO2. de t e r mining monthly and annua l l y aver aged reg ional net Modeling and synthesis (11) provide large-scale checks up t a k e or release in typical ecosystems. The flux networ k on inferred oceanic and global fluxes,especially when will be able to define the systematic differ ences betwee n exercised to provide global constraints using data from flight days and non-flight days for the airborne profi l e the global surface and airborne networks (elements (2) me a s u re m e n t s . Flux data can help account for CO2 ne t and (3). Models allow tests of our understanding through exch a n g e on days when analysis of the reg ional net simulations of past disturbances,such as major El Niño- exch a n g e is not possible from atmospheric data, e . g . ,d u r- Southern Oscillation (ENSO) events. ing frontal passages or large weather even t s . Improved carbon and land use inventories (4) provide a first-order check on the inferences from atmospheric Goal 3: Program Elements 1, 4, 5, 7, 11 measurements. By telling us where the carbon is going The program elements (5) (reconstruct historical land (or coming from),program element 4 also contributes sig- use) and (4) (expand global terrestrial carbon and land nificantly to the second part of Goal 1,to understand the use inventories) are specifically designed to address Goal underlying mechanisms that regulate the Northern 3: to determine the impact of historical and current land Hemisphere terrestrial sink,and more generally, global ter- use. The expanded flux network (1),long-term terrestrial restrial sinks and sources.The long-term terrestrial obser- observations (7),and terrestrial process studies and vations (7),process studies and manipulations (8) provide manipulations (8) will provide integral checks and con- fundamental tools to develop new understanding of ter- straints on the interpretation of results from (4) and (5). restrial sinks and sources. Modeling and synthesis (11) will be the major tools bring- Estimates of the Northern Hemisphere terrestrial car- ing together the data and concepts developed by these program elements. bon sink made from atmospheric CO2 observations are very sensitive to the magnitude of the carbon sink in the North Atlantic and North Pacific. Oceanic observations (10) in the North Atlantic and Pacific Oceans thus provide Goal 4: Program Element 11 (integrating ele- important constraints on the Northern Hemisphere terres- ments 1-10) trial carbon sink. The program element modeling and synthesis (11) rep- Modeling and synthesis (11) provide a large-scale check resents the most comprehensive and integrating tool for on inferred Northern and global fluxes when combined Goal 4,projecting future atmospheric concentrations of with global network (3) and airborne (2) data,and allow CO2. This goal is a major scientific undertaking,in which tests of our understanding through simulations of past and the models and analysis must be closely integrated with all present conditions. VII A U.S. Carbon Cycle Science Plan other elements of the CCSP.To succeed, responsible agen- integrated carbon cycle research consisting of a scientific cies will need to develop a managerial framework with a steering committee (SSC),interagency working group unified vision of the program and with greatly enhanced (IWG),and Carbon Cycle Science Program office. The mutual collaboration and strategic planning. intent is to strengthen relevant federal agency activities by improved coordination,integration,coherence,prioritization, focus,and adherence to conceptual goals. The pro- Goal 5: The Entire CCSP (all elements, posed structure will provide the basis for issuing intera- integrated and coordinated) gency research announcements to stimulate a broad range of important new research,with agreed-upon goals,using Goal 5—developing the scientific basis for evaluating a coordinated interagency merit review process of propos- management decisions relating to CO in many critical 2 als and overall agency programs. The importance of fos- ways represents the culmination of the entire Carbon tering partnerships among Federal laboratories,the extra- Cycle Science Plan. The ultimate measure of a successful mural research community, and the private sector is carbon cycle research program will be found in its ability stressed,as is the need to communicate effectively the to provide practical answers to both scientific and evolving understanding of carbon sources and sinks to the societal questions. public and policy makers.

Implementation Principles Initial Funding Priorities Past exp e r ience with large- s c a l e , mul t i d i s c i p l i n a r y global Most of the program elements outlined above serve cha n g e res e a r ch prog rams (e.g. the TOGA Prog ram) has more than one of the major long-term and five-year goals. de m o n s t r ated that a cohere n t ,i n t e grated approa c h to pro- Again,the program requires a coordinated,integrated gram implementation is essential for optimal exec u t i o n approach by the responsible agencies: the value of an and deliver y of products designed to serve societal needs. integrated program will greatly exceed the return from The principles for successful implementation of the CCSP uncoordinated program elements. pro g ram , discussed in detail in Chapter 6 of the re p o rt ,a re : The individual prog ram elements described in this plan, • A scientific vision shared by the broad communi- ho wever , ar e not at equal stages of maturity and rea d i n e s s . ty and by the participating agencies to develop Some prog ram elements rep r esent intellectually- re a d y consistency and focus. wor k that has been constrained by limited res o u r ces in the • Shared programmatic responsibility to insure past and could begin immediately with a near-term infu- coherence,coordination,and strategic pursuit of sion of funding. Ne w techn o l o g y enterpris e s , for exa m p l e , program goals. ha ve suffer ed disprop o rt i o n a t e l y in the recent past due to su c h res o u r ce constrai n t s . Since many of the prog ram ele- • Program integration to bring together interdiscipli- ments in the CCSP call for focused techn o l o g y devel o p - nary aspects of the strategy. ment prior to large-scale implementation, we rec o m m e n d • Scientific guidance and review to provide feedback that a high prio r ity be placed on funding those techn o l o g y and support to program managers,to help foster inno- de velopment effor ts that have been defer red due to prio r vation and creativity, and to create an environment in s u f ficient funding. Mo r e specific a l l y,the fol l o wing pro- where program elements and goals evolve as new gram elements should be considered as high prio r ities for knowledge is obtained. initial funding: • Links to international programs to maximize the • Both facility and technology development for the benefits from efforts in all countries. enhanced flux network,airborne sampling,and automated and streamlined ocean sampling for long • Access to data and communication of research time-series and underway measurements results to insure timely communication of knowledge to the general public and to enhance the scientific • Airborne CO2 monitoring programs,both dispersed utility of new knowledge. weekly measurements and in support of regional studies With these principles in mind,the Working Group has recommended in Chapter 6 specific steps to establish a • An expanded and enhanced surface monitoring collaborative management structure for the Carbon Cycle network for atmospheric CO2 Science Program,with strong interagency commitment to • Improved forest inventories (with carbon measure- joint implementation of the program,including common ments a key focus and an explicit goal) and develop- development of requests for proposals and coordination ment of new techniques for remote sensing of above- review and funding activities. The program is built around ground biomass a tripartite,collaborative management structure for VIII Executive Summary

• Analysis of World Ocean Circulation Experiment/Joint Global Ocean Flux Study (WOCE/JGOFS) data for CO2 uptake by the oceans • New ongoing program of air-sea carbon flux and ocean inventory measurements • Continued ocean process studies,such as air-sea exchange and enhanced manipulation experiments • Enhanced development of Earth system modeling to include interactive carbon and climate dynamics.

IX Chapter 1: Introduction

Rationale This document outlines a U.S.Carbon Cycle Science Plan (CCSP) with the view that cr edible predictions of Future concentrations of atmospheric carbon dioxide fu t u r e atmospheric carbon dioxide levels can be (CO2) must be known to characterize and predict the made given realistic emission and climate scenarios behavior of the Earth’s climate system on decadal to cen- that incorporate relevant interactions and feedbacks. tennial time scales. Predicting these future CO2 concentrations in turn requires understanding how the global The problems of understanding the carbon cycle and carbon cycle has functioned in the past and how it func- improving related predictive capabilities are complex. tions today. For these reasons,a variety of federal agen- These problems require coordinated interdisciplinary cies have funded scientific research into the oceanic, research that is scientifically rigorous and that at the same atmospheric,and terrestrial components of the global car- time advances knowledge and serves societal needs. bon cycle. This research has been an important element Achieving all this together is clearly a challenge. The pres- of the U.S.Global Change Research Program (USGCRP) ent plan aims at specifying an optimal mix of sustained since its inception.The carbon cycle consists of an inte- observations,modeling,and innovative process studies grated set of processes affecting closely coupled carbon and manipulative experiments such that the whole is reservoirs in the atmosphere and ocean and on land. greater than the sum of parts. This strategy for an inte- Successful predictions require considering all the impor- grated CCSP has two basic thrusts: tant processes that affect these reservoirs. Programs must • Developing a small number of new research initiatives therefore support a well-integrated approach,with links that are feasible,cost-effective,and compelling,to fostered among atmospheric,oceanic,terrestrial,and address difficult,linked scientific problems human sciences. • Strengthening the broad research agendas of the agen- Recently, interest in the global carbon cycle has intensi- cies through better coordination, focus,conceptual and fied. Progress in scientific understanding and technology strategic framework,and articulation of goals. has given rise to new scientific opportunities to address critical components of the atmosphere-ocean-land system. In sum,the rationale for a CCSP is severalfold. Future Policy makers have also been seeking the appropriate concentrations of atmospheric CO2 must be projected to responses to the United Nations Framework Convention characterize and predict the behavior of the Earth’s cli- on Climate Change and to the underlying societal and sci- mate system on decadal to centennial time scales. entific concerns. There is particular impetus to under- Moreover, the scientific community is in a good position stand the sources and sinks of carbon on continental and to make important progress in this area. But making such regional scales and the development of these sources and progress will require an unprecedented level of coordina- sinks over time. Also critical is elucidating how ocean car- tion among the scientists and government agencies that bon uptake and marine and terrestrial ecosystems might support this research. respond to changing climate and ocean circulation,or to the enhanced availability of CO2 and fixed nitrogen com- pounds associated with human activities such as the burn- Outline of the Research Program ing of fossil fuels. Changes in ecosystems may affect cli- The CCSP is designed to address two basic questions: mate and may have important resource and societal implications. Another topic of increasing interest is the pur- 1. What has happened to the CO2 that has already been poseful sequestration,or storage,of carbon by burial emitted through human activities (anthropogenic car- below ground or in the ocean,or by land management bon dioxide)? practices,to keep some amount of carbon from entering 2. What will be the future atmospheric carbon dioxide the atmosphere in the form of CO . 2 concentrations resulting from both past and future The ultimate measure of the success of the carbon emissions? cycle research program will be its ability to provide The first of these questions concerns the history and pragmatic answers to both scientific and societal ques- the present behavior of the carbon cycle. Information tions. Scientists and policy makers must be able to evalu- about the past and present provides us with the most ate alternative scenarios for future emissions from fossil powerful clues for understanding the behavior of anthro- fuels,effects of human land use,sequestration by carbon pogenic CO and the processes that control it,as well as sinks,and responses of carbon cycling to potential 2 its sensitivity to perturbations. Achieving a better climate change. 1 A U.S. Carbon Cycle Science Plan understanding of these phenomena will require a sus- ef fects of var ious management stra t e gi e s ,s u ch as no-till tained observational effort. agric u l t u re , long- vs short- r otation fore s t r y,and deep ocean CO in j e c t i o n ; and the effec t i v eness of res p o n s e / m i t i g a t i o n The second question directly concerns the goal of pre- 2 options such as controlling carbon emissions or enhancing dictability. The CCSP will study the essential processes carbon sinks. The CCSP addresses the interac t i o n s that influence how carbon cycling may change in the be t w een human systems and the carbon cycle in muc h the future. These studies will be integrated in a rigorous and same way that it does interactions between the climate sys- comprehensive effort to build and test models of carbon tem and the carbon cycle . While the prog ram does not cycle change,to evaluate and communicate uncertainties encompass either broad socioeconomic res e a r ch or cli - in alternative model simulations,and to make these simu- mate res e a r ch, it does consider prob lems of the interac - lations available for public scrutiny and application. tions of human systems and climate with the carbon cycle . The long-term goals and implementation objectives that flow from the two questions above are shown in Table 1.1 and reviewed in the sections that follow below on sus- Sustained Observations tained observations,manipulative experiments,and model A program of sustained observations of sufficient spa- development. tial and temporal resolution is essential for determining In addition to the physical,biogeochemical,and ecolog- interannual variability and long-term trends in terrestrial ical processes that are traditionally studied in carbon cycle and oceanic carbon sources and sinks,both globally and research,the integrated CCSP must address human influ- regionally. Such a program is also required to monitor the ences as well,with special emphasis on understanding the effectiveness and stability of any purposeful carbon consequences of land use and land cover changes;the sequestration activities,as well as to detect major shifts in

Table 1.1 US Carbon Cycle Science Plan

Scientific Question Long-Term Goals Implementation Objectives

What has happened to the carbon • Improve quantitative characterization of Develop sustained observational efforts to: dioxide already emitted by human past and present sources and sinks for • Accurately measure major net fluxes of CO2 from activities (anthropogenic CO2) CO2 terrestrial and oceanic regions of the world including that emitted through • Understand mechanisms over the full • Quantitatively determine the processes (long-term combustion of fossil fuels, range of relevant time scales transient) that control net sequestration of deforestation, and agriculture? • Improve understanding of the sinks anthropogenic CO2 and the temporal and spatial (ocean and land) for anthropogenic CO2 distribution of such sequestration • Monitor the efficacy and stability of purposeful carbon sequestration activities • Provide early detection of major shifts in carbon cycle function that may lead to rapid release of CO2

What will be the future atmospheric • Provide predictions of future sources • Provide a framework for rigorous, independent carbon dioxide concentrations resulting and sinks (ocean and land) with intercomparison and evaluation of climate and from both past and future emissions? enhanced credibility using models and carbon models using comprehensive data experiments incorporating the most developed in the Carbon Cycle Science Program and important mechanisms both formal and informal model assessment activities • Provide a scientific basis to evaluate • Develop models of the carbon cycle in the atmo- carbon sequestration strategies and sphere and ocean and on land to— measure net CO2 emissions from - Predict the lifetime, sustainability, and interannual/ major regions of the world decadal variability of terrestrial and ocean sinks - Predict how changes in human activities and the climate system will affect the global carbon cycle - Evaluate potential management strategies to enhance carbon sequestration • Evaluate measurement strategies for detecting emissions from major regions of the world

2 Chapter 1: Introduction the functioning of the carbon system that might lead to • Establishing accurate estimates of the oceanic carbon major changes in atmospheric CO2. sink,including its interannual variability and spatial distribution. In the near-term,attention should be Data on global and regional variability and trends in directed to the North Atlantic and North Pacific,to CO concentrations provide us with the most valuable 2 complement and optimize use of the terrestrial data on information on the response of the global carbon cycle to the Northern Hemisphere.In the long-term,emphasis climate changes and to processes such as forest regrowth should turn to the Southern Hemisphere oceans pole- or fertilization owing to increased CO and nitrogen 2 ward of approximately 30°S. oxides (NOx) from fossil fuels. Monitoring programs should take advantage of the recent insights from inverse • Establishing accurate estimates of the impact on the models to combine limited direct observations with evolving carbon budget of historical and current land model estimates to determine the spatial and temporal dis- use change,timber harvest,and deforestation in the tribution of carbon sources and sinks. The program of tropics and the Northern Hemisphere. observations must also include a strong focus on understanding critical processes that determine the long-term sequestration of anthropogenic CO2, as well as its interan- Manipulative Experiments nual variability. Manipulative experiments play a unique role in global The Mauna Loa and South Pole CO2 time series of C. D. change research. They allow direct study of many key Keeling provided the fir st unambiguous evidence of ecosystem processes that have strong control over the car- in c r easing atmospheric CO2. Time series data for CO2, bon cycle. These experiments can contribute to carbon combined with tra c e rs , constituents of the atmosphere cycle research in at least three ways. First,global changes that have a specific orig in or cha n g e in an understood way, in the coming decades will create a range of novel condi- 14 13 su c h as CO 2, CO 2, and O2, pr ovide strong quantitative tions,some of which will very likely be far enough outside co n s t r aints which help confir m the fac t o r s regulating the the envelope of current and past conditions that observa- global balance of carbon. Expanded atmospheric observational data alone cannot provide a sufficient basis for cred- tion netwo rk s ,i n cluding airborne measurements and a ible modeling. Experiments will be especially important growing number of flux towe rs ,w h i ch measure the in assessing responses to multiple,interacting changes, amount of CO2 going into or out of a certain area of land and in helping to assess slowly developing responses,such over a period of time, ha ve enabled scientists to begin as changes in biodiversity. Second,model development de f ining the reg ional distribution of carbon sources and and testing solely against observations of current patterns si n k s . This wor k is still rud i m e n t a r y,but alrea d y indicates cannot provide the rigor and level of credibility that great potential for assessing source and sink distrib u t i o n s comes from validation against carefully designed experi- at smaller (e.g., reg ional) scales. A major hypothesis ments. And third, explaining and illustrating the future tra- issuing from the past decade of res e a r ch is that jectory of the carbon cycle can be substantially enhanced th e r e exists a large terrestrial sink for anthro- through experimental manipulations. Even if the scientific pogenic CO2 in the Northern Hemisphere. Much of community believes the models,the illustrative value of a the near-term observational work proposed in this solid experiment to the broader public is invaluable. plan aims to test this hypothesis. For manipulative experiments to yield large payoffs, A critical task is to improve the modeling and statistical they will need to be tightly integrated with observations tools needed to infer sources and sinks and to otherwise and models. Targeted work must address key uncertain- interpret these observations. Additional oceanic and ter- ties. Interpreting results from manipulative experiments restrial observations must also be defined to complement has presented major challenges in the past,especially global monitoring data so that better temporal and spatial those relating to the experiments’small spatial scale,short resolution is achieved. Such observations should improve temporal scale,and incomplete coverage of relevant understanding of seasonal and year-to-year variability, and ecosystem processes. Interpretations will certainly contin- they should monitor regions identified as significant ue to be difficult,but these difficulties can be managed sources and/or sinks. through careful selection of study systems,precise defini- tion of key questions,and strong emphasis on integrating Three areas in particular—areas that the scientific com- the experiments with one another and with other compo- munity is positioned to address—require near-term nents of carbon cycle research. emphasis: Experiments can play an essential role in reducing • Establishing accurate estimates of the magnitude and uncertainties about the location of and the mechanisms partitioning of the current Northern Hemisphere ter- underlying current terrestrial and oceanic sinks. Critical restrial carbon sink. targets for experimental work include evaluating the

3 A U.S. Carbon Cycle Science Plan consequences of the CO2 increase from preindustrial lev- communities in modifying the carbon sink has only begun els to the present;the relative contributions of forest to be explored. Another major hypothesis that has regrowth,nitrogen deposition,and elevated CO2 to the guided the preparation of this plan is that the current carbon sink; and the interactions between ecologi- oceanic inventory of anthropogenic CO2 will contin- cal cha n g es and altered climate and atmosphere. To under- ue to increase in response to rising atmospheric stand the likel y future traj e c t o r y of the terres t r ial sink, CO2 concentrations, but the rate of increase will be ma n i p u l a t i v e exp e r iments should be used to exa m i n e modulated by changes in ocean circulation, biology, ecosystem responses to simultaneous var iation in mul t i p l e and chemistry. env i r onmental fac t o rs , and to integrate biogeo ch e m i c a l Better understanding of the terres t r ial sink is also need- responses with species ch a n ge s ,i n cluding alterations in bio- ed . Cu r rent estimates are deriv ed prim a ri l y from the mass di ve rs i t y . Pr ototype ocean manipulation exp e ri m e n t s balance of CO in the atmosphere combined with info rm a- sh o wed the importance of iron as a control on carbon 2 tion about fossil fuel input and ocean CO flu xe s . Th e r e is up t a k e by ocean biota in the “h i g h - nu t ri e n t ,l ow - ch l o r o- 2 growing consensus that a terres t r ial sink exi s t s , but the phy l l ” (HNLC) envi r onments of the Equatorial Pac i f ic (Coale ma gnitude and underlying causes are uncerta i n . Es t i m a t e s et al. 19 9 6 ) . A similar exp e r iment was rec e n t l y conducted of the magnitude of this sink var y by at least a factor of in the Southern Ocean. Su c h prototype exp e r iments show two . Potential underlying mechanisms include rec o ver y the potential for using ocean manipulation exp e r iments to fr om prior land use, lengthening of the growing season un d e r stand complex nut r ient and ecosystem interac t i o n s due to wa rm i n g ,C O and nitrog en fert i l i z a t i o n , and buria l related to the role of ocean biota in the carbon cycle . 2 of carbon in sediments and rice paddies. In ven t o r ies based Fu t u r e res e a r ch should also include manipulation exp e r i- on direct estimates of wood volume in U.S . for ests sugges t ments to elucidate the role of iron in biological nitrog en fix - a smaller sink than do atmospheric estimates. ation (Fal k owski et al. 19 9 8 , Karl et al. 19 9 7 ) . Ma n i p u l a t i v e exp e r iments should not be considered full simulations of Carbon sources and sinks,and especially those suscep- the future. Th e y are almost inevi t a bly imperfect as simul a - tible to human perturbation,whether inadvertent or inten- ti o n s , even when they provide unique and critical insights tional,are highly variable over subcontinental and sub- into underlying mecha n i s m s . basin scales. Hence, information to evaluate the effects of ongoing land use and possible options for mitigation and/or adaptation will be required to significantly improve Model Development: Understanding and models of regional CO2 exchanges. This endeavor Predicting Critical Processes requires much greater understanding of how overall pat- Quantitative understanding of critical processes, terns of change (natural or human-induced) in critical ter- processes that operate on time scales from seasons to restrial and marine ecosystems might affect fluxes of car- decades and longer, is essential to predict future atmos- bon to the atmosphere. For example,how might climate- pheric concentrations of CO .Policy decisions affecting related changes in productivity in coastal ecosystems 2 affect the role of marine ecosystems in the cycling of car- future CO2 emissions—whether to implement restraints or management,or to abstain from doing so—require the bon? What conditions and constraints are most important capability to predict terrestrial and oceanic carbon sinks. in influencing a region’s role as a source or sink for CO2, The models to serve these purposes are also necessary and how might these conditions change in the future? to analyze observations and determine optimal What is the long-term photosynthetic and species distribu- sampling strategies. tion response of terrestrial ecosystems to climatic anom- alies and long-term trends,and how does this determine Predicting these sinks will require the development of the carbon uptake? In this context,a focused carbon and coupled terrestrial biosphere–atmosphere–ocean climate climate research program should be viewed as an impor- models. Many scientific issues need to be solved to tant component of a broader scientific effort to under- improve current predictive capability. The magnitude of stand large-scale ecosystem dynamics and change (such as the present oceanic sink is reasonably well known (or the effort proposed in the recent NRC “Pathways” report constrained),but the spatial distribution and evolution of [NRC 1998]). the oceanic sink over time are poorly known. The ability to predict the future behavior of the ocean sink is compli- The development of carbon cycle models must be car- cated by the likelihood of large changes in ocean circula- ried out within a framework of rigorous comparison with tion as predicted by coupled climate models and observed observations,as well as inter-model comparisons. It is not in previous periods of rapid climate change. Particular enough that models make accurate predictions on the importance attaches to the Southern Ocean,a vast region whole;their crucial components must also be tested for which there is a paucity of data. The possible role of against the data. changing species composition of marine biotic

4 Chapter 1: Introduction

Structure of the Plan Ach i e ving the important goals of a good carbon cycle science prog ram will req u i r e a new level of scientific and pro g rammatic integrat i o n . Sc i e n t i fi c a l l y,it is necessary to ha ve a more complete picture of the ocean-land-atmosphere in t e r actions of the carbon and climate system. Also needed is more thorough understanding of the physical and biogeo - chemical processes that cha ra c t e r ize and control this sys- te m . This knowl e d g e must encompass rel a t i ve l y long-term pr ocesses that are difficult to study, su c h as natural and human-mediated disturbance, land use and their impacts on ec o l o g ical processes such as succession. Pro g ram m a t i c a l l y, it is necessary to implement a prog ram that effec t i ve l y integrates sustained monitoring activities, field and labo ra t o r y res e a r ch, modeling and analytical studies, and target e d assessment activities designed to meet the inform a t i o n needs of society. These goals are attainable . And their attainment has an urgency increa s i n g l y recognized in the policy aren a . This plan provides a blueprint for a carbon cycle science program that focuses this renewed commitment. The plan is organized into the following chapters: Chapter 2 describes the past and present carbon cycle,addressing the question:What has happened to the CO2 that has already been emitted through human activities (anthropogenic carbon dioxide)? Chapter 3 explores the critical scientific issues for predicting the future carbon cycle,addressing the question: What will be the future atmospheric carbon dioxide concentrations resulting from both past and future emissions? Chapter 4 describes the proposed Carbon Cycle Science Program. Chapter 5 estimates the resource requirements to implement the proposed program. Chapter 6 discusses the prog ram implementation prin - ciples and critical partn e r ships req u i r ed for a successful pro g ram (including issues related to mul t i d i s c i p l i n a r y res e a r ch ,i n t e ragency coord i n a t i o n ,l ab o ra t o ry – u n i ve rs i t y – pri v ate sector collab o ra t i o n s ,i n t e rnational prog ra m s ,a n d facilities and infras t ru c t u re ) .

5 Chapter 2: The Past and Present Carbon Cycle

Chapter 1 reviewed the two overarching scientific ocean,the terrestrial biosphere,or both. The ocean car- questions that the Carbon Cycle Science Plan (CCSP) is bon sink is the best known of these two remaining com- designed to address. This chapter turns more focused ponents of the carbon budget. According to the IPCC,the attention to the first of these two questions: ocean absorbed 2.0 ± 0.8 Gt C/yr. This finding suggests that the global net terrestrial biosphere sink was only 0.2 What has happened to the CO that has already 2 ± 1.0 Gt C/yr. However, the estimated loss of carbon from been emitted through human activities (anthro- the terrestrial biosphere due to deforestation is estimated pogenic carbon dioxide)? to have been 1.6 ± 1.0 Gt C/yr. Thus,there must have Carbon cycle research relating to climate change funda- been a compensatory storage of carbon in the terrestrial mentally concerns three large scientific issues. The first is biosphere of 1.8 ± 1.6 Gt C/yr. The latter figure largely the natural partitioning of carbon among the “mobile” compensates for the land use source,so that the global reservoirs—the ocean,atmosphere,and soil and terrestrial net storage in terrestrial ecosystems was close to zero. biosphere—partitioning that is influenced itself by cli- It has not been easy to arrive at these estimates. mate change. The second issue is the redistribution of fos- Regarding ocean uptake,if the CO entering the ocean sil fuel CO within those same three reservoirs,and 2 2 were mixed homogeneously, the cumulative increase in assessments of proposals to prevent emissions or the total carbon content after 10 years at the rate of 2 Gt sequester carbon through new technologies. The third C/yr would only be 1.2 mmol/kg. This equals the detec- issue concerns transfers between the terrestrial biosphere tion limit of the currently best analytical techniques. Most and the atmosphere induced by other human activities of the CO has been added to the upper few hundred such as forest clearing and regrowth,the management of 2 meters,however. The signal is therefore detectable,but it agricultural soils,and the feedback potential of anthro- must be extracted from large natural variations of up to pogenically driven changes in atmospheric chemistry and about one to two hundred mmol C/kg. For this reason, climate to alter these transfer rates. until very recently, ocean uptake has been estimated using Historical changes in the quasi–steady state of the car- models whose parameters are calibrated against the pene- bon system are clearly reflected in ice core and isotopic tration into the ocean of other tracers such as 14C and records,which also record the unprecedented changes tritium from nuclear tests,and chlorofluorocarbons. caused by anthropogenic CO emissions (Raynaud et al. 2 Estimates of terrestrial carbon loss have been based on 1993). The globally averaged atmospheric CO mole frac- 2 surveys of land use and changes in land use, above- and tion is now over 365 mmol/mol (or parts per million, below-ground average carbon densities for ecosystem ppm)—higher than it has been for hundreds of thousands classes,and models of the development of carbon storage of years. Atmospheric concentrations of CO had 2 after disturbance,under human management,and during remained between 270 and 290 ppm during the last sev- succession (e.g.,Houghton et al.1987). These assess- eral thousand years,but rose suddenly to the present level ments have varied greatly, as reflected in the range for during the second half of the 20th century. This increase tropical biomass destruction assessed by the IPCC to be is coincident with the rapid rise of fossil fuel burning. 1.6 ± 1.0 Gt C/yr.Part of the problem in this evaluation is The modern rate of atmospheric CO2 increase is accu- the great heterogeneity of carbon content on small spatial rately known from contemporary direct atmospheric scales. But also,surveys have been difficult to compare measurements as well as from air stored in ice and firn because of incompatible definitions,different counting (Battle et al.1996,Etheridge et al.1996). The Inter- methods,the treatment of secondary forest growth,and governmental Panel on Climate Change (IPCC) assessment similar reasons. has summarized the state of scientific knowledge in this Estimates of terrestrial carbon uptake have been area (Schimel et al.1995). During the decade of the obtained from mass balance considerations such as those 1980s,the rate of increase in the atmosphere was 3.3 ± described above,which give a total uptake of 1.8 ± 1.6 Gt 0.2 billions of metric tons (also called gigatons) of carbon C/yr. Other estimates come from direct observations of per year (Gt C/yr). The next best-known piece of the puz- forest carbon inventory changes (0.9 Gt C/yr in the zle is the rate of fossil fuel emissions,which was 5.5 ± 0.5 Northern Hemisphere),models of NO fertilization (0.6 to Gt C/yr during the 1980s (Marland et al.1994). Thus,it is x 0.9 Gt C/yr),and CO fertilization (0.5 to 2.0 Gt C/yr) clear that on average a large fraction of the emitted CO 2 2 (Schimel et al.1995). There are strong indications that did not remain in the atmosphere. Conservation of mass there is a large uptake in the Northern Hemisphere,but implies that the missing emissions must have entered the 7 A U.S. Carbon Cycle Science Plan

The principal anthropogenic carbon fluxes, against the background of some of the main ‘background’natural fluxes. The results are updated from the 1995 IPCC report and Schimel (1995) with updates from Sarmiento and Sundquist (1992) and Hansell and Carlson (1998).Fluxes shown are estimated averages for the decade of the 1980s. Where unbalanced fluxes are indicated (global primary production and respiration, land use and air-sea gas exchange) the imbalance reflects the anthropogenic fluxes. Note that this figure shows decadal average fluxes and we now know these fluxes to be quite variable from year to year.

8 Chapter 2: The Past and Present Carbon Cycle the location,magnitude,and mechanisms of this uptake (1) discriminates against uptake of 13C relative to 12C, are poorly understood. leaving the atmosphere enriched in 13C,and (2) pro- The following three sections summarize recent duces O2, enhancing the O2/N2 ratio. progress on the most uncertain components of the pertur- 5. A new technique is eddy covariance,which can meas- bations to the global carbon budget: (1) the terrestrial ure vertical transport in a turbulent atmosphere. Flux carbon sink,(2) the ocean sink,and (3) the land use measurements obtained by this technique in different source. Each section is concluded with a proposed major ecosystems have demonstrated the ability of some near-term (5 to 10 year) research initiative that will forests to act as significant net sinks for atmospheric address the most compelling scientific issues identified, CO2 (Wofsy et al.1993,Baldocchi et al.1996). and which is also scientifically feasible and cost-effective. However, the number of such measurements that is available is too limited to draw any strong conclusions. The Terrestrial Carbon Sink 6. The increase in amplitude of the seasonal cycle of atmospheric CO2, and especially the earlier onset of Perhaps the biggest recent change in thinking about the summer photosynthetic drawdown,is consistent the terrestrial sink has been that earlier estimates of enor- with net uptake by temperate land ecosystems (Myneni mous carbon losses from terrestrial ecosystems (Bolin et al.1997,Randerson et al.1997). 1977,Woodwell 1978) have given way to the idea that the terrestrial biosphere has been close to neutral with 7. Recent forestry surveys in various regions of the respect to carbon storage during the last decades. The Northern Hemisphere also tend to indicate net carbon observed destruction of forests appears to have been uptake,but not as large as the atmospheric data seem roughly compensated for by mechanisms of enhanced car- to imply (Dixon et al.1994). bon uptake. Eight independent lines of evidence support 8. During earlier decades (1970–1990) the measured this view: change in the oceanic 13C/12C ratio indicates relatively 1. The overall carbon budget estimated from the observed little net uptake by the terrestrial biosphere when ana- atmospheric increase and ocean uptake estimates using lyzed in the context of changes in atmospheric concen- calibrated ocean models requires modest terrestrial trations (Quay et al.1992). uptake to satisfy mass balance (Bacastow and Keeling The storage of organic carbon in terrestrial sediments 1973,Oeschger et al.1975). may be much more important than previously recognized 2. The smaller than expected north-south gradient of (Stallard 1998). Carbon storage through erosion may sequester carbon if significant amounts of eroded carbon at m o s p h e r ic CO2, combined with data on the parti a l pre s s u r e of CO in ocean surface wa t e rs ,s u g gests that are stored in sediments where they will decompose slow- 2 ly, and if regrowth of vegetation on eroded lands replaces th e r e is a large terres t r ial CO2 sink at temperate latitudes in the North e r n Hemisphere (Tans et al. 19 9 0 ) . The sink the lost carbon. This sink,combined with expansion of compensates for , or is larger than, the estimated rate of rice agriculture and postulated net storage of carbon in carbon loss due to defor estation in the trop i c s . rice paddy soils,could amount globally to 0.6-1.5 Gt C/yr (Stallard 1998). 3. The atmospheric ratio of oxygen to nitrogen has been declining due to the consumption of oxygen by fossil It is noteworthy that five of the above arguments are fuel burning. The ratio is decreasing as oxygen use is based entirely or partly on the atmospheric measurement required by this fossil fuel burning,though perhaps of fluxes,either directly or as deduced from spatial pat- slightly more slowly, indicating that there is no major terns of the concentration of gaseous species. The first argument is based on a global mass balance involving the net O2 sink other than fossil fuel burning. There may even be a small net source,which would suggest that ocean,and the first of the flux arguments makes use of air-sea fluxes. Estimates of aboveground terrestrial bio- biological fixation of CO2 may exceed rates of remineralization of organic matter (Battle et al.1996, Keeling et mass are being made routinely, but two-thirds of the global al.1996). terrestrial carbon is in soils,and it has been hard to obtain accurate inventories,let alone temporal change data for 4. The existence of a large terrestrial sink at northern lati- this pool. It is encouraging that the evidence for tudes is supported by 13C/12C ratio measurements in terrestrial sinks is internally consistent because each of atmospheric CO2 (Ciais et al.1995a) and by measure- the approaches has its own problems,to be briefly ments of the oxygen/nitrogen ratio (O2/N2) (Keeling et discussed next. 13 12 al.1996). At northern latitudes the C/ C and O2/N2 ratios are higher (relative to the Southern Hemisphere) Interpreting observations of the CO 2 mole fraction in than expected from fossil fuel burning alone. This terms of surface sources and sinks requires atmospheric suggests net uptake by photosynthesis,which transport models that portray large-scale circulation 9 A U.S. Carbon Cycle Science Plan accurately, and that also parameterize subgrid-scale mixing hundreds of meters (Wofsy et al.1993,Baldocchi et al. correctly. The recent atmospheric transport model com- 1996,Goulden et al.1998). There is also a network of parison study (TRANSCOM) exercises show that signifi- about 10 U.S.Department of Agriculture (USDA) cant improvement in both respects is sorely needed,and Agricultural Research Service grassland sites (tallgrass, that the atmospheric boundary layer is of particular mixed grass,shortgrass steppe,intermountain,and shrub- importance (Law et al.1996,Denning et al.1999). lands) equipped with Bowen Ratio/Energy Balance (BR) Because the atmospheric signature of low-latitude sources towers that are monitoring CO2 and H2O fluxes. In grass- is weak due to vigorous vertical mixing,and because there lands,BR systems work well and are less relatively expen- are no applicable observations in some important areas sive. Such measurements are used in conjunction with (such as tropical forests),the models currently work known disturbance history, and with climatic,plant physi- roughly as follows. Sources and sinks are derived from ological,ecological,and soil data,to investigate mecha- observed concentration patterns with some degree of nisms responsible for the uptake or loss of carbon for confidence for the temperate and high latitudes,and those whole forest ecosystems,including soils. for low latitudes then follow essentially from global mass The manageable spatial scale facilitates the search for conservation. Calibrations of certain aspects of the trans- flux mechanisms,but makes it necessary to test extrapola- port with other tracers with “known”sources,such as tions to regional scales with other methods. Relationships CFCs, 85Kr, or SF , are useful,but fail in that the strong 6 between fluxes and climatic variables have been found diurnal and seasonal cycles of CO sources are not includ- 2 (Goulden et al.1996,Goulden et al.1998). When turbu- ed,and also because these tracers have strong sources in lence is well developed,often during the day, the eddy- the tropics. covariance method is well demonstrated,but at night The interpretation of the isotopic data is subject to sub- under stable conditions,the measurements are less reli- stantial uncertainty. The main source of error is the con- able and heterotrophic respiratory fluxes (fluxes from tribution of purely isotopic exchange,often called the iso- non-photosynthetic organisms) may be underestimated. topic disequilibrium flux,which may occur with or with- This problem can be partially circumvented by selecting out an accompanying net exchange of total carbon (Tans measurement times when conditions for the method are et al.1993). A better determination of the rate of air-sea favorable,although this introduces the danger of aliasing, gas exchange is especially important to pin down the iso- or obtaining biased data because of non-representative topic disequilibrium flux between the ocean and atmo- sampling. An explicit quantification of the spatial error sphere. The isotopic disequilibrium between the atmo- associated with the measurement of net ecosystem pro- sphere and the terrestrial biosphere depends on the aver- duction (NEP) by the eddy flux method will allow a quan- age age of the respiratory carbon flux,which should be titatively defensible scaling-up from local to regional analy- better known as well. This age is important because older ses of NEP. This achievement will require replication of biomass was fixed during times that the atmosphere was eddy flux towers at both local and regional scales. The less depleted in 13C than today. In interpreting atmos- number and spacing of such towers is a topic for pheric 13C/12C trends, relatively small uncertainties in iso- research,but techniques are available that can be applied topic disequilibrium between the atmosphere and ocean to the task (e.g.,Harrison and Luther 1990). translate into large uncertainties in the partitioning of the A central theme of research on the terrestrial carbon total sink between ocean and land (e.g.,an uncertainty of sink over the last decade can be formulated as the follow- only 0.1 per mil,or parts per thousand,in isotopic disequi- ing hypothesis: librium gives an uncertainty of 0.5 Gt C in partitioning). Furthermore,the isotopic signature of terrestrial primary productivity, strongly influenced by the relative propor- Hypothesi s 1: Ther e i s a l ar ge ter r estr i al tions of the C3 and C4 photosynthetic pathways,needs to si nk for anthr o- pogeni c CO2 i n the be better defined.(Fung and al.1997,Lloyd and Farquahar Nor ther n Hemi spher e. 1994). The isotopic fractionation during C4 photosynthesis is not much different from that of air-sea exchange. This hypothesis is known by the term the Northern Hemisphere Terrestrial Carbon Sink or “Northern The atmospheric oxygen budget may be subject to Land Sink.” uncertainties concerning decadal variations in the ventilation of deeper, oxygen-poor waters of the ocean. It is The suggestion that there is such a carbon sink is reassuring that two completely different analytical tech- strongly supported by a wide range of observations,data niques give similar results (Bender et al.1994, Keeling and analysis,and modeling studies,many of which have been Shertz 1992). discussed above. The magnitude of this sink was estimated to be on the order of 1.5 + 1.0 Gt C/yr for 1980-1989 The new micrometeorological (eddy-covariance) flux (Schimel et al.1996),and as having increased to >2.5 Gt measurement methods have been developed to measure C/yr over the last decade,making it a major contributor to ecosystem exchange of trace gases on a spatial scale of

10 Chapter 2: The Past and Present Carbon Cycle

(A) Daily net ecosystem exchange for Harvard Forest, Central New England (top) and the BOREAS site, Manitoba, Canada (bottom). These data are illustrative of the rich information to be gained from flux towers.Both forest sites show the pattern of carbon exchange throughout the year, with carbon uptake occurring during the growing season (negative values) and release during the winter (positive values).Differences between sites can also be noted:the growing season starts earlier at the BOREAS site, yet the forest there is less efficient overall at storing carbon.Data from towers can also be analyzed in conjunction with temperature and precipitation patterns, for example, to determine the long-term responses of major ecosystems to climate variability. (B) Overall, the BOREAS site was a net source of carbon to the atmosphere during 1994 to 1998, despite a longer growing season than average. This result is partially due to the release of carbon from cold soils starting earlier in the year, around August, and persisting throughout the winter months.

(A)

(B)

11 A U.S. Carbon Cycle Science Plan perturbations of the carbon budget. However, in addition tools in understanding the future carbon cycle. to the large uncertainty in the magnitude of this sink,its The development of measurement techniques and location and interannual variability, and the mechanisms methods of data analysis have placed us in a position to that cause it are all highly speculative. propose a major new near-term CCSP initiative with the One important issue to address in this context is the following goal: role played by CO2 and nitrogen fertilization in the current carbon balance. Goal 1 : Establ i sh accur ate esti mates of These two types of fer tilization may stimulate biogen i c the magni tude of the potenti al up t a k e of carbon, thus providing a carbon sink. Th e y are Nor ther n Hemi spher e ter r estr i al car - gen e ra l l y believed to be most important at mid-north e r n bon si nk and the under l yi ng mechani sms latitudes (Schimel et al. 19 9 6 ) , but these mechanisms mus t that r egulate it. be further quantified and their possible future rol e s as s e s s e d . Ecosystem manipulations, in c luding studies with Large carbon sinks have been hypothesized to be asso- el e vated CO2 and nitrog en additions, and consideration of ciated with a variety of terrestrial phenomena (for exam- a ran g e of potentially interacting fac t o rs , will provi d e ple, greater erosion and sediment deposition,land use insights into phy s i o l o gi c a l ,e c o l o gi c a l , and biogeo ch e m i c a l changes,CO2 fertilization,and length of the growing sea- pr ocesses underlying the carbon storage in the terres t ri a l son). Clearly, additional observations and improved mod- bi o s p h e re . These same points apply to controlled applica- els of terrestrial processes are needed. tions of ozone. Pr ojects investigating these processes are Such terrestrial studies cannot be interpreted in isola- co o r dinated by the Global Change and Ter res t r ial Eco- tion. The Northern Land Sink hypothesis directly suggests systems (GCTE) project of the International Geosphere- additional linked hypotheses and observations. As Bi o s p h e r e (IGBP) Prog ram . The results from these exp e r i- research on land clarifies the distribution of particular ments will need to be ext r apolated from the exp e ri m e n t a l hypothesized sources and sinks,parallel improvements scale to larger regi o n s , req u i r ing understanding of the will be required in the temporal and spatial resolution of scale-dependence of the processes and their interac t i o n s . atmospheric and oceanic observations and models. For Extending the results of these experiments from the example,a direct corollary of the Northern Land Sink local to regional and global scales will require four com- hypothesis is the hypothesis that there is a large oceanic ponents. First,the experiments must be designed to pro- carbon sink in the Southern Hemisphere. This sink is vide access to the underlying mechanisms that control the hypothesized because the global CO2 budget requires a scaling. Second,the experiments must be repeated in large ocean sink somewhere,and Northern Hemisphere enough ecosystems to capture the full range of mecha- ocean CO2 uptake appears to be relatively small based on 13 nisms. Third,the regions where the responses are estimat- the atmospheric C and ocean surface pCO2 (partial ed must be sufficiently characterized to support the pressure of CO2) measurements (Tans et al.1990). These extrapolation. And fourth,the models used for the scaling are the same measurements and arguments that led to the must be robust. The terrestrial biosphere is too heteroge- hypothesis of the Northern Land Sink. neous to be characterized by simple extrapolation of a reasonable number of experiments,but the processes that regulate it are uniform enough to be effectively handled in The Ocean Carbon Sink mechanistic models. The total capacity of the ocean to dissolve anthro- Another fundamental issue in this area is how carbon pogenic CO2 is mainly a chemical property and can be storage is affected by changes in land use,disturbance,and calculated by considering the appropriate chemical equi- vegetation structure and composition. libria. The primary challenge to current understanding of the ocean carbon sink is in estimating the rate at which Carbon storage in vegetation is being continually anthropogenic CO dissolves in the ocean. It is necessary altered by land use and disturbance (e.g., fires) and with 2 to know the present rate of dissolution and how climate the recovery of vegetation from historic land use. In addi- change and changes in the biological pump will affect the tion,the competitive relationships among species and uptake rate and ultimate capacity in the future. A sepa- their interdependence are changing owing to multiple rate,but also important,problem is to determine the spa- causes, climate change among them. A change from grass- tial distribution of air-sea CO fluxes,a factor that is a land to forest, for example, would certainly change carbon 2 major constraint on inverse modeling analyses such as storage. Changes in vegetation assemblages are likely to those discussed in the previous section. Each of these be important drivers for carbon storage or release on issues is discussed in turn,beginning with present knowl- decadal and longer time scales. The development of edge of the rate of ocean carbon uptake and the spatial dynamic global vegetation models (DGVMs) is beginning distribution of carbon sources and sinks. Chapter 3 addi- to accelerate,and they will likely become important 12 Chapter 2: The Past and Present Carbon Cycle tionally discusses how climate change and related changes mentary methods have been proposed in order to con- in the biological pump might affect the rate of uptake in strain (or set the estimated outside boundaries of) the the future. oceanic uptake of anthropogenic carbon directly from ocean data: (1) repeat or time-series observations of Ocean carbon uptake has been estimated by consider- water column DIC to measure the current rate of change ing the change in carbon inventory through time,the mag- of the DIC inventory; (2) “preformed CO ” methods to nitude of the air-sea CO flux,and the magnitude of car- 2 2 calculate the integral change of the DIC inventory from bon transport within the ocean. The estimates used by the pre-industrial period to the present;and (3) air-sea the IPCC in constructing global carbon budgets are based CO flux methods to measure the present global net flux primarily on inventory calculations using ocean circula- 2 into the ocean. tion models calibrated or validated with tracer observations (e.g.,bomb radiocarbon;see Schimel et al.1996). One of the most important needs is to gain a better The magnitude of these estimates has remained consistent understanding of the spatial patterns of surface ocean over almost three decades of continuous testing with new CO2 concentrations and their variability. Seasonal and models and tracer data. Nevertheless,the estimates are interannual variability of CO2 concentrations in the sur- typically reported as having an uncertainty of ± 40 per- face ocean are one to two orders of magnitude greater cent. The large uncertainty reflects the fact that tracer than their annual increase due to uptake of anthropogenic measurements are imperfect analogs for fossil fuel CO2 carbon (e.g.,Bates et al.1996,Winn et al.1994). Because and have been relatively sparse until quite recently. the signal to be detected is much smaller than this variability, it takes a decade or longer to begin to see anthro- Further, ocean models have limitations in their repre- pogenic trends. In addition,the seasonal and interannual sentations of ocean circulation and mixing. An important variability in CO concentration gives information on how goal of research over the last decade has thus been to 2 the carbon cycle functions,and can be used in conjunc- expand the tracer data set and improve ocean carbon tion with other methods to help understand regional and models. The skill of ocean carbon cycle simulations is global patterns of carbon uptake. A very promising devel- directly linked with improving models of the physical opment of the last decade is new instrumentation that transport of the underlying ocean general circulation. will make it possible to measure pCO (DeGrandpre et al. Considerable success has been achieved in the last decade 2 1995, Friederich et al.1995,Goyet et al.1992,Merlivat and through better parameterization of mesoscale eddy mix- Brault 1995) and other properties from moorings. The ing processes (Gent et al.1995). CCSP envisions that moorings with these capabilities will Additionally, there has been a major effort to develop allow establishment of many more time-series stations at methods and obtain measurements to assess the oceanic feasible cost in otherwise remote locations. uptake directly or indirectly from dissolved inorganic car- An alternative to long-term measurements is to use bon (DIC) observations. methods of analyzing the data that permit identifying the One of the most important advances has been the anthropogenic carbon component within the huge back- acquisition of a new global data set of ocean tracer and ground signal and variability that arises from seasonal and carbon system observations of unprecedented accuracy interannual change. The “preformed CO2” method uses through the World Ocean Circulation Experiment/Joint the correlation of data on carbon, nutrients, oxygen,and Global Ocean Flux Study (WOCE/JGOFS). The tracer physical variables to separate natural variability from measurements,together with the rapid improvement of changes due to uptake of anthropogenic CO2 (see Gruber ocean circulation models based on these and other WOCE et al.1996). In a second method, multiple linear regres- observations,are expected to lead to improved model sion coefficients are calculated for carbon versus other uptake estimates. DIC data sets accurate to 1 to 2 variables (Wallace et al.1995). These coefficients are used mmol/kg,which is equivalent to 1 to 2 years’uptake of to correct for the natural variability between measure- anthropogenic CO2 in near-surface waters,are now avail- ments made at two different times,thereby isolating the able for hydrographic transects representing most of the change due to addition of anthropogenic carbon. The two world’s ocean. The 13C/12C of DIC was measured on methods are somewhat independent and address different every sample collected for Accelerator Mass Spectrometry time scales. The preformed CO2 method gives an estimate 14 (AMS) C measurement during WOCE,which yields the of the oceanic uptake of anthropogenic CO2 since pre- first oceanwide high-quality 13C/12C data set (±0.03 parts industrial times;the multiple linear regression method has per thousand). These new data sets are already significant- been applied to estimate the increase in oceanic CO2 con- ly improving knowledge of the distribution of inorganic centration between the Geochemical Ocean Sections pro- carbon in the ocean. gram (GEOSECS) expeditions and WOCE/JGOFS,spaced about two decades apart (Wallace et al.1995). Syntheses A parallel occurrence has been the development of of Atlantic Ocean (Gruber 1998) and Indian Ocean methods to estimate the oceanic uptake of carbon by (Sabine et al.1999) DIC data from WOCE/JGOFS,analyzed direct analysis of the DIC observations. Three comple- 13 A U.S. Carbon Cycle Science Plan by the preformed CO2 method,show large regional dis- Atlantic is perhaps the best-constrained region. Models, crepancies between observed and modeled penetration of observation-based estimates of air-sea flux,and estimates anthropogenic carbon. It is already clear that the of the ocean inventory of anthropogenic carbon all observed penetration of anthropogenic CO2 can provide, converge on similar answers (Gruber 1998,Gruber et al. through detailed comparison with models,the basis and 1996,Sarmiento et al.1995,Takahashi et al.1997). impetus for improving models of oceanic CO2 uptake. However, difficulties remain in reconciling these estimates with independent estimates of the net meridional trans- The flux of carbon between surface waters and the ports of carbon (Brewer et al.1989;Broecker and Peng atmosphere can also be constrained using data of the par- 1992;Holfort et al.1998, Keeling and Peng 1995). The tial pressure difference between the air and the water, North and Equatorial Pacific Ocean have not been as thor- DpCO , combined with estimates of the gas exchange 2 oughly explored,though the measurements that are avail- coefficient (Takahashi et al.1997). Of course,this flux able show reasonable consistency. The rest of the world is includes both a natural and an anthropogenic component. mostly very poorly constrained. The value of the technique lies primarily in determining the spatial distribution and temporal variability of air-sea Analysis of CO2 exchange in the Southern Ocean CO2 fluxes. It is particularly useful where the signals are Ocean (waters around Antarctica poleward of approxi- large,as in the North Atlantic and Equatorial Pacific,or in mately 50°S latitude) is especially difficult because of the the contrast between El Niño and non–El Niño years. high spatial and temporal variability of its properties and the difficulty of working in that region. The Southern One curren t l y weak link in the air-sea CO flu x 2 Ocean appears to be especially important to levels of ap p ro a c h is poor understanding of the kinetics of the atmospheric CO . Deep ocean waters are mixed to shal- pr ocess of air-sea gas exch a n g e (see Wanninkhof 1992). 2 low depths there,providing one of the major pathways for The degree of nonlinearity in the relationship between the the uptake of anthropogenic CO into the deep ocean. ga s - ex ch a n g e and wind speed is poorly unders t o o d ,p a rt i c- 2 Wind speeds are high,enhancing rates of air-sea CO ul a r l y at high wind speeds where bubble effects may 2 exchange. The dissolved macronutrients available at the become importa n t . For instance, if the exch a n g e vel o c i t y surface for plant growth are significantly underutilized wer e to depend more stron g l y than is curren t l y believed (e.g., Falkowski et al.1998),suggesting a capacity for on vari a bles correlated with wind speed, our estimates of enhanced uptake of CO by photosynthesis. Macro- flu x es in reg ions with high wind speeds, su c h as the high 2 nutrients may be underutilized for several reasons;there is la t i t u d e s , would increa s e . It is time to determine throu g h less sunlight at high latitudes,biological processes are me a s u r ements a muc h better param e t e r ization of the slower because of low temperatures,there is less time for exch a n g e vel o c i t y , so that observations of the atmosphere biota to develop because surface water is mixed back into can be trul y inform a t i v e for oceanographic issues, and vice deeper layers more quickly, and necessary micronutrients ve rs a . Su r f ace sources and sinks driv e large-scale grad i e n t s such as iron may be less available (Baar et al.1995, of atmospheric CO co n c e n t r ations that could be, for 2 Falkowski 1995). All these factors may be affected by exa m p l e , an important constraint on CO up t a k e by the 2 climate change. So u t h e r n Hemisphere ocean polewar d of about 30°S. As at m o s p h e r ic tran s p o r t models continue to improve ,t h i s Models of the oceanic uptake of anthropogenic CO2 co n s t r aint will become more compelling as a major input (e.g.,Sarmiento et al.1992) predict that a substantial for inver se models of atmospheric CO2 ob s e r vat i o n s . amount is being absorbed south of 30°S. Yet there are large discrepancies between the observations and the Another analysis approach that makes use of oceanic models of oceanic CO uptake in the high southern lati- DIC data is estimation of horizontal carbon transport with- 2 tudes. Recent survey data for the southern Atlantic in and between ocean basins (e.g.,Brewer et al.1989, (Gruber 1998) and Indian (Sabine 1999) oceans show sur- Holfort et al.1998). This approach permits separate prisingly little uptake in these areas. The (sparse) atmos- assessments of the transport of natural and anthropogenic pheric data also point toward a small Southern Ocean carbon components in the ocean. The air-sea flux can be sink. Recently, observations of a composite atmospheric inferred from the divergence of the horizontal transport, tracer, effectively equal to the sum of atmospheric O and which would provide a means of coupling the ocean 2 CO concentrations, were compared with predictions inventory method with the surface DpCO method. While 2 2 from three different ocean models (Stephens et al.1998). horizontal carbon transport estimation is only starting to The largest discrepancy between models and observations be implemented in the ocean,the tremendous improve- was found over the Southern Oceans,suggesting that the ment in ocean DIC measurements and the large new data models may misrepresent exchanges of O and CO in set gathered by WOCE/JGOFS show great promise for the 2 2 this region. near future. There are few constraints on the temporal variability of Knowledge of the spatial distribution of the air-sea CO 2 the air-sea flux of CO . The Equatorial Pacific is being flux varies greatly from one region to another. The North 2 monitored reasonably well (e.g., Feely et al.1995, Feely 14 Chapter 2: The Past and Present Carbon Cycle

2 Distribution of the mean annual sea-air pCO2 flux (partial pressure of carbon dioxide, moles CO2/m /yr) over the global oceans estimated for a reference year 1995. These data show the global pattern of surface CO2 uptake and release by the ocean. Note that the major sink regions are the North Atlantic and Southern Oceans, and that the Equatorial Pacific is a large source region in a typical year.This map does not reflect the variability due to El Niño/Southern Oscillation (ENSO) cycles, for example, which can alter the size of the ocean sink on an interannual basis. This map has been constructed based on about 2 million measurements of sea-air pCO2 difference made over the past 25 years. These values have been corrected to a reference year of 1995 for the increase in pCO2 of the atmosphere and surface ocean water that has occurred since the measurements were made, and the measurements made during El Niño years in the equatorial Pacific have been excluded. Thus, the map represents a climatological mean for non-El Niño conditions. The net CO2 flux across the sea surface has been computed using the effect of wind speed on the CO2 gas transfer coefficient formulation by Wanninkhof (Equation 1, 1992) and the mean monthly wind speed of Esbensen and Kushnir (1981). The numerical method used for the construction of these maps has been described in Takahashi et al.(1997) and Takahashi et al.(in press). The map yields an annual CO2 flux for the oceans of 2.2 PgC/yr, in which the North Atlantic (N of 14°N) and the Southern Ocean (S of 50°S) are major CO2 sink areas taking up 0.8 and 0.6 PgC/yr respectively. et al.1994, Feely et al.1996),and the ongoing feasibility carbon uptake rate has increased or decreased over the studies to instrument the ATLAS moorings of the Tropical- past few decades. Moreover, uncertainties of this magni- Atmosphere-Ocean (TAO) array with CO2 sensors are an tude limit the ability to constrain the historical global CO2 important development (Friederich et al.1995). There are budget using the record of atmospheric CO2 concentra- also two time-series measurements near Bermuda and tions over the last 200 years. Hawaii (e.g.,Bates et al.1996,Winn et al.1994). However, However, the problem of understanding the ocean most of the ocean is unknown with regard to the tempo- should not be viewed as simply a comparison of current ral variability of CO fluxes. Analyses of stable carbon iso- 2 and future DIC “snapshots.” It is also important to identify topes in atmospheric CO suggest that sinks in both the 2 and understand the mechanisms that might cause future ocean and the terrestrial biosphere vary by large amounts changes in the ocean carbon sink. from year to year (Keeling et al.1989, Keeling et al.1995a, Ciais et al.1995b, Francey et al.1995). No oceanic mecha- One of the most important issues to address is which nism has been developed to explain such conspicuous mechanisms of ocean circulation that significantly shape variability. A related need is for the calculation of tempo- anthropogenic CO2 uptake are likely to be affected by a ral changes in oceanic carbon uptake on the global scale. changing climate? The global oceanic uptake rate of CO is currently not 2 Changes in ocean circulation resulting from climate known to better than about ± 40 percent. This knowl- change will immediately affect the way anthropogenic edge is not sufficient to determine whether the ocean 15 A U.S. Carbon Cycle Science Plan

CO2 is exchanged between the atmosphere and the ocean tra n s e c t s . ocean. The behavior of the present and past ocean, To summari z e , our current understanding is that, a s including the response to temporal variability, provides a t m o s p h e ric CO l evels increase through time, the ocean the strongest clues to the links between ocean circulation 2 responds by dissolving more CO2 in the surface mixe d and atmospheric CO2. l aye r, and by mixing the CO2- e n ri ched surface wa t e rs An additional question is how is the biological pump d ow n wa rd through exch a n g e with deeper wa t e rs . T h e affected by the thermohaline circulation and changing possibility also exists that ch a n ges in the biological pump climate? And is there any evidence that the C/N and C/P m ay affect the future air-sea balance of CO2. If our pre s- ratios of marine production are changing,or that “pre- ent understanding of these processes is corre c t ,t h e formed” nutrients (the nutrient concentration of water ocean should have the capacity to absorb large quantities that has cooled and sunk to depth) are changing? of anthro p o genic CO2 over time scales of decades to mil- l e n n i a . Most model simulations of future atmospheri c C O ex ch a n ge between the ocean and atmosphere nat- 2 C O l evels assume that pre s e n t - d ay fa c t o rs contro l l i n g u ra l ly occurs at the ocean surfa c e . Photosynthesis by 2 plant phy s i o l o gy, air-sea ex ch a n ge , and ocean mixing will o rganisms in the upper, sunlit layer of the ocean ke e p s remain constant into the future (Schimel et al. 1 9 9 6 ) . the CO c o n c e n t ration of the surface wa t e rs substantially 2 H oweve r, as prev i o u s ly noted, these assumptions have l ower than that of deep wa t e rs . It does so by pro d u c i n g been questioned, e s p e c i a l ly because of the potential fo r o rganic carbon that is ex p o rted to the deep ocean where s i g n i ficant responses to climate cha n ge . Air-sea gas it is conve rted to DIC. The excess deep ocean DIC that exch a n ge , ocean circ u l a t i o n , and marine photosynthesis thus results is analogous to the organic carbon stored in a re susceptible to ch a n ges in air tempera t u re s ,w i n d s o i l s , in that it is isolated from the atmosphere . Wi t h o u t ve l o c i t i e s , sea surface ro u g h n e s s , and wind and pre c i p i t a- photosynthesis in the ocean, and assuming no other sur- tion pattern s . The most powerful tool for unders t a n d i n g face ch a n ges due to organisms (such as calcifi c a t i o n ) ,t h e the mechanisms underlying potential ch a n ges is in study- excess deep ocean DIC would escape to the atmosphere ing long-term trends and short e r - t e rm fl u c t u a t i o n s . G i ve n and atmospheric CO c o n c e n t ration would be betwe e n 2 all these considera t i o n s , a major CCSP initiative in 900 and 1,000 ppm (the current va l u e , ag a i n , is 365 ocean carbon cycle re s e a rch is proposed to ach i eve p p m ) . I f, on the other hand, photosynthesis continu e d the fo l l ow i n g : eve ry w h e re until all of the plant nu t rients we re fully depleted in all surface wa t e rs ,a t m o s p h e ric CO2 would be b e t ween 110 and 140 ppm. (The actual pre - i n d u s t ri a l Goal 2 : Establ i sh accur ate esti mates of a t m o s p h e ric CO2 c o n c e n t ration was 280 ppm.) T h e s e the oceani c car bon si nk and the fi g u res indicate the great power of the oceanic under l yi ng mechani sms that r egu- b i o l o gical pump. l ate i t. L o n g - t e rm surface ocean time series that incl u d e detailed bioge o chemical and CO2 m e a s u rements are i m p roving the mechanistic understanding of pro c e s s e s Land Use that affect ocean-atmosphere carbon uptake and part i- One of the main driving factors in determining the t i o n i n g . Ti m e - s e ries data sets from Bermuda and Hawa i i Northern Land Sink may be land use,both past and pres- a re being used to develop model para m e t e rizations of ent. For example,there has been widespread reforestation ocean bioge o chemical processes affecting the ocean car- since 1900 in the eastern United States following the bon cycle (Doney et al. 1 9 9 6 , Fasham 1995). S u ch time movement of the center of agricultural production toward s e ries are also incre a s i n g ly being used as test beds fo r the Midwest. Also,less agricultural land is needed today n ovel high-resolution measurement instru m e n t s . N o t abl e than during the first half of this century;the productivity findings from the time-series sites include an incre a s e d of agriculture has improved so much that double the out- awa reness of the complexity of the ocean’s nitroge n put can be produced on half as much land. The heavy use c y cle (Karl et al. 1 9 9 7 ) . This finding has the potential to of fertilizer, together with improved tilling practices,may alter the view of the sensitivity of atmospheric CO c o n- 2 also lead to increased stores of organic matter in soils. c e n t rations to biological processes in the ocean (Fa l k owski 1997). On longer time scales, ocean biology The carbon balance in the tropics also affects estimates can have an impact on the atmosphere (see the fo l l ow i n g of the magnitude of the Northern Land Sink. As pointed ch a p t e r ) . The time-series sites are among the few loca- out before,the major constraint on the estimated magni- tions where seasonal va riations in upper ocean 1 3C /1 2C tude of the global terrestrial carbon sink comes from two ratios are measured (Bacastow and Keeling 1973). S u ch numbers. The first is the estimate of a global net terrestri- t i m e - re s o l ved info rmation is critical for corre c t ly inter- al uptake of 0.2 ± 0.9 Gt C/yr for the period from 1980 to p reting the large “ s n a p s h o t ” data sets collected along 1989. This number is obtained from calculating fossil fuel 16 Chapter 2: The Past and Present Carbon Cycle

Top panel: Managed forests in the Coast Range of Oregon. Conversion of old-growth forests to managed plantations in the Pacific Northwest has reduced the store of carbon to less than 25-35% of the maximum value in the last century. This has resulted in a substantial loss of carbon to the atmosphere. Altering management by increasing the interval between harvests and/or removing less carbon each harvest would “re-store” much of this carbon over the next century. Bottom panel: Deforestation in tropical regions has released a substantial amount of carbon in the last 50 years. Here the moist tropical forest of Los Tuxlas in Veracruz State, Mexico has been converted to maize and pasture agriculture. This conversion has reduced carbon stores on these sites at least 5-fold. 17 A U.S. Carbon Cycle Science Plan emissions minus atmospheric growth and ocean uptake. Tests of the Northern Land Sink hypothesis will thus The second number is the estimate of 1.6 ± 1.0 Gt C/yr require the further development and refinement of data released to the atmosphere from tropical deforestation. sets on historical land use changes,carbon stores per unit The difference between these numbers gives a required area,and models that can use this information. The goal terrestrial uptake of 1.8 ± 1.4 Gt C/yr (Schimel et al. should be to acquire and analyze accurate global invento- 1996), much of which appears to be in the Northern ries of highly fractionated land use change. Doing so, Hemisphere. If the net carbon flux from changes in tropi- however, requires a concerted effort to develop high-quali- cal land use were at the lower limit of 0.6 Gt C/yr, the ty, spatially explicit,long-time-series data sets on land use. Northern Land Sink would drop to 0.8 Gt C/yr. These data sets should be constructed from a variety of Maintaining the observed north-south gradient of CO2 in sources,including high-resolution, remotely sensed the atmosphere would require that the Southern Ocean imagery, explicit data on land use going back many would have to be a smaller sink than currently estimated. decades, government publications, research data,and The recent land use estimate of Houghton et al.(1998) reconstruction using proxy information. Side-by-side com- yields 2.0 ± 0.8 Gt C/yr for the tropical land use source. parison of population change and measured rates of defor- This value would require a compensatory terrestrial car- estation,where the data coexist,lends confidence to the bon sink of 2.2 Gt C/yr as well as a larger Southern Ocean reconstruction of rates of deforestation using relatively carbon sink. Gaining more certainty in the land use num- well-defined changes in population as the predictor. bers will reduce uncertainty in other contributing compo- A major new CCSP initiative is therefore proposed to nents of the Northern Land Sink,such as the fertilizing meet the following goal: effects of rising atmospheric CO2 and N deposition. The uncertainty of net carbon flux from land use stems Goal 3 : Establ i sh accur ate esti mates of the largely from incomplete and often incompatible databases i mpacts of hi stor i cal and cur r ent used to compile land use changes,and from the lack of l and use patter ns and tr ends on knowledge of carbon fluxes associated with specific activ- the evol vi ng car bon budget at l ocal to con- ities. These points are particularly true of tropical areas ti nental scal es. and for land uses involving economically nonproductive ecosystems (i.e.,nonproductive in a direct sense),such as wetlands, riparian forests,and natural grasslands. For example,Houghton et al.(1998) notes that no reliable data exist for Latin America on that portion of the loss of agricultural land to degraded land that is not recovering to forest. This situation has forced the omission of such land from the calculation of net carbon flux. The same is true for land subjected to shifting cultivation and the harvest of wood in Sub-Saharan Africa.

18 Chapter 3: Predicting the Future Carbon Cycle

The previous chapter addressed the first of the two that neglects the ocean is doomed to be nearly irrelevant overarching questions that the Carbon Cycle Science Plan for policy. However, direct human interventions in the (CCSP) must attempt to answer. In this chapter, we turn ocean carbon cycle have thus far been minimal. to the second of the two fundamental issues for carbon Furthermore,any future human interventions,such as cycle research: direct injection of CO2 in the deep ocean or enhancement of the biological flux of carbon by fertilization,will What will be the future atmospheric carbon likely be dwarfed by the magnitude of the ongoing pas- dioxide concentrations resulting from both sive uptake. On the other hand,humanity is already past and future emissions? manipulating the terrestrial biosphere on a global scale, The res e a r ch prog ram outlined in this rep o r t will ulti- and its influence on atmospheric CO2 is substantial. The ma t e l y be measured by its ability to provide rel i a ble esti- effect on the carbon cycle of ecological interventions on mates of future atmospheric CO2 co n c e n t r ations under dif- land has been mostly inadvertent to date. Most of the fer ent conditions. On l y with such knowl e d g e will it be interventions have resulted in decreasing the amount of po s s i b le to assess alterna t i v e scenarios of future emissions carbon in various terrestrial carbon reservoirs. In particu- fr om fossil fuels, ef fects of human land use, se q u e s t ra t i o n lar, woody biomass and active soil organic matter, because by carbon sinks, and responses of carbon cycling to poten- of their intermediate turnover times (30 to 100 years),are tial climate cha n ge . Th u s , the for emost reason for addition- the largest terrestrial pools affected by land use conver- al res e a r ch is to develop the ability to predict responses of sion and agricultural establishment. This past reduction in the global carbon cycle to var ious types of cha n ge . Th e terrestrial carbon storage,however, suggests the opportu- CCSP must be integrated in the for m of a rigo r ous and nity to increase carbon in terrestrial systems through co m p re h e n s i v e effor t to build and test models of carbon intentional management. The realization that sequestra- cy c le cha n ge , to evaluate and communicate uncerta i n t i e s tion of carbon in wood and soil may play a significant role in alterna t i v e model simul a t i o n s , and to make these simul a - in offsetting CO2 from fossil fuel burning is already evi- tions avai l a ble for public scrut i n y and application. Th e dent in international negotiations. models must also be capable of evaluating alterna t i v e Concerning how high atmospheric CO might go in sc e n a r ios for management of the carbon cycle . 2 the future,two general questions must be answered: There are grounds for optimism that in coming years • What will be the partitioning of carbon among the fate of CO in the ocean can be ascertained with rea- 2 the mobile reservoirs, and how will climate sonable accuracy. The global mass balance among emis- change affect this partitioning? sions,atmosphere,ocean,and terrestrial biosphere ensures that a better quantitative estimate of ocean • How can the future growth of atmospheric CO2 uptake also improves the estimate for changes in terrestri- be managed? al storage,if only in very coarse geographical detail. Each of these points is examined,and thereafter, a Direct measurements of terrestrial inventories and fluxes, major new research initiative is proposed to address the in conjunction with atmospheric measurements and mod- most compelling scientific issues,an initiative that is also els,will help to refine the geographical details. scientifically feasible and cost-effective. Unfortunately, improved knowledge of the environmental fate of historical CO2 emissions cannot by itself give us confident predictions of future atmospheric CO2. The Projecting Future Atmospheric CO2 CO2 concentration trajectories calculated by the Inter- Concentrations governmental Panel on Climate Change (IPCC) for scenar- The IPCC has provided some scientific basis for inter- ios of fossil fuel burning assume that the future carbon national policy decisions by extrapolating the historical cycle will continue operating exactly as it is thought to behavior of the terrestrial biosphere and ocean into the have operated in the past. This assumption is not likely to future. Again,the assumption is that carbon uptake by the be correct. terrestrial biosphere will continue to occur through the There is a fundamental difference between the ocean same mechanisms as at present,and that ocean circulation and the terrestrial biosphere for policy decisions relating and biology will remain constant through time (Houghton to greenhouse gases. The ocean remains the biggest long- et al.1996). However, from coupled atmosphere-ocean term player in the carbon cycle,and any research program simulations with time-dependent radiative forcing

19 A U.S. Carbon Cycle Science Plan

(e.g.,Haywood et al.1997),it appears possible that the ter- as carbon-rich peatlands and forests or carbon-poor restrial biosphere and ocean carbon cycle might already deserts could substantially affect carbon storage on land. be experiencing direct effects of climate change today. Although slow changes to long-lived vegetation are diffi- These effects could become even larger over the next cen- cult to study experimentally, and models of these process- tury (e.g.,Cao and Woodward 1998,Sarmiento et al.1998). es are in early developmental stages,an integrated pro- Further, it seems likely that both terrestrial and oceanic gram of manipulative experiments,paleoecological stud- ecosystems will undergo significant indirect responses to ies,and continuing observations can facilitate continuing climate change and human impacts on the environment, improvements in the models. such as NO [nitrogen oxides] fertilization,air and water x Significant integrative research is needed to improve pollution,and CO increase. Such shifts might include 2 predictive capability for terrestrial ecosystems,both man- changes in species distribution in addition to changes in aged and natural. First,the basic science and its represen- the supply of nutrients and other ecosystem components tation in models must be improved. Better representation that determine carbon cycling. is needed of key processes,such as responses to disturbances,CO2 warming, nutrient deposition,and atmospheric pollution,including their interactions. Important ques- Terrestrial Ecosystems tions are emerging that require new direct evidence from Terrestrial ecosystems have played and will continue to observational and manipulative studies. Analyses of atmos- play a significant role in the global carbon cycle. Release pheric observations also require better models of terrestri- of CO2 from land use change has been a significant flux to al ecosystems. Models are an important tool in inverse the atmosphere historically, and could well continue or analysis of atmospheric data (inverse analysis is a mathe- even accelerate. In addition,there appears to be a signifi- matical tool that infers unknown variables from a given cant sink of CO2 in land ecosystems arising from the syn- set of equations,data and assumptions).Models of the ergistic effects of past land use changes,increasing atmos- land biosphere are urgently needed for data analysis. pheric CO2, the deposition of fixed nitrogen,and,possibly, Beyond global models,detailed site-specific models are climate changes over the past century or so. To project also required to assess the long-term consequences of the consequences of given human activities (such as fossil management options. Such models can be used to fuel burning and land use change),it is essential to under- appraise the influence of a given forest or agricultural stand the responses of terrestrial ecosystems. It is like- management practice on both commodity production and wise important to understand how climate interacts with environmental impacts at the same time,including carbon terrestrial ecosystems. Several factors will control the bal- storage and trace gas exchange. Improving projections of ance of terrestrial ecosystems with respect to carbon. The the future carbon balance of terrestrial ecosystems current degree of uncertainty regarding all of these factors requires an integration of global carbon cycle and climate is high. models (to calculate climate-ecosystem feedbacks on atmospheric CO ) and management-oriented models for Increasing CO and fixed nitrogen from fossil fuel burn- 2 2 use in decision making about terrestrial ecosystems. ing can both act as fertilizers to ecosystems,increasing net Decisions about terrestrial ecosystems will influence— primary production (NPP, the amount of carbon processed perhaps heavily—the future effects of ecosystems on by photosynthesis in green plants minus that lost through the atmosphere. respiration),and possibly carbon storage. Air and water pollution can lead to degradation of the ecosystem and loss of carbon. Observational and manipulative studies The Ocean have not yet yielded unambiguous results about the mag- nitudes of these effects,and there are thus substantial dis- Coupled atmosphere-ocean model simulations (e.g., agreements in model predictions. Hay w ood et al. 1997) predict a large war ming of the su r f ace wat e r s of the ocean (e.g., of 2.5°C by mid next There is evidence that climate change and variability ce n t u ry ) . Th e y also predict increased strat i f ication of the over the past century may have influenced both the distri- su r f ace ocean from the higher temperat u r e at low latitudes bution of vegetation and its productivity. The potential and increased precipitation in high latitudes. One of the exists for significant positive climate feedback in which consequences of these cha n g es may be a reduced rate of warming of arctic soils and permafrost,high in organic for mation of North Atlantic Deep Wat e r , perhaps as early as carbon content,could stimulate the oxidation to CO of 2 the next decade. Ho wever , on time scales of a few much of that carbon (Goulden et al.1998,Oechel et al. de c a d e s , cha n g es in the rate of convection and vert i c a l 1993). Most models of vegetation dynamics suggest possi- mixing appear more important for the surface CO ba l a n c e ble major redistribution of vegetation zones with future 2 than cha n g es in advec t i o n . Fu rt h e r , in at least one simul a - climate change. Changes in the extent of ecosystems such ti o n , ocean CO2 up t a k e is parti c u l a r l y impacted by cha n ge s

20 Chapter 3: Predicting the Future Carbon Cycle

Global, annual net primary production (NPP) (g C/m2/y) for the land and ocean biosphere. This calculation is from models that use satellite data to calculate the absorption of visible radiation by photosynthetic pigments in plants, algae, and cyanobacteria.The land model (CASA) and the ocean model (VGPM) are similar in their reliance on broadly observed patterns to scale photosynthesis and growth from the individual to the ecosystem level.This calculation, based on ocean data for 1978–1983 and land date for 1982–1990 produces a global NPP of 104.9 Pg C/y (104.9 x 1015 g C/y), with approximately half (46.2%) contributed by the oceans and half (53.8%) contributed by the land. These approximately equal contributions to global NPP highlight the role of both land and ocean processes in the global carbon cycle. in the Southern Ocean compared with other reg ions of the decrease by 30 percent and the pH by more than 0.2 in oc e a n . These cha n g es in ocean mixing and circulation may the mixed layer by the middle of the next century. Some reduce cumul a t i v e oceanic uptake of CO2 by 10 to 30 per- changes in taxonomic composition and bulk physiology cent between now and the middle of the next century could have an impact on important carbon cycle parame- (Matear and Hirst in pre s s ,S a rmiento et al. 19 9 8 ) . ters. Among these are the ratio of calcium carbonate (CaCO ) to organic carbon production in coral reefs Climate change may also have a major impact on ocean 3 (Smith and Buddemeier 1992) and coccolithophorids,the biology, which in turn would affect the ocean’s uptake of ratio of organic carbon to nutrients in material exported CO . One change may be reduced or altered global pro- 2 from the surface,and the amount of carbon locked up as ductivity due to slower upward mixing of nutrients from dissolved organic carbon,which presently amounts to the thermocline,or increases in productivity associated about 90 billions of metric tons,or 90 gigatons (Gt C),in with anthropogenic nutrients or richer supply of micronu- the upper 500 meters. trients by dust transport. Another may be change in taxonomic composition and physiology. Environmental Time-series observations from the past decade show changes that could drive such ecological shifts include clearly that ocean biota respond dramatically to interannu- warming of surface waters and stabilization of the water al climate variability such as the El Niño–Southern column.The supply of micronutrients by dust transport Oscillation (ENSO) (Karl 1999). Ocean warming due to has an important impact on ecology, as in making it possi- climate change could lead to changes of comparable mag- ble for nitrogen fixers to exist in nutrient-poor regions nitude. Simulations with ocean general circulation models (Michaels et al.1996a). Also important are carbon chem- show that biological changes may increase the oceanic istry changes. The concentration of carbonate ion will uptake of CO2 by 5 to 25 percent,depending on what 21 A U.S. Carbon Cycle Science Plan assumptions are made about how to model the biological the potential responses of oceanic communities to response. This effect is large enough to counteract much climate change. if not most of the reduction in uptake from mixing and Continuing to identify the main hypotheses and goals circulation,but the magnitude and even the sign of this for the CCSP as in Chapter 2, we can then state another effect are highly uncertain. fundamental hypothesis for carbon cycle research: The relationship between marine ecosystem structure and the rate at which biological processes move carbon Hypothesi s 2 : The oceani c i nventor y of between surface and deep waters is an emerging research anthr opogeni c CO wi l l conti nue to i ncr ease i n theme. The most prevalent ecosystem structure of the 2 r esponse to r i si ng atmospher i c CO con- open sea,particularly in the equatorial and subtropical 2 centr ati ons, but the r ate of i ncr ease wi l l be regions,is one dominated by the microbial loop. The modul ated by changes i n ocean ci r cul a- microbial loop consists of very small organisms in a com- ti on, bi ol ogy, and chemi str y. plex trophic structure with efficient nutrient recycling,little accumulation of biomass,and little export of carbon This hypothesis of an increasing long-term ocean car- from surface waters to the deep sea (Landry and al.1997). bon inventory (or “the Increasing Ocean Inventory Experimental evidence (e.g.,Coale et al.1996) shows that hypothesis”) provides a critical focus for concerns about nutrient perturbations to the steady state in these ocean the future long-term effectiveness of CO sinks. One regions leads to enhanced growth and dominance of 2 value of a hypothesis concerning future CO sinks is that diatoms and other large phytoplankton. Increased new 2 it challenges us to anticipate when and how it can be production associated with diatom blooms, for example, tested. The estimation of oceanic CO uptake has been a increases carbon dioxide uptake from the atmosphere in 2 primary objective of chemical oceanographic programs surface waters and (eventually) leads to vertical export of for more than two decades (GEOSECS,TTO, SAVE,WOCE, particulate carbon to deep waters. JGOFS1). Using a variety of direct and tracer measure- At high latitudes,and particularly in the Southern ments,these programs have considerably improved the Ocean,increased stratification in response to climate ability to model the ocean CO2 sink. The recent advances change may lead to a more efficient biological export of discussed previously suggest that trends in long-term carbon from surface to deep waters,thereby reducing sur- oceanic CO2 uptake may soon be susceptible to much face nutrients and carbon. The evidence for this effect is more direct verification. based primarily on numerical models (e.g.,Sarmiento and Like the Northern Land Sink hypothesis,the Increasing Le Quéré 1996) and has not yet been confirmed by direct Ocean Inventory hypothesis cannot be considered in iso- observation. In subtropical and equatorial waters,howev- lation. For example,it explicitly encompasses the need to er, future warming may cause the opposite effect. understand the effects of variations in oceanic circulation, Increased stratification of the water column may lead to a biology, and chemistry. Such variations have been hypoth- reduced supply of either micro- or macronutrients,a shift esized to account for the seasonal and interannual variabil- to N-fixing organisms,and even lower carbon export than ity in oceanic CO exchange associated with the phasing occurs today (Falkowski et al.1998). Some evidence 2 of the El Niño–Southern Oscillation and North Atlantic implies that such changes are now occurring in the Oscillation (NAO). Because ENSO and the NAO are also Northern Hemisphere (Karl et al.1997,McGowan et al. implicated in short-term atmospheric CO variations,they 1998). The potential contributions of such changes to the 2 must be considered in the use of atmospheric transport ocean carbon sink are not yet understood. Hypotheses inverse models to constrain the Northern Land Sink. that link warming-induced changes to upper-ocean stratifi- Constraints on the spatial distribution of air-sea fluxes are cation (and nutrient flux) and changes to marine ecosys- critical to determining the spatial distribution of terrestrial tem structure and carbon export are testable using long fluxes. Determining the North American or Eurasian flux- time-series observations. They may also be tested by es with confidence requires knowing the North Pacific focused process studies on ecosystem responses to inter- and North Atlantic fluxes. Thus,the Northern Land Sink annual variability (e.g.,during warm versus cold years) and Increasing Ocean Inventory hypotheses are directly and other “natural” experiments. related and closely linked in a variety of important ways. Current understanding of marine ecology and the abili- The Northern Land Sink and Increasing Ocean ty to predict the oceanic response to climate change are Inventory hypotheses represent a wide array of perspec- extremely rudimentary. It will take at least a decade of tives that are inherent in carbon-cycle research:not only research,including long time-series data taken at a num- terrestrial and marine,but also short-term and long-term, ber of stations,to significantly improve understanding of

1 Geochemical Ocean Sections program (GEOSECS);Transient Tracers in the Ocean (TTO);South Atlantic Ventilation Experiment (SAVE);World Ocean Circulation Experiment (WOCE); Joint Global Ocean Flux Study (JGOFS). 22 Chapter 3: Predicting the Future Carbon Cycle

Data-based and model estimates of uptake of carbon (mmol/kg) released to the atmosphere by human activities, or “anthropogenic carbon,” by the Western Atlantic Ocean along a transect illustrated in top left panel.Most of the anthropogenic carbon in the oceans is found in the upper 1 to 2 kilometers except in regions of deep water formation, such as the North and, to a lesser extent, South Atlantic.The ocean has a large capacity to dissolve anthropogenic carbon, but it takes many centuries to millenia for this capacity to be realized.Predicting the future uptake of anthropogenic carbon by the oceans thus requires models of ocean circulation such as those shwn in the bottom four panels of the figure. These models are able to reproduce the general features of the observed anthropogenic carbon distribution, such as the shallower penetration in the low latitudes, and deeper penetration in higher latitudes. However, there are also many differences among the models, and between the models and the observations. For example, all of the models fail to simulate a sufficiently deep penetration between about 30°N and 50°N, and three of them simulate too much penetration in the South Atlantic. This figure was constructed by the Ocean Carbon-Cycle Intercomparison Project team (OCMIP) using a variety of data and model sources including:Orr et al.in preparation, Gruber (1998), Sarmiento et al.(1995), Toggweiler et al. (1989), Taylor (1995), Maier-Reimer (1993), Madec and Imbard (1996), and Aumont et al.(1998). 23 A U.S. Carbon Cycle Science Plan regional and global,spatial and temporal,diagnostic and since the beginning of the industrial revolution (e.g., prognostic. They illustrate the extent of integration that is Maier-Reimer and Hasselmann 1987,Sarmiento et al. necessary in a framework of evolving hypotheses concern- 1992). The fraction that can dissolve in the ocean decreasing the global carbon cycle. es as the total amount added to the atmosphere increases. An additional 5 to 10 percent of the combined inventory To summarize,predicting the future role of oceanic and will be absorbed by the reaction of CO with CaCO in terrestrial biospheres in determining atmospheric CO 2 3 2 sediments on a time scale of 10,000 to 100,000 years needs to consider the feedback among climate change and (e.g.,Archer et al.1997). One way of managing atmo- terrestrial and marine biogeochemical processes. Climate spheric CO is to accelerate the uptake of carbon by the change will affect the terrestrial biosphere and ocean;and 2 ocean by pumping it into the slowly ventilated deep changes in the terrestrial biosphere and ocean will affect waters. These calculations assume that surface ocean tem- atmospheric CO . Capturing these mechanisms will 2 perature,salinity, and alkalinity remain constant,and that require more highly developed models of the terrestrial the marine cycle of CO fixation and sedimentation of and oceanic biospheres,ocean mixing and circulation,and 2 organic matter and carbonate skeletons (the “biological their response to warming. A major near-term goal of the pump”) does not modify the oceanic carbon distri- CCSP is therefore the following: bution significantly. In the terrestrial biosphere reservoir, photosynthesis Goal 4: Improve projections of future removes CO from the atmosphere,but much of the car- atmospher i c CO concentr ati ons thr ough a 2 2 bon is removed only temporarily (IGBP Terrestrial Carbon combination of manipulative exper iments and Working Group 1998). Approximately 50 percent of the model devel opment such that appr opr i ate bi o- initial uptake of carbon from photosynthesis is lost physi cal and ecol ogi cal mechani sms and car - through plant respiration. Again,the carbon remaining bon cycl eÐcl i mate feedbacks ar e i ncor por ated after respiration is net primary production (NPP). Part of i n gl obal cl i mate and car bon cycl e model s. NPP is lost as litter and enters the soil,where it decom- poses, releasing carbon to the atmosphere. On an annual basis,carbon remaining after decomposition is net ecosys- Management Strategies tem production (NEP). Further carbon losses occur by such processes as fire,insect consumption,and harvest. The goal of carbon management strategies is to abate These losses yield net biome production (NBP),the criti- the increase of anthropogenic carbon concentrations in cal carbon quantity in considering long-term carbon stor- the atmosphere. Two approaches for accomplishing this age. The IGBP Terrestrial Carbon Working Group (1998) goal are measures to reduce carbon emissions and meas- argues that belowground components of ecosystems are ures to increase the uptake of carbon by oceanic and especially important because belowground carbon gener- terrestrial biosphere sinks. There are no fundamental ally has slower turnover rates than aboveground carbon. physical or chemical barriers to using the energy poten- Slower turnover implies that carbon storage can be main- tially available in fuels much more effectively to bring tained over longer periods. down the rate of emissions.Relying on low or non-CO2 emitting energy sources is another way of reducing An important outcome of better information on land atmospheric CO2 concentrations in the long term. use change (Goal 3 for the CCSP, as outlined in Chapter 2) However, in the interim,it is also necessary to consider will be a stronger scientific basis to analyze strategies for the option of managing carbon uptake by the oceanic managing global carbon fluxes in compliance with inter- and terrestrial reservoirs. national treaties aimed at increasing terrestrial carbon sequestration. We need to know which management Exchange with the large geological reservoirs of car- strategies for terrestrial biological processes are cost-effec- bon,such as limestone,is slow. The carbon added to the tive in forests, grasslands,and croplands. Terrestrial atmosphere will stay confined to the three “mobile” reser- ecosystems are partially constrained by the legacy of land voirs mentioned earlier (i.e.,atmosphere,terrestrial bio- use practices in the past. Land use practices are discontin- sphere,and ocean) for hundreds of thousands of years. In uous in time and space. Hence,in the case of forests,dif- essence,the burning of coal,oil,and natural gas represents ferent stands in the same region will rarely be of the same an artificial transfer of carbon from the geological reser- age class and species composition. They will sequester voirs to the mobile reservoirs. carbon at different rates and in different amounts in The ultimate long-term sink for carbon of the three response to a given change in management (e.g., changes mobile reservoirs is the ocean. Model calculations show in harvest rate or predation control) or climate. that,on a time scale of 1,000 to 10,000 years,the ocean Delineating these effects will require close coupling will absorb about 85 percent of the anthropogenic carbon between observed land use patterns and trends and corre- that has been added to the combined atmosphere-ocean sponding estimates of carbon flux densities,to create

24 Chapter 3: Predicting the Future Carbon Cycle spatially explicit gradients of carbon flux at regional when trying to balance the global carbon budget with scales. This effort will require combining satellite and oceanic and terrestrial sources and sinks. The financial ground-based observation of land use patterns and trends, and economic stakes in correct carbon accounting are of flux measurements,and process-based modeling. great importance for any country in a world where emis- Together, information from all these sources will allow the sions permits can be traded. Some measurement strategy scientific assessment of management scenarios to increase is needed,independent from statistical and accounting sequestration of carbon on the land. methods,to determine the magnitude of fossil fuel CO2 emissions on regional and national scales. The only The scientific evaluation of management options nearly unequivocal (characteristic and stable) tracer for requires the direct involvement of social scientists in con- fossil fuel CO in the atmosphere is its complete lack of structing land use histories. Thus,there is a permeable 2 14C isotopes. boundary between this science plan and the relevant social sciences. Careful assessments are needed of the These considerations lead to another major near-term social and economic costs and benefits of proposed goal of the CCSP: carbon sequestration policies. One critical element will be the development of credible land use scenarios for Goal 5 : Devel op a sci enti fi c basi s to eval - the future. uate potenti al management str ategi es for Another important component of carbon management enhanci ng car bon sequestr ati on i n the envi - strategies will be the ability to verify commitments. Other r onment and for captur e/ di sposal str ategi es. sections of this report discuss the estimation of oceanic and terrestrial sinks. An important additional issue is the estimation of fossil fuel emissions. If the global estimate of fossil fuel emissions based on economic accounting methods is off by 10 percent,that would represent an error of 0.65 Gt C/yr—which is more than the reductions envisioned in the Kyoto Protocol. This error is quite large

25 Chapter 4: An Integrated Carbon Cycle Science Program

The previous two chapters of this report summarized our present understanding of the global carbon cycle and Hypothesi s 1 : Ther e i s a l ar ge ter r es- the basic questions that confront present-day efforts to tr ial sink for anthr opogenic CO2 i n the better understand it. This chapter presents an integrated Nor ther n Hemi spher e. plan to achieve the major goals that have been identified:

Hypothesi s 2 : The oceani c i nventor y of Near - Ter m Goal s of the U.S. Car bon Cycl e anthr opogeni c CO wi l l conti nue to i ncr ease Sci ence Pl an 2 i n r esponse to r i si ng atmospher i c CO2 concentr ati ons, but the r ate of i ncr ease wi l l Goal 1 : Establ i sh accur ate esti mates of be modul ated by changes i n ocean the magni tude of the potenti al Nor ther n ci r cul ati on, bi ol ogy, and chemi str y. Hemi spher e ter r estr i al car bon si nk and the under l yi ng mechani sms that r egul ate i t. The research program outlined here will ultimately be judged by its ability to provide practical answers to both scientific and societal questions. Scientists and policy Goal 2 : Establ i sh accur ate esti mates of makers must be able to evaluate alternative scenarios for the oceani c car bon si nk and the under l yi ng future emissions from fossil fuels,effects of human land mechani sms that r egul ate i t. use,sequestration by carbon sinks,and responses of carbon cycling to potential climate change. Thus,a key moti- vation for further research is to develop the predictive Goal 3 : Establ i sh accur ate esti mates of capability to define responses of the global carbon cycle the i mpact of hi stor i cal and cur r ent l and use to change,as reflected in Goals 4 and 5. patter ns and tr ends on the evol vi ng car bon budget at l ocal to conti nental scal es. Recent assessments of global environmental research have emphasized the need for programs that are both integrated and focused (e.g.,National Research Council Goal 4: Improve projections of future 1998). This plan puts forth a program that focuses on key atmospher i c concentr ati ons of car bon di oxi de problems, yet maintains breadth to reveal new problems thr ough a combination of manipulative and priorities in those areas where the knowledge needed exper i ments and model devel opment that to define focused strategies is currently lacking. In i ncor por ates appr opr i ate bi ophysi cal and Chapter 6,this report also proposes a management struc- ecol ogi cal mechani sms and car bon cycl e- cl i - ture for implementation of the research plan and the mate feedbacks i nto gl obal cl i mate and car bon development of critical partnerships to ensure continuous cycl e model s. reassessment and prioritization of goals. Much recent progress in knowledge of the carbon cycle has resulted from studies at the global scale for time Goal 5 : Devel op a sci enti fi c basi s for periods of years to decades. However, to make significant eval uati ng potenti al management str ategi es progress in understanding and quantifying the critical for enhanci ng car bon sequestr ati on i n the mechanisms that will determine future levels of atmos- envi r onment and for captur e/ di sposal str ate- pheric CO , data must also be obtained for specific geo- gi es. 2 graphic regions over a range of time scales. The finger- prints of dominant processes are to be found by studying These goals are intended to guide research for the peri- regional carbon balances and temporal variability. Efforts od of the next 5 to 10 years. to address both intermediate spatial scales and longer time scales are thus essential components of the pro- Two general hypotheses were also identified in posed plan. Chapters 1 and 2 as those most critical for the U.S. Carbon Cycle Science Plan (CCSP) to address: 27 A U.S. Carbon Cycle Science Plan

The program will also study the main processes influ- Similarly, the northern oceans are relatively accessible, encing how carbon cycling may change in the future. and a solid foundation of oceanic data and knowledge is These studies will be integrated in a rigorous and compre- available to support integration of studies of North hensive effort to build and test models of carbon cycle American atmospheric CO2 exchange with studies of CO2 change, evaluate and communicate uncertainties in alter- exchange in the oceanic regions adjacent to it. These foci native model simulations,and make these simulations offer a unique opportunity to combine atmospheric, available for public scrutiny and use. Clearly, the systemat- oceanic,and terrestrial studies in a way that will constrain ic incorporation of newly understood mechanisms in major components of the global CO2 budget. models must be accompanied by model integration using Goal 1: Understanding the Northern high-quality standard inputs and rigorous consistency tests Hemisphere Terrestrial Carbon Sink against an array of benchmark data. Data management and data set construction are sometimes underrepresent- One principal goal of the CCSP is to establish accurate ed in hypothesis-oriented programs. This pitfall must be estimates of the magnitude of the potential Northern avoided because,ultimately, it weakens the ability to test Hemisphere terrestrial carbon sink and the biophysical hypotheses using comprehensive data and to develop mechanisms that regulate this sink. Several major activi- powerful generalizations and new hypotheses. ties should be conducted in this area: At its most basic level,the global carbon cycle must be • An expanded prog ram of atmospheric concentrat i o n viewed as a singular entity. Its various components are so me a s u r ements and modeling improvements in support of interactive—over so many different scales of time and in ver se calculations and global biogeo c hemical models space—that they cannot conveniently be “isolated”for • A network of integrated terrestrial research sites with independent study or modeling. Data are most valuable eddy-covariance flux measurements and associated when combined from a variety of measurements and process studies,manipulation experiments,and models, methods associated with different carbon-cycle compo- which as a whole are sufficient to reduce uncertainty nents; for example,when oceanic data are applied to help about the current and future carbon cycle to accept- interpret results for the atmosphere and terrestrial bios- able limits. phere,and vice versa. The present plan,then,proposes three different types of general approach: The two boxes here summarize the specific proposed program elements,which are discussed in detail in the • Extend observations over the important space and sections that follow. time scales of variability in all active carbon reservoirs • Develop manipulative experiments to probe key mechanisms and their interactions Goal 1a: At mospher ic Measur ement s and Models—Maj or Pr ogr am Element s and Act iv it ies • Integrate these data, analysis, and modeling approaches so that they are mutually supportive and ¥ Expand cur r ent at mospher ic monit or ing and can be focused on key problems. obser v at ional net w or ks t o acquir e— The observational strat e g y described in this chapter is designed to combine atmospheric measurements with - Ver t ical concent r at ion pr of iles in cont inen- ob s e r vations from space, ai r , la n d , and sea to reveal specific t alsour ce/ sink r egions pr ocesses that affect terres t r ial and oceanic carbon - Incr eased cont inuous measur ement s (as exch a n g e at reg ional as well as global scales. While the goa l cont r ast ed w it h w eekly f lask dat a) of CO2 and of the terres t r ial component of this strat e g y is ultimately to associat ed t r acer s at select ed cont inent al sur f ace un d e r stand the North e r n Hemisphere terres t r ial sink, th e st at ions continent of North Am e r ica is an excellent focus for the U.S . res e a r ch community in developing this res e a r ch objec- - Flask measur ement s in under sampled ti ve . No r th Am e r ican logistical capabilities are exc e l l e n t r egions, pr ov iding t he dat a needed f or cont inent al and cost-effec t i ve , and there are ext e n s i v e existing data sets and r egional appor t ionment of net car bon for North Am e r ican ecosystems, land use, so i l s ,i n d u s t ri a l ex change. ac t i v i t i e s , and history. Recent res e a r ch has pointed to the ¥ Enhance t he suit e of measur ement s at t he glob- pa r ticular importance of understanding terres t r ial carbon al net w or k sit es t o include ox y gen, car bon and exch a n g e in the North e r n Hemisphere, and North Am e ri c a ’ s ox y gen isot opes, r adon, and ot her par amet er s t o la r ge geopolitical units facilitate the development of an inte- const r ain t he locat ions and pr ocesses r esponsible grated continental analysis of carbon sources and sinks. f or t he car bon sour ces and sinks. Par allel res e a r ch by European and Asian colleagues will be en c o u r aged to expand the cover age into other parts of the No rt h e r n Hemisphere terres t r ial biosphere. 28 Chapter 4: An Integrated Carbon Cycle Science Program

Goal 1a (continued) Atmospheric Measurements and Modeling for Inverse Calculations ¥ Conduct focused field campaigns over North America using aircraft sampling techniques in combination with Atmospheric transport of CO2 integrates the effects of improved at mospheric transport models and an enhanced local sources and sinks,with mixing around a latitude cir- flux tower network, to confirm and refine estimates of cle in a few weeks,and between hemispheres in about a the magnitude of terrestrial sources and sinks of CO2. year.A primary test of Northern Land Sink hypotheses requires resolving the longitudinal structure of the surface ¥ Improve inverse models and strengthen connection CO2 flux,in addition to the variation in latitude. Clearly, between models and observations. the observations will have to define concentration gradients between the continents and the sea and between the planetary boundary layer and the middle and upper tropo- Goal 1b: Terrest rial Studies—Major Program Element s and sphere. These are inherently difficult measurements, Activities because atmospheric mixing is so much more rapid around latitude circles than across them,and measure- ¥ Synthesize results of recent and ongoing terrestrial ments near sources or sinks are highly variable. Design of carbon flux studies and use the data to constrain global the necessary atmospheric sampling program requires carbon/ t errest rial/ atmosphere models. careful attention to the spatial and temporal distribution of sampling,the precision of the atmospheric data,and ¥ Conduct long-term monitoring of changes in above the details of some boundary-layer atmospheric processes and belowground carbon stocks on forest, agriculture, that are not well understood at present. and range lands using improved biomet ric inventories that explicitly address carbon issues, along with new types of The atmospheric observing system currently consists of remote sensing, to determine long-term changes in car- roughly 100 sites around the world at which air is collect- bon stocks. ed weekly in paired flasks for trace gas analysis at central laboratories,and few sites where observations are made ¥ Develop new technology to facilitate direct long-term continuously.The sites are intentionally located in remote flux measurement s of CO2, and install an expanded net- marine locations to avoid local “contamination”by indus- work of long-term flux measurements emphasizing the trial or terrestrial emissions or uptake,and are operated at acquisition of data for representative ensembles of undis- low cost using cooperative arrangements with volunteers. turbed, managed, and disturbed lands along major gradi- There are very few data acquired at altitude and at mid- ents of soils, land use history, and climate, and with a continental stations. number of towers and sites sufficient to allow quantifica- To provide meaningful constraints on net terrestrial tion of error in the spatial domain. CO2 exchange on the regional scale,the observing network will need to be strengthened considerably to charac- ¥ Quantify mechanisms controlling terrestrial sources terize spatial and temporal variations associated with the and sinks, their evolution in the geologically recent past, carbon fluxes that need to be measured. The present net- and the likely course of their evolution over the next work is designed to be insensitive to regional net decades to centuries. This quantification should be exchanges. A recent evaluation of 10 global tracer trans- achieved through regionally nested sets of manipulation port models used for CO inversions (Denning et al.1999) experiments, other process studies, and CO f lux meas- 2 2 found that the models converged when compared to the urements. Measurements should be integrat ed through a observed values for an inert tracer (SF ) at flask stations in parallel set of nested models, wit h analyses scaled up to 6 the remote marine boundary layer. However, they regions through use of regional inventory data. diverged where the data are sparse (aloft and over the continents). This problem is even worse for CO due to ¥ Conduct manipulations and focused process studies on 2 the covariance between diurnal and seasonal cycles of ecosystem, local, and regional scales, coordinated with CO net exchange and rates of atmospheric mixing,often CO flux measurements to quantify the mechanisms con- 2 2 called the “rectifier effect.” Expanding the atmospheric trolling t errest rial sources and sinks, their duration, and observing network to include routine sampling aloft,par- past and future evolution. ticularly over the continents,should be one of the highest priorities for the future. Atmospheric sampling over the ¥ Develop techniques for monitoring the dynamics of terrestrial surface must include vertical profiles through a belowground carbon stocks. depth sufficient to capture most of the vertical mixing of the surface signal. At a minimum,this vertical sampling must span the depth of the planetary boundary layer (1 to 3 km in warm sunny weather). A primary research goal should be to determine first the optimal sampling density, 29 A U.S. Carbon Cycle Science Plan supporting measurements,and combination of continuous the footprint of an eddy flux tower (Desjardins et al.1997, versus flask samples, for long-term airborne sampling. Goulden et al.1998). These types of studies directly Multiple species,such as tracers of industrial activity, and address the issues of scalability of tower fluxes,and can isotopic ratios must also be measured to obtain the infor- be used to design lower cost, routine sampling programs mation needed to interpret the observations (e.g., for larger scales. Potosnak et al.1999). Independent estimates of reg ional-scale carbon flu xe s The atmospheric boundary layer and the covariance by inver sion of atmospheric data will req u i r e a dense net- between terrestrial ecosystem processes and near-surface wor k of samples collected by light aircra f t . Light aircra f t turbulence are not resolved in most of the current genera- sampling must be dense enough to capture meaningful gration of global atmospheric models. Most atmospheric dients in surface flu xe s , and must sample both within and tracer transport codes used for CO2 inversion calculations abo ve the convec t i v e boundary layer .Ver tical prof iles from represent subgrid scale vertical transport very crudely, if light aircr aft over a continental reg ion could be coupled at all. Likewise, very few of these models include a diur- with high-altitude transects sampled from approp ri a t e l y nal cycle of CO2 exchange. in s t r umented commercial aircr aft (Marenco et al. 19 9 8 ) . Ho wever , even if a model could correc t l y rep r esent the Both inversion calculations and forward models will local covar iance struc t u r e of the flu x es and the turbu- benefit tremendously from additional constraints such as le n c e , the influence of the rec t i f ier effect on the observed regionally detailed emissions data and multiple tracers. co n c e n t r ation field at remote flask stations depends on the Samples should be analyzed for CO2, as well as CO, CH4, pe r sistence of the ver tical gradient as the air is tran s p o rt e d O2/N2, and stable isotopic ratios,all of which provide con- ho ri z o n t a l l y for hundreds or thousands of kilometers. Th i s straints on the carbon cycle. Ancillary data such as PBL pr ocess is ver y poorly res o l v ed in even the most detailed structure (from wind profiling and traditional sounding global models, and is not well understood theoret i c a l l y. systems),atmospheric transport (from four-dimensional data assimilation systems,4DDA),and the isotopic ratios of A major effort in understanding the local forcing,spa- other components of the land-atmosphere system (plants, tial scaling,and long-distance transport aspects of the rec- soils,precipitation,and groundwater) are needed. Such a tifier effect is required,through both observations and system has been proposed using automated sampling models. Testing the Northern Land Sink hypothesis also equipment and rental aircraft (Tans et al.1996). requires filling the gap at the crucial “middle scale”of the flux-transport-concentration problem. This middle scale Regional observing and modeling programs have been between local and large-scale observations is completely proposed in other countries on a “campaign”basis,and missing from the current observing system and models. the results of these studies can provide useful constraints on regional flux estimates using inverse modeling. The detailed design of the required atmospheric sam- Regional experiments quantifying carbon fluxes or tracer pling program,which must be coordinated with design of concentrations are currently underway or planned for the the terrestrial flux network and ocean measurements pro- near future in Europe,Siberia,Brazil,and Australia.The posed below, is a major scientific endeavor beyond the design and implementation of U.S.observing systems and scope of this document. However some general require- modeling programs should be optimized to take advantage ments are clear. A strategy must be developed for atmo- of these complementary programs. spheric sampling over continental regions which takes into account the differences in ecosystem exchanges in Ideally, inverse calculations of the carbon budget stable and convective conditions. This variation must be should subsume all available information,including flask explored over a range of ecosystems and meteorological samples,in situ data,aircraft sampling,air-sea flux meas- regimes by sampling from eddy flux towers,tall towers, urements and eddy covariance data.Carbon budgets calcu- balloons,and light aircraft. Continuous long-term meas- lated from inverse methods should not, for example,be urements of the vertical profile of CO2 and other trace inconsistent with measured diurnal cycles of CO2 data gases on tall transmission towers allows “representative” collected by regional sampling programs in other parts of conditions to be defined for the planetary boundary layer the world. The current global observing system is so (PBL) sampling at the local scale (Bakwin et al.1995, poorly constrained in the tropics, for example,that tropi- Bakwin et al.1998,Hurst et al.1997). Similar information cal fluxes are freely estimated in inversion models as a can be obtained from continuous long-term measurements residual,allowing unacceptable freedom of terrestrial flux- of CO2 and other trace gases in conjunction with eddy- es in higher latitudes without violating global mass bal- correlation fluxes defining rates of exchange between the ance.Inclusion of new regional data from experiments in surface layer and the planetary boundary layer (Potosnak Amazonia in these inverse models would provide stronger et al.1999).Light aircraft can be instrumented with con- constraints on the carbon budget of North America, tinuous analyzers to determine the boundary-layer budgets directly addressing uncertainties in the Northern of trace gases over areas orders of magnitude larger than Hemisphere Land Carbon Sink hypothesis. 30 Chapter 4: An Integrated Carbon Cycle Science Program

New approaches to inverse modeling are needed to some degree . Th u s , great care must be used in interpret i n g apply highly resolved atmospheric data to constrain ob s e r vations and exp e r iments along grad i e n t s . In addition, regional fluxes. Atmospheric transport across regional the confounding of control va ri able s , to g ether with var i- areas is sufficiently rapid that concentration changes will abi l i t y , means that significant replication must be obtained have to be resolved on the order of hours to a few days at least at some points along envi r onmental grad i e n t s . rather than months or years. Trace gas transport will need The role of land use must be a central subject in any to be represented at much higher spatial and temporal plan for terrestrial carbon research. In the annual terres- resolutions than at present,possibly using “observed” trial CO budget summarized above,it is evident that a sig- meteorological fields from four-dimensional data assimila- 2 nificant portion of the terrestrial fluxes is related to pres- tion systems,or by incorporating carbon fluxes into mod- ent and/or past land use. Additional evidence from atmos- els used in operational weather forecasting. Significant pheric and oceanic measurements suggests that most of improvements in the land-surface parameterizations used the ~1.8 Gt C/yr land sink may be occurring in the in numerical weather prediction would be required. Northern Hemisphere (Ciais et al.1995a). Some recent An objective of the global observational and inverse analyses suggest that an appreciable fraction of the total modeling system should be to provide meaningful integral terrestrial sink may reside in North America (Fan et al. constraints on spatially extrapolated estimates of carbon 1998,Rayner et al.1998). Inventory information is also fluxes derived by “upscaling”local fluxes using process- accumulating to suggest significant sinks of carbon in based models and remote sensing.These observing and North America,although the inventoried sinks are typical- modeling programs,described in the next section, would ly smaller than those suggested by the evidence from be extremely valuable in the context of top-down esti- atmospheric measurements. The vast majority of land in mates of flux derived independently from the global the United States and southern Canada was disturbed in observing program. the past and is managed,intensively or extensively for human use. North American carbon sinks,as well as uptake by European or Asian ecosystems,are strongly Terrestrial Observations, Experiments, affected by human activities. Studies of the role of land and Models use history in determining the fluxes are discussed below in this chapter in section “Goal 3: Land Use.” Studies to refine understanding of terrestrial CO2 exchange confront fundamental questions. What are the A deliberate sampling and experimental design is fluxes of carbon into today’s ecosystems? Which systems required,aimed at characterizing fluxes and processes are taking up how much carbon? What factors influence controlling carbon storage in forests, grasslands, agricultur- changes in past and contemporary ecosystem carbon stor- al lands and soils. The design should emphasize not the age (e.g.,CO2 itself, nitrogen deposition,other pollutants, identification of “typical”sites for a vegetation regime,but climate,management practices)? How has the rate of car- the identification of a network of sites within vegetation bon storage changed in the past centuries and decades? types that sample the principal axes of variation. These What systems and management practices cause net losses axes of variation would include not only climate and soils or gains of carbon? How will fluxes and storage of car- (Schimel et al.1997),but also disturbance type and time bon in the terrestrial biosphere change with changes in since disturbance. A high priority is the development of climate and the chemical composition of the atmosphere? new methods for measuring carbon fluxes belowground. Studies of terres t r ial carbon cycling must focus on A principal focus of studies within this network would systematic sampling strat e g ies designed to cha ra c t e ri z e be systematic observation of carbon exchange fluxes qu a n t i t a t i ve l y essential processes and to reject or confir m along environmental gradients. This is now possible to an sp e c i f ic hypotheses concerning responses along grad i e n t s unprecedented degree. The key gradients (e.g., climate, of principal controlling fac t o rs . Th e r e are sever al curren t nutrient deposition, forest age,plant functional types,land hypotheses concerning the particular ecosystems or use) can be defined. Rates of carbon exchange can be pr ocesses that take up CO2 in response to envi ro n m e n t a l measured. Previous efforts to estimate net CO2 exchange cha n ge . For exa m p l e , var iations in temperat u r e and soil have been hindered by pervasive small-scale heterogeneity mo i s t u r e (Dai and Fung 1993), growing season length in terrestrial carbon storage,and by difficulties in assess- (Myneni et al. 19 9 7 ) , N avai l a bility (Holland et al. 19 9 7 ) , ing changes in belowground carbon storage. Forest inven- CO 2 fer tilization (Friedlingstein et al. 19 9 5 ) , and fore s t tories represent a critical first step in quantifying storage, reg rowth (Tur ner et al. 1995) have all been suggested to be but they need to be upgraded to provide better informa- in vol v ed (see, e. g . , Goulden et al. 19 9 6 , Fan et al. 19 9 8 ) . tion on carbon,coordinated to help scale results from flux Although measurements along gradients are a power f u l stations and airborne regional measurements,and te c hnique for assessing ecosystem responses in systemati- in t e g rated to provide consistent, continental-scale estimates ca l l y differ ent conditions, in practice the fac t o r s that deter- of net carbon exch a n ge . Ed d y flux data are providing high mine the cha n g es along the gradient are confounded to resolution on cha n g es in carbon storage on the scale of 31 A U.S. Carbon Cycle Science Plan he c t a r es (e.g.,Goulden et al. 19 9 6 , Goulden et al. 19 9 8 ) . to providing information that ultimately can be used to Ho wever , fu r ther techn o l o g ical developments are needed to reduce uncertainty in estimates of current and future br ing down costs and improve the accessibility of the tech- regional carbon flux and storage. Such replication may nique and the rel i a bility of the method. (C o m m e r cial devel - depend to a large degree on the development of low-cost, opment of the techn o l o g y now appears to be underway.) stable, reliable,semiautomated instrument packages that Fin a l l y,he c t a r e- to kilometer-scale-resolution data can be will greatly reduce the manpower and logistical costs ext r apolated to reg ional (and ultimately global) domains associated with the measurements. using advanced remote-sensing techniques and veri fi e d The envisioned approach would combine process stud- th r ough expanded atmospheric concentration measure- ies and experiments (designed to increase basic knowl- ments and models described earlier in this cha p t e r . edge and improve predictive models) with flux measure- A networ k of flux measurements along wel l - d e fi n e d ments designed to allow models to be tested. A well- env i r onmental gradients provides sever al val u a ble prod - designed network of study sites would serve as a focal uc t s . Fl u x es can be direc t l y ext r apolated using area point for many different types of research. Process studies weighting from remote-sensing and inven t o r y inform a t i o n . on nutrient interactions,on feedbacks from species diver- Flux observations can provide infor mation about res p o n s - sity and changes to biogeochemistry, and on climate es to envi r onmental for cing (such as temperat u r e and soil effects are all needed. Experimental manipulations can mo i s t u re ) . Better understanding of the response to envi - help untangle complex mechanisms and test hypotheses ronmental for cing can then be used in ext r apolations and for ecosystem responses to conditions outside the current an a ly s i s . Flux measurements allow estimates of carbon envelope. For example,studies using preindustrial CO2 se q u e s t r ation from inven t o r ies to be compared to meas- levels can yield critical information on carbon storage ur ed carbon uptake. Fl u x es can also be ext r apolated using from past changes. Studies manipulating CO2, nitrogen mo d e l s . Estimates of seasonal or interan n ual var iations in deposition,and climate at sites at a range of times since flu x es can be compared to cha n g es infer red from atmos- disturbance are crucial for quantifying interactions among ph e r ic measurem e n t s . App l ying this approa c h to temporal these key global change drivers. It is essential that the var iations is especially importa n t . For exa m p l e , while inter- research network use common approaches and methods an n ual var iations in local climate and carbon flu x es may with the highest degree of standardization of methods and su g g est hypotheses about large-scale re g u l a t i o n ,t h ey pro- instruments as possible. Particular attention must be vide limited insight without direct large-scale mass-balance given to quality assurance in the operation of monitoring co n s t ra i n t s . Co n ve rs e ly,estimates of the large-scale atmos- equipment and conduct of manipulation experiments. ph e r ic CO2 seasonal cycle and sources and sinks provide a Formal protocols will certainly be required at an early vital global constraint on models. Ho wever , these estimates stage. Recognizing that new and better methods and pr ovide limited infor mation about the modeled mecha - instruments will be developed,the networks need to be nisms and sensitivity without greater spatial resolution and able to accommodate innovations so that the innovations pr ecision in estimated flu xe s . can be applied across the entire system,not piecemeal. The ability to apply eddy-covariance flux measure- The United States and other nations already invest sig- ments to regions will be limited by knowledge of errors in nificantly in natural resource inventories for management both temporal and spatial scales. Because the technique purposes. These inventories should be designed to pro- accumulates data at high frequency, there is essentially lit- vide better information on carbon. Only large-scale opera- tle problem in the temporal resolution of the specific tional inventories—such as those maintained by govern- measurements. It is critical,however, that the variability ment agencies—can provide data from the hundreds to of NEP over seasons and years be captured by continuous, thousands of sites needed for direct spatial integration. high-quality operation of each of the sites. It is in the spa- The inventory data need to be more effectively integrated tial domain that problems with regional measures of NEP with other sources of information,including eddy flux using the eddy-covariance flux method may arise. A and remote sensing. In addition,it is critical to extend the region (or biome) can be thought of as a unit of observa- inventory approach to cover the fate of carbon after it is tion from which samples can be drawn to allow the harvested from forests. This carbon includes not only region to be characterized quantitatively with explicit logging residues,but also manufactured products and statements of error for NEP in space. This characterization waste streams. can be achieved if eddy-covariance measurements are Finally, eddy-covariance flux time-series are needed to replicated within regions,not only across the axis of a par- provide closure on carbon budgets at key sites,to measure ticular gradient (as discussed above) but also normal to CO uptake or loss as a function of the location in the the gradient axis in order to quantify variance at each 2 experimental design. gradient level. The number of replications needed cannot be stated a priority as it is itself a research issue. The proposed terrestrial carbon research network Nevertheless,adequate replication is absolutely essential poses several significant challenges. For measurements 32 Chapter 4: An Integrated Carbon Cycle Science Program along gradients to have power in rejecting hypotheses or Manipulations including ecosystem-scale climate parameterizing models,large sample sizes are required. change,nitrogen additions,and elevated or decreased Today’s networks of flux sites number in the tens of instal- CO2 can provide critical insights on interactions,on lations. Globally this approach will require significantly responses to conditions in the past or the future,on more measurement sites,an arrangement that requires sig- ecosystem-to-ecosystem variation in responses,and on nificant prior investment in autonomous measurement interactions with other anthropogenic impacts,includ- technology and theory to make the technique simpler, ing harvesting,other land use change,biological inva- more robust,and less expensive. The program must be sions,and altered biological diversity. sustained for a significant period,because the measure- The next gen e r ation of manipulative exp e ri m e n t s ments become valuable only when reasonably long time- designed to understand and quantify the current terres - series have been collected,and they will become more tr ial sink should emphasize processes at the ecosystem valuable over time. s c a l e ,i n cluding responses of both biogeo ch e m i s t r y and Clearly, both the design of the initial network and its ec o l o g ical dyn a m i c s . Studies with prei n d u s t r ial CO2 ar e implementation will also require extraordinary care,statis- cri t i c a l , but will req u i r e new tech n o l o gi e s ,e s p e c i a l ly for tical rigor, and investment in technology. Enormous exp e r iments at the ecosystem scale. Ex p e r iments to resources could be expended on process studies, experi- quantify effects of multiple fac t o rs , alone and in combi- ments, flux measurements,and inventories without mate- n a t i o n ,a re also cruc i a l . Cu r ren t l y,we have little idea of rially reducing uncertainty about either today’s CO2 budg- the extent to which the carbon sink in for est reg rowt h et or simulations of future trends. For the nation’s scientif- in c ludes signals from increasing CO2, N deposition, or ic resources to be efficiently deployed,an initial synthesis climate cha n ge . Si m i l a r l y,we have no idea of the conse- and analysis of existing in situ data (including soil and sed- quences for carbon storage of veg etation cha n ge s , for iment surveys, forest inventories,and observations at exist- ex a m p l e ,i n c reasing shrub abundance in many of the ing Long-Term Ecological Research [LTER] and AmeriFlux wor l d ’ s grasslands (Archer et al. 19 9 5 ) . Ex p e ri m e n t s sites (LTER maintained by the National Science Fou n d a t i o n could be used to probe both the drive r s and the conse- and AmeriFlux maintained by the Department of Energy, quences of veg etation ch a n ge s ,e s p e c i a l ly exp e ri m e n t s National Aeronautics and Space Administration,and on ecosystems where the transitions can occur quickl y. National Oceanic and Atmospheric Administration), The next generation of experiments should also remote sensing,and model results is needed to define include pilot studies to evaluate deliberate carbon patterns of variability. Then,the research community must sequestration strategies. Issues concerning the limited be engaged in the effort to use this synthesis as a basis for spatial and temporal scale of manipulative experiments site selection in a network designed to understand pat- should receive intensive attention,so that lessons from terns at large scales. the experiments can be effectively interpreted and Once an analysis of existing data and models is done, a incorporated into global-scale models. critical set of measurements and experiments can be 2. Long-term terrestrial observations. E x p a n s i o n designed to efficiently sample the space identified. and enhancement of the LTER netwo rk will pay huge Different sets of measurements may be appropriate for dif- dividends by defining the ecological and current and ferent suites of sites. The design should take advantage of h i s t o rical land use fa c t o rs that regulate sequestra t i o n remote observations of land cover and land cover change, and release of carbon from major ecosystems. The net- which are quite comprehensive for the United States. wo rk expansion is envisioned to provide new sites, Similarly, inventory data are already available,and with sys- ro u g h ly equal in number to the ro u g h ly 20 curre n t ly tem and data management upgrades,management data ex i s t i n g ,s t ra t e gi c a l ly located to examine ecotones, o r may be made highly useful. With some effort in technolo- b o u n d a ry zones between re gions with diffe rent vege t a - gy and theory development,the present network of tions or biomes,like ly to play a significant role in re g u- roughly 20 flux measurement sites can be increased by a lating global CO . Enhancements are needed in two factor of 2 to 10. 2 d i m e n s i o n s . Q u a n t i t a t i ve ,e c o s y s t e m - l evel wo rk on car- Six other types of studies are proposed to complement bon stores and turn over should become a major com- the proposed flux measurements at network sites: ponent of each site, and LTERs should become key points for large-scale manipulations and for the 1. Experimental manipulations of CO , temperature, 2 expanded flux network . By doubling the budget and nitrogen,and other key controlling factors. Mani- number of sites in the current netwo rk , and adding pulations are an ongoing line of research that must be i m p o rtant new re s e a rch tasks, t h e re should be a enhanced. A number of specific questions about the s t rong synergy between the carbon focus and pre s e n t nature,quantitative importance,and persistence of e c o l o gical and pro c e s s - o riented goals of the LT E R s , mechanisms driving the current terrestrial carbon sink enhancing both. can be best addressed through direct experimentation.

33 A U.S. Carbon Cycle Science Plan

3. Intensified flux measurements at sites where terrestrial carbon cycling should exploit interaction detailed process studies are coordinated with with programs such as the World Climate Research eddy-covariance measurements. Intensive observa- Program’s Global Energy and Water Cycle Experiment tional studies can be conducted where carbon uptake Continental Scale International Project (GEWEX/GCIP). and many of its controls (N availability, soil moisture, This program has already demonstrated success in inte- microclimate,light) and mediating variables (Rubisco grating remotely sensed and in situ measurements of content of leaves,conductance,stem flow, below- energy and water fluxes. To the extent possible,carbon ground processes) are measured. In essence,such stud- research plans should be structured so that they can ies allow natural spatial and temporal variability to per- take advantage of improved satellite data products form the experiments that test hypotheses. expected in the near future. These products will Hypothesized control processes can be evaluated if a include new 30-meter Landsat Thematic Mapper- suitable suite of associated measurements (climatic, derived land cover data,high-resolution data sets atmospheric,and biological) is made. These studies expected in association with the ETA Mesoscale Model, share some of the advantages and disadvantages of and data products anticipated from the Mission to deliberate manipulations. The major advantage is that Planet Earth. Of particular interest is the possible appli- the perturbations are“natural,” including all time scales. cation of advanced algorithms to the upcoming Earth The disadvantage is that the conditions are not under Observing System (EOS) sensors to derive plant canopy control of the experimenter. This “natural approach”is functional properties. that followed by the present AmeriFlux network. 6. Integration of observations with model develop- 4. Flux scaling studies in which tall towers, bound- ment. Understanding terrestrial processes requires ary layer measurements (Convective boundary that ongoing observations be linked to the continued layer budgets) and aircraft profiles address the development and testing of models. It is extremely dif- scaling of land surface fluxes to their signatures ficult to develop terrestrial carbon models that include in the atmosphere. Opportunistic use should be state-of-the-art process representations for all of the made of tall transmission towers (e.g.,Bakwin et al. needed processes on multiple temporal and spatial 1995),which allow micrometeorological fluxes to be scales. Often the limitations of models serve as sign- determined over much larger footprints than typical posts in formulating and testing new hypotheses. canopy towers. Additionally, tall towers allow the verti- Examples of current model frontiers are the effects of cal profile of the eddy flux to be measured,testing scal- CO2 on a full suite of plant processes (including alloca- ing strategies by varying the footprint of the measure- tion of carbon to different parts of a plant), dynamic ment. With appropriate inclusion of meteorological interactions between carbon and nitrogen budgets, data collection (radar wind profilers or balloon son- hydrologic changes (such as drying or thawing of bore- des),tall towers also allow the direct measurement of al peat),and vegetation dynamics such as successional the local forcing of the atmospheric CO2 rectifier changes over long time scales. Models must also be effect,which will facilitate larger scale atmospheric improved to systematically incorporate information modeling. Aircraft observing campaigns should use about human and natural disturbance of the land sur- eddy-covariance sites as anchor points. These programs face. Current models emphasize physiological and bio- can be designed to provide flux estimates over a much geochemical processes and largely neglect the carbon larger footprint than that of even a tall flux tower, storage dynamics induced by cultivation, forest harvest, through measurements of continental boundary layer fire, fire suppression,and other intensive disturbances budgets for scales of several tens of kilometers (Chou as well as biological invasions, changes in biological 1999,Desjardins et al.1997,Lloyd et al.1996) to tran- diversity, and other ecological processes sects measured by flux aircraft for scales of tens to The chief requirement for the progress of modeling— hundreds of kilometers (ABLE-2/3 experiments, particularly in hypothesis testing—is better integration of BOREAS,FIFE2). These data allow a quantitative evalua- models and data. Model development is currently sup- tion of the relationship among intensive flux measure- ported by a variety of programs,including NOAA Carbon ments,land surface cover, and ancillary process data; Modeling Consortium (CMC), Terrestrial Ecology and and would facilitate the development of scaling Global Change program (TECO),Vegetation/Ecosystem strategies by providing spatially extensive snapshots Modeling and Analysis project (VEMAP),NSF’s Methods as context. and Models for Integrated Assessment (MMIA),and numer- 5. Remote sensing of terrestrial properties. Remote ous disciplinary programs. Although this diverse range of sensing must be a critical component of any plan for support encourages innovations,more effort is needed in terrestrial carbon research. Efforts to characterize integration. The assembly of observational data into

2Atmospheric Boundary Layer (ABLE);BOReal Ecosystem-Atmosphere Study (BOREAS);First ISCLIP (International Satellite Land Surface Climatology program) Field Experiment (FIFE) 34 Chapter 4: An Integrated Carbon Cycle Science Program

Photograph of the 447-meter tall WLEF-TV transmitter tower,Park Falls, Wisconsin. The tower is owned by the State of Wisconsin Educational Communications Board, and is being used for measurements of CO2 mixing ratios (see Bakwin et al., 1998) and atmosphere/surface exchange of CO2 by eddy covariance. Transmitter towers up to 610 meters tall are located in many areas of the USA. 35 A U.S. Carbon Cycle Science Plan standard forms that can be used by models is a significant directed toward developing a rigorous statistical integrative task,without which rigorous testing and com- approach to placement of sites for atmospheric moni- parison of models cannot be achieved. Thus, very differ- toring and manipulation experiments. ent models may appear to have equal “validity,” incorrect • A continued effort is needed on modeling,on the inte- hypotheses may not be rejected,and the improvement of gration of new experimental knowledge into models, models and hypotheses may be inhibited. Standard data and on sustained testing of models in increasingly rigor- sets are urgently required for terrestrial model input (cli- ous model-data comparisons. Models are required mate,soils,disturbance regimes,N deposition) at regional both for the integration of knowledge about the pres- and global scales. Without these standardized inputs, ent,and for prediction of the response of systems to model intercomparisons are meaningless. Standard data future changes. sets must be assembled in ways that are compatible with all models simulating particular processes and scales. • Enhanced manipulative experiments at the ecosystem There is ample precedent for this approach. In meteorolo- scale are critical for exploring ecosystem responses to gy, the production of reanalysis data sets has resulted in environmental conditions outside the current envelope long time series of physical variables that are critical in and for assessing responses to interactions among cli- evaluating climate models. These data sets might be mate,ecological processes,and anthropogenic changes. valuable as climate inputs to carbon models,but they Manipulative experiments should be designed to sup- must be checked for consistency with known carbon- port model development, facilitate scaling in space and energy-water relationships. time, evaluate proposals for managed carbon sequestration,and enhance broad communication about the role Recent Intergovernmental Panel on Climate Change of the terrestrial biosphere in the global carbon cycle. (IPCC) intercomparisons of global carbon-cycle models were made only after all the models were required to • Long-term ecological research,with greater emphasis meet consistency criteria based on input of a standard set on carbon-related issues,will provide the critical data of historical atmospheric CO2 and emissions data. Clearly, on ecological and land use historical factors regulating the systematic incorporation of newly understood mecha- sequestration of carbon. nisms in terrestrial models must be accompanied by model integration using high-quality standard inputs and rigorous consistency tests against an array of Goal 2: Understanding the Ocean benchmark data. Carbon Sink

The implementation of the terrestrial studies described Long-term goals for CO2 research in the ocean are, first, above might proceed as follows: to quantify the uptake of anthropogenic CO2 by the ocean,including its interannual variability and spatial dis- • An analysis of existing spatial data, model re s u l t s ,a n d tribution;and,second,to understand and model the remote-sensing products must be initiated as a basis for processes that control the ocean’s uptake of CO . Uptake an exp e r imental and observing system design that will 2 of anthropogenic CO can be quantified by measuring al l o w the testing of hypotheses and the ext r apolation of 2 either the flux itself or the resulting change in carbon si t e - s p e c i f ic studies. A wor king group with modest inventory. Both should be carried out,with a strong funding should be assembled to underta k e this activity. emphasis on disaggregating the global uptake into contri- The project should assess the U. S .d a t a , results and prod - butions from major ocean regions and monitoring tempo- ucts in detail and global infor mation at lower res o l u t i o n. ral variability. The process and modeling goals can be • The nation’s forest, agricultural,and aquatic monitoring attained by research on the rate-limiting steps for uptake, programs should be evaluated with the goal of identify- the causes of spatial and temporal variability, and the bet- ing low-cost/high-leverage enhancements to the exist- ter integration of models with data to predict long-term ing programs. Collaboration among agency scientists trends. These goals are similar to those for the terrestrial and managers and the broader carbon science commu- environment,but the major research challenges are dis- nity is crucial. tinct. Goals in understanding the ocean carbon sink are the following: • Flux measurements are an essential part of the program,because they may provide closure on carbon and • Develop new technology to facilitate systematic and water budgets,which is difficult with conventional lower cost long-term observations sampling,especially closure on belowground fluxes. • Carry out air-sea carbon flux measurements with a However, there are problems with the existing technol- near-term focus on the North Atlantic and North ogy and theory. Resources need to be invested to Pacific,to coordinate and integrate results most closely reduce uncertainties in measuring nighttime fluxes, with the terrestrial research on the Northern allow use in more complex terrain,and develop cheap- Hemisphere and tropics er and more autonomous systems. Effort needs to be 36 Chapter 4: An Integrated Carbon Cycle Science Program

• Conduct global surveys of oceanic inventories of fossil ¥ Init iat e ongoing pr ogr am t o r epeat global CO2 CO2 along with relevant tracers,including their evolu- and t r acer sur v ey s ev er y 10 t o15 y ear s t o mon- tion with time,in support of inverse calculations and it or t he oceanic CO2 inv ent or y and it s ev olut ion global models of carbon uptake in space and t ime. • Realize studies of the physical and biogeochemical d. Pr ocess st udies, models, and sy nt hesis processes controlling the air-sea flux of carbon in the oceans,including manipulation experiments and devel- ¥ Vigor ously pur sue ocean pr ocess st udies, opment of models. including manipulat iv e ex per iment s, t o impr ov e The following bullets summarize the proposed mechanist ic under st anding of pr ocesses cont r ol- program elements. ling car bon upt ake and t heir sensit iv it y t o climat e. This r esear ch includes dir ect measur e-

ment s of air - sea f lux of CO2, f act or s r egulat ing Goal 2: Maj or Pr ogr am Element s and Act iv it ies biological f lux es, and lar ge- scale t r acer r elease and t r acking ex per iment s t o def ine quant it at iv ely a. Requir ed new t echnology t he cont r ols on upt ake of ant hr opogenic car bon by t he oceans. ¥ Inst r ument s f or measur ement of CO and 2 ¥ Dev elop impr ov ed models of phy sical, chemi- r elat ed quant it ies on moor ings, dr if t er s, and cal, and biological pr ocesses t o analy ze pr ocess t ow ed v er t ical sampler s obser v at ions and t o pr oj ect how t hese pr ocesses

¥ Rapid w at er sampling t echniques may af f ect ocean upt ake of CO2 in t he f ut ur e. ¥ High t hr oughput mult i- element analy zer s f or ¥ Dev elop new basin and global ocean car bon shipboar d car bon sy st em measur ement s (of dis- cy cle pr edict iv e and inv er se models, including solv ed inor ganic car bon [ DIC] , dissolv ed or ganic impr ov ed pr ocess models, t o analy ze air - sea f lux mat t er , par t iculat e or ganic mat t er , alkalinit y , and ocean car bon inv ent or y and t r acer obser v a- t emper at ur e, salinit y , nut r ient s, O2, ) and r elat ed t ions and t o pr oj ect f ut ur e ocean upt ake of CO2. t r acer s (CFC’ s, 14C, 13C, et c). ¥ Dev elop t he use of r emot e- sensing t ools f or b. Air - sea car bon f lux es ocean monit or ing of phy sical and biological pr op- er t ies t o gain mechanist ic insight and t o ex t r apo- ¥ Apply new t echnology t o inst all an ex panded lat e local obser v at ions t o lar ger scales. net w or k of long- t er m st at ions, dr if t er s, and under w ay measur ement s, emphasizing acquisit ion of dat a f or r epr esent at iv e ensembles of oceanic Until recently, the primary tool used for studying the r egions and condit ions, t o def ine by obser v at ion oceanic uptake of anthropogenic carbon has been models t he spat ial dist r ibut ion and t empor al v ar iabilit y validated with observations of tracers,particularly the dis- of t he air - sea f lux . tribution of natural and bomb radiocarbon. The absence of carbon measurements to verify model estimates has ¥ Conduct f ocused f ield campaigns at basin scale been a serious limitation in testing our understanding of ov er bot h t he Pacif ic and At lant ic using air cr af t oceanic uptake. Recent improvements in the precision of sampling combined w it h shipboar d and gr ound- measurements of DIC and associated tracers such as O2 based measur ement s and impr ov ed at mospher ic have given us greatly increased confidence in estimating t r anspor t models. The goal is t o conf ir m and the inventory of anthropogenic carbon directly from DIC. r ef ine est imat es of t he magnit ude of oceanic New techniques discussed in Chapter 2 allow filtering out sour ces and sinks of CO2 f r om pCO2 (par t ial the background DIC concentration and the effects of sea- pr essur e of car bon diox ide) dat a, and t o pr ov ide sonal and interannual variability. These achievements t he t ools and dat a necessar y f or impr ov ed make it possible to detect the total anthropogenic inven- inv er se model est imat es of Nor t her n Hemispher e tory and its change over time scales on the order of a t er r est r ial sinks. decade. The measurements are also being used to estimate transport of carbon across sections in ocean basins. c. Oceanic inv ent or y measur ement s Both the tracer-based modeling approach and the invento- ¥ Sy nt hesize r esult s of r ecent global ocean car - ry method will continue to be important tools in monitor- bon sur v ey s in suppor t of global ing the oceanic uptake of anthropogenic carbon. car bon/ ocean/ at mospher e inv er se and pr edict iv e Nevertheless,neither modeling nor the inventory model dev elopment and planning f or f ut ur e global approach can give us the air-sea flux of CO2 at a given sur v ey s. time and place. These methods are thus of limited use in 37 A U.S. Carbon Cycle Science Plan determining the annual and interannual variability of the obtain with dedicated global measurement programs carbon cycle,and in constraining inverse models of atmos- using current methods. It is very important to assure the pheric observations. However, such spatial and temporal development of lower cost,more efficient methods for resolution is required for atmospheric inversions to detect collecting water samples and making at-sea measurements the Northern Land Sink as well as to better understand of the carbon system (dissolved inorganic carbon,dis- the processes that control the air-sea flux of carbon and solved organic matter, particulate organic matter, alkalinity, 14 its variation in time. The best way to obtain adequate spa- T, S, nutrients, O2, and related tracers such as CFC’s, C, tial and temporal resolution is by measuring the air-sea etc). A major need also exists for the continued provision flux. This can be done either directly using techniques of high-quality DIC standards,which was a major element that were recently tested successfully for the first time at in the success of the World Ocean Circu l a t i o n sea,or indirectly using measurements of DpCO2 multiplied Ex p e ri m e n t / J oint Global Ocean Flux Study (WOC E / J G O F S ) by an estimate of the gas exchange coefficient. The direct carbon measurement program. Standards should be devel- technique,though,does not lend itself to the large num- oped for other important tracers as well,with particular ber of measurements that would be necessary to obtain importance attached to carbon isotopes and nutrients. adequate temporal and spatial resolution. Its primary Autonomous measurement capability for carbon system application will be in the essential task of improving our parameters is also strongly needed. Time-series measure- understanding of air-sea flux processes. The indirect ments are presently limited to locations near island ports measurement technique is the most promising for obtain- such as Bermuda and Hawaii. The present U.S.ocean car- ing the time-dependent air-sea flux over a given period of bon-cycle time-series stations at Hawaii and Bermuda are time and region of the ocean. The application of the indi- run at an annual cost of about $1 million each. Activities rect flux measurement technique faces formidable obsta- supported by the operating budgets include studies of the cles,as discussed in Chapter 2. However, recent progress processes affecting the carbon cycle as well as monitoring in techniques for measuring DpCO and in our under- 2 of the carbon cycle per se. Significant improvements in standing of gas exchange and ability to study it gives the coverage and the economics of the carbon cycle mon- cause for optimism that air-sea flux measurements will itoring component of the present activities are possible. contribute significantly to our understanding of the global The scientific community presently has the technical carbon cycle over the next decade. capability to deploy instruments for measuring the air-sea In conjunction with measurements of the air-sea flux difference of pCO2 on commercial and research ships. and ocean carbon inventory, it is essential to study the There has been initial success in deploying reliable instru- physical and biogeochemical processes that control the ments on autonomous buoys (Friederich et al.1995). air-sea flux and its spatial and temporal variability. It is Further technical developments may allow the addition of not sufficient to know within rough bounds the global in situ monitoring of related variables such as nutrients rate of carbon uptake. Simulations of climate change and oxygen. show significant warming of the surface low latitudes as A major program of technological development is an well as freshening of the high latitudes. These trends essential component of ocean carbon research over the would increase the vertical stratification and have a major next 5 years. impact on ocean circulation. It is very likely that these changes will have a major impact on the transport of carbon from the surface ocean to the abyss,as well as affect- Air-Sea Carbon Fluxes ing the biological pump. Projections of future trends in atmospheric CO2 must be based on an adequate under- We recommend the following programs to characterize standing of relevant oceanic processes and the incorpora- the climatological uptake of CO2 by the oceans,the geo- tion of this understanding into models. Considerable graphic distribution of uptake of CO2, and, eventually, its work remains to be done in these areas,building on the interannual variability. promise demonstrated by recent progress. Time Series Observations. The air-sea flux of CO2 is The identified program elements are discussed in more extraordinarily variable in space and time due to changes detail in the following subsections. in circulation,temperature,and salinity, as well as biology. The key to determining this flux and understanding its variations is continuous in situ monitoring,including both Needed New Technology time-series stations and regular measurements along transects using ships of opportunity. The processes that control carbon cycling in the ocean exhibit large spatial variance,which itself is not stationary, Temporal variations in some areas,such as the but changes significantly on diurnal,seasonal,annual, Equatorial Pacific,have been identified as major causes of decadal,and even longer time scales. Comprehensive spa- variability in air-sea CO2 exchange. Existing time-series tial and temporal coverage would be very expensive to studies are located mostly in the subtropical ocean gyres, 38 Chapter 4: An Integrated Carbon Cycle Science Program

while there are major gaps in data on regions of active Thorough analysis of recent ocean carbon obser- ocean mixing and high biological variability, especially in vations. At present,the most robust constraint on ocean subpolar and polar latitudes. Temporal variability is great- uptake of CO2 comes from surveys and time-series of car- est in surface and subsurface layers,locations where bio- bon system measurements carried out with instruments logical and physical feedbacks are most likely to alter the and methods now available and analyzed by techniques ocean’s ability to absorb CO2. Characterization and like those discussed in Chapter 2. Essential complementa- understanding of temporal variance is a prerequisite for ry measurements include those of carbon isotopes,the understanding the processes that limit rates of ocean standard hydrographic and nutrient measurements,and CO2 uptake. transient tracers of ocean circulation. These surveys and time-series measurements provide direct in situ con- Mo o r ed and underway time-series observations should, straints on ocean CO uptake that additionally provide wh e n e ver possible , map a ran g e of prop e r ties associated 2 strong constraints on the role of the terrestrial biosphere with biological prod u c t i v i t y . Extending measurem e n t s in the global carbon budget. be yond pCO2, te m p e ra t u re , and salinity (for example) can giv e both rate and process infor mation about biologic a l Over the past decade,the WOCE/JGOFS programs have pro d u c t i v i t y . Co n c e n t r ation measurements of CO2 an d provided an invaluable baseline on the current state of the nut r ients are val u a ble in this reg a rd . Pr ecise measurem e n t s ocean that will be useful for assessing future changes. of O2 co n c e n t r ations in the mixed layer , coupled where Unquestionably, however, survey and time-series measure- po s s i b le with measurements of inert gas concentrat i o n s ments must continue beyond the WOCE/JGOFS era to and wind speed, yield infor mation on net production and answer fundamental questions,such as whether oceanic ven t i l a t i o n . The measurement of isotopic prop e r ties give s CO2 uptake is increasing or decreasing globally in the 13 rate infor mation of interes t : the d C of CO2 co n s t ra i n s future. Such work is the analogue of the current network net production (Zhang and Quay 1997), while the trip l e of atmospheric measurements,but substantially more diffi- isotope measurement of dissolved O2 co n s t r ains gross pro- cult to implement. To plan the best possible program of duction (Luz et al. 19 9 9 ) . Me a s u r ements of atmospheric continuing ocean observations,needed spatial resolutions O2 co n c e n t r ation giv e val u a ble infor mation about instanta- must be determined,along with which variables to meas- neous and annua l l y aver aged ocean carbon flu xe s . ure at what accuracy. A high priority over the next few On g oing flask sampling can ref lect interan n ual vari ab i l i t y years must be to use the current vastly increased oceanic and long-term trends in ocean carbon flu xe s . Co n t i nu o u s data set to answer these questions and to design an opti- co n c e n t r ation measure m e n t s ,n ow becoming possible , mal survey program. should be developed and used for more detailed studies. Monitoring the evolving fossil CO2 inventory in Focused Field Campaigns. An important lesson of car- the oceans along with related tracers. Monitoring the bon cycle res e a r ch has been how muc h knowl e d g e of one evolving fossil CO2 inventory is an essential long-term carbon cycle component depends on knowl e d g e of others. component of any effort to understand oceanic CO2 One of the major constraints on the magnitude of the ter- exchange. Because the costs of conducting comprehen- res t r ial sink and its spatial distribution has been knowl e d g e sive surveys are presently very high, we propose that the of the air-sea carbon flux and its spatial distrib u t i o n . In par- needed measurements be accomplished through a ti c u l a r ,ad d r essing the North e r n Land Sink hypothesis will reduced-level but continuous effort rather than the global req u i r e focused field campaigns over both the North “snapshot”paradigm of previous expeditions. This activity Atlantic and North Pac i f ic to estimate the air-sea flux of car- would be an ongoing and focused on specific limited bon during the time terres t r ial flu x es are analyz e d . Th e s e areas each year. The goal would be to cover the entire campaigns should include aircr aft sampling as well as ship- world every 10 to 15 years. This approach would cost less bo a r d and autonomous measurem e n t s , and will req u i r e per year and would assure maintenance of the required im p r oved atmospheric tran s p o r t models for analys i s . measurement expertise and capability for the long term. The ongoing survey could exploit strong linkages with Fu t u r e field campaigns should focus on areas of crit i c a l other efforts such as CLIVAR (Climate Variability and im p o r tance to our understanding of the global carbon cycle . Predictability programme) and GCOS (Global Climate In parti c u l a r ,the Southern Ocean air-sea flux is pres e n t l y Observing System) to make efficient use of ship time. It is po o r l y constrained and its temporal vari a bility is unknown . thus critical to ensure that tracers relevant to the carbon cycle (e.g.,DIC,O2, temperature,salinity) are covered in Oceanic Inventory Measurements programs like CLIVAR that have the goal to monitor temporal trends in ocean properties. We recommend two programs to characterize the The measurement suite should include total alkalinity oceanic inventory of fossil CO and its evolution 2 and should frequently include a third CO -system proper- over time: 2 ty such as pH to assure internal consistency. The

39 A U.S. Carbon Cycle Science Plan measurement of 13C and TOC [total organic carbon] are One challenge is characterizing marine ecosystems well es s e n t i a l , as are the development and use of standards . enough to model their effects quantitatively on air-sea car- Me a s u r ements should include transient tra c e rs ,w h i ch pro- bon fluxes. Process studies include direct manipulation of vide temporal infor mation about ocean mixing and wat e r the environment,such as the large-scale iron enrichment studies (Coale et al.1996),as well as systematic observa- mass history that is essential to interpreting fossil CO2 di s t r i- bu t i o n s . Fin a l l y, th e y should include bioactive che m i c a l s tions of the effects of “natural experiments,” such as El su c h as iron whose oceanic distribution is poorly known Niño. Process studies,and in particular ecosystem-level but is likel y an important influence on ocean carbon flu xe s . manipulations (along the lines of the recent iron fertilization studies) should be used to evaluate the host of bio- A complementary approach exploits the fact that the logical mechanisms proposed as potential feedbacks on invasion of anthropogenic CO2 into the ocean looks,in ocean carbon uptake (Denman et al.1996). Possible feed- aggregate,like a vertical transport problem that can be backs that can be tested include modifications of iron fer- effectively attacked through studies of the distribution of tilization rates of HNLC (high-nutrient,low-chlorophyll) transient tracers. Various tracers are now available to span 3 3 14 regions,particularly in the Southern Ocean;subtropical a variety of mixing scales ( H/ He, C,CFC-12,SF6, nitrogen fixation rates;Redfield carbon to nutrient ratios HFCs). Some of these tracers reveal mixing over the criti- of export material;community allocation of sinking versus cal longer (decadal and century) time scales;and some suspended/dissolved material;calcification rates;and sub- help identify current short-term invasion rates for compar- surface remineralization depth scales. The effects of ison with older data. Large-scale tracer release experi- coastal eutrophication and elevated UV radiation have also ments provide an independent means to assess vertical been hypothesized as significant,but are rather poorly transport rates and mechanisms. Likewise,direct eddy constrained. flux measurements are just starting to become available to test parameterizations of air-sea exchange. An emerging research theme is the relationship between ecosystem structure and the export flux of carbon. Ecosystems dominated by the microbial loop recycle Process Studies, Models, and Synthesis carbon efficiently and show very little export of carbon, whereas other regions dominated by diatoms and other We recommend the following four programs: large phytoplankton tend to show a large carbon export. Selected studies of processes that control the It is important to understand how changes in ocean stratification and circulation in response to changes in climate mechanisms and rates of air-sea CO2 exchange, vertical transport within the seas, and cycling of DIC will affect the ecosystem structure and uptake of anthro- and alkalinity. Ocean process studies yield understand- pogenic carbon. ing of the basic mechanisms controlling the ocean carbon Improved process models. Mathematical models are cycle. Process submodels are also required in models of required to express the relationships between physical the time-dependent air-sea flux. Ideally, process studies forcing and biogeochemical processes,to evaluate the and observational systems should be linked as intimately implications of results and observations for upper ocean as possible to help determine the response of ocean car- pCO2 variations and ocean carbon uptake. Seasonally bon fluxes to interannual variability in climate forcing,and resolved models of the circulation and physical properties to identify feedback mechanisms. Some important of the upper ocean will improve the ability to take into processes that need to be studied are gas exchange and account the influence of temperature,salinity, and trans- the physical processes that determine the rate of anthro- port on sea surface pCO2.To consider interannual and pogenic carbon transport from the mixed layer into the longer term variability, as well as the effects of climate thermocline and deep ocean. Also important is the role of change,is essential to explore how changes in sea surface biological processes in determining the spatial and tempo- temperature,ocean circulation,stratification,and sea ice ral variability of air-sea fluxes and anthropogenic carbon formation all affect the atmosphere-ocean balance of car- uptake. A major issue is the sensitivity of physical trans- bon. Biological models embedded into models of physical port and biological processes to ocean properties that are circulation will help further explain the large role that sea- likely to change with climate change. Knowledge of seasonally and interannually forced biological processes play sonal and interannual variability as well as direct manipu- in regulating upper ocean chemistry and CO2 partial pres- lation experiments will be an important source of infor- sure in the mixed layer. Models with sufficient resolution mation. A deeper understanding of what controls the bio- to simulate meso-scale processes will be needed to study logical pump in the ocean is required,including the role interactions that determine many of the ecosystem of micronutrients such as iron and zinc,whose input to processes. A close collaboration with physical oceanogra- the ocean may be affected by climate,land use,or pollu- phers studying ocean variability is an essential component tion. Program oversight must assure that priority is given of this program. to studies that can make a genuine contribution to understanding future atmospheric CO2 levels. 40 Chapter 4: An Integrated Carbon Cycle Science Program

Improved basin and global carbon cycle models. use of satellite altimeter and scatterometer data to A new generation of models must be developed for long- increase spatial and temporal coverage,as well as in situ term projections of carbon uptake,integrated with the studies to develop appropriate equations relating transfer observations to allow data to be assimilated and to test velocity and surface roughness. An essential complement the models. Conversely, models must be able to simulate to satellite estimates of net carbon fluxes comes from the distribution of important quantities that can be tested atmospheric O2/N2 data. These provide a fundamental in the field and must otherwise help in designing field constraint on seasonal new production (that portion of experiments. For example,because variability in ocean primary production that does not represent internal hydrographic structure and ocean carbon cycling is in mixed-layer recycling) on a hemispheric or basin scale, most instances associated with circulation changes, varia- and the interannual variability in these rates. tions of carbon and hydrographic parameters will be cor- Current estimates of mean ocean primary production related. An analysis that does not account for this covari- must be improved at basin to global spatial scales and ance can be seriously in error, and a model that does not daily to interannual temporal scales. Satellite-derived simulate the observed covariance will be wrong. Such fields of near-surface phytoplankton chlorophyll concen- problems will require a synthesis activity involving both trations are essential to developing the calculations and model simulations and measurements. Program structures models needed to do this (Behrenfeld and Falkowski must be implemented to force both modelers and experi- 1997). An essential focus of this work should be the menters to make this integration happen. assessment of new production. New production is more Remote-sensing observations and numerical models are closely related to the net flux of CO2 from surface to deep a natural marriage inasmuch as ocean in situ observations waters than is total primary production. The relationship are often too sparse to initiate or validate model fields. between new production and net primary production is Satellite data are often better matched to the time and not a simple proportionality. Estimating new production space scales of models,but satellites generally measure requires knowledge of nutrient budgets of surface water, only very simple indices of biological and physical as well as net primary production. Direct measurement of processes and distributions. Thus,models are necessary to nutrient concentrations is not possible from space,but extrapolate simple satellite indices to more complex bio- sometimes concentrations of major plant nutrients such as logical and physical processes and distributions consistent nitrate can be related by regression to sea surface temper- with the accuracy provided by in situ measurements and ature on regional scales (e.g.,Kamykowski and Zentara as required for carbon cycle research. 1986). Mineral aerosols are one of the primary sources of iron and other trace elements that limit new production Remote sensing of ocean conditions and and net primary production in some important ocean processes. Remote sensing is an important tool for regions,including the Southern Ocean. Aerosol plumes extrapolating measurements and calculations in time and and sea surface temperatures are both detectable by space space and for providing estimates of key environmental sensors.It may therefore be possible to use satellite data parameter values needed by coupled physical and biogeo- to help quantify nutrient budgets in the upper ocean,and chemical ocean models. Remote-sensing observations can hence rates of new production. also provide critical information during process studies and large-scale manipulation experiments. Advanced ocean color sensors will help improve estimates of global primary production. For example,the One important application of satellite data is to relate satellite MODIS (Moderate Resolution Imaging the CO transfer velocity across the air-sea interface to 2 Spectrometer) and MERIS (Medium Resolution Imaging “sea surface roughness”on ocean-basin scales. Spectrometer) will provide estimates of chlorophyll fluo- Accumulated evidence suggests that estimates of wind rescence,which can then be used to estimate phytoplank- velocity alone are not sufficient to estimate transfer veloci- ton quantum efficiency or at least the phytoplankton pho- ty, owing to other factors that also influence surface toadaptive state. The additional spectral bands and higher roughness. These factors include the sea surface wave signal to noise ratio of these advanced sensors will also spectrum and resulting surface renewal and near-surface provide some capability to distinguish phytoplankton turbulence (Jähne et al.1987),which are affected by sur- functional groups and thus some capability to classify factants (Frew 1997) as well as wind forcing. In particu- ocean ecosystem structure on large scales. lar, small-scale waves (small gravity and capillary waves) are believed to play a primary role in promoting gas Se ver al advanced ocean color sensors are planned by exchange,and some of the scattering properties of these NASA and other international space agencies for the next waves are observable with space radar sensors (scatter- fiv e ye a rs , but plans are ver y uncertain for the post-2005 ometers and altimeters). One focus for future efforts ti m e f ra m e . By the end of the next decade, basic ocean should be understanding the relationship between radar color measurement capability may be tran s fe r red to opera- backscatter from small-scale surface waves and the trans- tional satellites (e.g., the NOAA / D O D / N ASA National Pol a r - fer velocity. This work will require the complementary Orbiting Operational Envi r onmental Satellite System, 41 A U.S. Carbon Cycle Science Plan

N P O E S S ,p ro j e c t ) , but it is not clear if the full capabi l i t i e s cropland, much of it unrecorded,is presumed to be a anticipated from MODIS and MERIS will continue beyon d major contributor to deforestation (FAO 1995). The the middle of the next decade. Co o p e r ation and coordi n a - unrecorded tropical cropland expansion tends to be high- tion among international space agencies should be encour- ly fragmented,making its monitoring by remote sensing aged to ensure the continuing remote-sensing capabi l i t i e s difficult with all but the highest resolution sensors. req u i r ed to support future carbon res e a r ch prog ram s . Afforestation in the Northern Hemisphere mid-latitudes is a significant component of the Northern Land Sink, yet there is a lack of uniformity in the criteria used to enu- Goal 3: Assessing the Role of Land Use merate afforested area and define associated carbon flux densities (Nilsson and Schopfhauser 1995). A major effort Chapter 2,in its final section,discussed the importance is needed to assemble land use histories for both tropical of land use to the overall carbon cycle,as well as to under- deforestation and Northern Hemisphere afforestation. The standing the time history and spatial distribution of car- basis for such histories includes a variety of potential data bon sources and sinks. Goal 3 of the CCSP is therefore to sources (e.g.,national censuses,provincial and local land establish accurate estimates of the impacts of historical use surveys and tax rolls, remotely sensed imagery, and and current land use patterns and trends on the evolving proxy measures such as changes in local population and carbon budget at local to continental scales. The follow- technology regimes). These land use histories must be ing are the major program elements and activities pro- spatially explicit (of county-level or finer resolution, posed to move toward this goal. depending on data availability) and should extend as far back in time as needed to capture the current impacts of Goal 3: Maj or Pr ogr am Element s and Act iv it ies past disturbances on carbon fluxes in a particular region. ¥ Document t he hist or y of agr icult ur al ex pansion A wide range of historical data have been applied to and abandonment and it s impact on t he cont empo- the estimation of CO2 emissions from land use (e.g., r ar y car bon balance of Nor t h Amer ica. Conduct Houghton et al.1983,Melillo et al.1988, Foster 1993). int ensiv e hist or ical land use analy ses acr oss Recent progress has focused on the merging of many env ir onment al gr adient s, in coor dinat ion w it h t he kinds of historical records with satellite-derived measures pr oposed Nor t h Amer ican net w or k of land cover, and on the use of increasingly sophisticated of eddy f lux measur ement sit es. models of ecosystem response (e.g., Foster 1995). This approach constrains historical reconstructions to be con- ¥ Impr ov e obser v at ional capabilit ies in t r opical sistent with the geographic distribution of current land r egions t hr ough cooper at ion w it h t r opical nat ions, cover, and provides the spatial resolution needed to test including int er nat ional ex pedit ions and ex changes hypothesized sources and sinks. As high-resolution atmos- and t r aining of scient ist s and manager s. pheric sampling continues and expands in the future, records of CO2 and its stable isotopes,combined with 14 ¥ Apply t he obser v at ional pr ogr ams discussed records of C,CH4, and other indicators of particular land under Goal 1 in t he t r opics, including long- t er m use effects (e.g., N2O),will become increasingly valuable obser v at ions and modeling. as historical constraints. Of particular concern in rec o n s t r ucting historical emis- ¥ Couple socioeconomic analy sis of land manage- sions is the question of a pre- a gric u l t u r al steady state. ment st r at egies w it h biogeochemical models t o Although ice core rec o r ds indicate that the global CO ev aluat e t he social and biophy sical v iabilit y of 2 bu d g et was near a steady state for thousands of yea r s f ut ur e car bon sequest r at ion opt ions. be fo r e the 19th century,this evidence does not assure that pa r ticular reg ions wer e not net sources or sinks for car- bo n . For ex a m p l e ,s ever al studies have suggested that high- The emphasis of the proposed program elements and latitude soils wer e still sequestering carbon as part of their activities is the tropics and North America. As discussed rec o ver y from the most recent deglaciation (Billings et al. in the final section of Chapter 2,the tropics are important 1 9 8 2 ,H a rden et al. 19 9 2 ) . Estimates of human emissions because of the magnitude of the postulated deforestation ty p i c a l l y assume a pre- a gric u l t u r al steady state (e.g., source in this region. The focus on North America com- Houghton 1993), wh e r eas direct assessments of CO flu xe s plements the proposed terrestrial sink program element 2 ma y ref lect local non-steady-state effec t s . T h u s ,i n t e grat i o n by providing information on the history of the observa- of emissions estimates and flux observations must inclu d e tion sites. cautious consideration of steady-state assumptions. The data needed to construct reliable land use histories Estimates of human CO emissions must account for have temporal and spatial gaps over large parts of the 2 CO fluxes associated with materials removed to locations world. In the tropics particularly, the expansion of 2 other than the sites of original growth or burial. The need 42 Chapter 4: An Integrated Carbon Cycle Science Program to account for the fate of forest products is widely recog- range from qualitative conceptual models to schemes for nized (Houghton et al.1983). However, confusion persists connecting or extrapolating observations,to simulations about what oxidation of products from previous harvests of multifactor processes. Models based on all these should be reported in accounting for the net effects of approaches can provide powerful tools for asking “what- forestry activities (Houghton 1993). Recently, Stallard if” questions about the consequences of future events. (1998) has suggested that eroded soil carbon constitutes Manipulative experiments can provide a mechanism for another “off-site” form of carbon that must be explicitly direct tests of critical model predictions at a range of spa- included in carbon budgets. Because agricultural activities tial and temporal scales. are known to have vastly increased historical rates of erosion in northern temperate latitudes,the fate of eroded carbon is directly relevant to testing the Northern Land Models Sink hypothesis. The following provides an overview of the types of models and experiments needed for prediction,including Goal 4: Improving Projections of the necessary scientific foundations for credibility. Future Atmospheric CO2 Models for Data Analysis. A vast array of observations of the natural carbon cycle and results from experi- Better projections and information are needed to meet ments that manipulate controlling variables will emerge future societal needs: from the proposed program. Models designed for data assimilation and analysis represent a critical step in devel- Goal 4: Maj or Pr ogr am Element s and Act iv it ies oping more refined experiments.Inverse and diagnostic models (the latter used to probe data sets) are the bridge ¥ Impr ov e r epr esent at ion of phy sical and biologi- between the data collection,conceptual understanding, cal pr ocesses in models of t he car bon cy cle t o and predictive models. For example,tracer transport mod- r ef lect obser v at ions and new know ledge mor e els use atmospheric obser vations to test rates of transport accur at ely . derived from general circulation models (GCMs) or assimilated meteorological fields. Inverse models use atmos- ¥ Pr ov ide a f r amew or k f or r igor ous analy sis of pheric and oceanic observations of concentrations to esti- obser v at ion and independent compar isons and mate surface fluxes,providing a test for hypotheses about ev aluat ions of climat e and car bon models using the time-space variability of fluxes.Inverse modeling thus compr ehensiv e dat a and bot h f or mal and inf or mal forms a link between data,theory, and prediction,provid- model assessment act iv it ies. ing constraints on today’s fluxes that should be consistent in models of the potential future. ¥ Dev elop new gener at ions of t er r est r ial bios- pher e and ocean car bon ex change models, includ- Advanced Component Models and Scaling. ing t he r oles of bot h nat ur al and ant hr opogenic Process and theoretical studies lead to new understanding dist ur bances, succession, and t he f eedbacks and,in turn,improvement in model output,as mecha- t hr ough climat e change and ant hr opogenic per t ur - nisms are incorporated into models. The impact of bat ions t o car bon diox ide and nut r ient s. processes on the whole system must be evaluated by comparing the improved models to earlier formulations ¥ Dev elop coupled ear t h sy st em models incor po- and to observations,and by conducting sensitivity analy- r at ing t er r est r ial and oceanic biogeochemical ses to determine their importance. pr ocesses in climat e models of t he at mospher e Some new scientific results must be scaled in time and and ocean. s p a c e . For ex a m p l e , the re c t i fier effect (ag a i n , the cova ri- ance between diurnal and seasonal cycles of CO n e t ¥ Pr edict ions t he f ut ur e ev olut ion of CO concen- 2 2 exch a n g e and rates of atmospheric mixing) can be under- t r at ions, f or use in ev aluat ing t he consequences stood by micro m e t e o ro l o gical and plant phy s i o l o gi c a l of societ al decisions. t h e o ry. H oweve r, as it occurs on finer spatial and temporal scales than those re s o l ved in global models, a n The foundation for predicting future changes in the a p p ro a ch for ag gregating the effects of the re c t i fier has to carbon cycle is understanding the mechanisms that regu- be deve l o p e d . S i m i l a r ly, the biochemical effects of CO2 late it,the way those mechanisms respond to environmen- on photosynthesis have been understood for over a tal changes,and the ways those mechanisms interact. Like d e c a d e , but the impact on long-live d , whole plants, p l a n t current understanding,advances in the future will be c o m mu n i t i e s , and ecosystems is uncl e a r. To ach i eve the based on a combination of observations,manipulative o b j e c t i ves listed for Goal, 4 a vigo rous pro gram linking experiments,and synthesis via models. Useful models will p rocess studies to modeling must be maintained for the 43 A U.S. Carbon Cycle Science Plan n ew science to be incorporated in local and global Earth system models build on the knowledge of mecha- p re d i c t i ve models. nisms gained in process studies and advanced component modeling studies,and on the evaluation of knowledge Models for Decision Analysis. The great majority of derived from diagnostic studies and historical and paleo- terrestrial systems are currently managed. Management records. They will allow us to examine questions about choices have implications for carbon storage (e.g., choices the relationships among policy choices,sources and sinks, such as no-till agriculture, fire suppression,length of forest and resulting concentrations. These models demand rigor- harvest rotation,and fertilization). Carbon storage is one ous examination of knowledge, extensive computing factor to consider in making land management decisions, resources,and sustained efforts of testing and refining and reliable tools are needed for evaluating the carbon using integrated knowledge from the Carbon Cycle storage consequences of various land management scenar- Science Program overall. ios. There is an urgent need for local- or regional-scale models in agriculture, forestry, urban ecosystems,and All of the above models must be painstakingly com- marine systems that can be used to evaluate the effects of pared to observations, requiring strong commitment to specific interventions (e.g.,no-till agriculture,deep ocean testing and refining models. It can be very difficult to test CO2 injection) to enhance carbon storage,along with large-scale models against real data sets,obtained in specif- assessing associated environmental and economic conse- ic locations and at specific times. A coordinated effort to quences.These models will form a link between the basic promote rigorous comparisons of models and measure- science of the carbon research community and the needs ments is needed on the part of the agencies involved in of stakeholders and decision makers. the CCSP. Carbon System Models and Carbon-Climate Models. The carbon system is intimately coupled to the Experiments physical climate system. Indeed,the focus on carbon derives from the role of CO2 as a greenhouse gas. To proj- Experimental studies on the future of the terrestrial ect the large-scale consequences of changes to the carbon carbon cycle should be used to probe controlling process- cycle,it is necessary to link knowledge of the carbon sys- es,especially when the nature of the controls may be tem to that of the climate system. For example,if obscured by spurious correlations or long time constants. enhanced concentrations of greenhouse gases disrupted These studies should explore the behavior of key mecha- oceanic circulation,the ocean carbon system could be nisms to forcing outside the range of current ambient con- affected by changes in both physics and biology. These ditions,and evaluate responses to forcing from simultane- changes would help shape the effect of any anthro- ous changes in multiple environmental and ecological fac- pogenic increase in CO2 emissions on atmospheric con- tors. In addition, well-designed experiments can be pow- centrations,thus modulating the climate response. erful tools for making results from carbon cycle research accessible to a broad range of stakeholders. Manipulative Similarly, feedbacks could occur through terrestrial sys- experiments are,however, complements to models,not tems,with terrestrial carbon exchange possibly modified full alternatives. The processes controlling the future ter- by climate then affecting atmospheric CO , in turn modi- 2 restrial carbon cycle operate over a broad range of spatial fying the climate impact of any given anthropogenic emis- and temporal scales. Many processes with the potential to sion scenario. These feedbacks are presently uncertain, dominate future changes in terrestrial carbon storage—for depending on the rate,magnitude,and spatial distribution example, changes in the tree species that dominate forests of climatic changes. For example,if climate changes are or changes in the frequency of wildfire—are beyond greatest in peatlands at high latitudes,CO emissions 2 access through short-term,small-scale experiments. Even would likely increase. More equable warming could result if it were feasible to make an aggressive commitment of in either increases or decreases of carbon storage in the resources to work at greater temporal and spatial scales, terrestrial biosphere. manipulative experiments lose their utility as the time To analyze the potential effects of future industrial and scale of the experiment merges into the tempo of global land use emission scenarios,Earth system models are changes,the far from fully replicatable grand experiment required that couple physical models of the atmosphere, at the global scale. ocean,and cryosphere to biogeochemical models of ter- Manipulative experiments can provide powerful access restrial and marine ecosystems. Earth system models pre- to key uncertainties about the future trajectory of the dict partitioning of emissions of CO (and other green- 2 global carbon cycle. However, many aspects of the terres- house gases) between the mobile terrestrial,oceanic,and trial carbon cycle are very difficult to study with manipu- atmospheric reservoirs,and then predict climatic lative experiments. Consequently, the research targets response and the effects on biophysical processes in the need to be selected with careful attention to the likely land and ocean. importance of the focal processes,the efficiency with 44 Chapter 4: An Integrated Carbon Cycle Science Program

Interannual variations in temperature, the Southern Oscillation Index (SOI) (which correlates with El Niño cycles), and the growth rate of CO2. Recent work has suggested that observed variations in the growth rate of atmospheric CO2 may bear the signature of climate effects on ecosystems and the oceans. Understanding the mechanisms that have caused global terrestrial and ocean carbon exchange to vary over the period of observation provides a vital test for the “scaling”of process level knowledge and a basis for predictive modeling.

45 A U.S. Carbon Cycle Science Plan which the experiments probe these processes,and the ments should focus explicitly on the way that differences feasibility of obtaining the critical information using other in ecosystem characteristics influence responses to global approaches. In addition, experiments must be designed changes. For example,matched manipulations on sub- for effective integration with observational studies and strates of different age or with ecosystems of different bio- models. The following approaches should be pursued logical diversity can greatly enhance the ability to general- and enhanced. ize. Some of these experiments should focus on the interface between land use and atmospheric change,with Natural Experiments. Frequently, nature presents a manipulations of CO , temperature,and other environ- ready-made experiment,providing investigators an oppor- 2 mental factors at sites of different ages since disturbance, tunity to study ecosystem responses to a major environ- with different management practices,and with different mental forcing. Natural experiments are often permanent levels of anthropogenically stimulated biological invasion. or at least persistent features of the landscape,allowing Multifactor experiments should address a broad range of unique access to responses with long time constants,usu- mechanisms that may influence future carbon storage. ally with little or no cost for maintaining the natural These mechanisms range from direct responses of plant manipulation. Recent studies with ecosystems near natu- growth,to enhanced success of one or more plant or ral CO vents are a good example of the applicability of 2 microbial species,to altered decomposition,to changes in these features for understanding the future of the global sensitivity to fire,pests,or pathogens. Integrating results carbon cycle (Raschi et al.1997). The Hawaiian lava flows from these multifactor experiments in models and general with a range of different ages and local climates represent interpretations will present many challenges,but it is a parallel example for basic ecological research that is fun- much wiser to confront the challenges than to ignore damentally relevant to the carbon cycle (Chadwick et al. the processes to which these experiments provide 1999,Torn et al.1997). unique access. Natural experiments are invaluable links between long- Experiments with Model Ecosystems. The agenda term observations and purposeful manipulations. They of multifactor experiments is challenging,complex,and are natural system analogs of research designed to take essential. Many of the processes that may play critical advantage of recent patterns of land use change. Because roles in ecosystem responses have large spatial scales and these experiments provide unique opportunities to long response times. This circumstance places a high pri- observe long-term,and sometimes large-scale, responses to ority on optimizing the trade-offs between experimental controlling factors,they should receive greater emphasis. tractability and relevance to the ecosystems with the Natural experiments that should be evaluated include not largest potential impact on the carbon cycle. As in many only CO vents and gradients of substrate age,but also 2 fields of science,progress can be rapid if a sizeable frac- steep,topographic gradients of temperature and precipitation of the resources is invested in studies on experimen- tion, geothermal features,and sites with unusual flora and tal models. These are systems chosen with an emphasis fauna as a result of isolation. Natural experiments form a on tractability and access to the key processes,and with continuum with gradient studies. Some of the most pow- much less emphasis on the model system’s quantitative erful uses of the natural experiment approach may lie in contribution to the global carbon cycle. For example, combining common gradients with much less common studies on annual grasslands can address multiple global natural experiments. change factors while allowing reasonable replication and Multifactor Manipulations. Results from global large numbers of individuals per replicate. For other pur- change experiments are often difficult to interpret in poses,studies on artificial ecosystems in meso-cosm facili- more general terms. Ecosystems targeted by different ties may provide a useful balance of tractability and rele- experiments typically differ in a number of features,creat- vance. With sufficient resources and planning,useful ing a barrier to partitioning control for the different experiments on model systems might operate on a large responses among the different features. Multifactor scale,addressing, for example,the relative success of manipulations can play a critical role in enhancing the grassland species and rapidly growing trees under a range foundation for generalization and reducing uncertainties of manipulations. about future responses. The next generation of multifac- A res e a r ch prog ram with an emphasis on model systems tor experiments should address two kinds of questions. will req u i r e careful planning and coordination acros s First,how do ecosystem-scale responses to global change exp e ri m e n t s . Scaling interpretations from the models to components change when the global changes occur in the systems with the most direct rel e vance to the carbon combination? Experiments should consider interactions cy c le may req u i r e exp e r iments deliberat e l y designed to among elevated CO , warming,and altered levels of pre- 2 ad d r ess uncertainties in the scaling. In the longer term, we cipitation,nitrogen deposition,and ozone. Second,how should expect a continuing iteration among exp e ri m e n t s , do the characteristics of ecosystems themselves affect ob s e r vat i o n s , and models. Evidence from one of these response to global change? A new generation of experi- ap p ro a c hes will rec e i v e added value from extension acros s 46 Chapter 4: An Integrated Carbon Cycle Science Program the others, and hypotheses suggested by one approa c h will be tested by evidence from all three approa ch e s . Human land use may be directed toward enhancing terrestrial carbon sinks through a variety of management Goal 5: Evaluating Management Strategies options. Focusing on the United States,such options include manipulation of forestry practices (such as harvest dynamics,and pest and pathogen control) and agricultural Goal 5: Maj or Pr ogr am Element s and Act iv it ies land management practices (such as tillage,crop rotations, and fertilizer use;see Lal et al.1998). To evaluate the ¥ Sy nt hesize r esult s of global t er r est r ial car bon potential of management strategies to sequester carbon, sur v ey s, f lux measur ement s, and ot her r elev ant there must be improved process understanding of the r esult s, and use t his analy sis t o dev elop t he sci- relationships between changes in land cover and biophysi- ent if ic basis f or management st r at egies t o cal controls of carbon exchange. One way to gain such enhance car bon sequest r at ion. understanding is the close coupling of historical information on land use with eddy-correlation experiments across ¥ Det er mine t he f easibilit y , env ir onment al a network of sites encompassing large temporal and spa- impact s, st abilit y , and ef f ect iv e t ime scale f or tial variability of fluxes and land uses (beginning with the AmeriFlux network). The land use information is needed capt ur e and disposal of indust r ial CO2 in t he deep ocean and in geological r eser v oir s. to infer historical ecologies (e.g.,species composition,soil organic matter dynamics) that help explain current car- ¥ Def ine pot ent ial st r at egies f or max imizing car - bon fluxes. Where possible in this network, experimental bon st or age t hat simult aneously enhance economic designs should allow for the observation of the effects of and r esour ce v alues in f or est s, soils, agr icul- alternative management strategies (e.g., fire suppression, t ur e, and w at er r esour ces. age-selective forest harvesting). Efforts should be made to ensure that new tower sites have highly variable land use ¥ Ident if y t he cr it er ia t o ev aluat e t he v ulner abil- histories. This will broaden the base of knowledge from it y and st abilit y of sequest r at ion sit es t o climat e which to project future fluxes. Further coupling of land change and ot her per t ur bat ions. use and carbon flux information with biogeochemical models,such as those participating in the Vegetation/ ¥ Document t he pot ent ial sust ainabilit y , lif e- Ecosystem Modeling and Analysis project (VEMAP),may t imes, and int er annual/ decadal v ar iabilit y of dif - be used to predict carbon uptake by whole ecosystems. f er ent managed sequest r at ion st r at egies. Results from the U.S. Forest Service Forest Inventory Analysis (FIA;Schroeder et al.1997),of periodic measure- ¥ Pr ov ide est imat es of t he uncer t aint ies r elat ed ments of wood volume on thousands of plots of land t o managed sequest r at ion t hat ar e r equir ed f or classed as “forest,” indicate lower values of net carbon policy discussions and decisions. accumulation in forests (about 0.3 Gt C/yr) for North America than are inferred from inverse analyses (0.9 to 1 ¥ Dev elop t he monit or ing t echniques and st r at egy Gt C/yr) or tower flux data (~0.7 Gt C/yr). However, the t o v er if y t he ef f icacy and sust ainabilit y of car - FIA data set has not been optimized nor sufficiently ana- bon sequest r at ion pr ogr ams. lyzed as yet to ensure that the estimated rates for net carbon sequestration derived from it are accurate. The focus ¥ Cr eat e a consist ent dat abase of env ir onment al, has been to define the wood volume potentially available economic, and social per f or mance measur es f or for harvest. Thus, wood volumes on land reclassified as t he pr oduct ion, ex ploit at ion, and f at e of w ood and nonforest and wood harvested as forest products are both soil car bon “ f r om cr adle t o gr av e” (f or w ood, removed from the inventory, but a very large quantity of f r om st and est ablishment t hr ough f inal disposal; this wood may remain in the form of wood for long peri- f or soil car bon, f r om f or mat ion t hr ough bur ial). ods. Properly accounting for such “removals”in the carbon budget should increase corresponding estimates of ¥ Dev elop and par amet er ize an analy t ic f r ame- net carbon sequestration by 50 to 100 percent. In the w or k f or pr edict ing and ev aluat ing management CCSP, we propose broadening the FIA to incorporate an and policy alt er nat iv es consider ing mat er ial explicit focus on carbon,including upgrading the sam- t r ansf er , economics, societ al f act or s, w ast es pling scheme,increasing the number of parameters meas- and emissions, r aw mat er ial, labor and ener gy ured,and increasing the frequency of measurements. input s, car bon sour ces, sinks, and f lux es, and out put f or mass car bon balance. In deep ocean sequestration of CO2, industrial CO2 recovered from fossil fuel power plants may be injected directly into the deep ocean for disposal. This arrange- 47 A U.S. Carbon Cycle Science Plan ment would circumvent the slow natural transfer rate of assumptions as competing hypotheses for carbon cycling CO2 from the surface to the deep ocean. Laboratory stud- in the future,but none stands out as pervasively important ies and in situ experiments show that liquid CO2 is denser or useful in the sense that the Northern Land Sink and than seawater at water depths below about 3,000 meters. Increasing Ocean Inventory hypotheses dominate current Additionally, rapid growth of CO2 hydrate crystals along studies. Questions about future atmospheric CO2 levels the water-liquid CO2 interface has been observed. Thus, highlight the need for a comprehensive and balanced pro- liquid CO2, when piped onto the seafloor, is likely to settle gram of observations,manipulative experiments,and mod- there and be quickly transformed into solid hydrate forms. els. These must contribute to the focused testing of The liquid CO2 and the hydrate crystals dissolve gradually hypotheses while remaining comprehensive enough to into the overlying seawater and mix into the surrounding entrain the new ideas and discoveries that will guide deep sea. The retention time of CO2 in the deep ocean hypotheses for the future. Through testing current depends on ocean transport and chemical processes such hypotheses,unforeseen questions,as well as answers,will as neutralization reactions with calcareous sediments. surely emerge. Once the water returns to the ocean surface,some CO 2 In particular, current attempts to project the future can escape back to the atmosphere. Hence,disposal in require assumptions and speculation that extend well relatively isolated regions of the ocean is preferred beyond the range of issues of any program focused on because of longer retention times. The effects of acidified presently testable hypotheses. A future could be hypothe- seawaters on benthic ecosystems must be evaluated sized for atmospheric CO based on reasonable emissions before any large-scale disposal can be contemplated. 2 scenarios,consistent with the notion that future trends Near-field as well as far-field distributions of CO around a 2 will follow the pattern of past trends (in biological and disposal site must be investigated. The scientific investiga- ocean uptake,emissions,etc.). However, this assumption tions outlined in the CCSP, such improvements in ocean encompasses many untested hypotheses. For example, general circulation models (GCMs),will be directly appli- the IPCC scenarios for future CO levels are based on cable in evaluating CO sequestration in the oceans. 2 2 models that have been calibrated using a “fertilization Not all management strategies will be economically effect”to account for the past and present “missing”car- viable or socially acceptable. For example, over the long bon sink. Yet this effect is just one of many proposed run,a rising global demand for timber products may pro- mechanisms for terrestrial carbon uptake,and large-scale vide a strong inducement to keep forest harvest cycles efforts to test it are just beginning. It is unknown how shorter than the optimal time for sequestering the greatest stable terrestrial uptake mechanisms are;the fertilization amount of carbon. Collaboration among biogeochemical effect might be expected to continue for much longer modelers,and economists and other social scientists will than some other contributing mechanisms,such as refor- be necessary to examine the economic efficiency and estation in the United States and Europe. Likewise,the social desirability of various management strategies. Such IPCC CO2 scenarios depend on ocean models that assume collaboration will require information flows across a per- a steady climate. These assumptions seem increasingly meable boundary between the initiatives proposed in this precarious as more is learned about the interplay between plan and associated work in the human dimensions oceanic and atmospheric dynamics,and as the prospect of research community. significant changes in the global climate system become more likely. Clearly, even as the scientific community works toward resolving uncertainties through focused Observing and Discovering hypothesis testing,strong action is necessary to assure that new ideas and discoveries continue to help solve An essential complementary perspective to testing questions about the future behavior of the Earth system. defined hypotheses is suggested by the second central scientific question for the CCSP: “What will be the future As this report has made clear, an essential component atmospheric CO2 loading resulting from both past and of the research strategy is a sustained observational effort. future emissions?” It is very difficult to frame this ques- This endeavor should provide improved,quantitative esti- tion within the context of a single hypothesis or even a mates of natural and anthropogenic carbon distributions, framework of closely linked hypotheses. Scientists must sources,and sinks,and their temporal and spatial variabili- look toward the future with considerable humility, bearing ty at regional as well as global scales. Efforts to develop in mind the significant uncertainties involved in identify- more focused studies of particular problems,or tests of ing the processes responsible for a carbon sink as large as particular hypotheses,should not obscure the need to the hypothesized Northern Land Sink,or in trying to pre- continue and expand a program of comprehensive basic dict how the ocean sink may develop over time. Some observations,with parallel modeling to assimilate the data. issues concerning future carbon cycling can be addressed Because these observations will be useful in assessing car- through such hypotheses as the Increasing Ocean bon sequestration activities,policy needs must be treated Inventory. It is also possible to formulate alternative with utmost attention during planning. The design and 48 Chapter 4: An Integrated Carbon Cycle Science Program implementation of such a program,in a cost-effective et al.1996) for the bottle experiments. It has been manner and in concert with the international research hypothesized that iron supply played an important role in community, is a major task. the changes of atmospheric CO2 content of the last ice ages (Martin 1990). Iron also also appears to be a major In addition to a program of comprehensive observa- factor in nitrogen fixation in the low-latitude surface tions,support is suggested for four types of studies out- ocean (e.g.,Karl et al.1997,Michaels et al.1996a,Michaels side the realm of programs to test current hypotheses. et al.1996b). A major source of iron to the ocean is dust Because all these study types are inherently innovative, delivered through the atmosphere. One notable question they are defined by example here, rather than by descrip- that has arisen,then,is whether changes in dust tions that might prove too restrictive. transport that accompany climate change could affect Studies That May Anticipate “Surprises.” Some oceanic photosynthesis. studies may help reveal unexpected interactions between Studies of Past Geologic Variations in the Carbon the global carbon cycle and the climate system. Recent Cycle. It is widely recognized that the global carbon studies of paleoclimate have revealed the past occurrence cycle has varied in the geologic past. Analyses of gases of surprising short-term instabilities in the Earth’s climate sampled from ice cores have shown that atmospheric CO system. Concerns about instabilities in future climate 2 and CH concentrations varied in concert with the pro- must be extended to the potential effects of these instabil- 4 nounced fluctuations between glacial and interglacial cli- ities on carbon cycling. matic conditions over the past several hundred thousand An example of one such surprise was the finding of years (Barnola et al.1987,Stauffer et al.1988, Jouzel et al. coupled atmosphere-ocean simulations of climate change, 1993). Larger past CO2 variations—comparable to those which predict large increases in the vertical stratification anticipated from future human activities—have been of the ocean with warming in lower latitudes and freshen- inferred from models based on the geologic record of the ing in higher latitudes. These changes would lead to large last 100 million years and longer (Berner 1994). Many of reductions in ocean thermohaline circulation as well as in the processes involved in these changes are so slow that vertical mixing in general (Manabe and Stouffer 1994). they have little relevance to time scales of usual human The impact of these changes on ocean biology would like- concern. However, the geologic record is a vital source of ly be extremely large. The effects on the air-sea balance of information about the dynamic nature of carbon cycling carbon are difficult to predict,,because of the counteract- and its interactions with the global climate system over a ing effects of slowed circulation and a potentially more broad spectrum of time scales. Geologic research can pro- efficient biological pump taking up a higher fraction of vide important and unexpected insights that are directly surface nutrients (Sarmiento et al.1998). relevant to the principal questions addressed in this science plan. Climatic warming appears to promote carbon storage in mid-latitude forests,but possibly to promote release of Measurements from the paleorecord suggest that much larger stores of organic carbon from boreal forests exchanges between reservoirs of the carbon system are and peatlands through increased rates of fire and mineral- not static on millennial time scales. For example,analyses ization (Goulden et al.1998,Kasischke et al.1995) of an ice core from Taylor Dome,Antarctica,show that atmospheric CO concentrations changed over time dur- Studies That Suggest Innovative Hypotheses or 2 ing the Holocene,with a steady increase from 260-280 Ideas. Creative ideas often do not fit preconceived ppm over a period of 7000 years (Indermühle et al.1999). hypotheses or priorities. Important new concepts may These changes may have been due to changes in the arise from unexpected sources,often from scientific sub- amount of terrestrial biomass and/or ocean CO storage. disciplines not normally associated with carbon research. 2 Numerical models of anthropogenic CO2 generally assume A recent paper (Stallard 1998) uses evidence from geo- that a steady state existed before human influence. This morphology and sedimentology to hypothesize that large assumption appears to be reasonable for global models amounts of carbon may be buried on land as a result of because the rate of CO2 change during the Holocene was enhanced erosion and subsequent deposition accompa- very slow compared to the anthropogenic CO2 perturba- nied by fertilization associated with cultivation. This study tion. However, the changes indicated by the ice core data has focused attention on the need to account for the fate may be quite significant for studies of carbon budgets on a of large quantities of eroded soil carbon. regional scale,because the inferred imbalances in the carbon cycle are not likely to have been uniformly distrib- The hypothesis that iron might play an important role uted over the Earth’s surface. in oceanic photosynthesis has a long history. However, the first trustworthy in vitro tests of the hypothesis had to Using precise correlations among ice and sediment await the development of accurate techniques for meas- cores,paleoclimatologists have shown that the most urements in seawater (Martin 1991). Recent in situ iron recent glacial cycles were punctuated by large and fertilization experiments provide dramatic support (Coale extremely abrupt climate events (of years to decades) 49 A U.S. Carbon Cycle Science Plan

(Blunier et al.1998,Dansgaard et al.1993). Many of these goals will permit us to carry out several critical tasks: rapid events appear to correlate with abrupt changes in • Present a new generation of models, rigorously tested atmospheric methane (Brook et al.1993,Chappellaz et al. using time-dependent,three-dimensional data sets,suit- 1993),and some may have been associated with modest able for predicting future changes in atmospheric CO2. changes in atmospheric CO2 (Stauffer et al.1998). Thus, even on time scales of years to decades,there is geologic • Develop management and mitigation strategies that evidence of close interaction between climate and the car- optimize carbon sequestration opportunities using ter- bon cycle. The evidence suggests that atmospheric restrial ecosystems,primarily forest and agricultural sys- methane may be a more sensitive indicator of certain tems,in addition to evaluating the efficiency of ocean changes than atmospheric CO2. carbon sequestration. Studies of the “Human Dimensions” of Future Currently, estimates of terrestrial sequestration and Carbon Cycle Trends. Future terrestrial carbon budgets oceanic uptake of CO2 vary significantly, depending on will be strongly influenced by human land use. The best the data used and the analytical approach (e.g., Forest attempts to forecast future trends may be confounded by Inventory Analysis,direct flux measurements in major uncertainties about future human activities that are influ- ecosystems,inverse model analysis of CO2 data from sur- enced by factors beyond the usual scope of scientific face stations, changes over time of global CO2,O2, 13 12 interests in carbon cycling. CO2/ CO2, etc). Future rates of tropical deforestation will depend large- The proposed research program under Goals 1,2,and ly on the expansion of regional agricultural capacity. Most 3,will reconcile these estimates to a precision sufficient of the developed countries,which are usually outside the for policy decisions by providing new types of data and tropics,will depend almost exclusively on technology-driv- new, stringent tests for models and assessments. The spe- en increases in productivity per unit land area to increase cific deliverables from these three goals are as following. future capacity. In these countries,new changes in the Goal 1 is to establish accurate estimates of the magni- amount of cropland in production are likely to be negligi- tude of the potential Northern Hemisphere terrestrial car- ble.The developing countries will depend on a mix of bon sink and the underlying mechanisms that regulate it. increases in productivity along with expansion of cropland to increase future capacity. Excluding China,these • Define the existence and magnitude of past and pres- countries have about 2.5 billion hectares of land on which ent terrestrial carbon sinks. rain-fed crops could give reasonable yields,with approxi- • Elucidate the impacts of climate variations (e.g.,length mately 80 percent of it in tropical Africa and Latin America of growing season,soil moisture,long-term temperature (FAO 1995). Much of the deforestation taking place in the changes),of fertilization (CO , NO , etc.),and of land tropics is the result of the disorderly, unrecorded expan- 2 x use changes on atmospheric CO . sion of cropland (FAO 1995). The FAO (1995) projects 2 that deforestation from this unrecorded expansion of • Document and constrain uncertainties related to the cropland will not only likely continue,but will probably potential Northern Hemisphere terrestrial carbon sink. do so at a rate exceeding that required to meet national Goal 2 is to establish accurate estimates of the oceanic agricultural capacity targets. Quantification of rates of carbon sink and the underlying mechanisms that regulate unrecorded conversion of tropical forest to cropland is a it. In the near-term,the focus will be on the North critical need for projecting future rates of deforestation. Atlantic and North Pacific oceans to best complement the terrestrial focus on the Northern Hemisphere and tropics. Expected Results • Incorporate better understanding of ocean processes in currently undersampled regions,such as the Southern The ultimate goal of the U.S.CCSP is to provide the Ocean,into general circulation models (GCMs). scientific understanding required to predict future concentrations of atmospheric CO2 and evaluate alterna- • De t e r mine the ex i s t e n c e ,m ag n i t u d e , and interan nu a l tive scenarios for future emissions from fossil fuels,effects va ri ability of oceanic carbon sinks and sources on a of human land use,sequestration by carbon sinks,and reg ional scale through assimilation of new observat i o n s . responses of carbon cycling to possible climate change. • Explain observed changes in the ocean carbon sink due These issues are addressed by Goals 4 and 5—improv- to variations in circulation and biological and chemical ing projections of atmospheric CO2 through complemen- changes using enhanced models tested rigorously tary methodologies and by incorporating regulating mech- against new observations. anisms,and developing a scientific basis for carbon • Incorporate improved oceanic CO flux estimates to sequestration management strategies. Achieving these two 2 improve constraints on inverse models for estimating 50 Chapter 4: An Integrated Carbon Cycle Science Program

terrestrial sinks. Goal 3 is to establish accurate estimates of the impacts of historical and current land use patterns and trends on the evolving carbon budget at local to continental scales. • Document the existence,location,and magnitude of carbon sources and sinks resulting from historical and current land use changes and land management practices in critical regions. • Provide improved assessments of the uncertainties related to land use,particularly for agriculture in North America and the tropics. The results of all the studies under goals 1 through 5 will be communicated to the public and policy makers in papers published in peer-reviewed journals.

51 Chapter 5: Summary of Program Elements and Resource Requirements

Table 5.1 breaks out the Carbon Cycle Science Plan • Continued ocean process studies and enhanced manip- (CCSP) program elements and summarizes the research ulation experiments components and approximate costs. The elements are • Enhanced development of Earth system modeling to also mapped onto the CCSP goals as described below. include interactive carbon and climate dynamics.

Initial Funding Priorities Mapping of Program Elements to Most of the program elements outlined below serve CCSP Goals more than one of the major long-term and five-year goals. The following paragraphs map the program elements Again,the program requires a coordinated,integrated shown in Table 5.1 to the goals and objectives of the approach by the responsible agencies: the value of an Carbon Cycle Science Plan. integrated program will greatly exceed the return from uncoordinated program elements. The individual program elements described in this Goal 1: Program Elements 1-4, 6-9, 11 plan,however, are not at equal stages of readiness. Some The program elements (1) expanded terrestrial flux program elements represent work that has been con- network,(2) airborne CO observation program,(3) global strained by limited resources in the past and could begin 2 CO monitoring network,and (4) land use inventories, immediately with a near-term infusion of funding. New 2 taken together, address the first part of the CCSP Goal 1: technology enterprises, for example,have suffered dispro- to quantify the Northern Hemisphere terrestrial sink and portionately in the recent past due to such resource con- more generally, to quantify the global terrestrial sink for straints. Since many of the program elements in the CCSP CO . Program elements (1),(2),and (3) together provide call for focused technology development prior to large- 2 direct long-term measurements defining sources and sinks scale implementation, we recommend that a high priority on regional scales. The proposed airborne observations be placed on funding those technology development (2) are capable of providing integrated measures of efforts that have been deferred due to insufficient fund- regional net exchange several times per month;the con- ing. More specifically, the following program elements ceptual framework for interpreting these observations should be considered as high priorities for initial funding: needs to be strengthened and tested through a series of • Both facility and technology development for the strategically planned regional observation experiments enhanced flux network,airborne sampling,and auto- (6).The expanded flux network (1) complements the mated and streamlined ocean sampling for long time- other elements by determining monthly and annually aver- series and underway measurements aged regional net uptake or release in typical ecosystems. The flux network will be able to define the systematic dif- • Ai r b o r ne CO mo n i t o r ing prog ram s , both disperse d 2 ferences between flight days and non-flight days for the wee k l y measurements and in support of reg ional studies airborne profile measurements. Flux data can help • An expanded and enhanced surface monitoring net- account for CO2 net exchange on days when analysis of work for atmospheric CO2 the regional net exchange is not possible from atmospheric data,e.g.,during frontal passages or large • Improved forest inventories (with carbon measure- weather events. ments a key focus and an explicit goal) and development of new techniques for remote sensing of above- Improved carbon and land use inventories (4) provide ground biomass a first-order check on the inferences from atmospheric measurements. By telling us where the carbon is going • Analysis of World Ocean Circulation Experiment/Joint (or coming from),program element 4 also contributes sig- Global Ocean Flux Study (WOCE/JGOFS) data for CO 2 nificantly to the second part of Goal 1,to understand the uptake by the oceans underlying mechanisms that regulate the Northern • New ongoing program of air-sea carbon flux and ocean Hemisphere terrestrial sink,and more generally, global ter- inventory measurements restrial sinks and sources.The long-term terrestrial observations (7),process studies and manipulations (8) provide

53 54 Chapter 5: Summary of Program Elements and Resource Requirements

fundamental tools to develop new understanding of ter- use. The expanded flux network (1),long-term terrestrial restrial sinks and sources. observations (7), and terrestrial process studies and manipulations (8) will provide integral checks and con- Estimates of the Northern Hemisphere terrestrial car- straints on the interpretation of results from (4) and (5). bon sink made from atmospheric CO observations are 2 Modeling and synthesis (11) will be the major tools bring- very sensitive to the magnitude of the carbon sink in the ing together the data and concepts developed by these North Atlantic and North Pacific. Oceanic observations program elements. (10) in the North Atlantic and Pacific Oceans thus provide important constraints on the Northern Hemisphere terrestrial carbon sink. Goal 4: Program Element 11 (integrating Modeling and synthesis (11) provide a large-scale check elements 1-10) on inferred Northern and global fluxes when combined The program element modeling and synthesis (11) rep- with global network (3) and airborne (2) data,and allow resents the most comprehensive and integrating tool for tests of our understanding through simulations of past and Goal 4,projecting future atmospheric concentrations of present conditions. CO2. This goal is a major scientific undertaking,in which the models and analysis must be closely integrated with all other elements of the CCSP.To succeed, responsible agen- Goal 2: Program Elements 2, 3, 6, 9-11 cies will need to develop a managerial framework with a The elements (9) surveys,time series, remote sensing, unified vision of the program and with greatly enhanced (10) a quantitative understanding of air-sea exchange mutual collaboration and strategic planning. processes,(2) airborne CO2 observation program,and (3) global CO2 monitoring network together address the first part of Goal 2: to quantify the oceanic uptake of CO2: Goal 5: The Entire CCSP (all elements, global ocean measurements.These elements provide integrated and coordinated) direct long-term measurements defining sources and sinks Goal 5—developing the scientific basis for evaluating on the scale of major ocean regions. A critically important management decisions relating to CO in many critical task is to successfully integrate expanded observations of 2 ways represents the culmination of the entire Carbon time series at key locations,observations of atmosphere- Cycle Science Plan. The ultimate measure of a successful ocean exchange,periodic global ocean surveys, remote carbon cycle research program will be found in its ability sensing of the oceans,large-scale airborne measurements to provide practical answers to both scientific and over the oceans,and atmospheric data from island sta- societal questions. tions. The conceptual framework for interpreting these observations will be developed and tested in element (6), strategically planned regional observation experiments. Ocean process studies and manipulations (10) tell us why the carbon is going (or coming from) major ocean regions and provide the basis for predicting long-term trends. Program elements (10) and (6) thus contribute significantly to the second part of Goal 2:to understand the mechanisms of oceanic uptake of CO2. Modeling and synthesis (11) provide large-scale checks on inferred oceanic and global fluxes,especially when exercised to provide global constraints using data from the global surface and airborne networks (elements (2) and (3). Models allow tests of our understanding through simulations of past disturbances,such as major El Niño- Southern Oscillation (ENSO) events.

Goal 3: Program Elements 1, 4, 5, 7, 11 The program elements (5) (reconstruct historical land use) and (4) (expand global terrestrial carbon and land use inventories) are specifically designed to address Goal 3: to determine the impact of historical and current land

55 Chapter 6: Program Implementation and Critical Partnerships

Implementation Principles Shared Programmatic Responsibility Past experience with large-scale, multidisciplinary glob- The National Research Council’s review of the TOGA al change research programs has provided invaluable Program (NRC 1996) points to the importance of imple- insights into principles for successful program implemen- menting this shared scientific vision through a dynamic, tation. During its deliberations,the authors of this report interactive partnership between the scientific community (the Carbon and Climate Working Group) and sponsoring and responsible federal agencies. In this partnership, agencies of the U.S.Global Change Research Program responsibility for program direction,implementation,and (USGCRP) identified many aspects of the successful review is shared,and all partners commit to the program- Tropical Ocean–Global Atmosphere (TOGA) Program as matic discipline needed to implement the shared vision of possible models for the Carbon Cycle Science Plan. Many the program. As seen in TOGA,commitment to a shared of the principles and strategies described in this chapter scientific vision requires that,once a Carbon Cycle have their roots in the TOGA Program and are drawn from Science Plan has been adopted,the federal agencies agree the experiences of Working Group members,sponsoring that resources will be secured and allocated in accor- agency representatives,and community colleagues,includ- dance with the plan. The scientific community must also ing a 1996 National Research Council review of the TOGA accept an appropriate level of responsibility for setting Program (NRC 1996). Working Group deliberations and priorities,sustaining relevant existing commitments,and discussions at the August 1998 Carbon Cycle Science making a compelling case for new resources. Planning Workshop identified the following key principles As was the case in TOGA,the programmatic partner- for program implementation. ship proposed here requires a strong mechanism for interagency coordination to take inventory of individual agency assets and ongoing programs. Agencies must also Shared Scientific Vision agree on appropriate agency roles and responsibilities for This plan rep r esents the critical fir st step in any success- implementing elements of the Carbon Cycle Science ful global cha n g e res e a r ch prog ram—the development of a Program,based on the agencies’individual capabilities and sh a r ed prog ram vision focused on cle a r l y defined probl e m s missions. Resources can,and probably should, remain in with identifia ble deliver ables of value to both the scientific individual agencies,provided that participating agencies co m m unity and the potential benefic i a r ies in the publi c agree to a single plan,a single committee for the pro- and priv ate sectors. De velopment of this shared vision gram’s scientific oversight and review, a unified propos- should invol v e broad community participation in an open al/program review process,and a single address for scien- pro c e s s , with prog rammatic fle xibility to evol v e continua l l y tific community and public access to information about as new scientific insights emerge ,n ew techn o l o g ies are the program. When successfully implemented,this de vel o p e d , and new infor mation needs are identifie d . ensures that funding decisions reflect scientific merit and programmatic relevance and are essentially “independent” In this context,the Working Group felt it essential to of agency boundaries. The program management struc- provide a clear statement of scientific priorities for a ture proposed later in this chapter is based on this model focused Carbon Cycle Science Program,while recognizing of interagency coordination. the critical contributions of related national and international global change research programs. These programs also address such issues as ecosystem dynamics,land Program Integration use/land cover change, climate variability and change,and atmospheric chemistry. The Carbon Cycle Science Like the TOGA Program,the Carbon Cycle Science Program proposed here describes the essential elements Program necessitates an integrated approach,combining of a program designed specifically to improve the quanti- sustained measurements and data analysis and synthesis; tative characterization of past and present CO2 sources targeted process studies and field research campaigns; and sinks;to develop models to predict future sources organized modeling and prediction efforts;and informa- and sinks;and to provide a scientific basis for evaluating tion management. In addition to requiring good integra- potential carbon sequestration strategies and for measur- tion among these programmatic activities,a successful ing net emissions from major regions of the world. Carbon Cycle Science Program will require a 57 A U.S. Carbon Cycle Science Plan multidisciplinary approach. It is critical to integrate stud- req u i r ement calls for a new level of collabo r ation among ies of atmospheric,oceanic,and terrestrial components of scientists engaged in modeling, me a s u r ement prog ram s , the carbon cycle,together with research on the human and field res e a r ch. In this context , the Wor king Grou p dimensions of carbon cycle changes. notes that early and continuous investment in data-model in t e g ration and synthesis is essential to providing a full un d e r standing of past, pre s e n t , and future carbon cycle Scientific Guidance and Review be h av i o r . A commitment to easy and open access also implies greater responsibilities for infor mation manage- Ex p e r ience with num e r ous global cha n g e res e a r ch pro- ment strat e g y to ensure timely provision of data, res e a r ch grams has demonstrated the importance of establi s h i n g fin d i n g s , and model res u l t s ; to assimilate/integrate observa- clear proc e d u r es for scientific guidance and revi e w of the tions and data from differ ent platfo rm s ,i n s t ru m e n t s ,a n d Carbon Cycle Science Prog ram from the outset. In the case field campaigns; to adhere to approp r iate standard s ,p ro t o- of TOG A , an NRC panel (under the auspices of the Climate co l s , and guidelines; and to provide metadata critical to the Re s e a r ch Committee of the Board on At m o s p h e r ic Sciences an a l ysis from num e r ous individual scientists and parti c i p a t - and Climate) provided a source of ongoing scientific guid- ing institutions. The continuation of support for any ance and revi e w throughout the prog ram ’ s lifet i m e . Th e res e a r ch group under this plan will depend on the timely Wor king Group recommends separating the scientific guid- and complete avai l a bility of data and models gen e ra t e d . ance and revi e w processes in a struc t u r e that draws from both the TOGA exp e r ience and the World Ocean Cir- Experience with other multidisciplinar y, global change culation Experiment (WOCE) and the Joint Global Ocean research programs like WOCE and TOGA highlights the Flux Study (JGOFS) exp e ri e n c e s . We recommend that an technical challenges and opportunities associated with independent Scientific Steering Committee be res p o n s i bl e open access and exchange within the scientific communi- for scientific guidance. (The for mation and composition of ty. The Carbon Cycle Science Program described in this this committee is addressed in the section on prop o s e d plan,however, places an additional burden of responsibili- ma n a gement struc t u r e below.) In addition, the Work i n g ty on the scientific community and sponsoring agencies: Gr oup recommends a periodic (e.g., ever y three to fiv e effective communication of research results to intended yea r s) ext e r nal revi e w of the Carbon Cycle Science users such as businesses,communities,land management Pro g ram . The Wor king Group suggests that the NRC agencies,and government officials. The U.S.Global Committee on Global Change Research is an approp ri a t e Change Research Program uses the term “assessment” bo d y to over see the conduct of these periodic revi e ws of when describing such an organized effort to convey scien- issues related to scientific quality,rel e van c e ,a c c o m p l i s h - tific results to potential beneficiaries in useful forms and m e n t s ,f u t u re plans, and prog ram implementation. to establish a continuing,interactive dialogue with those users. Current national and international discussions relating to the Framework Convention on Climate Change Links to International Programs highlight the importance of sustaining a dialogue that provides government and private sector decision makers with As the TOGA Prog ram demonstra t e d ,e s t ablishing stron g reliable,quantitative information on the sources and sinks ties to the related scientific effor ts of other countries and of carbon dioxide,both present and future. One element for mal international res e a r ch prog rams (the World Climate of this dialogue will involve scientific support for formal Re s e a r ch Prog ram in the case of TOGA) can yield signifi- assessment programs such as the work of the cant benefit s ; it rep r esents another key principle for suc- Intergovernmental Panel on Climate Change. In addition, cessful implementation of the Carbon Cycle Science participants in the Carbon Cycle Science Program should, Pro g ram . The case for international collabo r ation is even from the beginning,plan an organized program of commu- st ro n g er for carbon cycle res e a r ch in light of the ongoi n g nication,outreach,and education that supports the under- ef for ts of the Intergover nmental Panel on Climate Change standing and dissemination of new scientific insights and and current national and international deliberations rel a t e d research results to interested parties in and beyond the to the United Nations Fram e wor k Convention on Climate scientific community. Ch a n ge . Es t a blishing and sustaining effec t i v e links with related international prog rams is described in further detail be l o w under “C r itical Part n e rs h i p s . ” Proposed Program Management Structure With these principles in mind,the Working Group rec- Access to Data and Communication of ommends establishment of a collaborative management Research Results structure for the Carbon Cycle Science Program as depict- ed in Figure 6.1. Ach i e ving the goals described in this plan req u i r es a cul- tu r e of open exch a n g e of observations and associated data This trip a r tite management struc t u r e ref lects the shared pro d u c t s , res e a r ch fin d i n g s , and model res u l t s . In part ,t h i s responsibilities of the scientific community and the spon- 58 Chapter 6: Program Implementation and Critical Partnerships

Figure 6.1 Carbon Cycle Science Program management structure. so r ing fed e r al agen c i e s , and provides a mechanism to coor- goals and programmatic objectives described in the plan. dinate their contrib u t i o n s . While each component of the The SSC would achieve this by continually evaluating ma n a gement struc t u r e has prim a r y responsibility for differ - progress; by reviewing priorities and revising plans as ent elements of the prog ram , success req u i r es that these needed to reflect new scientific insights,technology, and ma n a gement components wor k in unison, in t e r acting on a information needs; by evaluating proposals for major day - t o - d a y basis, exch a n g ing infor mation free l y and openly, adjustments to measurement, field research,and modeling and jointly resolving critical issues and probl e m s . The pro- projects,ensuring a strong scientific review process for gram will stand only if all three “ l e g s ”a re balanced, st ro n g individual projects;and by periodically submitting the and positioned to support their special res p o n s i b i l i t i e s . Carbon and Climate Science Program to rigorous review by an outside body. The Working Group recommends that Responsibility for continuous scientific guidance would the sponsoring federal agencies look to the NRC’s be provided by a Scientific Steering Committee (SSC) Committee on Global Change Research to organize peri- comprised of an expert panel actively engaged in various odic (e.g., every three to five years) outside reviews of the aspects of carbon cycle science. Individual members of Carbon Cycle Science Program. the SSC would be selected by the sponsoring federal agencies in consultation with the scientific community and An Interagency Wor king Group (IWG) comprised of other interested parties. The NRC,through the Committee agencies contributing res o u r ces to support the Carbon on Global Change Research,could be a source of nomina- Cy c le Science Prog ram would have responsibility for inter- tions and/or review of a membership slate for the pro- agency coordination and prog ram managem e n t . The IWG posed SSC. Members of the steering committee should would have prim a r y responsibility for identifying and sus- have a vested interest in the success of the program,and taining rel e vant existing commitments and securing appro- therefore,committee membership should likely include pr iate res o u r ces for new activities in support of the pro- funded participants in the Carbon Cycle Science Program. gram . As for the TOGA Prog ram , the IWG for the Carbon Membership should reflect both individual Cy c le Science Prog ram should be comprised of prog ram disciplinary/programmatic expertise and the special chal- ma n a ger s with the responsibility and authority to commit lenges posed by the integrated nature of the program res o u r ces in support of the prog ram . The IWG should described in this plan. In addition,consideration should ha ve access to the scientific exp e r tise residing in individ- be given to including individuals who represent potential ual agen c i e s , and IWG members should have the res p o n s i - beneficiaries of the program in government,the private bility of ensuring that their agency assets—fi s c a l ,h u m a n , sector, and public interest groups. and capital—are brought to bear approp ri a t e l y in support of the prog ram . In addition to their responsibilities for Working with their colleagues in the sponsoring federal co m m unication within and among the participating scien- agencies,the SSC would have primary responsibility for ti f ic agen c i e s , IWG members will also be res p o n s i b le for ensuring that the detailed implementation of the Carbon en s u r ing that prog ram results are incorporated approp r i- Cycle Science Program follows directly from the scientific at e l y into fed e r al policy for mulation and mission agen c y 59 A U.S. Carbon Cycle Science Plan decision making. Si m i l a r l y,as public serva n t s ,m e m b e rs of science community and government policy officials and the IWG have a special responsibility for commun i c a t i n g resource management agencies in the federal government. the results of the Carbon Cycle Science Prog ram to inter- Managing the day-to-day implementation of this part- ested parties in the public and priv ate sector. nership between the scientific community and the spon- One of the IWG’s most important initial tasks will be soring federal agencies would be the responsibility of a identifying existing agency assets (human,programmatic, Carbon Cycle Science Program Office. Responsibility for and fiscal) and agreeing on individual agency roles and this component of the program management structure responsibilities in implementing the shared scientific could,as in the case of TOGA, reside within a federal vision described in this plan.Agency responsibilities for agency or, as in the WOCE program,be assigned through a the Carbon Cycle Science Program should properly reflect competitive process to a qualified scientific institution agency missions, expertise,and experience. By way of outside the government. As envisioned by this Working example,discussions during the August 1998 Workshop on Group,the function of the Carbon Cycle Science Program Carbon Cycle Science highlighted the following potential Office should not require establishing a large bureaucracy roles for individual agencies.The National Science or the expenditure of a significant level of resources. Foundation would continue its tradition of support for Some investment of resources (human and fiscal) will be critical long-term field studies and laboratory experiments required,however, to ensure that the Program Office can to illuminate key processes for carbon cycle and Earth sys- effectively provide the programmatic and institutional tem modeling,and seek funding for exploratory research “glue”to sustain the critical interaction between the scien- by individual investigators representing a variety of view- tific community and federal agencies responsible for points and approaches. Operational mission agencies implementation of the program. The Carbon Cycle such as the National Oceanic and Atmospheric Science Program Office staff should have appropriate lev- Administration, U.S.Geological Survey, and U.S. els of scientific and programmatic expertise (e.g.,a Ph.D. Department of Energy might play critical roles in support- or equivalent experience). ing sustained measurements and integrated,predictive Sp e c i f ic responsibilities of the Prog ram Office wou l d modeling projects. Resource management agencies such in c lude providing a prog ramwide point of contact and as the Department of Interior, Department of Energy, and so u r ce of infor mation on prog ram dire c t i o n ,a c t i v i t i e s ,s t a- the Department of Agriculture might assume primary tu s , and accomplishments; pr oviding prog ramwide liaison responsibility for examining the roles of terrestrial reser- with the National Research Council’s Committee on Global voirs for carbon dioxide and the impacts of human activi- Ch a n g e Research and other rel e vant scientific orga n i z a - ties on terrestrial ecosystems. The National Aeronautics tions and res e a r ch prog ra m s ;s e rving as prim a r y liaison for and Space Administration might provide the lead in appli- the U. S .p ro gram with rel e vant international scientific pro- cations of space-based, remote-sensing techniques and gra m s ;s u p p o rting the day- t o - d a y aspects of prog ram imple- technologies for carbon cycle science. None of these me n t a t i o n , assisting in prog ram devel o p m e n t , res o u rc e responsibilities would be exclusive,but some identifica- ma n a gem e n t , and prog ram evaluation effort s ; and provi d - tion of appropriate,individual agency roles will be essen- ing secret a r iat services for both the SSC and the IWG. tial to successful implementation of an integrated interagency program. Once the sponsoring federal agencies and the scientific Critical Partnerships community have adopted a shared scientific vision for the Successful implementation of the Carbon Cycle Science program and reached agreement on individual agencies’ Program described in this plan will require the creation roles,the IWG would be responsible for overseeing the and maintenance of a number of critical partnerships. joint implementation of the Carbon Cycle Science First and foremost,the program should represent a Program on behalf of the participating federal agencies. dynamic and innovative partnership between the scientif- We would like to emphasize that joint implementation ic community and the sponsoring agencies in the federal does not necessarily require pooling of resources or the government. The proposed management structure designation of a single “lead agency.” However, it does described above reflects this partnership and a shared require a joint commitment and adherence to the shared responsibility for ensuring scientific quality and program- scientific vision (as described in the Carbon Cycle Science matic relevance,setting priorities,adhering to plans,secur- Plan) and to a primary source of scientific advice (the SSC ing required resources,and interpreting and disseminating described above);to the implementation of a consistent the scientific and information products that emerge from and unified approach to proposal and program review the Carbon Cycle Science Program. processes;and to the provision of a clear point of contact for and source of information on the Carbon Cycle The Working Group believes that implementation of Science Program. In addition,the IWG would likely con- the program described in this plan would benefit signifi- stitute the primary liaison between the carbon cycle cantly from efforts to build sustained partnerships 60 Chapter 6: Program Implementation and Critical Partnerships between federal laboratories and the extramural research IGBP).In addition,the sustained measurement compo- community, leveraging the special capabilities and expert- nents of the program described in this plan could provide ise of those partners. For example,partnership in the substantial contributions to emerging global observing design,development,testing,and deployment of measure- programs such as the Global Climate Observing System ment systems could help address a number of issues fac- (GCOS),the Global Ocean Observing System (GOOS),and ing the scientific community today. This partnership the Global Terrestrial Observing System (GTOS). could ensure an appropriate mix of researchers aware of Conversely, these emerging multinational endeavors could the scientific requirements,and engineers and technicians leverage resources (such as space- and ground-based familiar with system capabilities;and facilitate the observational platforms);develop and demonstrate new exchange of ideas and experience among groups working technologies;and provide a global,Earth system context on the development of systems to meet similar needs. In for carbon cycle measurements emphasizing large-scale short,it could create an environment of creative synergy ecosystem dynamics and carbon cycle interactions with rather than simply duplicate effort. The Carbon Cycle variability and change in the climate system. Science Program envisioned in this plan provides a frame- Fin a l l y,the Wor king Group believes that the Carbon work for creative new partnerships among various disci- Cy c le Science Prog ram described here would benefit from plines and programmatic areas of expertise—such as indian early and sustained partn e r ship with the priv ate sector. viduals involved in measurements,modeling,and process Some of the prog ram ’ s goa l s — n o t a bly, pr oviding a scientif- research—within the scientific community. Similarly, inno- ic basis for evaluating potential carbon sequestration strat e - vative partnerships among supporting federal agencies gies and measuring net emissions at a reg ional scale— should bring individual agency assets and expertise to should be of interest and value to businesses invol v ed in bear on a shared program in complementary ways that the energy sector, res o u r ce use and management (e.g., build on existing capabilities,leverage limited resources, fore s t ry ) , and agric u l t u re , for exa m p l e . P ri vate sector avoid duplication,and produce new opportunities for exp e r tise and assets in techn o l o g y R&D could accelerat e scientific progress. the development and demonstration of critical new meas- The global nature of the carbon cycle and the interna - ur ement techn o l o gi e s . Scientists interested in the design, tional context of current policy deliberations related to de vel o p m e n t , testing and deployment of measurement sys- CO 2 and other greenhouse gases highlight the critical need tems face a number of obstacles today,and this Work i n g for international collabo r ation in carbon cycle inves t i g a - Gr oup believes that innovat i v e partn e r ships among fed e ra l ti o n s . As has been the case in the past, a strong U. S .p ro- l ab o ra t o ri e s ,u n i ve rs i t i e s , and the priv ate sector (in the U. S. gram can serve as a catalyst for similar prog rams in other and abr oad) would be mut u a l l y benefic i a l . In vestments in co u n t r ies and provide a focal point for international coor- the development of new systems should be supplemented di n a t i o n . Su c h multinational collabo r ation offer s opportu - with res o u r ces to make use of commerci a l l y avai l a ble nities to lever age res o u r ces and take advan t a ge of compara- me a s u r ement systems in the fie l d . Decisions to support ble methodologies and joint proj e c t s . Ea r l y collabo ra t i o n the development of new systems should rep r esent a com- with Canada, for exa m p l e , could produce significant bene- mitment of the time and res o u r ces req u i r ed to take new fits for proposed investigations of the North Am e r ican ter- in s t r umentation from concept to rea l i t y . Ea r l y invol ve m e n t res t r ial sink. Si m i l a r l y, in vestigations of ocean sources and of the priv ate sector in this process could help ensure a sinks in the Pac i f ic will benefit from international collabo - smooth transition from limited exp e r imental prototype to ration in muc h the same way that TOG A ’s investigations of co m m e rc i a l l y avai l a ble (and afford a ble) techn o l o g y to sup- El Niño benefited from the contributions of partn e r s po r t broad community needs. ar ound the Pac i f ic Basin and throughout the wor l d . Joint implementation of carbon cycle measurement As has also been seen in programs like TOGA,WOCE, programs and collaborative development of integrated and JGOFS,one particularly beneficial approach to inter- modeling and assessment capabilities could result in early national partnerships involves the development and progress,with direct benefits to the scientific community, implementation of strong U.S.contributions to established public policy officials,and private sector interests directly international global change research programs. The affected by carbon cycle and climate policies. The Carbon Cycle Science Program should establish strong ties Working Group was unable to engage in detailed discus- to the World Climate Research Program (particularly the sion of the character of such a public-private partnership CLIVAR Program),the Global Energy Water Cycle in carbon cycle research during this initial planning Experiment (GEWEX),the International Geosphere– phase. But we strongly encourage a thorough exploration Biosphere Programme (IGBP, particularly JGOFS,GAIM, of the challenges and opportunities of partnership with and IGAC),and the International Human Dimensions of the private sector as an early implementation task. Global Change Program (particularly the Land Use/Land Cover Change project being implemented jointly with

61 Bibliography

Archer,D.,H.Kheshgi and E.Maier-Reimer, Multiple Behrenfeld,M. J. and P. G. Falkowski,A consumers guide to Timescales for the Neutralization of Fossil Fuel CO2, phytoplankton primary production models, Limnol. Geophys.Res.Lett., 24, 405-408,1997. Oceanogr., 42, 1479-1491,1997. Archer, S., D. S.Schimel and E.A.Holland,Mechanisms of Bender, M.L.,T. Sowers, J. N. Barnola and J. Chappellaz, shrubland expansion:Land use, climate or CO2, Changes in the O2/N2 ratio of the atmosphere during Climatic Change, 29, 91-99,1995. recent decades reflected in the composition of air in the firn at Vostok Station,Antarctica, Geophys.Res.Lett., Aumont, O, J. C.Orr, D. Jamous, P. Monfray, O. Marti,and G. 21, 189-192,1994. Madec,A degradation approach to accelerate simulations to steady-state in a 3-D tracer transport model of Berner, R.A.,Geocarb II: A revised model of atmospheric the global ocean, Climate Dynamics, 14, 101-116, CO2 over Phanerozoic time, Am. J. Sci., 294,56-91, 1998. 1994. Baar, H. J.W. d., J.T. M.d. Jong, D. C.E.Bakker, B.M.Löscher, Billings,W.D., J. O. Luken, D.A.Mortensen and K.M. C.Veth, U. Bathmann and V. Smetacek,Importance of Peterson,Arctic tundra:a source or sink for atmospher- iron for plankton blooms and carbon dioxide draw- ic carbon-dioxide in a changing environment, down in the Southern Ocean, Nature, 373, 412-415, Oecologia, 53, 7-11,1982. 1995. Blunier,T., J. Chappellaz, J. Schwander,A.Dallenback,B. Bacastow. R.and C.D.Keeling,Atmospheric carbon diox- Stauffer,T.F. Stocker, D. Raymond, J. Jouzel,H.B.Clausen, ide and radiocarbon in the natural carbon cycle. II, C. U. Hammer and S. J. Johnson,Asynchrony of Changes from A.D. 1700 to 2070 as deduced from a Antarctica and Greenland climate during the last gla- geochemical reservoir. In: G.M.Woodwell and E.V. cial, Nature, 1-18,1998. Pecan (eds.) Carbon and the Biosphere. U.S.Dept.of Bolin,B.Changes of land biota and their importance for Commerce (Springfield,Virginia),86-135,1973. the carbon cycle, Science, 196, 613-615,1977. Bakwin, P. S., P.P.Tans,C.Zhao,W. Ussler and E.Quesnell, Brewer, P.G.,C.Goyet and D. Dyrssen,Carbon dioxide Measurements of carbon dioxide on a very tall tower, transport by ocean currents at 25-N latitude in the Tellus, 47B, 535-549,1995. Atlantic Ocean, Science, 246, 477-479,1989. Bakwin, P.S., P.P.Tans, D. F. Hurst,and C.Zhao, Broecker,W. S.and T.-H. Peng,Interhemispheric transport Measurements of carbon dioxide on very tall towers: of carbon dioxide by ocean circulation, Nature, 356, results of the NOAA/CMDL program. Tellus, 50B, 401- 587-589,1992. 415,1998. Brook,E. J.,T. Blunier and e.al.,Rapid variations in atmos- Baldocchi, D.,Valentini,R.,Running,S.,Oechel,W. and R. pheric methane concentration during the past 110,000 Dahlman, Strategies for measuring and modeling car- years, Science, 273, 1087-1091,1993. bon dioxide and water vapor fluxes over terrestrial ecosystems, Global Change Biology 2, 159-169,1996. Cao,M.and F. I.Woodward,Dynamic response of terrestrial ecosystem carbon cycling to global climate change, Barnola, J. M., D. Raynaud,Y. S. Korotkevich and C.Lorius, Nature, 393, 249-252,1998. Vostok ice core provides 160,000-year record of atmospheric CO2, Nature, 329, 408-414,1987. Chadwick, O.A.,L.A.Derry, P. M.Vitousek,B. J. Huebert and L. O. Hedin,Changing sources of nutrients during four Bates, N. R.,A. F. Michaels and A.H.Knap,Seasonal and million years of ecosystem development, Nature, 397, interannual variability of the oceanic carbon dioxide 491-497,1999. system at the U. S.JGOFS Bermuda Atlantic time-series study site, Deep-Sea Res., Part II, 1996. Chappellaz, J.,T. Blunder, D. Raynaud, J. M.Barnola, J. Schwander and B.Stauffer, Synchronous changes in Battle,M.,M.Bender,T. Sowers, P.P.Tans, J. H.Butler, J.W. atmospheric CH and Greenland climate between 40 Elkins, J.T. Conway, N. Zhang, P. Lang and A.D. Clarke, 4 and kyr BP, Nature, 366, 443-455,1993. Atmospheric gas concentrations over the past centur y measured in air from firn at the South Pole, Nature, Chou,W.W.,Estimation of Net Regional Surface Fluxes of 383, 2312-235,1996. CO2 and O3 Over Amazonia from Aircraft Data,1999.

63 A U.S. Carbon Cycle Science Plan

Ciais, P., P.P.Tans,M.Trolier, J.W. C.White and R. J. Francey,A Doney, S.C., D. M.Glover and R. G. Najjar,A new coupled, large Northern Hemisphere terrestrial CO2 sink indi- one-dimensional biological-physical model for the cated by the 13C/12C ratio of atmospheric, Science, upper ocean:applications to the JGOFS Bermuda 269, 1098-1102,1995a. Atlantic Time Series (BATS) Site, Deep Sea Res., 43, 591- 624,1996. 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. Esbensen,S.K.,and Y.Kushnir,The heat budget of the Schimel, Partitioning of ocean and land uptake of CO2 global ocean:An atlas based on estimates from surface as inferred by _13C measurements from the NOAA marine observations, Climatic Research Institute Climate Monitoring and Diagnostics Laboratory Global Report 29, Oregon State University, Corvallis,OR. Air Sampling, J. Geophys.Res., 100, 5051-5070,1995b. Etheridge, D. M.,L. P. Steele,R.L.Langenfelds and R. J. Coale,K.H.,K.S. Johnson,S.E. Fitzwater, R.M.Gordon,S. Francey, Natural and anthropogenic changes in atmo- Tanner, F.P. Chavez,L. Ferioli,C.Sakamoto, P. Rogers, F. spheric CO2 over the last 1000 years from air in Millero, P. Steinberg, P. Nightingale, D. Cooper,W.P. Antarctic ice and firn, J. Geophys.Res., 101, 4115-4128, Cochlan,M.R.Landry,J. Constantinou, G. Rollwagen,A. 1996. Trasvina and R. Kudela,A massive phytoplankton bloom Falkowski, P.G.,Evolution of the nitrogen cycle and its induced by an ecosystem-scale iron fertilization experi- influence on the biological sequestration of CO in the ment in the equatorial Pacifc Ocean, Nature, 383, 495- 2 ocean, Nature, 387, 272-274,1997. 501,1996. Falkowski, P.G.,Ironing out what controls primary produc- Dai,A.and I.Y. Fung,Can climate variability contribute to tion in the nutrient rich waters of the open ocean, the “missing”CO sink?, Global Biogeochem.Cycles, 7, 2 Global Change Biology, 1, 161-163,1995. 599-609,1993. Falkowski, P.G.,R.T. Barber and V. Smetacek, Dansgaard,W.,S. J. Johnsen, J. B.Clausen, D. Dahl-Jensen, N. Biogeochemical controls and feedbacks on ocean pri- S.Gundestrup,C. U. Hammer, C.S.Hvidberg, J. P. mary production, Science, 281, 200-206,1998. Steffensen,A.E.Sveinbjornsfottir, J. Jouzel and G. Bond, Evidence for general instability of past climate from a Fan,S.-M.,M.Gloor, J. Mahlman,S. Pacala, J. L.Sarmiento,T. 250-kyr ice-core record, Nature, 364, 218-220,1993. Takahashi and P.Tans,A large terrestrial carbon sink in North America implied by atmospheric and oceanic DeGrandpre,M. D.,T. R.Hammar, S. P. Smith and F. L.Sayles, CO2 data and models, Science, 282, 442-446,1998. In situ measurements of seawater pCO2, Limnol. Oceanogr., 40, 969-975,1995. FAO,World Agriculture:Towards 2010 - An FAO Study, in vol.edited by 14-23, John Wiley and Sons, West Sussex, Denman,K.,E.Hofmann and H.Marchant,Marine biotic 1995. responses to environmental change and feedbacks to climate,in Climate Change 1995, vol.edited by J.T. Fasham,M. J. R.,Variations in the seasonal cycle of biologi- Houghton,L. G. M. Filho,B.A.Callander, N. Harris,A. cal production in subarctic oceans: A model sensitivity Kattenberg and K.Maskell,483-516,Cambridge analysis, Deep Sea Res., 42, 1111-1150,1995. University Press,New York,1996. Feely, R.A.,R.Wanninkhof, C.E.Cosca, P. M.Murphy, M. F. Denning,A.S.,M.Holzer, K.R.Gurney, M.Heimann,R.M. Lamb and M. D. Steckley, CO2 distributions in the equa- Law,P.J. Rayner, I.Y. Fung,S.-M. Fan,S.Taguchi, P. torial Pacific during the 1991-1992 ENSO event, Deep Friedlingstein,Y. Balkanski,M.Maiss and I.Levin,Three- Sea Res., 42, 365-386,1995. dimensional transport and concentration of SF6:A Feely, R.A.,R.Wanninkhof, C.E.Coscas,M. J. McPhaden,R. model intercomparison study (TransCom 2), Tellus, in H.Byrne, F.J. Millero, F.P. Chavez,T. Clayton, D. M. press, 1999. Campbell and P. P. Murphy,The effect of tropical insta- Desjardins,R.L., J. L.MacPherson,L.Mahrt, P. Schuepp,E. bility waves on CO2 species distributions along the Patty, H.Neumann, D. Baldocchi,S.Wofsy, D. Fitzjarrald, equator in the eastern equatorial Pacific during the H.McCaughey, and D.W.Joiner, Scaling up flux meas- 1992 ENSO event, Geophys.Res.Lett., 21, 277-280, urements for the boreal forest using aircraft-tower com- 1994. binations. J. Geophys.Res.102, 29125-29134,1997. Feely, R.A.,R.Wanninkhof, C.Goyet, D. E.Archer and T. Dixon,R.K.,S.Brown,R.A.Houghton,A.M.Solomon,M.C. Takahashi,Variability of CO2 Distributions and Sea-Air Trexler and J.Wisniewski,Carbon pools and flux of Fluxes in the Central and Eastern Equatorial Pacific global forest ecosystems, Science, 263, 185-190,1994. During 1991-94 El Niño, Deep Sea Res., 33, 1996.

64 Bibliography

Foster, D. R.,Land-use history and transformations of the Gruber,N.,Anthropogenic CO2 in the Atlantic Ocean, forest landscape of central New England,in Humans as Global Biogeochem.Cycles, 12, 165-192,1998. Components of Ecosystems: Subtle Human Effects Gruber,N., J. L.Sarmiento and T.F. Stocker,An improved and the Ecology of Populated Areas, vol.edited by S.T. method for detecting anthropogenic CO in the A.Pickett and M.McDonnell,91-110,Springer-Verlag, 2 oceans, Global Biogeochem.Cycles, 10, 809-837,1996. New York,1993. Hansell, D.A.and C.A.Carlson,Net community production Foster, D. R.,Land-use history and four hundred years of of dissolved organic carbon. Global Biogeochem. vegetation change in New England,in Principles, Cycles, 12, 443-453,1998. Patterns and Processes of Land Use Change: Some Legacies of the Columbian Encounter, vol.edited by B. Harden, J.W.,E.T. Sundquist,R. F. Stallard and R.K.Mark, L.Turner, 253-318, John Wiley and Sons, New York, Dynamics of soil carbon during deglaciation of the 1995. Laurentide Ice Sheet, Science, 258, 1921-1924,1992. Francey, R. J., P.P.Tans,C.E.Allison,I. G. Enting, J.W. C. Harrison, D. E.and D. S.Luther, Surface winds from tropical White and M.Troller, Changes in oceanic and terrestrial Pacific Islands - Climatological statistics, Journal of carbon uptake since 1982, Nature, 373, 326-330,1995. Climate, 3, 251-271,1990. Frew, N. M.,The role of organic films in air-sea gas Haywood, J. M.,R. J. Stouffer, R.T.Wetherald,S.Manabe and exchange,in The Sea Surface and Global Change, vol. V. Ramaswamy,Transient response of a coupled model edited by P. S.Liss and R.A.Duce,121-171,Cambridge to estimate changes in greehouse gas and sulfate con- University Press,Cambridge,1997. centrations, Geophys.Res.Lett., 24, 1335-1338,1997. Friederich, G. E., P.G. Brewer, R.Herlien and F. P. Chavez, Holfort, J.,K.M. Johnson,B.Schneider, G. Siedler and D.W. Measurements of sea surface partial pressure of CO2 R.Wallace,The meridional transport of dissolved inor- from a moored buoy, Deep Sea Res., 42, 1175-1186, ganic carbon in the South Atlantic Ocean, Global 1995. Biogeochem.Cycles, 12, 479-499,1998. Friedlingstein, P.,I.Fung,E.Holland, J. John, G. Brassuer, D. Holland,E.A.,B.H.Braswell, J.-F. Lamarque,A.Townsend, J. Erickson and D. Schimel,On the contribution of CO2 Sulzman, J.-F. Muller, F. Dentener, G. Brasseru,H.L.II, J. E. fertilization to the missing biospheric sink, Global Penner and G.-J. Roelofs,Variations in the predicted spa- Biogeochemical Cycles, 9, 541-556,1995. tial distribution of atmospheric nitrogen deposition and their impact on carbon uptake by terrestrial ecosys- Fung,I.,C.B. Field, J.A.Berry, M.V.Thompson, J.T. tems, J. Geophys.Res., 102, 15,849-15,866,1997. Randerson,C.M.Malmstrom, P.M.Vitousek, G.J. Collatz, P.J. Sellers, D.A.Randall,A.S.Denning, F. Badeck and J. Houghton, J.T.,L. G. M. Filho,B.A.Callander, N. Harris,A. John,Carbon-13 exchanges between the atmosphere Kattenberg and K.Maskell, Climate Change 1995, vol. and the terrestrial biosphere, Global Biogeochem. pp.,Cambridge University Press,Cambridge,1996. Cycles, 11, 507-533,1997. Houghton,R.A.,Is Carbon accumulating in the northern Gent, P. R., J.Willebrand,T.J. McDougall and J. C. temperate zone?, Global Biogeochem.Cycles, 7, 611- McWilliams,Parameterizing eddy-induced tracer trans- 617,1993. ports in ocean circulation models, J. Phys.Oceanogr., H o u g h t o n ,R .A . ,R . D. Bo o n e , J. R. Fruc i , J. E .H o bb i e , J. M. 25, 463-474,1995. M e l i l l o ,C .A . Pa l m ,B . J. Pet e rs o n , G. R .S h aver , G. M. Goulden,M.L., J.W. Munger, S.M. Fan,B. C.Daube and S.C. Woo d we l l ,B .M o o re , D. L .S kole and N. Mye rs , The flux of Wofsy, Exchange of carbon dioxide by a deciduous for- carbon from terres t r ial ecosystems to the atmosphere in est: response to interannual climate variability, Science, 1980 due to cha n g es in land use: geo g raphic distrib u - 271, 1576-1578,1996. tion of the global flu x , Tel l u s , 39 B , 1 2 2 - 1 3 9 ,1 9 8 7 . Goulden,M.L.,S.C.Wofsy, J.W. Harden,S.E.Trumbore, P. M. Houghton,R.A.,E.A.Davidson and G. M.Woodwell, Crill,S.T. Gower,T. Fries, B.C.Daube,S.-M. Fan, D. J. Missing sinks, feedbacks,and understanding the role of Sutton,A.Bazzaz and J.W. Munger, Sensitivity of boreal terrestrial ecosystems in the global carbon balance, forest carbon balance to soil thaw, Science, 279, 214- Global Biogeochem.Cycles, 12, 25-34,1998. 217,1998. Houghton,R.A., J. E.Hobbie, J. M.Melillo,B.Moore,B. J. Goyet,C., D. R.Walt and P. G. Brewer, Development of a Peterson, G. R.Shaver and G. M.Woodwell,Changes in fiber optic sensor for measurement of pCO2 in sea the carbon content of terrestrial biota and soils water: design criteria and sea trials, Deep Sea Res., 39, between 1860 and 1980: A net release of CO2 to the 1015-1026,1992. atmosphere, Ecological Monographs, 53(3), 235-262, 1983.

65 A U.S. Carbon Cycle Science Plan

Hurst, D.F., P. S.Bakwin,R.C.Meyers,and J.W. Elkins, Keeling,R. F.,S.C.Piper and M.Heimann,Global and hemi- Behavior of trace gas mixing ratios on a very tall tower spheric CO2 sinks deduced from changes in atmospher- in North Carolina, J. Geophys.Res., 102, 8825-8835, ic O2 concentration, Nature, 381, 218-221,1996. 1997. Keeling,R. F. and S.R.Shertz,Seasonal and interannual vari- IGBP Terrestrial Carbon Working Group,The terrestrial car- ations in atmospheric oxygen and implications for the bon cycle:implications for the Kyoto Protocol, Science, global carbon cycle, Nature, 358, 723-727,1992. 280, 1393-1394,1998. Lal,R., J. M.Kimble,R. F.Follett and C. V. Cole, The Indermühle,A., T.F. Stocker, F.Joos,H. Fischer, H. J. Smith,M. Potential of U. S.Cropland to Sequester Carbon and Wahlen,B.Deck, D. Mastroianni,T.Tschumi,T. Blunier, R. Mitigate the Greenhouse Effect, vol.pp.,Ann Arbor Meyer and B.Stauffer, Holocene carbon-cycle dynamics Press,Chelsea,Michigan,1998. based on CO trapped in ice at Taylor Dome,Antarctica, 2 Landry, M.R.et al.,Iron and grazing constraints on primary Nature, 398, 121-126,1999. production in the central equatorial Pacific. An EqPac Jähne,B.,K. O. Münnich,R.Bösinger,A.Dutzi,W. Huber and synthesis, Limnol.Oceanogr., 42, 405-418,1997. P. Libner, On parameters influencing air-water gas Law, R.M., P.J. Rayner,A.S.Denning, D. Erickson,I.Y. Fung, exchange, J. Geophys.Res., 92, 1937-1949,1987. M.heimann,S.C.Piper, M.Ramonet,S.Taguchi,J.A. Jouzel, J., N. I.Barkov,J. M.Barnola,M.Bender, J. Chappellaz, Taylor, C.M.Trudinger and I. G.Watterson,Variations in G. Genthon,V. M. Kotlyakov,V. Lipenkov, C.Lorius, J. R. modeled atmospheric transport of carbon dioxide and Petit, D. Raynaud, G. Raisbeck,C.Ritz,T. Sowers,M. the consequences for CO2 inversions, Global Stievenard, F.Yiou and P.Yiou,Extending the Vostok ice- Biogeochem.Cycles, 10, 783-796,1996. core record of paleoclimate to the penultimate glacial Lloyd, J.,B.Kruijt, D.Y. Hollinger, J. Grace,R. J. Francey, S.-C. period, Nature, 364, 407-412,1993. Wong, F. M. Kelliher,A.C.Miranda,and E.-D. Schwartz, Kamykowski, D. and S.-J. Zentara,Predicting plant nutrient Vegetation effects on the isotopic composition of concentrations from temperature and Sigma-T in the atmospheric CO2 at local and regional scales: upper kilometer of the world ocean, Deep Sea Res., 33, Theoretical aspects and a comparison between rain for- 89-105,1986. est in Amazonia and a boreal forest in Siberia, Australian J. Plant Physiology, 23, 371-399,1996. Karl, D.,R.Letelier, L.Tupas, J. Dore, J. Christian and D. 13 Hebel,The role of nitrogen fixation in biogeochemical Lloyd, J. and G.Farquahar, C Discrimination during CO 2 cycling in the subtropical North Pacific Ocean, Nature, assimilation by the terrestrial biosphere. Oecologia, 99, 388, 533-588,1997. 201-215,1994. Karl, D.M.,A sea of change:Biogeochemical variability in Luz,B.,E.Barkan,M.Bender, M.Thiemens and K.Boering, the North Pacific subtropical gyre, Ecosystems 2:181- Triple-isotope composition of atmospheric oxygen as a 214,1999. tracer of biosphere productivity, Nature, 400, 547-550, 1999. Kasischke,E.S., N. L.Christensen,and B.J. Stock, Fire, Global Warming,and the carbon balance of boreal for- Madec, G.,and M.Imbard,A global ocean mesh to over- est. Ecol.Appl. 5, 437-451,1995. come the North Pole singularity, Climate Dynamics, 12, 381-388,1996. Keeling,C. D.,R.B.Bacastow,A. F. Carter, S.C.Piper,T.P. Whorf, M.Heimann,W.G. Mook and H.Roeloffzen,A Maier-Reimer, E.,Geochemical cycles in an ocean general three-dimensional model of atmospheric CO2 transport circulation model:Preindustrial tracer distributions, based on observed winds: 1 Analysis of observational Global Biogeochem.Cycles, 7, 645-677,1993. data,in Aspects of Climate Variability in the Pacific Maier-Reimer, E.and K.Hasselmann,Transport and storage and the Western Americas AGU Monograph 55, vol. of CO in the ocean - an inorganic ocean-circulation edited by D. H. Peterson,165-236,AGU,Washington 2 carbon cycle model, Clim.Dyn., 2, 63-90,1987. D.C.,1989. Manabe,S.and R. J. Stouffer, Multiple century response of a Keeling,C. D.,T.P.Whorf, M.Wahlen and J.V. d.Plicht, coupled ocean-atmosphere model to an increase of Interannual extremes in the rate of rise of atmospheric atmospheric carbon dioxide, Journal of Climate, 7, carbon dioxide since 1980, Nature, 375, 666-670, 5-23,1994. 1995.

Keeling,R. F. and T.-H. Peng,Transport of heat,CO2 and O2 by the Atlantic’s thermohaline circulation, Philo- sophical Transactions of the Royal Society, 348, 133-142,1995. 66 Bibliography

Marenco,A.,V.Thouret, P. Nédélec,H.Smit,M.Helten, D. Oechel,W. C.,S. J. Hastings, G.Vourltis,M. Jenkins, G. Kley, F. Karcher, P. Simon,K.Law, J. Pyle, G. Poschmann, Riechers and N. Grulke,Recent Changes of Arctic R.Von Wrede,C.Hume,and T. Cook,Measurement of Tundra Ecosystems from a Net Carbon Dioxide Sink to ozone and water vapor by Airbus in-service aircraft:The a Source, Nature, 361, 520-523,1993. MOZAIC airborne program,An overview, J. Geophys. Oeschger, H., U. Siegenthaler, U. Schotterer and A. Res., 103, 631-625,1998. Gugelmann,A box diffusion model to study the carbon Marland, G.,R. J.Andres,and T.A.Boden. Global, regional, dioxide exchange in nature, Tellus, 27, 168-192,1975. and national CO emissions,In:Trends ‘93,T.A.Boden, 2 Potosnak,M. J.,S.C.Wofsy,A.S.Denning,T.J. Conway, and D. P. Kaiser, R. J. Sepanski,and R.W. Stoss,(Eds.),publica- D. H.Barnes,Influence of biotic exchange and combustion No.ORNL/CDIAC-65,Carbon Dioxide Information tion sources on atmospheric CO concentrations in Center, Oak Ridge National Laboratory, Oak Ridge,TN, 2 New England from observations at a forest flux tower. J. pp.505-581,1994. Geophys.Res., in press,1999. Martin, J. H.,Glacial-interglacial CO change: the iron 2 Quay, P.D.,B. Tilbrook,and C.S.Wong,Oceanic uptake of hypothesis, Paleoceanography, 5, 1-13,1990. fossil fuel CO2: Carbon-13 evidence, Science, 256, 74- Martin, J. H.,Iron,Liebig’s Law and the greenhouse, 79,1992. Oceanography, 4, 52-55,1991. Randerson, J.T.,M.V.Thompson,T.J. Conway, I.Y. Fung and Matear, R. J. and A.C.Hirst,Climate change feedback on C.B. Field,The contribution of terrestrial sources and the future oceanic CO2 uptake, Tellus, in press. sinks to trends in the seasonal cycle of atmospheric carbon dioxide, Global Biogeochem.Cycles, 11, 535- McGowan, J.A., D. R.Cayan,and L.M.Dorman,Climate 560,1997. ocean variability and ecosystem response in the Northeast Pacific, Science, 281, 210-217,1998. Raschi,A., F. Milietta,R.Tognetti and P. R. v. Gardigen (Ed.), Plant Responses to Elevated CO : Evidence from Melillo, J. et al.,Land-use change in the Soviet Union 2 Natural Springs, vol.Cambridge University Press, between 1850 and 1980:causes of a net release of CO 2 Cambridge,1997. to the atmosphere, Tellus, 40B, 116-128,1988. Raynaud, D., J. Jouzel, J. M.Barnola, J. Chappellaz,R. J. Merlivat,L.and P. Brault,CARIOCA buoy:Carbon dioxide Delmas and C.Lorius,The ice record of greenhouse monitor, Sea Technology, 10, 23-30,1995. gases, Science, 259, 926-934,1993. Michaels,A. F., D. Olsen, J. Sarmiento, J.Ammerman,K. Rayner, P.J.,I. G. Enting,R. J. Francey and R.Langenfelds, Fanning,R. Jahnke,A.H.Knap, F. Lipschultz and J. Reconstructing the recent carbon cycle from atmo- Prospero,Inputs,losses and transformations of nitrogen spheric CO , D13C and O /N observations, Tellus, In and phosphorus in the pelagic North Atlantic Ocean, 2 2 2 press,1998. Biogeochemistry, 35, 181-226,1996a. Sabine,C.L.,R.M. Key, K.M. Johnson, F.J. Millero,A. Michaels,A. F., J. L.Sarmiento and J. Prospero,Excess nitrate Poisson, J. L.Sarmiento, D.W. R.Wallace,and C. D.Winn, and the rate of nitrogen fixation in the Sargasso Sea, Anthropogenic CO inventory of the Indian Ocean, EOS, 76, OS84,1996b. 2 Global Biogeochem.Cycles, 13, 179-198,1999. Myneni,R.B.,C. D. Keeling,C. J.Tucker, G.Asrar and R.R. Sarmiento, J. L.,T. M.C.Hughes,R. J. Stouffer and S. Nemani,Increased plant growth in the northern high Manabe,Simulated response of the ocean carbon cycle latitudes from 1981 to 1991, Nature, 386, 698-702, to anthropogenic climate warming, Nature, 393, 245- 1997. 249,1998. National Research Council, Global Environmental Sarmiento, J. L.and C.Le Quéré,Oceanic carbon dioxide Change:Research pathways for the next decade, uptake in a model of century-scale global warming, National Academy Press,Washington, D.C.,1998. Science, 274, 1346-1350,1996. National-Research-Council, Learning to Predict Climate Sarmiento, J. L.,R.Murnane and C.L.Quéré,Air-Sea CO Variations Associated with El Niño and the Southern 2 transfer and the carbon budget of the North Atlantic, Oscillation:Legacies of the TOGA Program, vol.pp., Philosophical Transactions of the Royal Society, 348, National Academy Press,Washington, D.C.,1996. 211-218,1995. Nilsson,S.and W. Schopfhauser,The carbon-sequestration Sarmiento, J. L., J. C.Orr and U. Siegenthaler,A perturbation potential of a global afforestation program, Climatic simulation of CO uptake in an ocean general circula- Change, 30, 267-293,1995. 2 tion model, J. Geophys.Res., 97, 3621-3646,1992.

67 A U.S. Carbon Cycle Science Plan

Sarmiento, J. L.,and E.T. Sundquist,Revised budget for the Tak a h a s h i , T. R . ,R .H .Wan n i n k h o f , R .A . Fee l y, R. F.Wei s s , D. oceanic uptake of anthropogenic carbon dioxide. W. Ch i p m a n , N. Ba t e s , J. O l a f s s o n ,C .S ab i n e , and S. C. Nature, 356, 589-593,1992. Su t h e r l a n d , Net sea-air CO2 flux over the global oceans: An improved estimate based on the sea-air pCO di f fer - Schimel, D. S.,VEMAP participants,and B.H.Braswell, 2 e n c e .I n : The proceedings of the 2nd CO in the Oceans Spatial variability in ecosystem processes at the conti- 2 Sy m p o s i u m ,Ts u k u b a , Jap a n , Janu a r y 18-23, 19 9 9 . nental scale:models,data and the role of disturbance, Ecol.Monog., 67, 251-271,1997. Tans, P.P., P. S.Bakwin and D.W. Guenther,A feasible global carbon cycle observing system:a plan to decipher Schimel, D., D.Alves,I.Enting,M.Heimann, F.Joos, D. today’s carbon cycle based on observations, Global Raynaud and T.Wigley, CO and the Carbon Cycle,in 2 Change Biology, 2, 309-318,1996. Climate Change 1995, vol.edited by J.T. Houghton,L. G. M. Filho,B.A.Callander, N. Harris,A.Kattenberg and Tans, P.P., J.A.Berry and R. F.Keeling,Oceanic 13C/12C K.Maskell, 76-86,Cambridge University Press, Observations: A new Window on Ocean CO2 Uptake, Cambridge,1996. Global Biogeochem.Cycles, 7, 353-368,1993. Schimel, D.,I. G. Enting,M.Heimann,T. M.L.Wigley, D. Tans, P.P.,I.Y. Fung and T.Takahashi,Observational con- Raynaud, D.Alves and U. Siegenthaler, CO2 and the straints on the global atmospheric CO2 budget, Carbon Cycle,in Climate Change 1994, vol.edited by Science, 247, 1431-1438,1990. J.T. Houghton,L. G. M. Filho, J. Bruce,H.Lee,B.A. Taylor, N. K.Seasonal uptake of anthropogenic CO in an Callander, E.Haites, N. Harris and K.Maskell,35-71, 2 ocean general circulation model, Tellus, 47B, 145-169, Cambridge University Press,Cambridge,1995. 1995. Schroeder,P.,S.Brown, J. M.Mo,R.Birdsey and C. Toggweiler, J. R.,K.Dixon,and K.Bryan,Simulation of Cieszewski,Biomass estimation for temperate broadleaf radiocarbon in a coarse-resolution world ocean model, forests of the United Staes using inventory data, Forest 1.Steady state pre-bomb distributions, J. Geophys.Res., Science, 42, 424-434,1997. 94, 8217-8242,1989. Smith,S.V. and R.W. Buddemeier, Global change and coral Torn,M.S.,S.E.Trumbore, O.A.Chadwick, P. M.Vitousek reef ecosystems, Ann.Rev. Ecol.Syst., 23, 89-118,1992. and D. M.Hendricks,Mineral control of soil organic car- Stallard,R. F.,Terrestrial sedimentation and the carbon bon storage and turnover, Nature, 389, 170-173,1997. cycle: Coupling weathering and erosion to carbon Turner, D. P., G. J. Koerper, M.E.Harmon and J. J. Lee,A car- burial, Global Biogeochem.Cycles, 12, 231-258,1998. bon budget for forests of the coterminous United Stauffer, B.,T. Blunier,A.Dallenback,A.Indermuhle, J. States, Ecol.Appl., 5, 421-436,1995. Schwander,T.F. Stocker, J.Tschumi, J. Chappellaz, D. Wallace, D.W. R.,Monitoring global ocean inventories, Raynaud,C. U. Hammer and H.B.Clausen,Atmospheric OOSDP Background Rep.5,Ocean Observ. Syst.Dev. CO concentration and millennial-scale climate change 2 Panel,Texas A&M Univ.,1995. during the last glacial period, Nature, 392, 59-62,1998. Wanninkhof, R.,Relationship between wind speed and gas Stauffer, B.,E.Lochbronner, H.Oeschger and J. Schwander, exchange over the ocean, J. Geophys.Res., 97, 7373- Methane concentration in the glacial atmosphere was 7383,1992. only half that of the preindustrial Holocene, Nature, 332, 812-814,1988. Winn,C. D., F.T. Mackenzie,C. J. Carillo,C.L.Sabine and D. M.Karl,Air-sea carbon dioxide exchange in the North Stephens,B.B.,R. F.Keeling,M.Heimann,K. D. Six,R. Pacific Subtropical Gyre: Implications for the global Murnane and K.Caldeira,Testing global ocean carbon carbon budget, Global Biogeochem.Cycles, 8, 157-164, cycle models using mesurements of atmospheric O 2 1994. and CO2 concentrations, Global Biogeochem.Cycles, 12, 213-230,1998. Wofsy, S.C.,M.L.Goulden, J. M.Munger, S.M. Fan, P. S. Bakwin et al.,Net Exchange of CO in a mid-Latitude Takahashi,T.,R.A. Feely, R. F.Weiss,R.H.Wanninkhof, D.W. 2 Forest, Science, 260, 1314-1317,1993. Chipman,S.C.Sutherland and T.T.Takahashi,Global air- sea flux of CO2: An estimate based on measurements Woodwell, G. M,R.H.Whittaker,W.A.Reiners, G. E.Likens, of sea-air pCO2 difference, Proc. Natl.Acad.Sci., 94, C.C.Delwiche,and D. B.Botkin,The biota and the 8992-8299,1997. world carbon budget, Science, 199, 141-146,1978. Zhang, J. R.and P. D. Quay,The total organic carbon export rate based on C-13 and C-13 of DIC budgets in the equaotrial Pacific region, Deep-Sea Research Part II, 44, 2163-2190,1997 68 Acronyms and Abbreviations

ARS Agricultural Research Service (USDA) NAO North Atlantic Oscillation BOREAS BOReal Ecosystem-Atmosphere Study NASA National Aeronautics and Space BR Bowen Ratio/Energy Balance (flux Administration measurement towers) NBP net biome production CCSP Carbon Cycle Science Plan NEP net ecosystem production CFCs chlorofluorocarbons NOAA National Oceanic and Atmospheric CLIVAR Climate Variability and Predictability Administration Programme (WCRP) NRC National Research Council DIC dissolved inorganic carbon NSF National Science Foundation DOE Department of Energy POM Particulate organic matter DOM dissolved organic matter TOC total organic carbon FAO Food and Agriculture Organization TOGA Tropical Ocean–Global Atmosphere FIA Forest Inventory Analysis (program of the Program U.S. Forest Service) TRANSCOM Atmospheric transport model ENSO El Niño–Southern Oscillation comparison study (GAIM project) GAIM Global Analysis,Interpretation and USDAU.S.Department of Agriculture Modelling (IGBP program) USGCRP U.S.Global Change Research Program GCOS Global Climate Observing System VEMAP Vegetation/Ecosystem Modeling and GCTE Global Change and Terrestrial Ecosystems Analysis Project (IGBP program) WCRP World Climate Research Programme GEOSECS Geochemical Ocean Sections program WOCE World Ocean Circulation Experiment GEWEX/GCIP Global Energy and Water Cycle (WCRP program) Experiment/Continental Scale International Project (World Climate Research Program project) GOOS Global Ocean Observing System Gt C/yr Billions of metric tons (or “gigatons”) of carbon per year GTOS Global Terrestrial Observing System IGAC International Global Atmospheric Chemistry Program IGBP International Geosphere–Biosphere Program IPCC Intergovernmental Panel on Climate Change JGOFS Joint Global Ocean Flux Study (IGBP program) LTER Long-Term Ecological Research (study sites maintained by the National Science Foundation) MERIS Medium Resolution Imaging Spectrometer MODIS Moderate-Resolution Imaging Spectroradiometer