A U.S. CARBON CYCLE SCIENCE PLAN...porting government agencies. The nature of the problem demands it. Carbon dioxide is exchanged among three major active reservoirs,the ocean,land,and

A U.S.CARBON

CYCLESCIENCE

PLAN

A U.S.CARBON

CYCLESCIENCE

PLAN

A Report of theCarbon and Climate Working Group

Jorge L. Sarmiento and Steven C. Wofsy, Co-Chairs


Jorge L. Sarmiento and Steven C. Wofsy, Co-Chairs

A U.S.CARBON CYCLESCIENCE PLAN


Jorge L. Sarmiento and Steven C.Wofsy, Co-Chairs

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

1999

ii

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 UniversityWilliam Easterling Pennsylvania State University Chris Field Carnegie InstitutionInez Fung University of California,BerkeleyRalph Keeling Scripps Institution of Oceanography, University of California, San DiegoJames McCarthy Harvard UniversityStephen Pacala Princeton UniversityW. M. Post Oak Ridge National LaboratoryDavid Schimel National Center for Atmospheric ResearchEric Sundquist United States Geological SurveyPieter Tans NOAA,Climate Monitoring and Diagnostics LaboratoryRay Weiss Scripps Institution of Oceanography, University of California,San DiegoJames Yoder University of Rhode Island

Preparation of the report was sponsored by the following agencies of the United States Global ChangeResearch Program: the Department of Energy, the National Aeronautics and Space Administration,the NationalOceanic and Atmospheric Administration,the National Science Foundation,and the United States GeologicalSurvey of the Department of Interior.

The Working Group began its work in early 1998,holding meetings in March and May 1998. A Carbon CycleScience Workshop was held in August 1998, in Westminster, Colorado,to solicit input from the scientific com-munity, and from interested Federal agencies. The revised draft Plan was made available for input by a broadsegment 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 Program400 Virginia Avenue,SWSuite 750Washington, DC 20024 USA phone:(202) 488-8630fax:(202) 488-8681

iii

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

Rationale

Carbon on earth is stored primarily in rocks and sedi-ments. Only a tiny fraction resides in mobile reservoirs(the atmosphere,oceans,and soil and terrestrial bios-phere) and is thus available to play a role in biological,physical and chemical processes at the earth’s surface.The small fraction of carbon present in the atmosphere ascarbon dioxide (CO2) is especially important: it is essen-tial for photosynthesis,and its abundance is a major regu-lator of the climate of the planet.

The major gases,nitrogen, oxygen,argon,which com-prise over 98% of the atmosphere,are transparent to far-infrared (heat) radiation. Trace gases such as carbon diox-ide, water vapor, nitrous oxide,and methane absorb heatradiation from the surface, warming the atmosphere andradiating heat back to the surface. This process,called thegreenhouse effect, is a natural phenomenon withoutwhich the Earth’s surface would be 30°C cooler than it isat present.

Throughout the climate extremes of the past 400,000years,during which there were four major glacial cycles,atmospheric CO2 concentrations varied by no more thantwenty percent from a mean of about 240 parts per mil-lion (ppm). Concentrations of CO2 and methane are nowhigher by more than 30% and 250%, respectively, than theprevious maxima. This very rapid increase has occurredsince the industrial revolution, raising concerns that thetemperature of the atmosphere may rise as a consequenceof an increase in the greenhouse effect.

The rapid increase in atmospheric concentrations of CO2over the past 150 ye a rs , re a ching current concentrations ofabout 370 ppm,c o rresponds with combustion of fossil fuelssince the beginning of the industrial age . C o nve rsion offo rested land to agri c u l t u ral use has also re d i s t ributed car-bon from plants and soils to the atmosphere . T h e re hasbeen growing concern in recent ye a rs that these high leve l sof carbon dioxide not only may lead to ch a n ges in thee a rt h ’s climate system but may also alter ecological balancest h rough phy s i o l o gical effects on ve ge t a t i o n .

Only about half of the CO2 released into the atmo-sphere by human activity (“anthropogenic”CO2 fromcombustion of fossil and biomass fuels and from land usechanges) currently resides in the atmosphere. Over thelast 10-20 years,more than half of the CO2 released byburning fossil fuels has been absorbed on land and in theoceans. These uptake and storage processes are called“sinks” for CO2, although the period over which the car-bon will be sequestered is unclear. The efficiency of

Executive Summary

global sinks has been observed to change from year toyear and decade to decade,due to a variety of mecha-nisms,only partly understood.

The understanding of carbon sources and sinks hasa d vanced enorm o u s ly in the last decade. T h e re is nowclear evidence that global uptake of anthro p o genic CO2o c c u rs by both land plants and by the ocean. The mag n i-tude of the oceanic sink, p rev i o u s ly infe rred from modelsand observations of chemical tra c e rs such as oceanic ra d i o-carbon and tritium distri b u t i o n s , has re c e n t ly been con-fi rmed by direct observation of the increase in dissolve di n o rganic carbon. A n a lysis of new tra c e rs such as ch l o ro fl u-o rocarbons provide further re finements to our unders t a n d-ing of carbon uptake by the oceans. The importance of thesink due to the terre s t rial biosphere has emerged fro ma n a lysis of the global carbon budge t ,i n cluding improve destimates of the ocean carbon uptake , as well as data on1 3C O2/1 2C O2 isotopic ratios and from ch a n ges in the ab u n-dance of O2 re l a t i ve to N2. Isotopes can gi ve info rm a t i o non the terre s t rial sink, for ex a m p l e , since plants pre fe re n t i a l-ly select certain isotopes during photosynthesis and leave aglobal signature in the isotopic ratio of carbon dioxide inthe atmosphere . Fo rest inve n t o ries and remote sensing ofve getation appear to confi rm a significant land sink in theN o rt h e rn Hemisphere and provide insight into the underly-ing mech a n i s m s . H oweve r, we cannot yet quantitative lyd e fine the global effects of human activities such as agri c u l-t u re and fo re s t ry or the influence of climate va riations suchas El Niño. Studies to determine these effects havee m e rged as critical for understanding long-term ch a n ges ina t m o s p h e ric concentration in the past, and will help to dra-m a t i c a l ly enhance understanding of how the eart h ’s cl i m a t ewill evo l ve in the future .

The Carbon Cycle Science Plan (CCSP) presented inthis document has several fundamental motivations. First,it is clear that the oceans and land ecosystems haveresponded in measurable ways to the atmosphericincrease in carbon dioxide,but the associated mecha-nisms are still not well quantified. Second,the land andocean sinks and sources appear to fluctuate naturally agreat deal over time and space,and will likely continue tovary in ways that are still unknown. Third,to predict thebehavior of Earth’s climate system in the future, we mustbe able to understand the functioning of the carbon sys-tem and predict the evolution of atmospheric CO2.Finally, scientific progress over the past decade hasenabled a new level of integrated understanding that isdirectly relevant to critical societal questions associatedwith the economic and environmental effects of forestry,agriculture,land use and energy use practices.

I

The development of the CCSP has been strongly influ-enced by the view that carbon cycle science requires anunprecedented coordination among scientists and sup-porting government agencies. The nature of the problemdemands it. Carbon dioxide is exchanged among threemajor active reservoirs,the ocean,land,and atmosphere,and through a variety of physical, chemical,and biologicalmechanisms,including both living and inanimate compo-nents. Research on atmospheric CO2 therefore encom-passes the full Earth system and involves many differentresearch disciplines and approaches.

Consequently, a large number of government agenciesand programs are involved in supporting research on thecarbon cycle,including data-gathering, field research,analysis,and modeling. Thus it is clear that there isextraordinary value to be gained by coordinating researchand encouraging disciplinary and organizational cross-fertilization through effective program integration. Inaddition,a recent report evaluating research programs inglobal environmental change by the National ResearchCouncil highlighted the importance of developing a coor-dinated, focused,scientific strategy for conducting carboncycle research (NRC 19981).

The Carbon Cycle Science Plan presents a strategy for aprogram to deliver credible predictions of future atmos-pheric carbon dioxide levels, given realistic emission andclimate scenarios, by means of approaches that can incor-porate relevant interactions and feedbacks of the carboncycle-climate system. The program will yield better under-standing of past changes in CO2 and will strengthen thescientific foundation for management decisions in numer-ous areas of great public interest.

The intent of the CCSP is to develop a stra t e gic andoptimal mix of essential components, w h i ch include sus-tained observa t i o n s ,m o d e l i n g , and innova t i ve process stud-i e s ,c o o rdinated to make the whole greater than the sum ofits part s .The design of the CCSP calls for coord i n a t e d , ri g-o ro u s ,i n t e rd i s c i p l i n a ry scientific re s e a rch that is stra t e gi-c a l ly pri o ritized to address societal needs. The plannedactivities must not only enhance understanding of the car-bon cycl e , but also improve capabilities to anticipate futureconditions and to make info rmed management decisions.

Basic Strategy

There are two components to the CCSP strategy:

• Developing a small number of potent new research ini-tiatives that are feasible,cost-effective,and compelling,to improve understanding of carbon dynamics in eachcarbon reservoir and of carbon interactions amongthe reservoirs.

• Strengthening the broad research agendas of the agen-cies through better coordination, focus,conceptual andstrategic framework,and articulation of goals.

Scientific Questions

In the very broadest terms,the present plan addressestwo fundamental scientific questions:

• What has happened to the carbon dioxide that hasalready been emitted by human activities (past anthro-pogenic CO2)?

• What will be the future atmospheric CO2 con-centration trajectory resulting from both past andfuture emissions?

The first of these questions deals with the past andpresent behavior of the carbon cycle. Information aboutthis behavior provides the most powerful clues for under-standing the disposition of carbon released as a result ofhuman activities,the underlying processes in this disposi-tion,and their sensitivity to perturbations. Improvedunderstanding will require targeted historical studies,thedevelopment of a sustained and coordinated observationaleffort of the atmosphere,land and sea,and associatedanalysis,synthesis,and modeling.

The second question focuses directly on the goal ofpredicting future concentrations of atmospheric CO2. Thestudy of essential processes in the carbon cycle will beintegrated with a rigorous and comprehensive effort tobuild and test models of carbon cycle change,to evaluateand communicate uncertainties in alternative model simu-lations,and to make these simulations available for publicscrutiny and application. The research program outlinedin the CCSP will depend on parallel initiatives in humandimensions programs to fully achieve its goals,especiallythat of predicting future atmospheric CO2 concentrations.

Current estimates of terrestrial sequestration andoceanic uptake of CO2 vary significantly depending onthe data and analytical approach used (e.g.,the ForestInventory Analysis,direct flux measurements in majorecosystems,inverse model analysis of CO2 data from sur-face stations, changes over time of global CO2, O2,13CO2/12CO2, measurement based estimates of the air-seaCO2 flux and dissolved inorganic carbon inventory, andocean and terrestrial vegetation model simulations). Theproposed program is designed to reconcile these esti-mates to a precision adequate for policy decisions bydelivering new types of data;applying new, stringent testsfor models and assessments;and providing quantitativeunderstanding of the factors that control sequestration ofCO2 in the ocean and on land. A sound basis for policydebates and decisions will be provided by the program asfully implemented.

A U.S. Carbon Cycle Science Plan

II

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

Executive Summary

Long-Term Goals

The elements of the proposed program (see the sec-tions that follow) have the following long term objectives:

Scientific Knowledge and Understanding

• Develop an observational infrastructure capable ofaccurately measuring net emissions (sources and sinks)of CO2 from major regions of the world.

• Document the partitioning of carbon dioxide sourcesand sinks among oceanic and terrestrial regions.

• U n d e rstand the processes that control the tempora lt rends and spatial distribution of past and current CO2s o u rces and sinks, and how these might ch a n ge underf u t u re conditions of climate and atmospheric ch e m i s t ry.

• Determine quantitatively the factors (long term andtransient) that regulate net sequestration of anthro-pogenic CO2.

• Develop the ability to predict and interpret changes inatmospheric CO2 in response to climate change andinputs of CO2, and including changes in the functionalaspects of the carbon cycle.

Application of Scientific Knowledge toSocietal Needs

• D evelop atmosphere-ocean-land models of the carbonc y cle to (1) predict the life t i m e s ,s u s t a i n ab i l i t y, and inter-a n nual/decadal va ri ability of terre s t rial and ocean sinks,and (2) provide a scientific basis for evaluating potentialm a n agement stra t e gies to enhance carbon sequestra t i o n .

• Develop the ability to monitor the efficacy and stabilityof purposeful carbon sequestration activities.

• Develop the capability for early detection of majorshifts in carbon cycle function that may lead to rapidrelease of CO2 or other unanticipated phenomena.

Program Goals for Fiscal Years 2000-2005and Associated Program Elements

Achieving these goals requires new mechanisms tointegrate information on the carbon cycle and coordinateinteragency efforts as well as additional funding in criticalareas as described below.The following goals will providehigh-value deliverables upon full implementation of theproposed program.

Goal 1: Quantify and understand theNorthern Hemisphere terrestrial carbon sink.

Goal 2: Quantify and understand the uptake of anthropogenic CO2 in the ocean.

Goal 3: Determine the impacts of past andcurrent land use on the carbon budget.

Goal 4: Provide greatly improved pro-jections of future atmospheric concentrations of CO2.

Goal 5: Develop the scientific basis for societal decisions about management of CO2 and the carbon cycle.

The potential for near-term progress is demonstratedby the identification of two general hypotheses thatare testable by integration of the program elementsdescribed below.

Hypotheses:

1. There is a large terrestrial sink for anthropogenic CO2in the Northern Hemisphere.

2. The oceanic inventory of anthropogenic CO2 willcontinue to increase in response to rising atmosphericCO2 concentrations,but the rate of increase will bemodulated by changes in ocean circulation,biology,and chemistry.

Goal 1: Understanding the NorthernHemisphere Land Sink

Goal 1 is intended to establish accurate estimates of themagnitude of the hypothesized Northern Hemisphere ter-restrial carbon sink and the underlying mechanisms thatregulate it.

Major Program Elements and Activities

• Expand atmospheric monitoring: vertical concentrationdata,column CO2 inventories,continuous measure-ments,new tracers (oxygen,carbon and oxygenisotopes, radon,CO, and other traces gases).

• Conduct field campaigns over North America,and even-tually over the adjacent oceans,using aircraft linked toenhanced flux tower networks and improved atmo-spheric transport models.

• Improve inverse models and strengthen connectionsbetween atmospheric model inferences and directterrestrial and oceanic observations.

III

• Expand the network of long-term flux measurementsfor representative ensembles of undisturbed,managed,and disturbed land along major gradients of soils,landuse history, and climate,using new technology tolower costs.

• Conduct manipulative experiments and process studieson ecosystem,local,and regional scales,coordinatedwith flux measurements to quantify terrestrialsources/sinks and changes over time.

• Obtain remote-sensing data defining changes in vegeta-tion cover and phenology.

Deliverables

• Define the existence and magnitude of past and pres-ent terrestrial carbon sinks.

• Elucidate the contributions of climate variations,suchas rainfall,length of growing season,soil moisture,andlong-term temperature changes,of fertilization (CO2,NOx, etc.),and of past and present land use changesand land use management to the contemporaryCO2 sink.

• Document and constrain the uncertainties related tothe potential Northern Hemisphere terrestrialcarbon sink.

Goal 2: Understanding the OceanCarbon Sink

Goal 2 is intended to establish accurate estimates of theoceanic carbon sink,including interannual variability, spa-tial distribution,sensitivity to changes in climate,andunderlying mechanisms. In the near-term, focus should begiven to the North Atlantic and North Pacific to comple-ment the focus on the Northern Hemisphere in Goal 1.In the long term, focus should be on important regionswith major gaps in current knowledge,such as theSouthern Ocean.


• Develop new technology for measuring oceanic CO2and related quantities on various observational plat-forms,including moorings,drifters,and new shipboardsampling techniques.

• Acquire air-sea carbon flux measurements,includingdata from long-term stations and from field campaignscombining aircraft,shipboard,and shore-based measure-ments;use improved ocean and atmosphere transportmodels to refine estimates of the magnitude of oceanicsources and sinks of CO2 based on pCO2 (partial pres-sure of carbon dioxide) data.

• Synthesize results of recent global ocean carbon sur-veys to elucidate the processes for carbon storage inthe sea,and conduct ongoing oceanic inventory andtracer measurements at a level that assures coverage ofall ocean areas every 10 to 15 years.

• Develop and implement process studies,models,andsynthesis,including manipulation experiments,large-scale tracer releases,and direct measurement of air-seafluxes of CO2; focus should be on defining physical andbiological factors regulating air-sea exchanges andexport to long-term storage in deep ocean waters.

• Develop remote sensing capability to monitor physicaland biological properties on larger scales.

Deliverables

• Understand ocean processes in critical regions such asthe Southern Ocean.

• Determine the existence,magnitude,and interannualvariability of oceanic carbon sinks and sources onregional scales.

• Attribute observed changes in the ocean carbon sink tovariations in circulation,biology, and chemistry.

• Incorporate improved oceanic CO2 flux estimates forbetter constraints on inverse models in estimatingterrestrial sinks.

Goal 3: Assessing the Role of Land Use

More fully stated,Goal 3 is to obtain accurate esti--mates of the ef fects of historic and current land use pat-terns on atmospheric CO2, emphasizing the tropics andNorthern Hemisphere.


• Document the history of agricultural expansion andabandonment and its impact on the contemporary car-bon balance of North America by conducting intensivehistorical land use and present-day land managementanalyses across environmental gradients in coordinationwith the proposed North American network of eddyflux measurement sites.

• Improve observational capabilities in tropical regionsthrough cooperation with tropical nations,includinginternational expeditions and exchanges and training ofscientists and managers.

• Apply the observational programs discussed in Goal 1in the tropics,including long-term observations andmodeling.


IV

• Couple historical and socioeconomic analysis of landmanagement strategies with biogeochemical models toevaluate the integrated effects of human activities thatsequester carbon (see Goal 5).

Deliverables

• Document the existence,location,and magnitude ofcarbon sources and sinks resulting from historical andcurrent land use change and land management.

• Provide the scientific basis for evaluating short-termand long-term impacts of deliberate modification of theterrestrial biosphere

• Provide better assessments of carbon balance uncer-tainties related to land use,particularly agriculture inNorth America and the tropics.

Goal 4: Improving Projections ofFuture Atmospheric CO2

More fully stated,Goal 4 is to improve projections offuture atmospheric concentrations of CO2 through a com-bination of manipulative experiments and model develop-ment that incorporates appropriate biophysical and eco-logical mechanisms and carbon cycle-climate feedbackinto global climate and carbon cycle models.


• Improve representation of physical and biologicalprocesses in carbon and climate models.

• Provide a framework for rigorous analysis of observa-tions and for independent comparisons and evaluationsof climate and carbon models using comprehensivedata and model assessment activities.

• Develop new generations of terrestrial biosphere andocean carbon exchange models that include the rolesof both natural and anthropogenic disturbances,succes-sion,and the feedback through changes in climate andatmospheric composition.

• Develop coupled Earth system models incorporatingterrestrial and oceanic biogeochemical processes inclimate models.

• Provide projections of the future evolution of CO2 con-centrations to evaluate the consequences of differentemission scenarios and to provide a scientific basis forsocietal decisions.

(See under Goal 5, for deliverables.)

Goal 5: Evaluating Management Strategies

Goal 5 is to develop a scientific basis for evaluatingpotential management strategies to enhance carbonsequestration in the environment and capture/disposal strategies.

Major program elements and activities

• Synthesize the results of global terrestrial carbon sur-veys, flux measurements,and related information;strengthen the scientific basis for management strate-gies to enhance carbon sequestration.

• Determine the feasibility, environmental impacts,stabili-ty, and effective time scale for capture and disposal ofindustrial CO2 in the ocean and geological reservoirs.

• Define potential strategies for maximizing carbonstorage that simultaneously enhance economic andresource values in forests,soils, agriculture,andwater resources.

• Identify criteria to evaluate the vulnerability andstability of sequestration sites to climate change andother perturbations.

• Document the potential sustainability of differentsequestration strategies and the storage time forsequestered CO2.

• Develop the monitoring techniques and strategy toverify the efficacy and sustainability of carbonsequestration programs.

• Develop scientific uncertainty estimates that arerequired for policy discussions and decisions related tomanaged carbon sequestration.

• Create a consistent database of environmental,econom-ic,and other performance measures for the production,exploitation,and fate of wood and soil carbon “fromcradle to grave”(for wood,from stand establishmentthrough final disposal; for soil carbon,from formationthrough burial).

Deliverables (Goals 4 and 5)

• Present a new generation of models, rigorously testedusing time-dependent 3-dimensional data sets,suitablefor predicting future changes in atmospheric CO2.

• Support the development of management/mitigationstrategies that optimize carbon sequestration opportu-nities using terrestrial ecosystems,primarily forest andagricultural systems,as well as evaluating the efficiencyof sequestration in the ocean

• Use these new models to identify critical gaps inour understanding and help design futureobservational strategies.

Executive Summary

V

VI

Executive Summary

Mapping of Program Elements toCCSP Goals

The following paragraphs map the program elementsshown in the table above to the goals and objectives ofthe Carbon Cycle Science Plan.

Goal 1: Program Elements 1-4, 6-9, 11

The pro gram elements (1) expanded terre s t rial flux net-wo rk , (2) airborne CO2 o b s e rvation pro gra m , (3) globalC O2 m o n i t o ring netwo rk , and (4) land use inve n t o ri e s ,t a ken toge t h e r, a d d ress the fi rst part of the CCSP Goal 1:to quantify the Nort h e rn Hemisphere terre s t rial sink andm o re ge n e ra l ly, to quantify the global terre s t rial sink fo rC O2. P ro gram elements (1), ( 2 ) , and (3) together prov i d ed i rect long-term measurements defining sources and sinkson re gional scales. The proposed airborne observations (2)a re capable of providing integrated measures of re gi o n a lnet ex ch a n ge seve ral times per month; the conceptualf ra m ewo rk for interpreting these observations needs to bes t rengthened and tested through a series of stra t e gi c a l lyplanned re gional observation ex p e riments (6).The ex p a n d-ed flux netwo rk (1) complements the other elements byd e t e rmining monthly and annu a l ly ave raged re gional netu p t a ke or release in typical ecosystems. The flux netwo rkwill be able to define the systematic diffe rences betwe e nflight days and non-flight days for the airborne pro fi l em e a s u re m e n t s . Flux data can help account for CO2 n e tex ch a n ge on days when analysis of the re gional netex ch a n ge is not possible from atmospheric data, e . g . ,d u r-ing frontal passages or large weather eve n t s .

Improved carbon and land use inventories (4) providea first-order check on the inferences from atmosphericmeasurements. By telling us where the carbon is going(or coming from),program element 4 also contributes sig-nificantly to the second part of Goal 1,to understand theunderlying mechanisms that regulate the NorthernHemisphere terrestrial sink,and more generally, global ter-restrial sinks and sources.The long-term terrestrial obser-vations (7),process studies and manipulations (8) providefundamental tools to develop new understanding of ter-restrial sinks and sources.

Estimates of the Northern Hemisphere terrestrial car-bon sink made from atmospheric CO2 observations arevery sensitive to the magnitude of the carbon sink in theNorth Atlantic and North Pacific. Oceanic observations(10) in the North Atlantic and Pacific Oceans thus provideimportant constraints on the Northern Hemisphere terres-trial carbon sink.

Modeling and synthesis (11) provide a large-scale checkon inferred Northern and global fluxes when combinedwith global network (3) and airborne (2) data,and allowtests of our understanding through simulations of past andpresent conditions.

Goal 2: Program Elements 2, 3, 6, 9-11

The elements (9) global ocean measurements (surveys,time series, remote sensing),(10) a quantitative under-standing of air-sea exchange processes,(2) airborne CO2observation program,and (3) global CO2 monitoring net-work together address the first part of Goal 2: to quantifythe oceanic uptake of CO2. These elements provide directlong-term measurements defining sources and sinks onthe scale of major ocean regions. A critically importanttask is to successfully integrate expanded observations oftime series at key locations,observations of atmosphere-ocean exchange,periodic global ocean surveys, remotesensing of the oceans,large-scale airborne measurementsover the oceans,and atmospheric data from island sta-tions. The conceptual framework for interpreting theseobservations will be developed and tested in element (6),strategically planned regional observation experiments.

Ocean process studies and manipulations (10) tell uswhy the carbon is going (or coming from) major oceanregions and provide the basis for predicting long-termtrends. Program elements (10) and (6) thus contributesignificantly to the second part of Goal 2:to understandthe mechanisms of oceanic uptake of CO2.

Modeling and synthesis (11) provide large-scale checkson inferred oceanic and global fluxes,especially whenexercised to provide global constraints using data fromthe global surface and airborne networks (elements (2)and (3). Models allow tests of our understanding throughsimulations 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 landuse) and (4) (expand global terrestrial carbon and landuse inventories) are specifically designed to address Goal3: to determine the impact of historical and current landuse. The expanded flux network (1),long-term terrestrialobservations (7),and terrestrial process studies andmanipulations (8) will provide integral checks and con-straints on the interpretation of results from (4) and (5).Modeling and synthesis (11) will be the major tools bring-ing together the data and concepts developed by theseprogram elements.

Goal 4: Program Element 11 (integrating ele-ments 1-10)

The program element modeling and synthesis (11) rep-resents the most comprehensive and integrating tool forGoal 4,projecting future atmospheric concentrations ofCO2. This goal is a major scientific undertaking,in whichthe models and analysis must be closely integrated with all

VII

other elements of the CCSP. To succeed, responsible agen-cies will need to develop a managerial framework with aunified vision of the program and with greatly enhancedmutual collaboration and strategic planning.

Goal 5: The Entire CCSP (all elements,integrated and coordinated)

Goal 5—developing the scientific basis for evaluatingmanagement decisions relating to CO2 in many criticalways represents the culmination of the entire CarbonCycle Science Plan. The ultimate measure of a successfulcarbon cycle research program will be found in its abilityto provide practical answers to both scientific andsocietal questions.

Implementation Principles

Past ex p e rience with large - s c a l e , mu l t i d i s c i p l i n a ry globalch a n ge re s e a rch pro grams (e.g. the TOGA Pro gram) hasd e m o n s t rated that a cohere n t ,i n t e grated appro a ch to pro-gram implementation is essential for optimal exe c u t i o nand delive ry of products designed to serve societal needs.The principles for successful implementation of the CCSPp ro gra m , discussed in detail in Chapter 6 of the re p o rt ,a re :

• A scientific vision shared by the broad communi-ty and by the participating agencies to developconsistency and focus.

• Shared programmatic responsibility to insurecoherence,coordination,and strategic pursuit ofprogram goals.

• Program integration to bring together interdiscipli-nary aspects of the strategy.

• Scientific guidance and review to provide feedbackand support to program managers,to help foster inno-vation and creativity, and to create an environmentwhere program elements and goals evolve as newknowledge is obtained.

• Links to international programs to maximize thebenefits from efforts in all countries.

• Access to data and communication of researchresults to insure timely communication of knowledgeto the general public and to enhance the scientificutility of new knowledge.

With these principles in mind,the Working Group hasrecommended in Chapter 6 specific steps to establish acollaborative management structure for the Carbon CycleScience Program,with strong interagency commitment tojoint implementation of the program,including commondevelopment of requests for proposals and coordinationreview and funding activities. The program is built arounda tripartite,collaborative management structure for

integrated carbon cycle research consisting of a scientificsteering committee (SSC),interagency working group(IWG),and Carbon Cycle Science Program office. Theintent is to strengthen relevant federal agency activities byimproved coordination,integration,coherence,prioritiza-tion, focus,and adherence to conceptual goals. The pro-posed structure will provide the basis for issuing intera-gency research announcements to stimulate a broad rangeof important new research,with agreed-upon goals,usinga coordinated interagency merit review process of propos-als and overall agency programs. The importance of fos-tering partnerships among Federal laboratories,the extra-mural research community, and the private sector isstressed,as is the need to communicate effectively theevolving understanding of carbon sources and sinks to thepublic and policy makers.

Initial Funding Priorities

Most of the program elements outlined above servemore than one of the major long-term and five-year goals.Again,the program requires a coordinated,integratedapproach by the responsible agencies: the value of anintegrated program will greatly exceed the return fromuncoordinated program elements.

The individual pro gram elements described in this plan,h oweve r, a re not at equal stages of maturity and re a d i n e s s .Some pro gram elements re p resent intellectually - re a dywo rk that has been constrained by limited re s o u rces in thepast and could begin immediately with a near-term infu-sion of funding. N ew tech n o l o gy enterpri s e s , for ex a m p l e ,h ave suffe red dispro p o rt i o n a t e ly in the recent past due tos u ch re s o u rce constra i n t s . Since many of the pro gram ele-ments in the CCSP call for focused tech n o l o gy deve l o p-ment prior to large-scale implementation, we re c o m m e n dthat a high pri o rity be placed on funding those tech n o l o gyd evelopment effo rts that have been defe rred due to pri o ri n s u fficient funding. M o re specifi c a l ly, the fo l l owing pro-gram elements should be considered as high pri o rities fo rinitial funding:

• Both facility and technology development for theenhanced flux network,airborne sampling,and auto-mated and streamlined ocean sampling for long time-series and underway measurements

• Airborne CO2 monitoring programs,both dispersedweekly measurements and in support of regionalstudies

• An expanded and enhanced surface monitoringnetwork for atmospheric CO2

• Improved forest inventories (with carbon measure-ments a key focus and an explicit goal) and develop-ment of new techniques for remote sensing of above-ground biomass


VIII

• Analysis of World Ocean Circulation Experiment/JointGlobal Ocean Flux Study (WOCE/JGOFS) data for CO2uptake by the oceans

• New ongoing program of air-sea carbon flux and oceaninventory measurements

• Continued ocean process studies,such as air-seaexchange and enhanced manipulation experiments

• Enhanced development of Earth system modeling toinclude interactive carbon and climate dynamics.

Executive Summary

IX

Rationale

Future concentrations of atmospheric carbon dioxide(CO2) must be known to characterize and predict thebehavior of the Earth’s climate system on decadal to cen-tennial time scales. Predicting these future CO2 concen-trations in turn requires understanding how the globalcarbon cycle has functioned in the past and how it func-tions today. For these reasons,a variety of federal agen-cies have funded scientific research into the oceanic,atmospheric,and terrestrial components of the global car-bon cycle. This research has been an important elementof the U.S.Global Change Research Program (USGCRP)since its inception.The carbon cycle consists of an inte-grated set of processes af fecting closely coupled carbonreservoirs in the atmosphere and ocean and on land.Successful predictions require considering all the impor-tant processes that affect these reservoirs. Programs musttherefore support a well-integrated approach,with linksfostered among atmospheric,oceanic,terrestrial,andhuman sciences.

Recently, interest in the global carbon cycle has intensi-fied. Progress in scientific understanding and technologyhas given rise to new scientific opportunities to addresscritical components of the atmosphere-ocean-land system.Policy makers have also been seeking the appropriateresponses to the United Nations Framework Conventionon Climate Change and to the underlying societal and sci-entific concerns. There is particular impetus to under-stand the sources and sinks of carbon on continental andregional scales and the development of these sources andsinks over time. Also critical is elucidating how ocean car-bon uptake and marine and terrestrial ecosystems mightrespond to changing climate and ocean circulation,or tothe enhanced availability of CO2 and fixed nitrogen com-pounds associated with human activities such as the burn-ing of fossil fuels. Changes in ecosystems may affect cli-mate and may have important resource and societal impli-cations. Another topic of increasing interest is the pur-poseful sequestration,or storage,of carbon by burialbelow ground or in the ocean,or by land managementpractices,to keep some amount of carbon from enteringthe atmosphere in the form of CO2.

The ultimate measure of the success of the carboncycle research program will be its ability to providepragmatic answers to both scientific and societal ques-tions. Scientists and policy makers must be able to evalu-ate alternative scenarios for future emissions from fossilfuels,effects of human land use,sequestration by carbonsinks,and responses of carbon cycling to potentialclimate change.

Chapter 1: Introduction

1

This document outlines a U.S.Carbon Cycle SciencePlan (CCSP) with the view that c redible predictions off u t u re atmospheric carbon dioxide levels can bemade given realistic emission and climate scenariosthat incorporate relevant interactions and feedbacks.

The problems of understanding the carbon cycle andimproving related predictive capabilities are complex.These problems require coordinated interdisciplinaryresearch that is scientifically rigorous and that at the sametime advances knowledge and serves societal needs.Achieving all this together is clearly a challenge. The pres-ent plan aims at specifying an optimal mix of sustainedobservations,modeling,and innovative process studiesand manipulative experiments such that the whole isgreater than the sum of parts. This strategy for an inte-grated CCSP has two basic thrusts:

• Developing a small number of new research initiativesthat are feasible,cost-effective,and compelling,toaddress difficult,linked scientific problems

• Strengthening the broad research agendas of the agen-cies through better coordination, focus,conceptual andstrategic framework,and articulation of goals.

In sum,the rationale for a CCSP is severalfold. Futureconcentrations of atmospheric CO2 must be projected tocharacterize and predict the behavior of the Earth’s cli-mate system on decadal to centennial time scales.Moreover, the scientific community is in a good positionto make important progress in this area. But making suchprogress will require an unprecedented level of coordina-tion among the scientists and government agencies thatsupport this research.

Outline of the Research Program

The CCSP is designed to address two basic questions:

1. What has happened to the CO2 that has already beenemitted through human activities (anthropogenic car -bon dioxide)?

2. What will be the future atmospheric carbon dioxideconcentrations resulting from both past and futureemissions?

The first of these questions concerns the history andthe present behavior of the carbon cycle. Informationabout the past and present provides us with the mostpowerful clues for understanding the behavior of anthro-pogenic CO2 and the processes that control it,as well asits sensitivity to perturbations. Achieving a better

understanding of these phenomena will require a sus-tained observational effort.

The second question directly concerns the goal of pre-dictability. The CCSP will study the essential processesthat influence how carbon cycling may change in thefuture. These studies will be integrated in a rigorous andcomprehensive effort to build and test models of carboncycle change,to evaluate and communicate uncertaintiesin alternative model simulations,and to make these simu-lations available for public scrutiny and application.

The long-term goals and implementation objectives thatflow from the two questions above are shown in Table 1.1and reviewed in the sections that follow below on sus-tained observations,manipulative experiments,and modeldevelopment.

In addition to the physical,biogeochemical,and ecolog-ical processes that are traditionally studied in carbon cycleresearch,the integrated CCSP must address human influ-ences as well,with special emphasis on understanding theconsequences of land use and land cover changes;the

e ffects of va rious management stra t e gi e s ,s u ch as no-tillagri c u l t u re , long- vs short - rotation fo re s t ry, and deep oceanC O2 i n j e c t i o n ; and the effe c t i veness of re s p o n s e / m i t i g a t i o noptions such as controlling carbon emissions or enhancingcarbon sinks. The CCSP addresses the intera c t i o n sb e t ween human systems and the carbon cycle in mu ch thesame way that it does interactions between the climate sys-tem and the carbon cycl e . While the pro gram does notencompass either broad socioeconomic re s e a rch or cl i-mate re s e a rch , it does consider pro blems of the intera c-tions of human systems and climate with the carbon cycl e .

Sustained Observations

A program of sustained observations of sufficient spa-tial and temporal resolution is essential for determininginterannual variability and long-term trends in terrestrialand oceanic carbon sources and sinks,both globally andregionally. Such a program is also required to monitor theeffectiveness and stability of any purposeful carbonsequestration activities,as well as to detect major shifts in


2

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 fromactivities (anthropogenic CO2) CO2 terrestrial and oceanic regions of the worldincluding that emitted through • Understand mechanisms over the full • Quantitatively determine the processes (long-termcombustion of fossil fuels, range of relevant time scales transient) that control net sequestration ofdeforestation, 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, independentcarbon 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 theclimate 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


3

the functioning of the carbon system that might lead tomajor changes in atmospheric CO2.

Data on global and regional variability and trends inCO2 concentrations provide us with the most valuableinformation on the response of the global carbon cycle toclimate changes and to processes such as forest regrowthor fertilization owing to increased CO2 and nitrogenoxides (NOx) from fossil fuels. Monitoring programsshould take advantage of the recent insights from inversemodels to combine limited direct observations withmodel estimates to determine the spatial and temporal dis-tribution of carbon sources and sinks. The program ofobservations must also include a strong focus on under-standing critical processes that determine the long-termsequestration of anthropogenic CO2, as well as its interan-nual variability.

The Mauna Loa and South Pole CO2 time series of C. D.Keeling provided the fi rst unambiguous evidence ofi n c reasing atmospheric CO2. Time series data for CO2,combined with tra c e rs , constituents of the atmospherethat have a specific ori gin or ch a n ge in an understood way,s u ch as 1 4C O2, 1 3C O2, and O2, p rovide strong quantitativec o n s t raints which help confi rm the fa c t o rs regulating theglobal balance of carbon. Expanded atmospheric observa-tion netwo rk s ,i n cluding airborne measurements and agrowing number of flux towe rs ,w h i ch measure theamount of CO2 going into or out of a certain area of landover a period of time, h ave enabled scientists to begi nd e fining the re gional distribution of carbon sources ands i n k s . This wo rk is still ru d i m e n t a ry, but alre a dy indicatesgreat potential for assessing source and sink distri b u t i o n sat smaller (e.g., re gional) scales. A major hypothesisissuing from the past decade of re s e a rch is thatt h e re exists a large terrestrial sink for anthro-pogenic CO2 in the Northern Hemisphere. Much ofthe near-term observational work proposed in thisplan aims to test this hypothesis.

A critical task is to improve the modeling and statisticaltools needed to infer sources and sinks and to otherwiseinterpret these observations. Additional oceanic and ter-restrial observations must also be defined to complementglobal monitoring data so that better temporal and spatialresolution is achieved. Such observations should improveunderstanding of seasonal and year-to-year variability, andthey should monitor regions identified as significantsources and/or sinks.

Three areas in particular—areas that the scientific com-munity is positioned to address—require near-termemphasis:

• Establishing accurate estimates of the magnitude andpartitioning of the current Northern Hemisphere ter-restrial carbon sink.

• Establishing accurate estimates of the oceanic carbonsink,including its interannual variability and spatialdistribution. In the near-term,attention should bedirected to the North Atlantic and North Pacific,tocomplement and optimize use of the terrestrial data onthe Northern Hemisphere.In the long-term,emphasisshould turn to the Southern Hemisphere oceans pole-ward of approximately 30°S.

• Establishing accurate estimates of the impact on theevolving carbon budget of historical and current landuse change,timber harvest,and deforestation in thetropics and the Northern Hemisphere.

Manipulative Experiments

Manipulative experiments play a unique role in globalchange research. They allow direct study of many keyecosystem processes that have strong control over the car-bon cycle. These experiments can contribute to carboncycle research in at least three ways. First,global changesin the coming decades will create a range of novel condi-tions,some of which will very likely be far enough outsidethe envelope of current and past conditions that observa-tional data alone cannot provide a sufficient basis for cred-ible modeling. Experiments will be especially importantin assessing responses to multiple,interacting changes,and in helping to assess slowly developing responses,suchas changes in biodiversity. Second,model developmentand testing solely against observations of current patternscannot provide the rigor and level of credibility thatcomes from validation against carefully designed experi-ments. And third, explaining and illustrating the future tra-jectory of the carbon cycle can be substantially enhancedthrough experimental manipulations. Even if the scientificcommunity believes the models,the illustrative value of asolid experiment to the broader public is invaluable.

For manipulative experiments to yield large payoffs,they will need to be tightly integrated with observationsand models. Targeted work must address key uncertain-ties. Interpreting results from manipulative experimentshas presented major challenges in the past,especiallythose relating to the experiments’small spatial scale,shorttemporal scale,and incomplete coverage of relevantecosystem processes. Interpretations will certainly contin-ue to be difficult,but these difficulties can be managedthrough careful selection of study systems,precise defini-tion of key questions,and strong emphasis on integratingthe experiments with one another and with other compo-nents of carbon cycle research.

Experiments can play an essential role in reducinguncertainties about the location of and the mechanismsunderlying current terrestrial and oceanic sinks. Criticaltargets for experimental work include evaluating the


4

consequences of the CO2 increase from preindustrial lev-els to the present;the relative contributions of forestregrowth,nitrogen deposition,and elevated CO2 to thecurrent carbon sink; and the interactions between ecologi-cal ch a n ges and altered climate and atmosphere . To under-stand the like ly future tra j e c t o ry of the terre s t rial sink,m a n i p u l a t i ve ex p e riments should be used to ex a m i n eecosystem responses to simultaneous va riation in mu l t i p l ee nv i ronmental fa c t o rs , and to integrate bioge o ch e m i c a lresponses with species ch a n ge s ,i n cluding alterations in bio-d i ve rs i t y. P rototype ocean manipulation ex p e ri m e n t ss h owed the importance of iron as a control on carbonu p t a ke by ocean biota in the “ h i g h - nu t ri e n t ,l ow - ch l o ro-p hy l l ” (HNLC) env i ronments of the Equatorial Pa c i fic (Coaleet al. 1 9 9 6 ) . A similar ex p e riment was re c e n t ly conductedin the Southern Ocean. S u ch prototype ex p e riments showthe potential for using ocean manipulation ex p e riments tou n d e rstand complex nu t rient and ecosystem intera c t i o n srelated to the role of ocean biota in the carbon cycl e .F u t u re re s e a rch should also include manipulation ex p e ri-ments to elucidate the role of iron in biological nitro gen fi x-ation (Fa l kowski et al. 1 9 9 8 , Karl et al. 1 9 9 7 ) . M a n i p u l a t i veex p e riments should not be considered full simulations ofthe future . T h ey are almost inev i t ably imperfect as simu l a-t i o n s , even when they provide unique and critical insightsinto underlying mech a n i s m s .

Model Development: Understanding andPredicting Critical Processes

Quantitative understanding of critical processes,processes that operate on time scales from seasons todecades and longer, is essential to predict future atmos-pheric concentrations of CO2. Policy decisions affectingfuture CO2 emissions—whether to implement restraintsor management,or to abstain from doing so—require thecapability to predict terrestrial and oceanic carbon sinks.The models to serve these purposes are also necessaryto analyze observations and determine optimalsampling strategies.

Predicting these sinks will require the development ofcoupled terrestrial biosphere–atmosphere–ocean climatemodels. Many scientific issues need to be solved toimprove current predictive capability. The magnitude ofthe present oceanic sink is reasonably well known (orconstrained),but the spatial distribution and evolution ofthe oceanic sink over time are poorly known. The abilityto predict the future behavior of the ocean sink is compli-cated by the likelihood of large changes in ocean circula-tion as predicted by coupled climate models and observedin previous periods of rapid climate change. Particularimportance attaches to the Southern Ocean,a vast regionfor which there is a paucity of data. The possible role ofchanging species composition of marine biotic

communities in modifying the carbon sink has only begunto be explored. Another major hypothesis that hasguided the preparation of this plan is that theoceanic inventory of anthropogenic CO2 will contin-ue to increase in response to rising atmosphericCO2 concentrations, but the rate of increase will bemodulated by changes in ocean circulation, biology,and chemistry.

Better understanding of the terre s t rial sink is also need-e d . C u rrent estimates are deri ved pri m a ri ly from the massbalance of CO2 in the atmosphere combined with info rm a-tion about fossil fuel input and ocean CO2 fl u xe s . T h e re isgrowing consensus that a terre s t rial sink ex i s t s , but them agnitude and underlying causes are uncert a i n . E s t i m a t e sof the magnitude of this sink va ry by at least a factor oft wo . Potential underlying mechanisms include re c ove ryf rom prior land use, lengthening of the growing seasondue to wa rm i n g ,C O2 and nitro gen fe rt i l i z a t i o n , and buri a lof carbon in sediments and rice paddies. I nve n t o ries basedon direct estimates of wood volume in U. S . fo rests sugge s ta smaller sink than do atmospheric estimates.

Carbon sources and sinks,and especially those suscep-tible to human perturbation,whether inadvertent or inten-tional,are highly variable over subcontinental and sub-basin scales. Hence, information to evaluate the effects ofongoing land use and possible options for mitigationand/or adaptation will be required to significantly improvemodels of regional CO2 exchanges. This endeavorrequires much greater understanding of how overall pat-terns of change (natural or human-induced) in critical ter-restrial and marine ecosystems might affect fluxes of car-bon to the atmosphere. For example,how might climate-related changes in productivity in coastal ecosystemsaffect the role of marine ecosystems in the cycling of car-bon? What conditions and constraints are most importantin influencing a region’s role as a source or sink for CO2,and how might these conditions change in the future?What is the long-term photosynthetic and species distribu-tion response of terrestrial ecosystems to climatic anom-alies and long-term trends,and how does this determinethe carbon uptake? In this context,a focused carbon andclimate research program should be viewed as an impor-tant component of a broader scientific effort to under-stand large-scale ecosystem dynamics and change (such asthe effort proposed in the recent NRC “Pathways” report[NRC 1998]).

The development of carbon cycle models must be car-ried out within a framework of rigorous comparison withobservations,as well as inter-model comparisons. It is notenough that models make accurate predictions on thewhole;their crucial components must also be testedagainst the data.


5

Structure of the Plan

A ch i eving the important goals of a good carbon cycl escience pro gram will re q u i re a new level of scientific andp ro grammatic integra t i o n . S c i e n t i fi c a l ly, it is necessary toh ave a more complete picture of the ocean-land-atmospherei n t e ractions of the carbon and climate system. Also neededis more thorough understanding of the physical and bioge o-chemical processes that ch a ra c t e rize and control this sys-t e m . This know l e d ge must encompass re l a t i ve ly long-termp rocesses that are difficult to study, s u ch as natural andhuman-mediated disturbance, land use and their impacts one c o l o gical processes such as succession. P ro gra m m a t i c a l ly,it is necessary to implement a pro gram that effe c t i ve ly inte-grates sustained monitoring activities, field and lab o ra t o ryre s e a rch ,modeling and analytical studies, and targe t e dassessment activities designed to meet the info rm a t i o nneeds of society. These goals are attainabl e . And theirattainment has an urgency incre a s i n g ly recognized in thepolicy are n a .

This plan provides a blueprint for a carbon cycle sci-ence program that focuses this renewed commitment.The plan is organized into the following chapters:

Chapter 2 describes the past and present carboncycle,addressing the question:What has happened to theCO2 that has already been emitted through human activi-ties (anthropogenic carbon dioxide)?

Chapter 3 explores the critical scientific issues for pre-dicting the future carbon cycle,addressing the question:What will be the future atmospheric carbon dioxide con-centrations resulting from both past and future emissions?

Chapter 4 describes the proposed Carbon CycleScience Program.

Chapter 5 estimates the resource requirements toimplement the proposed program.

Chapter 6 discusses the pro gram implementation pri n-ciples and critical part n e rships re q u i red for a successfulp ro gram (including issues related to mu l t i d i s c i p l i n a ryre s e a rch ,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 –p ri vate sector collab o ra t i o n s ,i n t e rnational pro gra m s ,a n dfacilities and infra s t ru c t u re ) .

Chapter 1 reviewed the two overarching scientificquestions that the Carbon Cycle Science Plan (CCSP) isdesigned to address. This chapter turns more focusedattention to the first of these two questions:

What has happened to the CO2 that has alreadybeen emitted through human activities (anthro-pogenic carbon dioxide)?

Carbon cycle research relating to climate change funda-mentally concerns three large scientific issues. The first isthe natural partitioning of carbon among the “mobile”reservoirs—the ocean,atmosphere,and soil and terrestrialbiosphere—partitioning that is influenced itself by cli-mate change. The second issue is the redistribution of fos-sil fuel CO2 within those same three reservoirs,andassessments of proposals to prevent emissions orsequester carbon through new technologies. The thirdissue concerns transfers between the terrestrial biosphereand the atmosphere induced by other human activitiessuch as forest clearing and regrowth,the management ofagricultural soils,and the feedback potential of anthro-pogenically driven changes in atmospheric chemistry andclimate to alter these transfer rates.

Historical changes in the quasi–steady state of the car-bon system are clearly reflected in ice core and isotopicrecords,which also record the unprecedented changescaused by anthropogenic CO2 emissions (Raynaud et al.1993). The globally averaged atmospheric CO2 mole frac-tion is now over 365 µmol/mol (or parts per million,ppm)—higher than it has been for hundreds of thousandsof years. Atmospheric concentrations of CO2 hadremained between 270 and 290 ppm during the last sev-eral thousand years,but rose suddenly to the present levelduring the second half of the 20th century. This increaseis coincident with the rapid rise of fossil fuel burning.

The modern rate of atmospheric CO2 increase is accu-rately known from contemporary direct atmosphericmeasurements as well as from air stored in ice and firn(Battle et al.1996,Etheridge et al.1996). The Inter-governmental Panel on Climate Change (IPCC) assessmenthas summarized the state of scientific knowledge in thisarea (Schimel et al.1995). During the decade of the1980s,the rate of increase in the atmosphere was 3.3 ±0.2 billions of metric tons (also called gigatons) of carbonper year (Gt C/yr). The next best-known piece of the puz-zle is the rate of fossil fuel emissions,which was 5.5 ± 0.5Gt C/yr during the 1980s (Marland et al.1994). Thus,it isclear that on average a large fraction of the emitted CO2did not remain in the atmosphere. Conservation of massimplies that the missing emissions must have entered the

Chapter 2: The Past and Present Carbon Cycle

7

ocean,the terrestrial biosphere,or both. The ocean car-bon sink is the best known of these two remaining com-ponents of the carbon budget. According to the IPCC,theocean absorbed 2.0 ± 0.8 Gt C/yr. This finding suggeststhat the global net terrestrial biosphere sink was only 0.2± 1.0 Gt C/yr. However, the estimated loss of carbon fromthe terrestrial biosphere due to deforestation is estimatedto have been 1.6 ± 1.0 Gt C/yr. Thus,there must havebeen a compensatory storage of carbon in the terrestrialbiosphere of 1.8 ± 1.6 Gt C/yr. The latter figure largelycompensates for the land use source,so that the globalnet storage in terrestrial ecosystems was close to zero.

It has not been easy to arrive at these estimates.Regarding ocean uptake,if the CO2 entering the oceanwere mixed homogeneously, the cumulative increase inthe total carbon content after 10 years at the rate of 2 GtC/yr would only be 1.2 µmol/kg. This equals the detec-tion limit of the currently best analytical techniques. Mostof the CO2 has been added to the upper few hundredmeters,however. The signal is therefore detectable,but itmust be extracted from large natural variations of up toabout one to two hundred µmol C/kg. For this reason,until very recently, ocean uptake has been estimated usingmodels whose parameters are calibrated against the pene-tration into the ocean of other tracers such as 14C andtritium from nuclear tests,and chlorofluorocarbons.

Estimates of terrestrial carbon loss have been based onsurveys of land use and changes in land use, above- andbelow-ground average carbon densities for ecosystemclasses,and models of the development of carbon storageafter disturbance,under human management,and duringsuccession (e.g.,Houghton et al.1987). These assess-ments have varied greatly, as reflected in the range fortropical biomass destruction assessed by the IPCC to be1.6 ± 1.0 Gt C/yr. Part of the problem in this evaluation isthe great heterogeneity of carbon content on small spatialscales. But also,surveys have been difficult to comparebecause of incompatible definitions,different countingmethods,the treatment of secondary forest growth,andsimilar reasons.

Estimates of terrestrial carbon uptake have beenobtained from mass balance considerations such as thosedescribed above,which give a total uptake of 1.8 ± 1.6 GtC/yr. Other estimates come from direct observations offorest carbon inventory changes (0.9 Gt C/yr in theNorthern Hemisphere),models of NOx fertilization (0.6 to0.9 Gt C/yr),and CO2 fertilization (0.5 to 2.0 Gt C/yr)(Schimel et al.1995). There are strong indications thatthere is a large uptake in the Northern Hemisphere,but


8

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. Whereunbalanced fluxes are indicated (global primary production and respiration, land use and air-sea gas exchange) theimbalance reflects the anthropogenic fluxes. Note that this figure shows decadal average fluxes and we now knowthese fluxes to be quite variable from year to year.


9

the location,magnitude,and mechanisms of this uptakeare poorly understood.

The following three sections summarize recentprogress on the most uncertain components of the pertur-bations to the global carbon budget: (1) the terrestrialcarbon sink,(2) the ocean sink,and (3) the land usesource. Each section is concluded with a proposed majornear-term (5 to 10 year) research initiative that willaddress the most compelling scientific issues identified,and which is also scientifically feasible and cost-effective.

The Terrestrial Carbon Sink

Perhaps the biggest recent change in thinking aboutthe terrestrial sink has been that earlier estimates of enor-mous carbon losses from terrestrial ecosystems (Bolin1977,Woodwell 1978) have given way to the idea that theterrestrial biosphere has been close to neutral withrespect to carbon storage during the last decades. Theobserved destruction of forests appears to have beenroughly compensated for by mechanisms of enhanced car-bon uptake. Eight independent lines of evidence supportthis view:

1. The overall carbon budget estimated from the observedatmospheric increase and ocean uptake estimates usingcalibrated ocean models requires modest ter restrialuptake to satisfy mass balance (Bacastow and Keeling1973,Oeschger et al.1975).

2 . The smaller than expected north-south gradient ofa t m o s p h e ric CO2, combined with data on the part i a lp re s s u re of CO2 in ocean surface wa t e rs ,s u g gests thatt h e re is a large terre s t rial CO2 sink at temperate latitudesin the Nort h e rn Hemisphere (Tans et al. 1 9 9 0 ) . The sinkcompensates fo r, or is larger than, the estimated rate ofcarbon loss due to defo restation in the tro p i c s .

3. The atmospheric ratio of oxygen to nitrogen has beendeclining due to the consumption of oxygen by fossilfuel burning. The ratio is decreasing as oxygen use isrequired by this fossil fuel burning,though perhapsslightly more slowly, indicating that there is no majornet O2 sink other than fossil fuel burning. There mayeven be a small net source,which would suggest thatbiological fixation of CO2 may exceed rates of reminer-alization of organic matter (Battle et al.1996, Keeling etal.1996).

4. The existence of a large terrestrial sink at northern lati-tudes is supported by 13C/12C ratio measurements inatmospheric CO2 (Ciais et al.1995a) and by measure-ments of the oxygen/nitrogen ratio (O2/N2) (Keeling etal.1996). At northern latitudes the 13C/12C and O2/N2ratios are higher (relative to the Southern Hemisphere)than expected from fossil fuel burning alone. Thissuggests net uptake by photosynthesis,which

(1) discriminates against uptake of 13C relative to 12C,leaving the atmosphere enriched in 13C,and (2) pro-duces O2, enhancing the O2/N2 ratio.

5. A new technique is eddy covariance,which can meas-ure vertical transport in a turbulent atmosphere. Fluxmeasurements obtained by this technique in differentecosystems have demonstrated the ability of someforests to act as significant net sinks for atmosphericCO2 (Wofsy et al.1993,Baldocchi et al.1996).However, the number of such measurements that isavailable is too limited to draw any strong conclusions.

6. The increase in amplitude of the seasonal cycle ofatmospheric CO2, and especially the earlier onset ofthe summer photosynthetic drawdown,is consistentwith net uptake by temperate land ecosystems (Myneniet al.1997,Randerson et al.1997).

7. Recent forestry surveys in various regions of theNorthern Hemisphere also tend to indicate net carbonuptake,but not as large as the atmospheric data seemto imply (Dixon et al.1994).

8. During earlier decades (1970–1990) the measuredchange in the oceanic 13C/12C ratio indicates relativelylittle net uptake by the terrestrial biosphere when ana-lyzed in the context of changes in atmospheric concen-trations (Quay et al.1992).

The storage of organic carbon in terrestrial sedimentsmay be much more important than previously recognized(Stallard 1998). Carbon storage through erosion maysequester carbon if significant amounts of eroded carbonare stored in sediments where they will decompose slow-ly, and if regrowth of vegetation on eroded lands replacesthe lost carbon. This sink,combined with expansion ofrice agriculture and postulated net storage of carbon inrice paddy soils,could amount globally to 0.6-1.5 Gt C/yr(Stallard 1998).

It is noteworthy that five of the above arguments arebased entirely or partly on the atmospheric measurementof fluxes,either directly or as deduced from spatial pat-terns of the concentration of gaseous species. The firstargument is based on a global mass balance involving theocean,and the first of the flux arguments makes use ofair-sea fluxes. Estimates of aboveground terrestrial bio-mass are being made routinely, but two-thirds of the globalterrestrial carbon is in soils,and it has been hard to obtainaccurate inventories,let alone temporal change data forthis pool. It is encouraging that the evidence forterrestrial sinks is internally consistent because each ofthe approaches has its own problems,to be brieflydiscussed next.

Interpreting observations of the CO2 mole fraction interms of surface sources and sinks requires atmospherictransport models that portray large-scale circulation

accurately, and that also parameterize subgrid-scale mixingcorrectly. The recent atmospheric transport model com-parison study (TRANSCOM) exercises show that signifi-cant improvement in both respects is sorely needed,andthat the atmospheric boundary layer is of particularimportance (Law et al.1996,Denning et al.1999).Because the atmospheric signature of low-latitude sourcesis weak due to vigorous vertical mixing,and because thereare no applicable observations in some important areas(such as tropical forests),the models currently workroughly as follows. Sources and sinks are derived fromobserved concentration patterns with some degree ofconfidence for the temperate and high latitudes,and thosefor low latitudes then follow essentially from global massconservation. Calibrations of certain aspects of the trans-port with other tracers with “known”sources,such asCFCs, 85Kr, or SF6, are useful,but fail in that the strongdiurnal and seasonal cycles of CO2 sources are not includ-ed,and also because these tracers have strong sources inthe tropics.

The interpretation of the isotopic data is subject to sub-stantial uncertainty. The main source of error is the con-tribution of purely isotopic exchange,often called the iso-topic disequilibrium flux,which may occur with or with-out an accompanying net exchange of total carbon (Tanset al.1993). A better determination of the rate of air-seagas exchange is especially important to pin down the iso-topic disequilibrium flux between the ocean and atmo-sphere. The isotopic disequilibrium between the atmo-sphere and the terrestrial biosphere depends on the aver-age age of the respiratory carbon flux,which should bebetter known as well. This age is important because olderbiomass was fixed during times that the atmosphere wasless depleted in 13C than today. In interpreting atmos-pheric 13C/12C trends, relatively small uncertainties in iso-topic disequilibrium between the atmosphere and oceantranslate into large uncertainties in the partitioning of thetotal sink between ocean and land (e.g.,an uncertainty ofonly 0.1 per mil,or parts per thousand,in isotopic disequi-librium gives an uncertainty of 0.5 Gt C in partitioning).Furthermore,the isotopic signature of terrestrial primaryproductivity, strongly influenced by the relative propor-tions of the C3 and C4 photosynthetic pathways,needs tobe better defined.(Fung and al.1997,Lloyd and Farquahar1994). The isotopic fractionation during C4 photosynthe-sis is not much dif ferent from that of air-sea exchange.

The atmospheric oxygen budget may be subject touncertainties concerning decadal variations in the ventila-tion of deeper, oxygen-poor waters of the ocean. It isreassuring that two completely different analytical tech-niques give similar results (Bender et al.1994, Keeling andShertz 1992).

The new micrometeorological (eddy-covariance) fluxmeasurement methods have been developed to measureecosystem exchange of trace gases on a spatial scale of

hundreds of meters (Wofsy et al.1993,Baldocchi et al.1996,Goulden et al.1998). There is also a network ofabout 10 U.S.Department of Agriculture (USDA)Agricultural Research Service grassland sites (tallgrass,mixed grass,shortgrass steppe,intermountain,and shrub-lands) equipped with Bowen Ratio/Energy Balance (BR)towers that are monitoring CO2 and H2O fluxes. In grass-lands,BR systems work well and are less relatively expen-sive. Such measurements are used in conjunction withknown disturbance history, and with climatic,plant physi-ological,ecological,and soil data,to investigate mecha-nisms responsible for the uptake or loss of carbon forwhole forest ecosystems,including soils.

The manageable spatial scale facilitates the search forflux mechanisms,but makes it necessary to test extrapola-tions to regional scales with other methods. Relationshipsbetween fluxes and climatic variables have been found(Goulden et al.1996,Goulden et al.1998). When turbu-lence is well developed,often during the day, the eddy-covariance method is well demonstrated,but at nightunder stable conditions,the measurements are less reli-able and heterotrophic respiratory fluxes (fluxes fromnon-photosynthetic organisms) may be underestimated.This problem can be partially circumvented by selectingmeasurement times when conditions for the method arefavorable,although this introduces the danger of aliasing,or obtaining biased data because of non-representativesampling. An explicit quantification of the spatial errorassociated with the measurement of net ecosystem pro-duction (NEP) by the eddy flux method will allow a quan-titatively defensible scaling-up from local to regional analy-ses of NEP. This achievement will require replication ofeddy flux towers at both local and regional scales. Thenumber and spacing of such towers is a topic forresearch,but techniques are available that can be appliedto the task (e.g.,Harrison and Luther 1990).

A central theme of research on the terrestrial carbonsink over the last decade can be formulated as the follow-ing hypothesis:

Hypothesi s 1 : Ther e i s a l ar ge ter r estr i alsi nk for anthr o- pogeni c CO2 i n theNor ther n Hemi spher e.

This hypothesis is known by the term the NorthernHemisphere Terrestrial Carbon Sink or “NorthernLand Sink.”

The suggestion that there is such a carbon sink isstrongly supported by a wide range of observations,dataanalysis,and modeling studies,many of which have beendiscussed above. The magnitude of this sink was estimat-ed to be on the order of 1.5 + 1.0 Gt C/yr for 1980-1989(Schimel et al.1996),and as having increased to >2.5 GtC/yr over the last decade,making it a major contributor to


10


11

(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 sitesshow 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:thegrowing season starts earlier at the BOREAS site, yet the forest there is less efficient overall at storing carbon.Datafrom towers can also be analyzed in conjunction with temperature and precipitation patterns, for example, to deter-mine 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 longergrowing season than average. This result is partially due to the release of carbon from cold soils starting earlier inthe year, around August, and persisting throughout the winter months.

(A)

(B)

perturbations of the carbon budget. However, in additionto the large uncertainty in the magnitude of this sink,itslocation and interannual variability, and the mechanismsthat cause it are all highly speculative.

One important issue to address in this context is therole played by CO2 and nitrogen fertilization in the cur-rent carbon balance.

These two types of fe rtilization may stimulate bioge n i cu p t a ke of carbon, thus providing a carbon sink. T h ey arege n e ra l ly believed to be most important at mid-nort h e rnlatitudes (Schimel et al. 1 9 9 6 ) , but these mechanisms mu s tbe further quantified and their possible future ro l e sa s s e s s e d . Ecosystem manipulations, i n cluding studies withe l evated CO2 and nitro gen additions, and consideration ofa ra n ge of potentially interacting fa c t o rs , will prov i d einsights into phy s i o l o gi c a l ,e c o l o gi c a l , and bioge o ch e m i c a lp rocesses underlying the carbon storage in the terre s t ri a lb i o s p h e re . These same points apply to controlled applica-tions of ozone. P rojects investigating these processes arec o o rdinated by the Global Change and Te rre s t rial Eco-systems (GCTE) project of the International Geosphere -B i o s p h e re (IGBP) Pro gra m . The results from these ex p e ri-ments will need to be ex t rapolated from the ex p e ri m e n t a lscale to larger re gi o n s , re q u i ring understanding of thescale-dependence of the processes and their intera c t i o n s .

Extending the results of these experiments from thelocal to regional and global scales will require four com-ponents. First,the experiments must be designed to pro-vide access to the underlying mechanisms that control thescaling. Second,the experiments must be repeated inenough ecosystems to capture the full range of mecha-nisms. Third,the regions where the responses are estimat-ed must be sufficiently characterized to support theextrapolation. And fourth,the models used for the scalingmust be robust. The terrestrial biosphere is too heteroge-neous to be characterized by simple extrapolation of areasonable number of experiments,but the processes thatregulate it are uniform enough to be effectively handled inmechanistic models.

Another fundamental issue in this area is how carbonstorage is affected by changes in land use,disturbance,andvegetation structure and composition.

Carbon storage in vegetation is being continuallyaltered by land use and disturbance (e.g., fires) and withthe recovery of vegetation from historic land use. In addi-tion,the competitive relationships among species andtheir interdependence are changing owing to multiplecauses, climate change among them. A change from grass-land to forest, for example, would certainly change carbonstorage. Changes in vegetation assemblages are likely tobe important drivers for carbon storage or release ondecadal and longer time scales. The development ofdynamic global vegetation models (DGVMs) is beginningto accelerate,and they will likely become important

tools in understanding the future carbon cycle.

The development of measurement techniques andmethods of data analysis have placed us in a position topropose a major new near-term CCSP initiative with thefollowing goal:

Goal 1 : Establ i sh accur ate est i mates ofthe magni tude of the potent i alNor ther n Hemi spher e ter r est r i al car -bon si nk and the under l y i ng mechani sms that r egul ate i t .

Large carbon sinks have been hypothesized to be asso -ciated with a variety of terrestrial phenomena (for exam-ple, greater erosion and sediment deposition,land usechanges,CO2 fertilization,and length of the growing sea-son). Clearly, additional observations and improved mod-els of terrestrial processes are needed.

Such terrestrial studies cannot be interpreted in isola-tion. The Northern Land Sink hypothesis directly suggestsadditional linked hypotheses and observations. Asresearch on land clarifies the distribution of particularhypothesized sources and sinks,parallel improvementswill be required in the temporal and spatial resolution ofatmospheric and oceanic observations and models. Forexample,a direct corollary of the Northern Land Sinkhypothesis is the hypothesis that there is a large oceaniccarbon sink in the Southern Hemisphere. This sink ishypothesized because the global CO2 budget requires alarge ocean sink somewhere,and Northern Hemisphereocean CO2 uptake appears to be relatively small based onthe atmospheric 13C and ocean surface pCO2 (partialpressure of CO2) measurements (Tans et al.1990). Theseare the same measurements and arguments that led to thehypothesis of the Northern Land Sink.

The Ocean Carbon Sink

The total capacity of the ocean to dissolve anthro-pogenic CO2 is mainly a chemical property and can becalculated by considering the appropriate chemical equi-libria. The primary challenge to current understanding ofthe ocean carbon sink is in estimating the rate at whichanthropogenic CO2 dissolves in the ocean. It is necessaryto know the present rate of dissolution and how climatechange and changes in the biological pump will affect theuptake rate and ultimate capacity in the future. A sepa-rate,but also important,problem is to determine the spa-tial distribution of air-sea CO2 fluxes,a factor that is amajor constraint on inverse modeling analyses such asthose discussed in the previous section. Each of theseissues is discussed in turn,beginning with present knowl-edge of the rate of ocean carbon uptake and the spatialdistribution of carbon sources and sinks. Chapter 3 addi-


12

tionally discusses how climate change and related changesin the biological pump might affect the rate of uptake inthe future.

Ocean carbon uptake has been estimated by consider-ing the change in carbon inventory through time,the mag-nitude of the air-sea CO2 flux,and the magnitude of car-bon transport within the ocean. The estimates used bythe IPCC in constructing global carbon budgets are basedprimarily on inventory calculations using ocean circula-tion models calibrated or validated with tracer observa-tions (e.g.,bomb radiocarbon;see Schimel et al.1996).The magnitude of these estimates has remained consistentover almost three decades of continuous testing with newmodels and tracer data. Nevertheless,the estimates aretypically reported as having an uncertainty of ± 40 per-cent. The large uncertainty reflects the fact that tracermeasurements are imperfect analogs for fossil fuel CO2and have been relatively sparse until quite recently.

Further, ocean models have limitations in their repre-sentations of ocean circulation and mixing. An importantgoal of research over the last decade has thus been toexpand the tracer data set and improve ocean carbonmodels. The skill of ocean carbon cycle simulations isdirectly linked with improving models of the physicaltransport of the underlying ocean general circulation.Considerable success has been achieved in the last decadethrough better parameterization of mesoscale eddy mix-ing processes (Gent et al.1995).

Additionally, there has been a major effort to developmethods and obtain measurements to assess the oceanicuptake directly or indirectly from dissolved inorganic car-bon (DIC) observations.

One of the most important advances has been theacquisition of a new global data set of ocean tracer andcarbon system observations of unprecedented accuracythrough the World Ocean Circulation Experiment/JointGlobal Ocean Flux Study (WOCE/JGOFS). The tracermeasurements,together with the rapid improvement ofocean circulation models based on these and other WOCEobservations,are expected to lead to improved modeluptake estimates. DIC data sets accurate to 1 to 2mmol/kg,which is equivalent to 1 to 2 years’uptake ofanthropogenic CO2 in near-surface waters,are now avail-able for hydrographic transects representing most of theworld’s ocean. The 13C/12C of DIC was measured onevery sample collected for Accelerator Mass Spectrometry(AMS) 14C measurement during WOCE,which yields thefirst oceanwide high-quality 13C/12C data set (±0.03 partsper thousand). These new data sets are already significant -ly improving knowledge of the distribution of inorganiccarbon in the ocean.

A parallel occurrence has been the development ofmethods to estimate the oceanic uptake of carbon bydirect analysis of the DIC observations. Three comple-

mentary methods have been proposed in order to con-strain (or set the estimated outside boundaries of) theoceanic uptake of anthropogenic carbon directly fromocean data: (1) repeat or time-series observations ofwater column DIC to measure the current rate of changeof the DIC inventory; (2) “preformed CO2” methods tocalculate the integral change of the DIC inventory fromthe pre-industrial period to the present;and (3) air-seaCO2 flux methods to measure the present global net fluxinto the ocean.

One of the most important needs is to gain a betterunderstanding of the spatial patterns of surface oceanCO2 concentrations and their variability. Seasonal andinterannual variability of CO2 concentrations in the sur-face ocean are one to two orders of magnitude greaterthan their annual increase due to uptake of anthropogeniccarbon (e.g.,Bates et al.1996,Winn et al.1994). Becausethe signal to be detected is much smaller than this vari-ability, it takes a decade or longer to begin to see anthro-pogenic trends. In addition,the seasonal and interannualvariability in CO2 concentration gives information on howthe carbon cycle functions,and can be used in conjunc-tion with other methods to help understand regional andglobal patterns of carbon uptake. A very promising devel-opment of the last decade is new instrumentation thatwill make it possible to measure pCO2 (DeGrandpre et al.1995, Friederich et al.1995,Goyet et al.1992,Merlivat andBrault 1995) and other properties from moorings. TheCCSP envisions that moorings with these capabilities willallow establishment of many more time-series stations atfeasible cost in otherwise remote locations.

An alternative to long-term measurements is to usemethods of analyzing the data that permit identifying theanthropogenic carbon component within the huge back-ground signal and variability that arises from seasonal andinterannual change. The “preformed CO2” method usesthe correlation of data on carbon, nutrients, oxygen,andphysical variables to separate natural variability fromchanges due to uptake of anthropogenic CO2 (see Gruberet al.1996). In a second method, multiple linear regres-sion coefficients are calculated for carbon versus othervariables (Wallace et al.1995). These coefficients are usedto correct for the natural variability between measure-ments made at two different times,thereby isolating thechange due to addition of anthropogenic carbon. The twomethods are somewhat independent and address differenttime scales. The preformed CO2 method gives an estimateof the oceanic uptake of anthropogenic CO2 since pre-industrial times;the multiple linear regression method hasbeen applied to estimate the increase in oceanic CO2 con-centration between the Geochemical Ocean Sections pro-gram (GEOSECS) expeditions and WOCE/JGOFS,spacedabout two decades apart (Wallace et al.1995). Synthesesof Atlantic Ocean (Gruber 1998) and Indian Ocean(Sabine et al.1999) DIC data from WOCE/JGOFS,analyzed


13

by the preformed CO2 method,show large regional dis-crepancies between observed and modeled penetration ofanthropogenic carbon. It is already clear that theobserved penetration of anthropogenic CO2 can provide,through detailed comparison with models,the basis andimpetus for improving models of oceanic CO2 uptake.

The flux of carbon between surface waters and theatmosphere can also be constrained using data of the par-tial pressure difference between the air and the water,∆pCO2, combined with estimates of the gas exchangecoefficient (Takahashi et al.1997). Of course,this fluxincludes both a natural and an anthropogenic component.The value of the technique lies primarily in determiningthe spatial distribution and temporal variability of air-seaCO2 fluxes. It is particularly useful where the signals arelarge,as in the North Atlantic and Equatorial Pacific,or inthe contrast between El Niño and non–El Niño years.

One curre n t ly weak link in the air-sea CO2 fl u xa p p ro a ch is poor understanding of the kinetics of thep rocess of air-sea gas ex ch a n ge (see Wanninkhof 1992).The degree of nonlinearity in the relationship between theg a s - ex ch a n ge and wind speed is poorly unders t o o d ,p a rt i c-u l a r ly at high wind speeds where bubble effects maybecome import a n t . For instance, if the ex ch a n ge ve l o c i t ywe re to depend more stro n g ly than is curre n t ly believe don va ri ables correlated with wind speed, our estimates offl u xes in re gions with high wind speeds, s u ch as the highl a t i t u d e s , would incre a s e . It is time to determine thro u g hm e a s u rements a mu ch better para m e t e rization of theex ch a n ge ve l o c i t y, so that observations of the atmospherecan be tru ly info rm a t i ve for oceanographic issues, and viceve rs a . S u r face sources and sinks dri ve large-scale gra d i e n t sof atmospheric CO2 c o n c e n t rations that could be, fo rex a m p l e , an important constraint on CO2 u p t a ke by theS o u t h e rn Hemisphere ocean polewa rd of about 30°S. A sa t m o s p h e ric tra n s p o rt models continue to improve ,t h i sc o n s t raint will become more compelling as a major inputfor inve rse models of atmospheric CO2 o b s e rva t i o n s .

Another analysis approach that makes use of oceanicDIC data is estimation of horizontal carbon transport with-in and between ocean basins (e.g.,Brewer et al.1989,Holfort et al.1998). This approach permits separateassessments of the transport of natural and anthropogeniccarbon components in the ocean. The air-sea flux can beinferred from the divergence of the horizontal transport,which would provide a means of coupling the oceaninventory method with the surface ∆pCO2 method. Whilehorizontal carbon transport estimation is only starting tobe implemented in the ocean,the tremendous improve-ment in ocean DIC measurements and the large new dataset gathered by WOCE/JGOFS show great promise for thenear future.

Knowledge of the spatial distribution of the air-sea CO2flux varies greatly from one region to another. The North

Atlantic is perhaps the best-constrained region. Models,observation-based estimates of air-sea flux,and estimatesof the ocean inventory of anthropogenic carbon allconverge on similar answers (Gruber 1998,Gruber et al.1996,Sarmiento et al.1995,Takahashi et al.1997).However, difficulties remain in reconciling these estimateswith independent estimates of the net meridional trans-ports of carbon (Brewer et al.1989;Broecker and Peng1992;Holfort et al.1998, Keeling and Peng 1995). TheNorth and Equatorial Pacific Ocean have not been as thor-oughly explored,though the measurements that are avail-able show reasonable consistency. The rest of the world ismostly very poorly constrained.

Analysis of CO2 exchange in the Southern OceanOcean (waters around Antarctica poleward of approxi-mately 50°S latitude) is especially dif ficult because of thehigh spatial and temporal variability of its properties andthe difficulty of working in that region. The SouthernOcean appears to be especially important to levels ofatmospheric CO2. Deep ocean waters are mixed to shal-low depths there,providing one of the major pathways forthe uptake of anthropogenic CO2 into the deep ocean.Wind speeds are high,enhancing rates of air-sea CO2exchange. The dissolved macronutrients available at thesurface for plant growth are significantly underutilized(e.g., Falkowski et al.1998),suggesting a capacity forenhanced uptake of CO2 by photosynthesis. Macro-nutrients may be underutilized for several reasons;there isless sunlight at high latitudes,biological processes areslower because of low temperatures,there is less time forbiota to develop because surface water is mixed back intodeeper layers more quickly, and necessary micronutrientssuch as iron may be less available (Baar et al.1995,Falkowski 1995). All these factors may be affected byclimate change.

Models of the oceanic uptake of anthropogenic CO2(e.g.,Sarmiento et al.1992) predict that a substantialamount is being absorbed south of 30°S. Yet there arelarge discrepancies between the observations and themodels of oceanic CO2 uptake in the high southern lati-tudes. Recent survey data for the southern Atlantic(Gruber 1998) and Indian (Sabine 1999) oceans show sur-prisingly little uptake in these areas. The (sparse) atmos-pheric data also point toward a small Southern Oceansink. Recently, observations of a composite atmospherictracer, effectively equal to the sum of atmospheric O2 andCO2 concentrations, were compared with predictionsfrom three different ocean models (Stephens et al.1998).The largest discrepancy between models and observationswas found over the Southern Oceans,suggesting that themodels may misrepresent exchanges of O2 and CO2 inthis region.

There are few constraints on the temporal variability ofthe air-sea flux of CO2. The Equatorial Pacific is beingmonitored reasonably well (e.g., Feely et al.1995, Feely


14

et al.1994, Feely et al.1996),and the ongoing feasibilitystudies to instrument the ATLAS moorings of the Tropical-Atmosphere-Ocean (TAO) array with CO2 sensors are animportant development (Friederich et al.1995). There arealso two time-series measurements near Bermuda andHawaii (e.g.,Bates et al.1996,Winn et al.1994). However,most of the ocean is unknown with regard to the tempo-ral variability of CO2 fluxes. Analyses of stable carbon iso-topes in atmospheric CO2 suggest that sinks in both theocean and the terrestrial biosphere vary by large amountsfrom year to year (Keeling et al.1989, Keeling et al.1995a,Ciais et al.1995b, Francey et al.1995). No oceanic mecha-nism has been developed to explain such conspicuousvariability. A related need is for the calculation of tempo-ral changes in oceanic carbon uptake on the global scale.The global oceanic uptake rate of CO2 is currently notknown to better than about ± 40 percent. This knowl-edge is not sufficient to determine whether the ocean

carbon uptake rate has increased or decreased over thepast few decades. Moreover, uncertainties of this magni-tude limit the ability to constrain the historical global CO2budget using the record of atmospheric CO2 concentra-tions over the last 200 years.

However, the problem of understanding the oceanshould not be viewed as simply a comparison of currentand future DIC “snapshots.” It is also important to identifyand understand the mechanisms that might cause futurechanges in the ocean carbon sink.

One of the most important issues to address is whichmechanisms of ocean circulation that significantly shapeanthropogenic CO2 uptake are likely to be af fected by achanging climate?

Changes in ocean circulation resulting from climatechange will immediately affect the way anthropogenic


15

Distribution of the mean annual sea-air pCO2 flux (partial pressure of carbon dioxide, moles CO2/m2/yr) over theglobal oceans estimated for a reference year 1995. These data show the global pattern of surface CO2 uptake andrelease by the ocean. Note that the major sink regions are the North Atlantic and Southern Oceans, and that theEquatorial Pacific is a large source region in a typical year. This map does not reflect the variability due to ElNiño/Southern Oscillation (ENSO) cycles, for example, which can alter the size of the ocean sink on an interannualbasis. This map has been constructed based on about 2 million measurements of sea-air pCO2 difference made overthe past 25 years. These values have been corrected to a reference year of 1995 for the increase in pCO2 of theatmosphere and surface ocean water that has occurred since the measurements were made, and the measurementsmade during El Niño years in the equatorial Pacific have been excluded. Thus, the map represents a climatologicalmean for non-El Niño conditions. The net CO2 flux across the sea surface has been computed using the effect ofwind speed on the CO2 gas transfer coefficient formulation by Wanninkhof (Equation 1, 1992) and the mean month-ly wind speed of Esbensen and Kushnir (1981). The numerical method used for the construction of these maps hasbeen described in Takahashi et al.(1997) and Takahashi et al.(in press). The map yields an annual CO2 flux for theoceans of 2.2 PgC/yr, in which the North Atlantic (N of 14°N) and the Southern Ocean (S of 50°S) are major CO2sink areas taking up 0.8 and 0.6 PgC/yr respectively.

CO2 is exchanged between the atmosphere and theocean. The behavior of the present and past ocean,including the response to temporal variability, providesthe strongest clues to the links between ocean circulationand atmospheric CO2.

An additional question is how is the biological pumpaffected by the thermohaline circulation and changingclimate? And is there any evidence that the C/N and C/Pratios of marine production are changing,or that “pre-formed” nutrients (the nutrient concentration of waterthat has cooled and sunk to depth) are changing?

C O2 ex ch a n ge between the ocean and atmosphere nat-u ra l ly occurs at the ocean surfa c e . Photosynthesis byo rganisms in the upper, sunlit layer of the ocean ke e p sthe CO2 c o n c e n t ration of the surface wa t e rs substantiallyl ower than that of deep wa t e rs . It does so by pro d u c i n go rganic carbon that is ex p o rted to the deep ocean whereit is conve rted to DIC. The excess deep ocean DIC thatthus results is analogous to the organic carbon stored ins o i l s , in that it is isolated from the atmosphere . Wi t h o u tphotosynthesis in the ocean, and assuming no other sur-face ch a n ges due to organisms (such as calcifi c a t i o n ) ,t h eexcess deep ocean DIC would escape to the atmosphereand atmospheric CO2 c o n c e n t ration would be betwe e n900 and 1,000 ppm (the current va l u e , ag a i n , is 365p p m ) . I f, on the other hand, photosynthesis continu e deve ry w h e re until all of the plant nu t rients we re fullydepleted in all surface wa t e rs ,a t m o s p h e ric CO2 would beb e t ween 110 and 140 ppm. (The actual pre - i n d u s t ri a la t m o s p h e ric CO2 c o n c e n t ration was 280 ppm.) T h e s efi g u res indicate the great power of the oceanicb i o l o gical pump.

L o n g - t e rm surface ocean time series that incl u d edetailed bioge o chemical and CO2 m e a s u rements arei m p roving the mechanistic understanding of pro c e s s e sthat affect ocean-atmosphere carbon uptake and part i-t i o n i n g . Ti m e - s e ries data sets from Bermuda and Hawa i ia re being used to develop model para m e t e rizations ofocean bioge o chemical processes affecting the ocean car-bon cycle (Doney et al. 1 9 9 6 , Fasham 1995). S u ch times e ries are also incre a s i n g ly being used as test beds fo rn ovel high-resolution measurement instru m e n t s . N o t abl efindings from the time-series sites include an incre a s e dawa reness of the complexity of the ocean’s nitro ge nc y cle (Karl et al. 1 9 9 7 ) . This finding has the potential toalter the view of the sensitivity of atmospheric CO2 c o n-c e n t rations to biological processes in the ocean( Fa l kowski 1997). On longer time scales, ocean biologycan have an impact on the atmosphere (see the fo l l ow i n gch a p t e r ) . The time-series sites are among the few loca-tions where seasonal va riations in upper ocean 1 3C /1 2Cratios are measured (Bacastow and Keeling 1973). S u cht i m e - re s o l ved info rmation is critical for corre c t ly inter-p reting the large “ s n a p s h o t ” data sets collected along

ocean tra n s e c t s .

To summari z e , our current understanding is that, a sa t m o s p h e ric CO2 l evels increase through time, the oceanresponds by dissolving more CO2 in the surface mixe dl aye r, and by mixing the CO2- e n ri ched surface wa t e rsd ow n wa rd through ex ch a n ge with deeper wa t e rs . T h epossibility also exists that ch a n ges in the biological pumpm ay affect the future air-sea balance of CO2. If our pre s-ent understanding of these processes is corre c t ,t h eocean should have the capacity to absorb large quantitiesof anthro p o genic CO2 over time scales of decades to mil-l e n n i a . Most model simulations of future atmospheri cC O2 l evels assume that pre s e n t - d ay fa c t o rs contro l l i n gplant phy s i o l o gy, air-sea ex ch a n ge , and ocean mixing willremain constant into the future (Schimel et al. 1 9 9 6 ) .H oweve r, as prev i o u s ly noted, these assumptions havebeen questioned, e s p e c i a l ly because of the potential fo rs i g n i ficant responses to climate ch a n ge . Air-sea gasex ch a n ge , ocean circ u l a t i o n , and marine photosynthesisa re susceptible to ch a n ges in air tempera t u re s ,w i n dve 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-tion pattern s . The most powerful tool for unders t a n d i n gthe mechanisms underlying potential ch a n ges is in study-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 nall these considera t i o n s , a major CCSP initiative inocean carbon cycle re s e a rch is proposed to ach i evethe fo l l ow i n g :

Goal 2 : Establ i sh accur ate est i mates ofthe oceani c car bon si nk and theunder l y i ng mechani sms that r egu-l ate i t .

Land Use

One of the main driving factors in determining theNorthern Land Sink may be land use,both past and pres-ent. For example,there has been widespread reforestationsince 1900 in the eastern United States following themovement of the center of agricultural production towardthe Midwest. Also,less agricultural land is needed todaythan during the first half of this century;the productivityof agriculture has improved so much that double the out-put can be produced on half as much land. The heavy useof fertilizer, together with improved tilling practices,mayalso lead to increased stores of organic matter in soils.

The carbon balance in the tropics also affects estimatesof the magnitude of the Northern Land Sink. As pointedout before,the major constraint on the estimated magni-tude of the global terrestrial carbon sink comes from twonumbers. The first is the estimate of a global net terrestri-al uptake of 0.2 ± 0.9 Gt C/yr for the period from 1980 to1989. This number is obtained from calculating fossil fuel


16


Top panel: Managed forests in the Coast Range of Oregon. Conversion of old-growth forests to managed plantationsin the Pacific Northwest has reduced the store of carbon to less than 25-35% of the maximum value in the lastcentury. This has resulted in a substantial loss of carbon to the atmosphere. Altering management by increasingthe interval between harvests and/or removing less carbon each harvest would “re-store” much of this carbon overthe 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 agri-culture. This conversion has reduced carbon stores on these sites at least 5-fold.

17

emissions minus atmospheric growth and ocean uptake.The second number is the estimate of 1.6 ± 1.0 Gt C/yrreleased to the atmosphere from tropical deforestation.The difference between these numbers gives a requiredterrestrial uptake of 1.8 ± 1.4 Gt C/yr (Schimel et al.1996), much of which appears to be in the NorthernHemisphere. If the net carbon flux from changes in tropi-cal land use were at the lower limit of 0.6 Gt C/yr, theNorthern Land Sink would drop to 0.8 Gt C/yr.Maintaining the observed north-south gradient of CO2 inthe atmosphere would require that the Southern Oceanwould have to be a smaller sink than currently estimated.The recent land use estimate of Houghton et al.(1998)yields 2.0 ± 0.8 Gt C/yr for the tropical land use source.This value would require a compensatory terrestrial car-bon sink of 2.2 Gt C/yr as well as a larger Southern Oceancarbon sink. Gaining more certainty in the land use num-bers will reduce uncertainty in other contributing compo-nents of the Northern Land Sink,such as the fertilizingeffects of rising atmospheric CO2 and N deposition.

The uncertainty of net carbon flux from land use stemslargely from incomplete and often incompatible databasesused to compile land use changes,and from the lack ofknowledge of carbon fluxes associated with specific activ-ities. These points are particularly true of tropical areasand for land uses involving economically nonproductiveecosystems (i.e.,nonproductive in a direct sense),such aswetlands, riparian forests,and natural grasslands. Forexample,Houghton et al.(1998) notes that no reliabledata exist for Latin America on that portion of the loss ofagricultural land to degraded land that is not recovering toforest. This situation has forced the omission of such landfrom the calculation of net carbon flux. The same is truefor land subjected to shifting cultivation and the harvestof wood in Sub-Saharan Africa.

Tests of the Northern Land Sink hypothesis will thusrequire the further development and refinement of datasets on historical land use changes,carbon stores per unitarea,and models that can use this information. The goalshould be to acquire and analyze accurate global invento-ries of highly fractionated land use change. Doing so,however, requires a concerted effort to develop high-quali-ty, spatially explicit,long-time-series data sets on land use.These data sets should be constructed from a variety ofsources,including high-resolution, remotely sensedimagery, explicit data on land use going back manydecades, government publications, research data,andreconstruction using proxy information. Side-by-side com-parison of population change and measured rates of defor-estation,where the data coexist,lends confidence to thereconstruction of rates of deforestation using relativelywell-defined changes in population as the predictor.

A major new CCSP initiative is therefore proposed tomeet the following goal:

Goal 3 : Establ i sh accur ate est i mates of thei mpacts of hi stor i cal and cur r entl and use pat ter ns and t r ends onthe evol v i ng car bon budget at l ocal to con-t i nental scal es.


18

The previous chapter addressed the first of the twooverarching questions that the Carbon Cycle Science Plan(CCSP) must attempt to answer. In this chapter, we turnto the second of the two fundamental issues for carboncycle research:

What will be the future atmospheric carbondioxide concentrations resulting from bothpast and future emissions?

The re s e a rch pro gram outlined in this re p o rt will ulti-m a t e ly be measured by its ability to provide re l i able esti-mates of future atmospheric CO2 c o n c e n t rations under dif-fe rent conditions. O n ly with such know l e d ge will it bep o s s i ble to assess altern a t i ve scenarios of future emissionsf rom fossil fuels, e ffects of human land use, s e q u e s t ra t i o nby carbon sinks, and responses of carbon cycling to poten-tial climate ch a n ge . T h u s , the fo remost reason for addition-al re s e a rch is to develop the ability to predict responses ofthe global carbon cycle to va rious types of ch a n ge . T h eCCSP must be integrated in the fo rm of a ri go rous andc o m p re h e n s i ve effo rt to build and test models of carbonc y cle ch a n ge , to evaluate and communicate uncert a i n t i e sin altern a t i ve model simu l a t i o n s , and to make these simu l a-tions ava i l able for public scru t i ny and application. T h emodels must also be capable of evaluating altern a t i ves c e n a rios for management of the carbon cycl e .

There are grounds for optimism that in coming yearsthe fate of CO2 in the ocean can be ascertained with rea-sonable accuracy. The global mass balance among emis-sions,atmosphere,ocean,and terrestrial biosphereensures that a better quantitative estimate of oceanuptake also improves the estimate for changes in terrestri-al storage,if only in very coarse geographical detail.Direct measurements of terrestrial inventories and fluxes,in conjunction with atmospheric measurements and mod-els,will help to refine the geographical details.

Unfortunately, improved knowledge of the environmen-tal fate of historical CO2 emissions cannot by itself give usconfident predictions of future atmospheric CO2. TheCO2 concentration trajectories calculated by the Inter-governmental Panel on Climate Change (IPCC) for scenar -ios of fossil fuel burning assume that the future carboncycle will continue operating exactly as it is thought tohave operated in the past. This assumption is not likely tobe correct.

There is a fundamental dif ference between the oceanand the terrestrial biosphere for policy decisions relatingto greenhouse gases. The ocean remains the biggest long-term player in the carbon cycle,and any research program

Chapter 3: Predicting the Future Carbon Cycle

19

that neglects the ocean is doomed to be nearly irrelevantfor policy. However, direct human interventions in theocean carbon cycle have thus far been minimal.Furthermore,any future human interventions,such asdirect injection of CO2 in the deep ocean or enhance-ment of the biological flux of carbon by fertilization,willlikely be dwarfed by the magnitude of the ongoing pas-sive uptake. On the other hand,humanity is alreadymanipulating the terrestrial biosphere on a global scale,and its influence on atmospheric CO2 is substantial. Theeffect on the carbon cycle of ecological interventions onland has been mostly inadvertent to date. Most of theinterventions have resulted in decreasing the amount ofcarbon in various terrestrial carbon reservoirs. In particu-lar, woody biomass and active soil organic matter, becauseof their intermediate turnover times (30 to 100 years),arethe largest terrestrial pools af fected by land use conver-sion and agricultural establishment. This past reduction interrestrial carbon storage,however, suggests the opportu-nity to increase carbon in terrestrial systems throughintentional management. The realization that sequestra-tion of carbon in wood and soil may play a significant rolein offsetting CO2 from fossil fuel burning is already evi-dent in international negotiations.

Concerning how high atmospheric CO2 might go inthe future,two general questions must be answered:

• What will be the partitioning of carbon amongthe mobile reservoirs, and how will climatechange affect this partitioning?

• How can the future growth of atmospheric CO2be managed?

Each of these points is examined,and thereafter, amajor new research initiative is proposed to address themost compelling scientific issues,an initiative that is alsoscientifically feasible and cost-effective.

Projecting Future Atmospheric CO2Concentrations

The IPCC has provided some scientific basis for inter-national policy decisions by extrapolating the historicalbehavior of the terrestrial biosphere and ocean into thefuture. Again,the assumption is that carbon uptake by theterrestrial biosphere will continue to occur through thesame mechanisms as at present,and that ocean circulationand biology will remain constant through time (Houghtonet al.1996). However, from coupled atmosphere-oceansimulations with time-dependent radiative forcing

(e.g.,Haywood et al.1997),it appears possible that the ter-restrial biosphere and ocean carbon cycle might alreadybe experiencing direct effects of climate change today.These effects could become even larger over the next cen-tury (e.g.,Cao and Woodward 1998,Sarmiento et al.1998).Further, it seems likely that both terrestrial and oceanicecosystems will undergo significant indirect responses toclimate change and human impacts on the environment,such as NOx [nitrogen oxides] fertilization,air and waterpollution,and CO2 increase. Such shifts might includechanges in species distribution in addition to changes inthe supply of nutrients and other ecosystem componentsthat determine carbon cycling.

Terrestrial Ecosystems

Terrestrial ecosystems have played and will continue toplay a significant role in the global carbon cycle. Releaseof CO2 from land use change has been a significant flux tothe atmosphere historically, and could well continue oreven accelerate. In addition,there appears to be a signifi-cant sink of CO2 in land ecosystems arising from the syn-ergistic effects of past land use changes,increasing atmos-pheric CO2, the deposition of fixed nitrogen,and,possibly,climate changes over the past century or so. To projectthe consequences of given human activities (such as fossilfuel burning and land use change),it is essential to under-stand the responses of terrestrial ecosystems. It is like-wise important to understand how climate interacts withterrestrial ecosystems. Several factors will control the bal-ance of terrestrial ecosystems with respect to carbon. Thecurrent degree of uncertainty regarding all of these factorsis high.

Increasing CO2 and fixed nitrogen from fossil fuel burn-ing can both act as fertilizers to ecosystems,increasing netprimary production (NPP, the amount of carbon processedby photosynthesis in green plants minus that lost throughrespiration),and possibly carbon storage. Air and waterpollution can lead to degradation of the ecosystem andloss of carbon. Observational and manipulative studieshave not yet yielded unambiguous results about the mag-nitudes of these effects,and there are thus substantial dis-agreements in model predictions.

There is evidence that climate change and variabilityover the past century may have influenced both the distri-bution of vegetation and its productivity. The potentialexists for significant positive climate feedback in whichwarming of arctic soils and permafrost,high in organiccarbon content,could stimulate the oxidation to CO2 ofmuch of that carbon (Goulden et al.1998,Oechel et al.1993). Most models of vegetation dynamics suggest possi-ble major redistribution of vegetation zones with futureclimate change. Changes in the extent of ecosystems such

as carbon-rich peatlands and forests or carbon-poordeserts could substantially affect carbon storage on land.Although slow changes to long-lived vegetation are dif fi-cult to study experimentally, and models of these process-es are in early developmental stages,an integrated pro-gram of manipulative experiments,paleoecological stud-ies,and continuing observations can facilitate continuingimprovements in the models.

Significant integrative research is needed to improvepredictive capability for terrestrial ecosystems,both man-aged and natural. First,the basic science and its represen-tation in models must be improved. Better representationis needed of key processes,such as responses to distur-bances,CO2 warming, nutrient deposition,and atmospher-ic pollution,including their interactions. Important ques-tions are emerging that require new direct evidence fromobservational and manipulative studies. Analyses of atmos-pheric observations also require better models of terrestri-al ecosystems. Models are an important tool in inverseanalysis of atmospheric data (inverse analysis is a mathe-matical tool that infers unknown variables from a givenset of equations,data and assumptions).Models of theland biosphere are urgently needed for data analysis.Beyond global models,detailed site-specific models arealso required to assess the long-term consequences ofmanagement options. Such models can be used toappraise the influence of a given forest or agriculturalmanagement practice on both commodity production andenvironmental impacts at the same time,including carbonstorage and trace gas exchange. Improving projections ofthe future carbon balance of terrestrial ecosystemsrequires an integration of global carbon cycle and climatemodels (to calculate climate-ecosystem feedbacks onatmospheric CO2) and management-oriented models foruse in decision making about terrestrial ecosystems.Decisions about terrestrial ecosystems will influence—perhaps heavily—the future effects of ecosystems onthe atmosphere.

The Ocean

Coupled atmosphere-ocean model simulations (e.g.,H ay wood et al. 1997) predict a large wa rming of thes u r face wa t e rs of the ocean (e.g., of 2.5°C by mid nex tc e n t u ry ) . T h ey also predict increased stra t i fication of thes u r face ocean from the higher tempera t u re at low latitudesand increased precipitation in high latitudes. One of theconsequences of these ch a n ges may be a reduced rate offo rmation of North Atlantic Deep Wa t e r, perhaps as early asthe next decade. H oweve r, on time scales of a fewd e c a d e s , ch a n ges in the rate of convection and ve rt i c a lmixing appear more important for the surface CO2 b a l a n c ethan ch a n ges in adve c t i o n . F u rt h e r, in at least one simu l a-t i o n , ocean CO2 u p t a ke is part i c u l a r ly impacted by ch a n ge s


20


21

Global, annual net primary production (NPP) (g C/m2/y) for the land and ocean biosphere. This calculation is frommodels 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 onbroadly observed patterns to scale photosynthesis and growth from the individual to the ecosystem level.This calcula-tion, 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 bythe land. These approximately equal contributions to global NPP highlight the role of both land and ocean processesin the global carbon cycle.

in the Southern Ocean compared with other re gions of theo c e a n . These ch a n ges in ocean mixing and circulation mayreduce cumu l a t i ve oceanic uptake of CO2 by 10 to 30 per-cent between now and the middle of the next century(Matear and Hirst in pre s s ,S a rmiento et al. 1 9 9 8 ) .

Climate change may also have a major impact on oceanbiology, which in turn would affect the ocean’s uptake ofCO2. One change may be reduced or altered global pro-ductivity due to slower upward mixing of nutrients fromthe thermocline,or increases in productivity associatedwith anthropogenic nutrients or richer supply of micronu-trients by dust transport. Another may be change in taxo-nomic composition and physiology. Environmentalchanges that could drive such ecological shifts includewarming of surface waters and stabilization of the watercolumn.The supply of micronutrients by dust transporthas an important impact on ecology, as in making it possi-ble for nitrogen fixers to exist in nutrient-poor regions(Michaels et al.1996a). Also important are carbon chem-istry changes. The concentration of carbonate ion will

decrease by 30 percent and the pH by more than 0.2 inthe mixed layer by the middle of the next century. Somechanges in taxonomic composition and bulk physiologycould have an impact on important carbon cycle parame-ters. Among these are the ratio of calcium carbonate(CaCO3) to organic carbon production in coral reefs(Smith and Buddemeier 1992) and coccolithophorids,theratio of organic carbon to nutrients in material exportedfrom the surface,and the amount of carbon locked up asdissolved organic carbon,which presently amounts toabout 90 billions of metric tons,or 90 gigatons (Gt C),inthe upper 500 meters.

Time-series observations from the past decade showclearly that ocean biota respond dramatically to interannu-al climate variability such as the El Niño–SouthernOscillation (ENSO) (Karl 1999). Ocean warming due toclimate change could lead to changes of comparable mag-nitude. Simulations with ocean general circulation modelsshow that biological changes may increase the oceanicuptake of CO2 by 5 to 25 percent,depending on what


22

assumptions are made about how to model the biologicalresponse. This effect is large enough to counteract muchif not most of the reduction in uptake from mixing andcirculation,but the magnitude and even the sign of thiseffect are highly uncertain.

The relationship between marine ecosystem structureand the rate at which biological processes move carbonbetween surface and deep waters is an emerging researchtheme. The most prevalent ecosystem structure of theopen sea,particularly in the equatorial and subtropicalregions,is one dominated by the microbial loop. Themicrobial loop consists of very small organisms in a com-plex trophic structure with efficient nutrient recycling,lit-tle accumulation of biomass,and little export of carbonfrom surface waters to the deep sea (Landry and al.1997).Experimental evidence (e.g.,Coale et al.1996) shows thatnutrient perturbations to the steady state in these oceanregions leads to enhanced growth and dominance ofdiatoms and other large phytoplankton. Increased newproduction associated with diatom blooms, for example,increases carbon dioxide uptake from the atmosphere insurface waters and (eventually) leads to vertical export ofparticulate carbon to deep waters.

At high latitudes,and particularly in the SouthernOcean,increased stratification in response to climatechange may lead to a more efficient biological export ofcarbon from surface to deep waters,thereby reducing sur-face nutrients and carbon. The evidence for this ef fect isbased primarily on numerical models (e.g.,Sarmiento andLe Quéré 1996) and has not yet been confirmed by directobservation. In subtropical and equatorial waters,howev-er, future warming may cause the opposite effect.Increased stratification of the water column may lead to areduced supply of either micro- or macronutrients,a shiftto N-fixing organisms,and even lower carbon export thanoccurs today (Falkowski et al.1998). Some evidenceimplies that such changes are now occurring in theNorthern Hemisphere (Karl et al.1997,McGowan et al.1998). The potential contributions of such changes to theocean carbon sink are not yet understood. Hypothesesthat link warming-induced changes to upper-ocean stratifi-cation (and nutrient flux) and changes to marine ecosys-tem structure and carbon export are testable using longtime-series observations. They may also be tested byfocused process studies on ecosystem responses to inter-annual variability (e.g.,during warm versus cold years)and other “natural” experiments.

Current understanding of marine ecology and the abili-ty to predict the oceanic response to climate change areextremely rudimentary. It will take at least a decade ofresearch,including long time-series data taken at a num-ber of stations,to significantly improve understanding of

the potential responses of oceanic communities toclimate change.

Continuing to identify the main hypotheses and goalsfor the CCSP as in Chapter 2, we can then state anotherfundamental hypothesis for carbon cycle research:

Hypothesi s 2 : The oceani c i nventor y ofanthr opogeni c CO2 wi l l cont i nue to i ncr ease i nr esponse to r i si ng atmospher i c CO2 con-cent r at i ons, but the r ate of i ncr ease wi l l be modul ated by changes i n ocean ci r cul a-t i on, bi ol ogy, and chemi st r y.

This hypothesis of an increasing long-term ocean car-bon inventory (or “the Increasing Ocean Inventoryhypothesis”) provides a critical focus for concerns aboutthe future long-term ef fectiveness of CO2 sinks. Onevalue of a hypothesis concerning future CO2 sinks is thatit challenges us to anticipate when and how it can betested. The estimation of oceanic CO2 uptake has been aprimary objective of chemical oceanographic programsfor more than two decades (GEOSECS,TTO, SAVE,WOCE,JGOFS1). Using a variety of direct and tracer measure-ments,these programs have considerably improved theability to model the ocean CO2 sink. The recent advancesdiscussed previously suggest that trends in long-termoceanic CO2 uptake may soon be susceptible to muchmore direct verification.

Like the Northern Land Sink hypothesis,the IncreasingOcean Inventory hypothesis cannot be considered in iso-lation. For example,it explicitly encompasses the need tounderstand the ef fects of variations in oceanic circulation,biology, and chemistry. Such variations have been hypoth-esized to account for the seasonal and interannual variabil-ity in oceanic CO2 exchange associated with the phasingof the El Niño–Southern Oscillation and North AtlanticOscillation (NAO). Because ENSO and the NAO are alsoimplicated in short-term atmospheric CO2 variations,theymust be considered in the use of atmospheric transportinverse models to constrain the Northern Land Sink.Constraints on the spatial distribution of air-sea fluxes arecritical to determining the spatial distribution of terrestrialfluxes. Determining the North American or Eurasian flux-es with confidence requires knowing the North Pacificand North Atlantic fluxes. Thus,the Northern Land Sinkand Increasing Ocean Inventory hypotheses are directlyrelated and closely linked in a variety of important ways.

The Northern Land Sink and Increasing OceanInventory hypotheses represent a wide array of perspec-tives that are inherent in carbon-cycle research:not onlyterrestrial and marine,but also short-term and long-term,

1 Geochemical Ocean Sections program (GEOSECS);Transient Tracers in the Ocean (TTO);South Atlantic Ventilation Experiment (SAVE);World OceanCirculation Experiment (WOCE); Joint Global Ocean Flux Study (JGOFS).


23

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 theanthropogenic carbon in the oceans is found in the upper 1 to 2 kilometers except in regions of deep water forma-tion, such as the North and, to a lesser extent, South Atlantic.The ocean has a large capacity to dissolve anthro-pogenic carbon, but it takes many centuries to millenia for this capacity to be realized.Predicting the future uptakeof anthropogenic carbon by the oceans thus requires models of ocean circulation such as those shwn in the bottomfour panels of the figure. These models are able to reproduce the general features of the observed anthropogenic car-bon 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. Forexample, all of the models fail to simulate a sufficiently deep penetration between about 30°N and 50°N, and threeof 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 ofdata 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).

regional and global,spatial and temporal,diagnostic andprognostic. They illustrate the extent of integration that isnecessary in a framework of evolving hypotheses concern-ing the global carbon cycle.

To summarize,predicting the future role of oceanic andterrestrial biospheres in determining atmospheric CO2needs to consider the feedback among climate change andterrestrial and marine biogeochemical processes. Climatechange will affect the terrestrial biosphere and ocean;andchanges in the terrestrial biosphere and ocean will affectatmospheric CO2. Capturing these mechanisms willrequire more highly developed models of the terrestrialand oceanic biospheres,ocean mixing and circulation,andtheir response to warming. A major near-term goal of theCCSP is therefore the following:

Goal 4 : Impr ove pr oj ect i ons of futur eatmospher i c CO2 concent r at i ons thr ough acombi nat i on of mani pul at i ve exper i ments andmodel devel opment such that appr opr i ate bi o-physi cal and ecol ogi cal mechani sms and car -bon cycl e–cl i mate feedbacks ar e i ncor por atedi n gl obal cl i mate and car bon cycl e model s.

Management Strategies

The goal of carbon management strategies is to abatethe increase of anthropogenic carbon concentrations inthe atmosphere. Two approaches for accomplishing thisgoal are measures to reduce carbon emissions and meas-ures to increase the uptake of carbon by oceanic andterrestrial biosphere sinks. There are no fundamentalphysical or chemical barriers to using the energy poten-tially available in fuels much more ef fectively to bringdown the rate of emissions.Relying on low or non-CO2emitting energy sources is another way of reducingatmospheric CO2 concentrations in the long term.However, in the interim,it is also necessary to considerthe option of managing carbon uptake by the oceanicand terrestrial reservoirs.

Exchange with the large geological reservoirs of car-bon,such as limestone,is slow. The carbon added to theatmosphere will stay confined to the three “mobile” reser-voirs mentioned earlier (i.e.,atmosphere,terrestrial bio-sphere,and ocean) for hundreds of thousands of years. Inessence,the burning of coal,oil,and natural gas representsan artificial transfer of carbon from the geological reser-voirs to the mobile reservoirs.

The ultimate long-term sink for carbon of the threemobile reservoirs is the ocean. Model calculations showthat,on a time scale of 1,000 to 10,000 years,the oceanwill absorb about 85 percent of the anthropogenic carbonthat has been added to the combined atmosphere-ocean

since the beginning of the industrial revolution (e.g.,Maier-Reimer and Hasselmann 1987,Sarmiento et al.1992). The fraction that can dissolve in the ocean decreas-es as the total amount added to the atmosphere increases.An additional 5 to 10 percent of the combined inventorywill be absorbed by the reaction of CO2 with CaCO3 insediments on a time scale of 10,000 to 100,000 years(e.g.,Archer et al.1997). One way of managing atmo-spheric CO2 is to accelerate the uptake of carbon by theocean by pumping it into the slowly ventilated deepwaters. These calculations assume that surface ocean tem-perature,salinity, and alkalinity remain constant,and thatthe marine cycle of CO2 fixation and sedimentation oforganic matter and carbonate skeletons (the “biologicalpump”) does not modify the oceanic carbon distri-bution significantly.

In the terrestrial biosphere reservoir, photosynthesisremoves CO2 from the atmosphere,but much of the car-bon is removed only temporarily (IGBP Terrestrial CarbonWorking Group 1998). Approximately 50 percent of theinitial uptake of carbon from photosynthesis is lostthrough plant respiration. Again,the carbon remainingafter respiration is net primary production (NPP). Part ofNPP is lost as litter and enters the soil,where it decom-poses, releasing carbon to the atmosphere. On an annualbasis,carbon remaining after decomposition is net ecosys-tem production (NEP). Further carbon losses occur bysuch processes as fire,insect consumption,and harvest.These losses yield net biome production (NBP),the criti-cal carbon quantity in considering long-term carbon stor-age. The IGBP Terrestrial Carbon Working Group (1998)argues that belowground components of ecosystems areespecially important because belowground carbon gener-ally has slower turnover rates than aboveground carbon.Slower turnover implies that carbon storage can be main-tained over longer periods.

An important outcome of better information on landuse change (Goal 3 for the CCSP, as outlined in Chapter 2)will be a stronger scientific basis to analyze strategies formanaging global carbon fluxes in compliance with inter-national treaties aimed at increasing terrestrial carbonsequestration. We need to know which managementstrategies for terrestrial biological processes are cost-ef fec-tive in forests, grasslands,and croplands. Terrestrialecosystems are partially constrained by the legacy of landuse practices in the past. Land use practices are discontin -uous in time and space. Hence,in the case of forests,dif-ferent stands in the same region will rarely be of the sameage class and species composition. They will sequestercarbon at different rates and in dif ferent amounts inresponse to a given change in management (e.g., changesin harvest rate or predation control) or climate.Delineating these effects will require close couplingbetween observed land use patterns and trends and corre-sponding estimates of carbon flux densities,to create


24

spatially explicit gradients of carbon flux at regionalscales. This effort will require combining satellite andground-based observation of land use patterns and trends,flux measurements,and process-based modeling.Together, information from all these sources will allow thescientific assessment of management scenarios to increasesequestration of carbon on the land.

The scientific evaluation of management optionsrequires the direct involvement of social scientists in con-structing land use histories. Thus,there is a permeableboundary between this science plan and the relevantsocial sciences. Careful assessments are needed of thesocial and economic costs and benefits of proposedcarbon sequestration policies. One critical element willbe the development of credible land use scenarios forthe future.

Another important component of carbon managementstrategies will be the ability to verify commitments. Othersections of this report discuss the estimation of oceanicand terrestrial sinks. An important additional issue is theestimation of fossil fuel emissions. If the global estimateof fossil fuel emissions based on economic accountingmethods is off by 10 percent,that would represent anerror of 0.65 Gt C/yr—which is more than the reductionsenvisioned in the Kyoto Protocol. This error is quite large

when trying to balance the global carbon budget withoceanic and terrestrial sources and sinks. The financialand economic stakes in correct carbon accounting are ofgreat importance for any country in a world where emis-sions permits can be traded. Some measurement strategyis needed,independent from statistical and accountingmethods,to determine the magnitude of fossil fuel CO2emissions on regional and national scales. The onlynearly unequivocal (characteristic and stable) tracer forfossil fuel CO2 in the atmosphere is its complete lack of14C isotopes.

These considerations lead to another major near-termgoal of the CCSP:

Goal 5 : Devel op a sci ent i f i c basi s to eval -uate potent i al management st r ategi es forenhanci ng car bon sequest r at i on i n the envi -r onment and for captur e/ di sposal st r ategi es.


25

The previous two chapters of this report summarizedour present understanding of the global carbon cycle andthe basic questions that confront present-day ef forts tobetter understand it. This chapter presents an integratedplan to achieve the major goals that have been identified:

Near - Ter m Goal s of the U.S. Car bon Cycl eSci ence Pl an

Goal 1 : Establ i sh accur ate est i mates ofthe magni tude of the potent i al Nor ther nHemi spher e ter r est r i al car bon si nk and theunder l y i ng mechani sms that r egul ate i t .

Goal 2 : Establ i sh accur ate est i mates ofthe oceani c car bon si nk and the under l y i ngmechani sms that r egul ate i t .

Goal 3 : Establ i sh accur ate est i mates ofthe i mpact of hi stor i cal and cur r ent l and usepat ter ns and t r ends on the evol v i ng car bonbudget at l ocal to cont i nental scal es.

Goal 4 : Impr ove pr oj ect i ons of futur eatmospher i c concent r at i ons of car bon di oxi dethr ough a combi nat i on of mani pul at i veexper i ments and model devel opment thati ncor por ates appr opr i ate bi ophysi cal andecol ogi cal mechani sms and car bon cycl e- cl i -mate feedbacks i nto gl obal cl i mate and car boncycl e model s.

Goal 5 : Devel op a sci ent i f i c basi s foreval uat i ng potent i al management st r ategi esfor enhanci ng car bon sequest r at i on i n theenvi r onment and for captur e/ di sposal st r ate-gi es.

These goals are intended to guide research for the peri-od of the next 5 to 10 years.

Two general hypotheses were also identified inChapters 1 and 2 as those most critical for the U.S.Carbon Cycle Science Plan (CCSP) to address:

Chapter 4: An Integrated Carbon CycleScience Program

27

Hypothesi s 1 : Ther e i s a l ar ge ter r es-t r i al si nk for anthr opogeni c CO2 i n theNor ther n Hemi spher e.

Hypothesi s 2 : The oceani c i nventor y ofanthr opogeni c CO2 wi l l cont i nue to i ncr easei n r esponse to r i si ng atmospher i c CO2concent r at i ons, but the r ate of i ncr ease wi l lbe modul ated by changes i n ocean ci r cul at i on, bi ol ogy, and chemi st r y.

The research program outlined here will ultimately bejudged by its ability to provide practical answers to bothscientific and societal questions. Scientists and policymakers must be able to evaluate alternative scenarios forfuture emissions from fossil fuels,effects of human landuse,sequestration by carbon sinks,and responses of car-bon cycling to potential climate change. Thus,a key moti-vation for further research is to develop the predictivecapability to define responses of the global carbon cycleto change,as reflected in Goals 4 and 5.

Recent assessments of global environmental researchhave emphasized the need for programs that are bothintegrated and focused (e.g.,National Research Council1998). This plan puts forth a program that focuses on keyproblems, yet maintains breadth to reveal new problemsand priorities in those areas where the knowledge neededto define focused strategies is currently lacking. InChapter 6,this report also proposes a management struc-ture for implementation of the research plan and thedevelopment of critical partnerships to ensure continuousreassessment and prioritization of goals.

Much recent progress in knowledge of the carboncycle has resulted from studies at the global scale for timeperiods of years to decades. However, to make significantprogress in understanding and quantifying the criticalmechanisms that will determine future levels of atmos-pheric CO2, data must also be obtained for specific geo-graphic regions over a range of time scales. The finger-prints of dominant processes are to be found by studyingregional carbon balances and temporal variability. Effortsto address both intermediate spatial scales and longertime scales are thus essential components of the pro-posed plan.

The program will also study the main processes influ-encing how carbon cycling may change in the future.These studies will be integrated in a rigorous and compre-hensive effort to build and test models of carbon cyclechange, evaluate and communicate uncertainties in alter-native model simulations,and make these simulationsavailable for public scrutiny and use. Clearly, the systemat-ic incorporation of newly understood mechanisms inmodels must be accompanied by model integration usinghigh-quality standard inputs and rigorous consistency testsagainst an array of benchmark data. Data managementand data set construction are sometimes underrepresent-ed in hypothesis-oriented programs. This pitfall must beavoided because,ultimately, it weakens the ability to testhypotheses using comprehensive data and to developpowerful generalizations and new hypotheses.

At its most basic level,the global carbon cycle must beviewed as a singular entity. Its various components are sointeractive—over so many different scales of time andspace—that they cannot conveniently be “isolated”forindependent study or modeling. Data are most valuablewhen combined from a variety of measurements andmethods associated with different carbon-cycle compo-nents; for example,when oceanic data are applied to helpinterpret results for the atmosphere and terrestrial bios-phere,and vice versa. The present plan,then,proposesthree different types of general approach:

• Extend observations over the important space andtime scales of variability in all active carbon reservoirs

• Develop manipulative experiments to probe keymechanisms and their interactions

• Integrate these data, analysis, and modelingapproaches so that they are mutually supportive andcan be focused on key problems.

The observational stra t e gy described in this chapter isdesigned to combine atmospheric measurements witho b s e rvations from space, a i r, l a n d , and sea to reveal specifi cp rocesses that affect terre s t rial and oceanic carbonex ch a n ge at re gional as well as global scales. While the go a lof the terre s t rial component of this stra t e gy is ultimately tou n d e rstand the Nort h e rn Hemisphere terre s t rial sink, t h econtinent of North A m e rica is an excellent focus for theU. S . re s e a rch community in developing this re s e a rch objec-t i ve . N o rth A m e rican logistical capabilities are ex c e l l e n tand cost-effe c t i ve , and there are ex t e n s i ve existing data setsfor North A m e rican ecosystems, land use, s o i l s ,i n d u s t ri a la c t i v i t i e s , and history. Recent re s e a rch has pointed to thep a rticular importance of understanding terre s t rial carbonex ch a n ge in the Nort h e rn Hemisphere , and North A m e ri c a ’sl a rge geopolitical units facilitate the development of an inte-grated continental analysis of carbon sources and sinks.Pa rallel re s e a rch by European and Asian colleagues will bee n c o u raged to expand the cove rage into other parts of theN o rt h e rn Hemisphere terre s t rial biosphere .

Similarly, the northern oceans are relatively accessible,and a solid foundation of oceanic data and knowledge isavailable to support integration of studies of NorthAmerican atmospheric CO2 exchange with studies of CO2exchange in the oceanic regions adjacent to it. These focioffer a unique opportunity to combine atmospheric,oceanic,and terrestrial studies in a way that will constrainmajor components of the global CO2 budget.

Goal 1: Understanding the NorthernHemisphere Terrestrial Carbon Sink

One principal goal of the CCSP is to establish accurateestimates of the magnitude of the potential NorthernHemisphere terrestrial carbon sink and the biophysicalmechanisms that regulate this sink. Several major activi-ties should be conducted in this area:

• An expanded pro gram of atmospheric concentra t i o nm e a s u rements and modeling improvements in support ofi nve rse calculations and global bioge o chemical models

• A network of integrated terrestrial research sites witheddy-covariance flux measurements and associatedprocess studies,manipulation experiments,and models,which as a whole are sufficient to reduce uncertaintyabout the current and future carbon cycle to accept-able limits.

The two boxes here summarize the specific proposedprogram elements,which are discussed in detail in thesections that follow.

Goal 1a: At mospher ic Measur ement s andModels—Maj or Pr ogr am Element s and Act iv i t ies

• Ex pand cur r ent at mospher ic moni t or ing andobser v at ional net w or ks t o acqui r e—

- Ver t ical concent r at ion pr of i les in cont inen-t alsour ce/ sink r egions

- Incr eased cont inuous measur ement s (as cont r ast ed w i t h w eekly f lask dat a) of CO2 and associat ed t r acer s at select ed cont inent al sur f ace

st at ions

- Flask measur ement s in under sampledr egions, pr ov iding t he dat a needed f or cont inent aland r egional appor t ionment of net car bonex change.

• Enhance t he sui t e of measur ement s at t he glob-al net w or k si t es t o include ox y gen, car bon andox y gen isot opes, r adon, and ot her par amet er s t oconst r ain t he locat ions and pr ocesses r esponsiblef or t he car bon sour ces and sinks.


28

Goal 1a (cont inued)

• Conduct f ocused f ield campaigns over Nort h America

using aircraf t sampling t echniques in combinat ion wit h

improved at mospheric t ransport models and an enhanced

flux t ower net work, t o conf irm and ref ine est imat es of

t he magnit ude of t errest rial sources and sinks of CO2.

• Improve inverse models and st rengt hen connect ion

bet ween models and observat ions.

Goal 1b: Terrest rial St udies—Major Program Element s and

Act ivit ies

• Synt hesize result s of recent and ongoing t errest rial

carbon f lux st udies and use t he dat a t o const rain global

carbon/ t errest rial/ at mosphere models.

• Conduct long-t erm monit oring of changes in above

and belowground carbon st ocks on forest , agricult ure,

and range lands using improved biomet ric invent ories t hat

explicit ly address carbon issues, along wit h new t ypes of

remot e sensing, t o det ermine long-t erm changes in car-

bon st ocks.

• Develop new t echnology t o facilit at e direct long-t erm

f lux measurement s of CO2, and inst all an expanded net -

work of long-t erm f lux measurement s emphasizing t he

acquisit ion of dat a for represent at ive ensembles of undis-

t urbed, managed, and dist urbed lands along major gradi-

ent s of soils, land use hist ory, and climat e, and wit h a

number of t owers and sit es suf f icient t o allow quant if ica-

t ion of error in t he spat ial domain.

• Quant if y mechanisms cont rolling t errest rial sources

and sinks, t heir evolut ion in t he geologically recent past ,

and t he likely course of t heir evolut ion over t he next

decades t o cent uries. This quant if icat ion should be

achieved t hrough regionally nest ed set s of manipulat ion

experiment s, ot her process st udies, and CO2 f lux meas-

urement s. Measurement s should be int egrat ed t hrough a

parallel set of nest ed models, wit h analyses scaled up t o

regions t hrough use of regional invent ory dat a.

• Conduct manipulat ions and focused process st udies on

ecosyst em, local, and regional scales, coordinat ed wit h

CO2 f lux measurement s t o quant if y t he mechanisms con-

t rolling t errest rial sources and sinks, t heir durat ion, and

past and fut ure evolut ion.

• Develop t echniques for monit oring t he dynamics of

belowground carbon st ocks.

Atmospheric Measurements and Modeling forInverse Calculations

Atmospheric transport of CO2 integrates the effects oflocal sources and sinks,with mixing around a latitude cir-cle in a few weeks,and between hemispheres in about ayear.A primary test of Northern Land Sink hypothesesrequires resolving the longitudinal structure of the surfaceCO2 flux,in addition to the variation in latitude. Clearly,the observations will have to define concentration gradi-ents between the continents and the sea and between theplanetary boundary layer and the middle and upper tropo-sphere. These are inherently difficult measurements,because atmospheric mixing is so much more rapidaround latitude circles than across them,and measure-ments near sources or sinks are highly variable. Design ofthe necessary atmospheric sampling program requirescareful attention to the spatial and temporal distributionof sampling,the precision of the atmospheric data,andthe details of some boundary-layer atmospheric processesthat are not well understood at present.

The atmospheric observing system currently consists ofroughly 100 sites around the world at which air is collect-ed weekly in paired flasks for trace gas analysis at centrallaboratories,and few sites where observations are madecontinuously.The sites are intentionally located in remotemarine locations to avoid local “contamination”by indus-trial or terrestrial emissions or uptake,and are operated atlow cost using cooperative arrangements with volunteers.There are very few data acquired at altitude and at mid-continental stations.

To provide meaningful constraints on net terrestrialCO2 exchange on the regional scale,the observing net-work will need to be strengthened considerably to charac-terize spatial and temporal variations associated with thecarbon fluxes that need to be measured. The present net-work is designed to be insensitive to regional netexchanges. A recent evaluation of 10 global tracer trans-port models used for CO2 inversions (Denning et al.1999)found that the models converged when compared to theobserved values for an inert tracer (SF6) at flask stations inthe remote marine boundary layer. However, theydiverged where the data are sparse (aloft and over thecontinents). This problem is even worse for CO2 due tothe covariance between diurnal and seasonal cycles ofCO2 net exchange and rates of atmospheric mixing,oftencalled the “rectifier effect.” Expanding the atmosphericobserving network to include routine sampling aloft,par-ticularly over the continents,should be one of the highestpriorities for the future. Atmospheric sampling over theterrestrial surface must include vertical profiles through adepth sufficient to capture most of the vertical mixing ofthe surface signal. At a minimum,this vertical samplingmust span the depth of the planetary boundary layer (1 to3 km in warm sunny weather). A primary research goalshould be to determine first the optimal sampling density,

Chapter 4: An Integrated Carbon Cycle Science Program

29

supporting measurements,and combination of continuousversus flask samples, for long-term airborne sampling.Multiple species,such as tracers of industrial activity, andisotopic ratios must also be measured to obtain the infor-mation needed to interpret the observations (e.g.,Potosnak et al.1999).

The atmospheric boundary layer and the covariancebetween terrestrial ecosystem processes and near-surfaceturbulence are not resolved in most of the current genera-tion of global atmospheric models. Most atmospherictracer transport codes used for CO2 inversion calculationsrepresent subgrid scale vertical transport very crudely, ifat all. Likewise, very few of these models include a diur-nal cycle of CO2 exchange.

H oweve r, even if a model could corre c t ly re p resent thelocal cova riance stru c t u re of the fl u xes and the turbu-l e n c e , the influence of the re c t i fier effect on the observe dc o n c e n t ration field at remote flask stations depends on thep e rsistence of the ve rtical gradient as the air is tra n s p o rt e dh o ri z o n t a l ly for hundreds or thousands of kilometers . T h i sp rocess is ve ry poorly re s o l ved in even the most detailedglobal models, and is not well understood theore t i c a l ly.

A major effort in understanding the local forcing,spa-tial scaling,and long-distance transport aspects of the rec-tifier effect is required,through both observations andmodels. Testing the Northern Land Sink hypothesis alsorequires filling the gap at the crucial “middle scale”of theflux-transport-concentration problem. This middle scalebetween local and large-scale observations is completelymissing from the current observing system and models.

The detailed design of the required atmospheric sam-pling program,which must be coordinated with design ofthe terrestrial flux network and ocean measurements pro-posed below, is a major scientific endeavor beyond thescope of this document. However some general require-ments are clear. A strategy must be developed for atmo-spheric sampling over continental regions which takesinto account the dif ferences in ecosystem exchanges instable and convective conditions. This variation must beexplored over a range of ecosystems and meteorologicalregimes by sampling from eddy flux towers,tall towers,balloons,and light aircraft. Continuous long-term meas-urements of the vertical profile of CO2 and other tracegases on tall transmission towers allows “representative”conditions to be defined for the planetary boundary layer(PBL) sampling at the local scale (Bakwin et al.1995,Bakwin et al.1998,Hurst et al.1997). Similar informationcan be obtained from continuous long-term measurementsof CO2 and other trace gases in conjunction with eddy-correlation fluxes defining rates of exchange between thesurface layer and the planetary boundary layer (Potosnaket al.1999).Light aircraft can be instrumented with con-tinuous analyzers to determine the boundary-layer budgetsof trace gases over areas orders of magnitude larger than

the footprint of an eddy flux tower (Desjardins et al.1997,Goulden et al.1998). These types of studies directlyaddress the issues of scalability of tower fluxes,and canbe used to design lower cost, routine sampling programsfor larger scales.

Independent estimates of re gional-scale carbon fl u xe sby inve rsion of atmospheric data will re q u i re a dense net-wo rk of samples collected by light airc ra f t . Light airc ra f tsampling must be dense enough to capture meaningful gra-dients in surface fl u xe s , and must sample both within andab ove the conve c t i ve boundary laye r. Ve rtical pro files fro mlight airc raft over a continental re gion could be coupledwith high-altitude transects sampled from appro p ri a t e lyi n s t rumented commercial airc raft (Marenco et al. 1 9 9 8 ).

Both inversion calculations and forward models willbenefit tremendously from additional constraints such asregionally detailed emissions data and multiple tracers.Samples should be analyzed for CO2, as well as CO, CH4,O2/N2, and stable isotopic ratios,all of which provide con-straints on the carbon cycle. Ancillary data such as PBLstructure (from wind profiling and traditional soundingsystems),atmospheric transport (from four-dimensionaldata assimilation systems,4DDA),and the isotopic ratios ofother components of the land-atmosphere system (plants,soils,precipitation,and groundwater) are needed. Such asystem has been proposed using automated samplingequipment and rental aircraft (Tans et al.1996).

Regional observing and modeling programs have beenproposed in other countries on a “campaign”basis,andthe results of these studies can provide useful constraintson regional flux estimates using inverse modeling.Regional experiments quantifying carbon fluxes or tracerconcentrations are currently underway or planned for thenear future in Europe,Siberia,Brazil,and Australia.Thedesign and implementation of U.S.observing systems andmodeling programs should be optimized to take advantageof these complementary programs.

Ideally, inverse calculations of the carbon budgetshould subsume all available information,including flasksamples,in situ data,aircraft sampling,air-sea flux meas-urements and eddy covariance data.Carbon budgets calcu-lated from inverse methods should not, for example,beinconsistent with measured diurnal cycles of CO2 datacollected by regional sampling programs in other parts ofthe world. The current global observing system is sopoorly constrained in the tropics, for example,that tropi-cal fluxes are freely estimated in inversion models as aresidual,allowing unacceptable freedom of terrestrial flux-es in higher latitudes without violating global mass bal-ance.Inclusion of new regional data from experiments inAmazonia in these inverse models would provide strongerconstraints on the carbon budget of North America,directly addressing uncertainties in the NorthernHemisphere Land Carbon Sink hypothesis.


30

New approaches to inverse modeling are needed toapply highly resolved atmospheric data to constrainregional fluxes. Atmospheric transport across regionalareas is sufficiently rapid that concentration changes willhave to be resolved on the order of hours to a few daysrather than months or years. Trace gas transport will needto be represented at much higher spatial and temporalresolutions than at present,possibly using “observed”meteorological fields from four-dimensional data assimila -tion systems,or by incorporating carbon fluxes into mod-els used in operational weather forecasting. Significantimprovements in the land-surface parameterizations usedin numerical weather prediction would be required.

An objective of the global observational and inversemodeling system should be to provide meaningful integralconstraints on spatially extrapolated estimates of carbonfluxes derived by “upscaling”local fluxes using process-based models and remote sensing.These observing andmodeling programs,described in the next section, wouldbe extremely valuable in the context of top-down esti-mates of flux derived independently from the globalobserving program.

Terrestrial Observations, Experiments,and Models

Studies to refine understanding of terrestrial CO2exchange confront fundamental questions. What are thefluxes of carbon into today’s ecosystems? Which systemsare taking up how much carbon? What factors influencechanges in past and contemporary ecosystem carbon stor-age (e.g.,CO2 itself, nitrogen deposition,other pollutants,climate,management practices)? How has the rate of car-bon storage changed in the past centuries and decades?What systems and management practices cause net lossesor gains of carbon? How will fluxes and storage of car-bon in the terrestrial biosphere change with changes inclimate and the chemical composition of the atmosphere?

Studies of terre s t rial carbon cycling must focus onsystematic sampling stra t e gies designed to ch a ra c t e ri z eq u a n t i t a t i ve ly essential processes and to reject or confi rms p e c i fic hypotheses concerning responses along gra d i e n t sof principal controlling fa c t o rs . T h e re are seve ral curre n thypotheses concerning the particular ecosystems orp rocesses that take up CO2 in response to env i ro n m e n t a lch a n ge . For ex a m p l e , va riations in tempera t u re and soilm o i s t u re (Dai and Fung 1993), growing season length(Myneni et al. 1 9 9 7 ) , N ava i l ability (Holland et al. 1 9 9 7 ) ,C O2 fe rtilization (Friedlingstein et al. 1 9 9 5 ) , and fo re s tre growth (Tu rner et al. 1995) have all been suggested to bei nvo l ved (see, e . g . , Goulden et al. 1 9 9 6 , Fan et al. 1 9 9 8 ) .Although measurements along gradients are a powe r f u lt e chnique for assessing ecosystem responses in systemati-c a l ly diffe rent conditions, in practice the fa c t o rs that deter-mine the ch a n ges along the gradient are confounded to

some degre e . T h u s , great care must be used in interpre t i n go b s e rvations and ex p e riments along gra d i e n t s . In addition,the confounding of control va ri abl e s , t o gether with va ri-ab i l i t y, means that significant replication must be obtainedat least at some points along env i ronmental gra d i e n t s .

The role of land use must be a central subject in anyplan for terrestrial carbon research. In the annual terres-trial CO2 budget summarized above,it is evident that a sig-nificant portion of the terrestrial fluxes is related to pres-ent and/or past land use. Additional evidence from atmos-pheric and oceanic measurements suggests that most ofthe ~1.8 Gt C/yr land sink may be occurring in theNorthern Hemisphere (Ciais et al.1995a). Some recentanalyses suggest that an appreciable fraction of the totalterrestrial sink may reside in North America (Fan et al.1998,Rayner et al.1998). Inventory information is alsoaccumulating to suggest significant sinks of carbon inNorth America,although the inventoried sinks are typical-ly smaller than those suggested by the evidence fromatmospheric measurements. The vast majority of land inthe United States and southern Canada was disturbed inthe past and is managed,intensively or extensively forhuman use. North American carbon sinks,as well asuptake by European or Asian ecosystems,are stronglyaffected by human activities. Studies of the role of landuse history in determining the fluxes are discussed belowin this chapter in section “Goal 3: Land Use.”

A deliberate sampling and experimental design isrequired,aimed at characterizing fluxes and processescontrolling carbon storage in forests, grasslands, agricultur-al lands and soils. The design should emphasize not theidentification of “typical”sites for a vegetation regime,butthe identification of a network of sites within vegetationtypes that sample the principal axes of variation. Theseaxes of variation would include not only climate and soils(Schimel et al.1997),but also disturbance type and timesince disturbance. A high priority is the development ofnew methods for measuring carbon fluxes belowground.

A principal focus of studies within this network wouldbe systematic observation of carbon exchange fluxesalong environmental gradients. This is now possible to anunprecedented degree. The key gradients (e.g., climate,nutrient deposition, forest age,plant functional types,landuse) can be defined. Rates of carbon exchange can bemeasured. Previous efforts to estimate net CO2 exchangehave been hindered by pervasive small-scale heterogeneityin terrestrial carbon storage,and by difficulties in assess-ing changes in belowground carbon storage. Forest inven-tories represent a critical first step in quantifying storage,but they need to be upgraded to provide better informa-tion on carbon,coordinated to help scale results from fluxstations and airborne regional measurements,andi n t e grated to provide consistent, continental-scale estimatesof net carbon ex ch a n ge . E d dy flux data are providing highresolution on ch a n ges in carbon storage on the scale of


31

h e c t a res (e.g.,Goulden et al. 1 9 9 6 ,Goulden et al. 1 9 9 8 ) .H oweve r, f u rther tech n o l o gical developments are needed tob ring down costs and improve the accessibility of the tech-nique and the re l i ability of the method. ( C o m m e rcial deve l-opment of the tech n o l o gy now appears to be underway. )Fi n a l ly,h e c t a re- to kilometer-scale-resolution data can beex t rapolated to re gional (and ultimately global) domainsusing advanced remote-sensing techniques and ve ri fi e dt h rough expanded atmospheric concentration measure-ments and models described earlier in this ch a p t e r.

A netwo rk of flux measurements along we l l - d e fi n e de nv i ronmental gradients provides seve ral va l u able pro d-u c t s . F l u xes can be dire c t ly ex t rapolated using are aweighting from remote-sensing and inve n t o ry info rm a t i o n .Flux observations can provide info rmation about re s p o n s-es to env i ronmental fo rcing (such as tempera t u re and soilm o i s t u re ) . Better understanding of the response to env i-ronmental fo rcing can then be used in ex t rapolations anda n a ly s i s . Flux measurements allow estimates of carbons e q u e s t ration from inve n t o ries to be compared to meas-u red carbon uptake . F l u xes can also be ex t rapolated usingm o d e l s . Estimates of seasonal or intera n nual va riations infl u xes can be compared to ch a n ges infe rred from atmos-p h e ric measure m e n t s . Ap p lying this appro a ch to tempora lva riations is especially import a n t . For ex a m p l e , while inter-a n nual va riations in local climate and carbon fl u xes mays u g gest hypotheses about large-scale re g u l a t i o n ,t h ey pro-vide limited insight without direct large-scale mass-balancec o n s t ra i n t s . C o nve rs e ly, estimates of the large-scale atmos-p h e ric CO2 seasonal cycle and sources and sinks provide avital global constraint on models. H oweve r, these estimatesp rovide limited info rmation about the modeled mech a-nisms and sensitivity without greater spatial resolution andp recision in estimated fl u xe s .

The ability to apply eddy-covariance flux measure-ments to regions will be limited by knowledge of errors inboth temporal and spatial scales. Because the techniqueaccumulates data at high frequency, there is essentially lit-tle problem in the temporal resolution of the specificmeasurements. It is critical,however, that the variabilityof NEP over seasons and years be captured by continuous,high-quality operation of each of the sites. It is in the spa-tial domain that problems with regional measures of NEPusing the eddy-covariance flux method may arise. Aregion (or biome) can be thought of as a unit of observa-tion from which samples can be drawn to allow theregion to be characterized quantitatively with explicitstatements of error for NEP in space. This characterizationcan be achieved if eddy-covariance measurements arereplicated within regions,not only across the axis of a par-ticular gradient (as discussed above) but also normal tothe gradient axis in order to quantify variance at eachgradient level. The number of replications needed cannotbe stated a priority as it is itself a research issue.Nevertheless,adequate replication is absolutely essential

to providing information that ultimately can be used toreduce uncertainty in estimates of current and futureregional carbon flux and storage. Such replication maydepend to a large degree on the development of low-cost,stable, reliable,semiautomated instrument packages thatwill greatly reduce the manpower and logistical costsassociated with the measurements.

The envisioned approach would combine process stud-ies and experiments (designed to increase basic knowl-edge and improve predictive models) with flux measure-ments designed to allow models to be tested. A well-designed network of study sites would serve as a focalpoint for many different types of research. Process studieson nutrient interactions,on feedbacks from species diver-sity and changes to biogeochemistry, and on climateeffects are all needed. Experimental manipulations canhelp untangle complex mechanisms and test hypothesesfor ecosystem responses to conditions outside the currentenvelope. For example,studies using preindustrial CO2levels can yield critical information on carbon storagefrom past changes. Studies manipulating CO2, nitrogendeposition,and climate at sites at a range of times sincedisturbance are crucial for quantifying interactions amongthese key global change drivers. It is essential that theresearch network use common approaches and methodswith the highest degree of standardization of methods andinstruments as possible. Particular attention must begiven to quality assurance in the operation of monitoringequipment and conduct of manipulation experiments.Formal protocols will certainly be required at an earlystage. Recognizing that new and better methods andinstruments will be developed,the networks need to beable to accommodate innovations so that the innovationscan be applied across the entire system,not piecemeal.

The United States and other nations already invest sig-nificantly in natural resource inventories for managementpurposes. These inventories should be designed to pro-vide better information on carbon. Only large-scale opera-tional inventories—such as those maintained by govern-ment agencies—can provide data from the hundreds tothousands of sites needed for direct spatial integration.The inventory data need to be more effectively integratedwith other sources of information,including eddy fluxand remote sensing. In addition,it is critical to extend theinventory approach to cover the fate of carbon after it isharvested from forests. This carbon includes not onlylogging residues,but also manufactured products andwaste streams.

Finally, eddy-covariance flux time-series are needed toprovide closure on carbon budgets at key sites,to measureCO2 uptake or loss as a function of the location in theexperimental design.

The proposed terrestrial carbon research networkposes several significant challenges. For measurements


32

along gradients to have power in rejecting hypotheses orparameterizing models,large sample sizes are required.Today’s networks of flux sites number in the tens of instal-lations. Globally this approach will require significantlymore measurement sites,an arrangement that requires sig-nificant prior investment in autonomous measurementtechnology and theory to make the technique simpler,more robust,and less expensive. The program must besustained for a significant period,because the measure-ments become valuable only when reasonably long time-series have been collected,and they will become morevaluable over time.

Clearly, both the design of the initial network and itsimplementation will also require extraordinary care,statis-tical rigor, and investment in technology. Enormousresources could be expended on process studies, experi-ments, flux measurements,and inventories without mate-rially reducing uncertainty about either today’s CO2 budg-et or simulations of future trends. For the nation’s scientif-ic resources to be efficiently deployed,an initial synthesisand analysis of existing in situ data (including soil and sed-iment surveys, forest inventories,and observations at exist-ing Long-Term Ecological Research [LTER] and AmeriFluxsites (LTER maintained by the National Science Fo u n d a t i o nand AmeriFlux maintained by the Department of Energy,National Aeronautics and Space Administration,andNational Oceanic and Atmospheric Administration),remote sensing,and model results is needed to definepatterns of variability. Then,the research community mustbe engaged in the effort to use this synthesis as a basis forsite selection in a network designed to understand pat-terns at large scales.

Once an analysis of existing data and models is done, acritical set of measurements and experiments can bedesigned to efficiently sample the space identified.Different sets of measurements may be appropriate for dif-ferent suites of sites. The design should take advantage ofremote observations of land cover and land cover change,which are quite comprehensive for the United States.Similarly, inventory data are already available,and with sys-tem and data management upgrades,management datamay be made highly useful. With some ef fort in technolo-gy and theory development,the present network ofroughly 20 flux measurement sites can be increased by afactor of 2 to 10.

Six other types of studies are proposed to complementthe proposed flux measurements at network sites:

1. Experimental manipulations of CO2, temperature,nitrogen,and other key controlling factors. Mani-pulations are an ongoing line of research that must beenhanced. A number of specific questions about thenature,quantitative importance,and persistence ofmechanisms driving the current terrestrial carbon sinkcan be best addressed through direct experimentation.

Manipulations including ecosystem-scale climatechange,nitrogen additions,and elevated or decreasedCO2 can provide critical insights on interactions,onresponses to conditions in the past or the future,onecosystem-to-ecosystem variation in responses,and oninteractions with other anthropogenic impacts,includ-ing harvesting,other land use change,biological inva-sions,and altered biological diversity.

The next ge n e ration of manipulative ex p e ri m e n t sdesigned to understand and quantify the current terre s-t rial sink should emphasize processes at the ecosystems c a l e ,i n cluding responses of both bioge o ch e m i s t ry ande c o l o gical dy n a m i c s . Studies with pre i n d u s t rial CO2 a rec ri t i c a l , but will re q u i re new tech n o l o gi e s ,e s p e c i a l ly fo rex p e riments at the ecosystem scale. E x p e riments toquantify effects of multiple fa c t o rs , alone and in combi-n a t i o n ,a re also cru c i a l . C u rre n t ly, we have little idea ofthe extent to which the carbon sink in fo rest re grow t hi n cludes signals from increasing CO2, N deposition, o rclimate ch a n ge . S i m i l a r ly, we have no idea of the conse-quences for carbon storage of ve getation ch a n ge s , fo rex a m p l e ,i n c reasing shrub abundance in many of thewo r l d ’s grasslands (Archer et al. 1 9 9 5 ) . E x p e ri m e n t scould be used to probe both the dri ve rs and the conse-quences of ve getation ch a n ge s ,e s p e c i a l ly ex p e ri m e n t son ecosystems where the transitions can occur quick ly.

The next generation of experiments should alsoinclude pilot studies to evaluate deliberate carbonsequestration strategies. Issues concerning the limitedspatial and temporal scale of manipulative experimentsshould receive intensive attention,so that lessons fromthe experiments can be effectively interpreted andincorporated into global-scale models.

2. Long-term terrestrial observations. E x p a n s i o nand enhancement of the LTER netwo rk will pay hugedividends by defining the ecological and current andh i s t o rical land use fa c t o rs that regulate sequestra t i o nand release of carbon from major ecosystems. The net-wo rk expansion is envisioned to provide new sites,ro u g h ly equal in number to the ro u g h ly 20 curre n t lyex i s t i n g ,s t ra t e gi c a l ly located to examine ecotones, o rb o u n d a ry zones between re gions with diffe rent ve ge t a-tions or biomes,like ly to play a significant role in re g u-lating global CO2. Enhancements are needed in twod 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-bon stores and turn over should become a major com-ponent of each site, and LTERs should become keypoints for large-scale manipulations and for theexpanded flux netwo rk . By doubling the budget andnumber of sites in the current netwo rk , and addingi m p o rtant new re s e a rch tasks, t h e re should be as t rong synergy between the carbon focus and pre s e n te c o l o gical and pro c e s s - o riented goals of the LT E R s ,enhancing both.


33

3. Intensified flux measurements at sites wheredetailed process studies are coordinated witheddy-covariance measurements. Intensive observa-tional studies can be conducted where carbon uptakeand many of its controls (N availability, soil moisture,microclimate,light) and mediating variables (Rubiscocontent of leaves,conductance,stem flow, below-ground processes) are measured. In essence,such stud-ies allow natural spatial and temporal variability to per-form the experiments that test hypotheses.Hypothesized control processes can be evaluated if asuitable suite of associated measurements (climatic,atmospheric,and biological) is made. These studiesshare some of the advantages and disadvantages ofdeliberate manipulations. The major advantage is thatthe perturbations are“natural,” including all time scales.The disadvantage is that the conditions are not undercontrol of the experimenter. This “natural approach”isthat followed by the present AmeriFlux network.

4. Flux scaling studies in which tall towers, bound-ary layer measurements (Convective boundarylayer budgets) and aircraft profiles address thescaling of land surface fluxes to their signaturesin the atmosphere. Opportunistic use should bemade of tall transmission towers (e.g.,Bakwin et al.1995),which allow micrometeorological fluxes to bedetermined over much larger footprints than typicalcanopy towers. Additionally, tall towers allow the verti -cal profile of the eddy flux to be measured,testing scal-ing strategies by varying the footprint of the measure-ment. With appropriate inclusion of meteorologicaldata collection (radar wind profilers or balloon son-des),tall towers also allow the direct measurement ofthe local forcing of the atmospheric CO2 rectifiereffect,which will facilitate larger scale atmosphericmodeling. Aircraft observing campaigns should useeddy-covariance sites as anchor points. These programscan be designed to provide flux estimates over a muchlarger footprint than that of even a tall flux tower,through measurements of continental boundary layerbudgets for scales of several tens of kilometers (Chou1999,Desjardins et al.1997,Lloyd et al.1996) to tran-sects measured by flux aircraft for scales of tens tohundreds of kilometers (ABLE-2/3 experiments,BOREAS,FIFE2). These data allow a quantitative evalua-tion of the relationship among intensive flux measure-ments,land surface cover, and ancillary process data;and would facilitate the development of scalingstrategies by providing spatially extensive snapshotsas context.

5. Remote sensing of terrestrial properties. Remotesensing must be a critical component of any plan forterrestrial carbon research. Efforts to characterize

terrestrial carbon cycling should exploit interactionwith programs such as the World Climate ResearchProgram’s Global Energy and Water Cycle ExperimentContinental Scale International Project (GEWEX/GCIP).This program has already demonstrated success in inte-grating remotely sensed and in situ measurements ofenergy and water fluxes. To the extent possible,carbonresearch plans should be structured so that they cantake advantage of improved satellite data productsexpected in the near future. These products willinclude new 30-meter Landsat Thematic Mapper-derived land cover data,high-resolution data setsexpected in association with the ETA Mesoscale Model,and data products anticipated from the Mission toPlanet Earth. Of particular interest is the possible appli-cation of advanced algorithms to the upcoming EarthObserving System (EOS) sensors to derive plant canopyfunctional properties.

6. Integration of observations with model develop-ment. Understanding terrestrial processes requiresthat ongoing observations be linked to the continueddevelopment and testing of models. It is extremely dif-ficult to develop terrestrial carbon models that includestate-of-the-art process representations for all of theneeded processes on multiple temporal and spatialscales. Often the limitations of models serve as sign-posts in formulating and testing new hypotheses.Examples of current model frontiers are the effects ofCO2 on a full suite of plant processes (including alloca-tion of carbon to different parts of a plant), dynamicinteractions between carbon and nitrogen budgets,hydrologic changes (such as drying or thawing of bore-al peat),and vegetation dynamics such as successionalchanges over long time scales. Models must also beimproved to systematically incorporate informationabout human and natural disturbance of the land sur-face. Current models emphasize physiological and bio-geochemical processes and largely neglect the carbonstorage dynamics induced by cultivation, forest harvest,fire, fire suppression,and other intensive disturbancesas well as biological invasions, changes in biologicaldiversity, and other ecological processes

The chief requirement for the progress of modeling—particularly in hypothesis testing—is better integration ofmodels and data. Model development is currently sup-ported by a variety of programs,including NOAA CarbonModeling Consortium (CMC), Terrestrial Ecology andGlobal Change program (TECO),Vegetation/EcosystemModeling and Analysis project (VEMAP),NSF’s Methodsand Models for Integrated Assessment (MMIA),and numer-ous disciplinary programs. Although this diverse range ofsupport encourages innovations,more effort is needed inintegration. The assembly of observational data into


2Atmospheric Boundary Layer (ABLE);BOReal Ecosystem-Atmosphere Study (BOREAS);First ISCLIP (International Satellite Land Surface Climatologyprogram) Field Experiment (FIFE)

34


35

Photograph of the 447-meter tall WLEF-TV transmitter tower, Park Falls, Wisconsin. The tower is owned by the State ofWisconsin Educational Communications Board, and is being used for measurements of CO2 mixing ratios (seeBakwin et al., 1998) and atmosphere/surface exchange of CO2 by eddy covariance. Transmitter towers up to 610meters tall are located in many areas of the USA.

standard forms that can be used by models is a significantintegrative task,without which rigorous testing and com-parison of models cannot be achieved. Thus, very differ-ent models may appear to have equal “validity,” incorrecthypotheses may not be rejected,and the improvement ofmodels and hypotheses may be inhibited. Standard datasets are urgently required for terrestrial model input (cli -mate,soils,disturbance regimes,N deposition) at regionaland global scales. Without these standardized inputs,model intercomparisons are meaningless. Standard datasets must be assembled in ways that are compatible withall models simulating particular processes and scales.There is ample precedent for this approach. In meteorolo-gy, the production of reanalysis data sets has resulted inlong time series of physical variables that are critical inevaluating climate models. These data sets might bevaluable as climate inputs to carbon models,but theymust be checked for consistency with known carbon-energy-water relationships.

Recent Intergovernmental Panel on Climate Change(IPCC) intercomparisons of global carbon-cycle modelswere made only after all the models were required tomeet consistency criteria based on input of a standard setof historical atmospheric CO2 and emissions data. Clearly,the systematic incorporation of newly understood mecha-nisms in terrestrial models must be accompanied bymodel integration using high-quality standard inputsand rigorous consistency tests against an array ofbenchmark data.

The implementation of the terrestrial studies describedabove might proceed as follows:

• An analysis of existing spatial data, model re s u l t s ,a n dremote-sensing products must be initiated as a basis fo ran ex p e rimental and observing system design that willa l l ow the testing of hypotheses and the ex t rapolation ofs i t e - s p e c i fic studies. A wo rking group with modestfunding should be assembled to undert a ke this activity.The project should assess the U. S .d a t a , results and pro d-ucts in detail and global info rmation at lower re s o l u t i on.

• The nation’s forest, agricultural,and aquatic monitoringprograms should be evaluated with the goal of identify-ing low-cost/high-leverage enhancements to the exist -ing programs. Collaboration among agency scientistsand managers and the broader carbon science commu-nity is crucial.

• Flux measurements are an essential part of the pro-gram,because they may provide closure on carbon andwater budgets,which is difficult with conventionalsampling,especially closure on belowground fluxes.However, there are problems with the existing technol-ogy and theory. Resources need to be invested toreduce uncertainties in measuring nighttime fluxes,allow use in more complex terrain,and develop cheap-er and more autonomous systems. Effort needs to be

directed toward developing a rigorous statisticalapproach to placement of sites for atmospheric moni-toring and manipulation experiments.

• A continued effort is needed on modeling,on the inte-gration of new experimental knowledge into models,and on sustained testing of models in increasingly rigor-ous model-data comparisons. Models are requiredboth for the integration of knowledge about the pres-ent,and for prediction of the response of systems tofuture changes.

• Enhanced manipulative experiments at the ecosystemscale are critical for exploring ecosystem responses toenvironmental conditions outside the current envelopeand for assessing responses to interactions among cli-mate,ecological processes,and anthropogenic changes.Manipulative experiments should be designed to sup-port model development, facilitate scaling in space andtime, evaluate proposals for managed carbon sequestra-tion,and enhance broad communication about the roleof the terrestrial biosphere in the global carbon cycle.

• Long-term ecological research,with greater emphasison carbon-related issues,will provide the critical dataon ecological and land use historical factors regulatingsequestration of carbon.

Goal 2: Understanding the OceanCarbon Sink

Long-term goals for CO2 research in the ocean are, first,to quantify the uptake of anthropogenic CO2 by theocean,including its interannual variability and spatial dis-tribution;and,second,to understand and model theprocesses that control the ocean’s uptake of CO2. Uptakeof anthropogenic CO2 can be quantified by measuringeither the flux itself or the resulting change in carboninventory. Both should be carried out,with a strongemphasis on disaggregating the global uptake into contri-butions from major ocean regions and monitoring tempo-ral variability. The process and modeling goals can beattained by research on the rate-limiting steps for uptake,the causes of spatial and temporal variability, and the bet-ter integration of models with data to predict long-termtrends. These goals are similar to those for the terrestrialenvironment,but the major research challenges are dis-tinct. Goals in understanding the ocean carbon sink arethe following:

• Develop new technology to facilitate systematic andlower cost long-term observations

• Carry out air-sea carbon flux measurements with anear-term focus on the North Atlantic and NorthPacific,to coordinate and integrate results most closelywith the terrestrial research on the NorthernHemisphere and tropics


36

• Conduct global surveys of oceanic inventories of fossilCO2 along with relevant tracers,including their evolu-tion with time,in support of inverse calculations andglobal models of carbon uptake

• Realize studies of the physical and biogeochemicalprocesses controlling the air-sea flux of carbon in theoceans,including manipulation experiments and devel-opment of models.

The following bullets summarize the proposedprogram elements.

Goal 2 : Maj or Pr ogr am Element s and Act iv i t ies

a. Requi r ed new t echnology

• Inst r ument s f or measur ement of CO2 andr elat ed quant i t ies on moor ings, dr i f t er s, andt ow ed v er t ical sampler s

• Rapid w at er sampl ing t echniques

• High t hr oughput mul t i - element analy zer s f orshipboar d car bon sy st em measur ement s (of dis-solv ed inor ganic car bon [ DIC] , dissolv ed or ganicmat t er , par t iculat e or ganic mat t er , alkal ini t y ,t emper at ur e, sal ini t y , nut r ient s, O2, ) and r elat edt r acer s (CFC’ s, 14C, 13C, et c) .

b. A i r - sea car bon f lux es

• Apply new t echnology t o inst al l an ex pandednet w or k of long- t er m st at ions, dr i f t er s, andunder w ay measur ement s, emphasizing acquisi t ionof dat a f or r epr esent at iv e ensembles of oceanicr egions and condi t ions, t o def ine by obser v at iont he spat ial dist r ibut ion and t empor al v ar iabi l i t yof t he ai r - sea f lux .

• Conduct f ocused f ield campaigns at basin scaleov er bot h t he Paci f ic and At lant ic using ai r cr af tsampl ing combined w i t h shipboar d and gr ound-based measur ement s and impr ov ed at mospher ict r anspor t models. The goal is t o conf i r m andr ef ine est imat es of t he magni t ude of oceanicsour ces and sinks of CO2 f r om pCO2 (par t ialpr essur e of car bon diox ide) dat a, and t o pr ov idet he t ools and dat a necessar y f or impr ov edinv er se model est imat es of Nor t her n Hemispher et er r est r ial sinks.

c. Oceanic inv ent or y measur ement s

• Sy nt hesize r esul t s of r ecent global ocean car -bon sur v ey s in suppor t of globalcar bon/ ocean/ at mospher e inv er se and pr edict iv emodel dev elopment and planning f or f ut ur e globalsur v ey s.

• Ini t iat e ongoing pr ogr am t o r epeat global CO2and t r acer sur v ey s ev er y 10 t o15 y ear s t o mon-i t or t he oceanic CO2 inv ent or y and i t s ev olut ionin space and t ime.

d. Pr ocess st udies, models, and sy nt hesis

• Vigor ously pur sue ocean pr ocess st udies,including manipulat iv e ex per iment s, t o impr ov emechanist ic under st anding of pr ocesses cont r ol -l ing car bon upt ake and t hei r sensi t iv i t y t o cl i -mat e. This r esear ch includes di r ect measur e-ment s of ai r - sea f lux of CO2, f act or s r egulat ingbiological f lux es, and lar ge- scale t r acer r eleaseand t r acking ex per iment s t o def ine quant i t at iv elyt he cont r ols on upt ake of ant hr opogenic car bon byt he oceans.

• Dev elop impr ov ed models of phy sical , chemi-cal , and biological pr ocesses t o analy ze pr ocessobser v at ions and t o pr oj ect how t hese pr ocessesmay af f ect ocean upt ake of CO2 in t he f ut ur e.

• Dev elop new basin and global ocean car boncy cle pr edict iv e and inv er se models, includingimpr ov ed pr ocess models, t o analy ze ai r - sea f luxand ocean car bon inv ent or y and t r acer obser v a-t ions and t o pr oj ect f ut ur e ocean upt ake of CO2.

• Dev elop t he use of r emot e- sensing t ools f orocean moni t 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-lat e local obser v at ions t o lar ger scales.

Until recently, the primary tool used for studying theoceanic uptake of anthropogenic carbon has been modelsvalidated with observations of tracers,particularly the dis-tribution of natural and bomb radiocarbon. The absenceof carbon measurements to verify model estimates hasbeen a serious limitation in testing our understanding ofoceanic uptake. Recent improvements in the precision ofmeasurements of DIC and associated tracers such as O2have given us greatly increased confidence in estimatingthe inventory of anthropogenic carbon directly from DIC.New techniques discussed in Chapter 2 allow filtering outthe background DIC concentration and the effects of sea-sonal and interannual variability. These achievementsmake it possible to detect the total anthropogenic inven-tory and its change over time scales on the order of adecade. The measurements are also being used to esti-mate transport of carbon across sections in ocean basins.Both the tracer-based modeling approach and the invento-ry method will continue to be important tools in monitor-ing the oceanic uptake of anthropogenic carbon.

Nevertheless,neither modeling nor the inventoryapproach can give us the air-sea flux of CO2 at a giventime and place. These methods are thus of limited use in


37

determining the annual and interannual variability of thecarbon cycle,and in constraining inverse models of atmos-pheric observations. However, such spatial and temporalresolution is required for atmospheric inversions to detectthe Northern Land Sink as well as to better understandthe processes that control the air-sea flux of carbon andits variation in time. The best way to obtain adequate spa-tial and temporal resolution is by measuring the air-seaflux. This can be done either directly using techniquesthat were recently tested successfully for the first time atsea,or indirectly using measurements of ∆pCO2 multipliedby an estimate of the gas exchange coefficient. The directtechnique,though,does not lend itself to the large num-ber of measurements that would be necessary to obtainadequate temporal and spatial resolution. Its primaryapplication will be in the essential task of improving ourunderstanding of air-sea flux processes. The indirectmeasurement technique is the most promising for obtain-ing the time-dependent air-sea flux over a given period oftime and region of the ocean. The application of the indi-rect flux measurement technique faces formidable obsta-cles,as discussed in Chapter 2. However, recent progressin techniques for measuring ∆pCO2 and in our under-standing of gas exchange and ability to study it givescause for optimism that air-sea flux measurements willcontribute significantly to our understanding of the globalcarbon cycle over the next decade.

In conjunction with measurements of the air-sea fluxand ocean carbon inventory, it is essential to study thephysical and biogeochemical processes that control theair-sea flux and its spatial and temporal variability. It isnot sufficient to know within rough bounds the globalrate of carbon uptake. Simulations of climate changeshow significant warming of the surface low latitudes aswell as freshening of the high latitudes. These trendswould increase the vertical stratification and have a majorimpact on ocean circulation. It is very likely that thesechanges will have a major impact on the transport of car-bon from the surface ocean to the abyss,as well as affect-ing the biological pump. Projections of future trends inatmospheric CO2 must be based on an adequate under-standing of relevant oceanic processes and the incorpora-tion of this understanding into models. Considerablework remains to be done in these areas,building on thepromise demonstrated by recent progress.

The identified program elements are discussed in moredetail in the following subsections.

Needed New Technology

The processes that control carbon cycling in the oceanexhibit large spatial variance,which itself is not stationary,but changes significantly on diurnal,seasonal,annual,decadal,and even longer time scales. Comprehensive spa-tial and temporal coverage would be very expensive to

obtain with dedicated global measurement programsusing current methods. It is very important to assure thedevelopment of lower cost,more efficient methods forcollecting water samples and making at-sea measurementsof the carbon system (dissolved inorganic carbon,dis-solved organic matter, particulate organic matter, alkalinity,T, S, nutrients, O2, and related tracers such as CFC’s, 14C,etc). A major need also exists for the continued provisionof high-quality DIC standards,which was a major elementin the success of the World Ocean Circ u l a t i o nE x p e ri m e n t / Joint Global Ocean Flux Study (WO C E / J G O F S)carbon measurement program. Standards should be devel-oped for other important tracers as well,with particularimportance attached to carbon isotopes and nutrients.

Autonomous measurement capability for carbon systemparameters is also strongly needed. Time-series measure-ments are presently limited to locations near island portssuch as Bermuda and Hawaii. The present U.S.ocean car-bon-cycle time-series stations at Hawaii and Bermuda arerun at an annual cost of about $1 million each. Activitiessupported by the operating budgets include studies of theprocesses affecting the carbon cycle as well as monitoringof the carbon cycle per se. Significant improvements inthe coverage and the economics of the carbon cycle mon-itoring component of the present activities are possible.The scientific community presently has the technicalcapability to deploy instruments for measuring the air-seadifference of pCO2 on commercial and research ships.There has been initial success in deploying reliable instru-ments on autonomous buoys (Friederich et al.1995).Further technical developments may allow the addition ofin situ monitoring of related variables such as nutrientsand oxygen.

A major program of technological development is anessential component of ocean carbon research over thenext 5 years.

Air-Sea Carbon Fluxes

We recommend the following programs to characterizethe climatological uptake of CO2 by the oceans,the geo-graphic distribution of uptake of CO2, and, eventually, itsinterannual variability.

Time Series Observations. The air-sea flux of CO2 isextraordinarily variable in space and time due to changesin circulation,temperature,and salinity, as well as biology.The key to determining this flux and understanding itsvariations is continuous in situ monitoring,including bothtime-series stations and regular measurements along tran-sects using ships of opportunity.

Temporal variations in some areas,such as theEquatorial Pacific,have been identified as major causes ofvariability in air-sea CO2 exchange. Existing time-seriesstudies are located mostly in the subtropical ocean gyres,


38

while there are major gaps in data on regions of activeocean mixing and high biological variability, especially insubpolar and polar latitudes. Temporal variability is great-est in surface and subsurface layers,locations where bio-logical and physical feedbacks are most likely to alter theocean’s ability to absorb CO2. Characterization andunderstanding of temporal variance is a prerequisite forunderstanding the processes that limit rates of oceanCO2 uptake.

M o o red and underway time-series observations should,w h e n ever possibl e , map a ra n ge of pro p e rties associatedwith biological pro d u c t i v i t y. Extending measure m e n t sb eyond pCO2, t e m p e ra t u re , and salinity (for example) cangi ve both rate and process info rmation about biologi c a lp ro d u c t i v i t y. C o n c e n t ration measurements of CO2 a n dnu t rients are va l u able in this re g a rd . P recise measure m e n t sof O2 c o n c e n t rations in the mixed laye r, coupled wherep o s s i ble with measurements of inert gas concentra t i o n sand wind speed, yield info rmation on net production andve n t i l a t i o n . The measurement of isotopic pro p e rties gi ve srate info rmation of intere s t : the δ1 3C of CO2 c o n s t ra i n snet production (Zhang and Quay 1997), while the tri p l eisotope measurement of dissolved O2 c o n s t rains gross pro-duction (Luz et al. 1 9 9 9 ) . M e a s u rements of atmospheri cO2 c o n c e n t ration gi ve va l u able info rmation about instanta-neous and annu a l ly ave raged ocean carbon fl u xe s .O n going flask sampling can re flect intera n nual va ri ab i l i t yand long-term trends in ocean carbon fl u xe s . C o n t i nu o u sc o n c e n t ration measure m e n t s ,n ow becoming possibl e ,should be developed and used for more detailed studies.

Focused Field Campaigns. An important lesson of car-bon cycle re s e a rch has been how mu ch know l e d ge of onecarbon cycle component depends on know l e d ge of others .One of the major constraints on the magnitude of the ter-re s t rial sink and its spatial distribution has been know l e d geof the air-sea carbon flux and its spatial distri b u t i o n . In par-t i c u l a r, a d d ressing the Nort h e rn Land Sink hypothesis willre q u i re focused field campaigns over both the Nort hAtlantic and North Pa c i fic to estimate the air-sea flux of car-bon during the time terre s t rial fl u xes are analy z e d . T h e s ecampaigns should include airc raft sampling as well as ship-b o a rd and autonomous measure m e n t s , and will re q u i rei m p roved atmospheric tra n s p o rt models for analy s i s .

F u t u re field campaigns should focus on areas of cri t i c a li m p o rtance to our understanding of the global carbon cycl e .In part i c u l a r, the Southern Ocean air-sea flux is pre s e n t lyp o o r ly constrained and its temporal va ri ability is unknow n .

Oceanic Inventory Measurements

We recommend two programs to characterize theoceanic inventory of fossil CO2 and its evolutionover time:

Thorough analysis of recent ocean carbon obser-vations. At present,the most robust constraint on oceanuptake of CO2 comes from surveys and time-series of car-bon system measurements carried out with instrumentsand methods now available and analyzed by techniqueslike those discussed in Chapter 2. Essential complementa-ry measurements include those of carbon isotopes,thestandard hydrographic and nutrient measurements,andtransient tracers of ocean circulation. These surveys andtime-series measurements provide direct in situ con-straints on ocean CO2 uptake that additionally providestrong constraints on the role of the terrestrial biospherein the global carbon budget.

Over the past decade,the WOCE/JGOFS programs haveprovided an invaluable baseline on the current state of theocean that will be useful for assessing future changes.Unquestionably, however, survey and time-series measure-ments must continue beyond the WOCE/JGOFS era toanswer fundamental questions,such as whether oceanicCO2 uptake is increasing or decreasing globally in thefuture. Such work is the analogue of the current networkof atmospheric measurements,but substantially more diffi-cult to implement. To plan the best possible program ofcontinuing ocean observations,needed spatial resolutionsmust be determined,along with which variables to meas-ure at what accuracy. A high priority over the next fewyears must be to use the current vastly increased oceanicdata set to answer these questions and to design an opti-mal survey program.

Monitoring the evolving fossil CO2 inventory inthe oceans along with related tracers. Monitoring theevolving fossil CO2 inventory is an essential long-termcomponent of any effort to understand oceanic CO2exchange. Because the costs of conducting comprehen-sive surveys are presently very high, we propose that theneeded measurements be accomplished through areduced-level but continuous effort rather than the global“snapshot”paradigm of previous expeditions. This activitywould be an ongoing and focused on specific limitedareas each year. The goal would be to cover the entireworld every 10 to 15 years. This approach would cost lessper year and would assure maintenance of the requiredmeasurement expertise and capability for the long term.The ongoing survey could exploit strong linkages withother efforts such as CLIVAR (Climate Variability andPredictability programme) and GCOS (Global ClimateObserving System) to make efficient use of ship time. It isthus critical to ensure that tracers relevant to the carboncycle (e.g.,DIC,O2, temperature,salinity) are covered inprograms like CLIVAR that have the goal to monitor tem-poral trends in ocean properties.

The measurement suite should include total alkalinityand should frequently include a third CO2-system proper-ty such as pH to assure internal consistency. The


39

measurement of 13C and TOC [total organic carbon] aree s s e n t i a l , as are the development and use of standard s .M e a s u rements should include transient tra c e rs ,w h i ch pro-vide temporal info rmation about ocean mixing and wa t e rmass history that is essential to interpreting fossil CO2 d i s t ri-b u t i o n s . Fi n a l ly, t h ey should include bioactive ch e m i c a l ss u ch as iron whose oceanic distribution is poorly know nbut is like ly an important influence on ocean carbon fl u xe s .

A complementary approach exploits the fact that theinvasion of anthropogenic CO2 into the ocean looks,inaggregate,like a vertical transport problem that can beeffectively attacked through studies of the distribution oftransient tracers. Various tracers are now available to spana variety of mixing scales (3H/3He, 14C,CFC-12,SF6,HFCs). Some of these tracers reveal mixing over the criti-cal longer (decadal and century) time scales;and somehelp identify current short-term invasion rates for compar-ison with older data. Large-scale tracer release experi-ments provide an independent means to assess verticaltransport rates and mechanisms. Likewise,direct eddyflux measurements are just starting to become available totest parameterizations of air-sea exchange.

Process Studies, Models, and Synthesis

We recommend the following four programs:

Selected studies of processes that control themechanisms and rates of air-sea CO2 exchange, ver-tical transport within the seas, and cycling of DICand alkalinity. Ocean process studies yield understand-ing of the basic mechanisms controlling the ocean carboncycle. Process submodels are also required in models ofthe time-dependent air-sea flux. Ideally, process studiesand observational systems should be linked as intimatelyas possible to help determine the response of ocean car-bon fluxes to interannual variability in climate forcing,andto identify feedback mechanisms. Some importantprocesses that need to be studied are gas exchange andthe physical processes that determine the rate of anthro-pogenic carbon transport from the mixed layer into thethermocline and deep ocean. Also important is the role ofbiological processes in determining the spatial and tempo-ral variability of air-sea fluxes and anthropogenic carbonuptake. A major issue is the sensitivity of physical trans-port and biological processes to ocean properties that arelikely to change with climate change. Knowledge of sea-sonal and interannual variability as well as direct manipu-lation experiments will be an important source of infor-mation. A deeper understanding of what controls the bio-logical pump in the ocean is required,including the roleof micronutrients such as iron and zinc,whose input tothe ocean may be affected by climate,land use,or pollu-tion. Program oversight must assure that priority is givento studies that can make a genuine contribution to under-standing future atmospheric CO2 levels.

One challenge is characterizing marine ecosystems wellenough to model their effects quantitatively on air-sea car-bon fluxes. Process studies include direct manipulation ofthe environment,such as the large-scale iron enrichmentstudies (Coale et al.1996),as well as systematic observa-tions of the ef fects of “natural experiments,” such as ElNiño. Process studies,and in particular ecosystem-levelmanipulations (along the lines of the recent iron fertiliza-tion studies) should be used to evaluate the host of bio-logical mechanisms proposed as potential feedbacks onocean carbon uptake (Denman et al.1996). Possible feed-backs that can be tested include modifications of iron fer-tilization rates of HNLC (high-nutrient,low-chlorophyll)regions,particularly in the Southern Ocean;subtropicalnitrogen fixation rates;Redfield carbon to nutrient ratiosof export material;community allocation of sinking versussuspended/dissolved material;calcification rates;and sub-surface remineralization depth scales. The effects ofcoastal eutrophication and elevated UV radiation have alsobeen hypothesized as significant,but are rather poorlyconstrained.

An emerging research theme is the relationshipbetween ecosystem structure and the export flux of car-bon. Ecosystems dominated by the microbial loop recyclecarbon efficiently and show very little export of carbon,whereas other regions dominated by diatoms and otherlarge phytoplankton tend to show a large carbon export.It is important to understand how changes in ocean strati-fication and circulation in response to changes in climatewill affect the ecosystem structure and uptake of anthro-pogenic carbon.

Improved process models. Mathematical models arerequired to express the relationships between physicalforcing and biogeochemical processes,to evaluate theimplications of results and observations for upper oceanpCO2 variations and ocean carbon uptake. Seasonallyresolved models of the circulation and physical propertiesof the upper ocean will improve the ability to take intoaccount the influence of temperature,salinity, and trans-port on sea surface pCO2. To consider interannual andlonger term variability, as well as the effects of climatechange,is essential to explore how changes in sea surfacetemperature,ocean circulation,stratification,and sea iceformation all affect the atmosphere-ocean balance of car-bon. Biological models embedded into models of physicalcirculation will help further explain the large role that sea-sonally and interannually forced biological processes playin regulating upper ocean chemistry and CO2 partial pres-sure in the mixed layer. Models with sufficient resolutionto simulate meso-scale processes will be needed to studyinteractions that determine many of the ecosystemprocesses. A close collaboration with physical oceanogra-phers studying ocean variability is an essential componentof this program.


40

Improved basin and global carbon cycle models.A new generation of models must be developed for long-term projections of carbon uptake,integrated with theobservations to allow data to be assimilated and to testthe models. Conversely, models must be able to simulatethe distribution of important quantities that can be testedin the field and must otherwise help in designing fieldexperiments. For example,because variability in oceanhydrographic structure and ocean carbon cycling is inmost instances associated with circulation changes, varia-tions of carbon and hydrographic parameters will be cor -related. An analysis that does not account for this covari-ance can be seriously in error, and a model that does notsimulate the observed covariance will be wrong. Suchproblems will require a synthesis activity involving bothmodel simulations and measurements. Program structuresmust be implemented to force both modelers and experi-menters to make this integration happen.

Remote-sensing observations and numerical models area natural marriage inasmuch as ocean in situ observationsare often too sparse to initiate or validate model fields.Satellite data are often better matched to the time andspace scales of models,but satellites generally measureonly very simple indices of biological and physicalprocesses and distributions. Thus,models are necessary toextrapolate simple satellite indices to more complex bio-logical and physical processes and distributions consistentwith the accuracy provided by in situ measurements andas required for carbon cycle research.

Remote sensing of ocean conditions andprocesses. Remote sensing is an important tool forextrapolating measurements and calculations in time andspace and for providing estimates of key environmentalparameter values needed by coupled physical and biogeo-chemical ocean models. Remote-sensing observations canalso provide critical information during process studiesand large-scale manipulation experiments.

One important application of satellite data is to relatethe CO2 transfer velocity across the air-sea interface to“sea surface roughness”on ocean-basin scales.Accumulated evidence suggests that estimates of windvelocity alone are not sufficient to estimate transfer veloci-ty, owing to other factors that also influence surfaceroughness. These factors include the sea surface wavespectrum and resulting surface renewal and near-surfaceturbulence (Jähne et al.1987),which are affected by sur-factants (Frew 1997) as well as wind forcing. In particu-lar, small-scale waves (small gravity and capillary waves)are believed to play a primary role in promoting gasexchange,and some of the scattering properties of thesewaves are observable with space radar sensors (scatter-ometers and altimeters). One focus for future effortsshould be understanding the relationship between radarbackscatter from small-scale surface waves and the trans-fer velocity. This work will require the complementary

use of satellite altimeter and scatterometer data toincrease spatial and temporal coverage,as well as in situstudies to develop appropriate equations relating transfervelocity and surface roughness. An essential complementto satellite estimates of net carbon fluxes comes fromatmospheric O2/N2 data. These provide a fundamentalconstraint on seasonal new production (that portion ofprimary production that does not represent internalmixed-layer recycling) on a hemispheric or basin scale,and the interannual variability in these rates.

Current estimates of mean ocean primary productionmust be improved at basin to global spatial scales anddaily to interannual temporal scales. Satellite-derivedfields of near-surface phytoplankton chlorophyll concen-trations are essential to developing the calculations andmodels needed to do this (Behrenfeld and Falkowski1997). An essential focus of this work should be theassessment of new production. New production is moreclosely related to the net flux of CO2 from surface to deepwaters than is total primary production. The relationshipbetween new production and net primary production isnot a simple proportionality. Estimating new productionrequires knowledge of nutrient budgets of surface water,as well as net primary production. Direct measurement ofnutrient concentrations is not possible from space,butsometimes concentrations of major plant nutrients such asnitrate can be related by regression to sea surface temper-ature on regional scales (e.g.,Kamykowski and Zentara1986). Mineral aerosols are one of the primary sources ofiron and other trace elements that limit new productionand net primary production in some important oceanregions,including the Southern Ocean. Aerosol plumesand sea surface temperatures are both detectable by spacesensors.It may therefore be possible to use satellite datato help quantify nutrient budgets in the upper ocean,andhence rates of new production.

Advanced ocean color sensors will help improve esti-mates of global primary production. For example,thesatellite MODIS (Moderate Resolution ImagingSpectrometer) and MERIS (Medium Resolution ImagingSpectrometer) will provide estimates of chlorophyll fluo-rescence,which can then be used to estimate phytoplank-ton quantum efficiency or at least the phytoplankton pho-toadaptive state. The additional spectral bands and highersignal to noise ratio of these advanced sensors will alsoprovide some capability to distinguish phytoplanktonfunctional groups and thus some capability to classifyocean ecosystem structure on large scales.

S eve ral advanced ocean color sensors are planned byNASA and other international space agencies for the nex tfi ve ye a rs , but plans are ve ry uncertain for the post-2005t i m e f ra m e . By the end of the next decade, basic oceancolor measurement capability may be tra n s fe rred to opera-tional satellites (e.g., the NOA A / D O D / NASA National Po l a r -Orbiting Operational Env i ronmental Satellite System,


41

N P O E S S ,p ro j e c t ) , but it is not clear if the full capab i l i t i e santicipated from MODIS and MERIS will continue beyo n dthe middle of the next decade. C o o p e ration and coord i n a-tion among international space agencies should be encour-aged to ensure the continuing remote-sensing capab i l i t i e sre q u i red to support future carbon re s e a rch pro gra m s .

Goal 3: Assessing the Role of Land Use

Chapter 2,in its final section,discussed the importanceof land use to the overall carbon cycle,as well as to under-standing the time history and spatial distribution of car-bon sources and sinks. Goal 3 of the CCSP is therefore toestablish accurate estimates of the impacts of historicaland current land use patterns and trends on the evolvingcarbon budget at local to continental scales. The follow-ing are the major program elements and activities pro-posed to move toward this goal.


• Document t he hist or y of agr icul t ur al ex pansionand abandonment and i t s impact on t he cont empo-r ar y car bon balance of Nor t h Amer ica. Conductint ensiv e hist or ical land use analy ses acr ossenv i r onment al gr adient s, in coor dinat ion w i t h t hepr oposed Nor t h Amer ican net w or kof eddy f lux measur ement si t es.

• Impr ov e obser v at ional capabi l i t ies in t r opicalr egions t hr ough cooper at ion w i t h t r opical nat ions,including int er nat ional ex pedi t ions and ex changesand t r aining of scient ist s and manager s.

• Apply t he obser v at ional pr ogr ams discussedunder Goal 1 in t he t r opics, including long- t er mobser v at ions and model ing.

• Couple socioeconomic analy sis of land manage-ment st r at egies w i t h biogeochemical models t oev aluat e t he social and biophy sical v iabi l i t y off ut ur e car bon sequest r at ion opt ions.

The emphasis of the proposed program elements andactivities is the tropics and North America. As discussedin the final section of Chapter 2,the tropics are importantbecause of the magnitude of the postulated deforestationsource in this region. The focus on North America com-plements the proposed terrestrial sink program elementby providing information on the history of the observa-tion sites.

The data needed to construct reliable land use historieshave temporal and spatial gaps over large parts of theworld. In the tropics particularly, the expansion of

cropland, much of it unrecorded,is presumed to be amajor contributor to deforestation (FAO 1995). Theunrecorded tropical cropland expansion tends to be high-ly fragmented,making its monitoring by remote sensingdifficult with all but the highest resolution sensors.Afforestation in the Northern Hemisphere mid-latitudes isa significant component of the Northern Land Sink, yetthere is a lack of uniformity in the criteria used to enu-merate afforested area and define associated carbon fluxdensities (Nilsson and Schopfhauser 1995). A major effortis needed to assemble land use histories for both tropicaldeforestation and Northern Hemisphere afforestation. Thebasis for such histories includes a variety of potential datasources (e.g.,national censuses,provincial and local landuse surveys and tax rolls, remotely sensed imagery, andproxy measures such as changes in local population andtechnology regimes). These land use histories must bespatially explicit (of county-level or finer resolution,depending on data availability) and should extend as farback in time as needed to capture the current impacts ofpast disturbances on carbon fluxes in a particular region.

A wide range of historical data have been applied tothe estimation of CO2 emissions from land use (e.g.,Houghton et al.1983,Melillo et al.1988, Foster 1993).Recent progress has focused on the merging of manykinds of historical records with satellite-derived measuresof land cover, and on the use of increasingly sophisticatedmodels of ecosystem response (e.g., Foster 1995). Thisapproach constrains historical reconstructions to be con-sistent with the geographic distribution of current landcover, and provides the spatial resolution needed to testhypothesized sources and sinks. As high-resolution atmos-pheric sampling continues and expands in the future,records of CO2 and its stable isotopes,combined withrecords of 14C,CH4, and other indicators of particular landuse effects (e.g., N2O),will become increasingly valuableas historical constraints.

Of particular concern in re c o n s t ructing historical emis-sions is the question of a pre - agri c u l t u ral steady state.Although ice core re c o rds indicate that the global CO2b u d get was near a steady state for thousands of ye a rsb e fo re the 19th century, this evidence does not assure thatp a rticular re gions we re not net sources or sinks for car-b o n . For ex a m p l e ,s eve ral studies have suggested that high-latitude soils we re still sequestering carbon as part of theirre c ove ry from the most recent deglaciation (Billings et al.1 9 8 2 ,H a rden et al. 1 9 9 2 ) . Estimates of human emissionst y p i c a l ly assume a pre - agri c u l t u ral steady state (e.g.,Houghton 1993), w h e reas direct assessments of CO2 fl u xe sm ay re flect local non-steady-state effe c t s . T h u s ,i n t e gra t i o nof emissions estimates and flux observations must incl u d ecautious consideration of steady-state assumptions.

Estimates of human CO2 emissions must account forCO2 fluxes associated with materials removed to locationsother than the sites of original growth or burial. The need


42

to account for the fate of forest products is widely recog-nized (Houghton et al.1983). However, confusion persistsabout what oxidation of products from previous harvestsshould be reported in accounting for the net effects offorestry activities (Houghton 1993). Recently, Stallard(1998) has suggested that eroded soil carbon constitutesanother “off-site” form of carbon that must be explicitlyincluded in carbon budgets. Because agricultural activitiesare known to have vastly increased historical rates of ero-sion in northern temperate latitudes,the fate of erodedcarbon is directly relevant to testing the Northern LandSink hypothesis.

Goal 4: Improving Projections ofFuture Atmospheric CO2

Better projections and information are needed to meetfuture societal needs:


• Impr ov e r epr esent at ion of phy sical and biologi -cal pr ocesses in models of t he car bon cy cle t or ef lect obser v at ions and new know ledge mor eaccur at ely .

• Pr ov ide a f r amew or k f or r igor ous analy sis ofobser v at ion and independent compar isons andev aluat ions of cl imat e and car bon models usingcompr ehensiv e dat a and bot h f or mal and inf or malmodel assessment act iv i t ies.

• Dev elop new gener at ions of t er r est r ial bios-pher e and ocean car bon ex change models, includ-ing t he r oles of bot h nat ur al and ant hr opogenicdist ur bances, succession, and t he f eedbackst hr ough cl imat e change and ant hr opogenic per t ur -bat ions t o car bon diox ide and nut r ient s.

• Dev elop coupled ear t h sy st em models incor po-r at ing t er r est r ial and oceanic biogeochemicalpr ocesses in cl imat e models of t he at mospher eand ocean.

• Pr edict ions t he f ut ur e ev olut ion of CO2 concen-t r at ions, f or use in ev aluat ing t he consequencesof societ al decisions.

The foundation for predicting future changes in thecarbon cycle is understanding the mechanisms that regu-late it,the way those mechanisms respond to environmen-tal changes,and the ways those mechanisms interact. Likecurrent understanding,advances in the future will bebased on a combination of observations,manipulativeexperiments,and synthesis via models. Useful models will

range from qualitative conceptual models to schemes forconnecting or extrapolating observations,to simulationsof multifactor processes. Models based on all theseapproaches can provide powerful tools for asking “what-if” questions about the consequences of future events.Manipulative experiments can provide a mechanism fordirect tests of critical model predictions at a range of spa-tial and temporal scales.

Models

The following provides an overview of the types ofmodels and experiments needed for prediction,includingthe necessary scientific foundations for credibility.

Models for Data Analysis. A vast array of observa-tions of the natural carbon cycle and results from experi-ments that manipulate controlling variables will emergefrom the proposed program. Models designed for dataassimilation and analysis represent a critical step in devel-oping more refined experiments.Inverse and diagnosticmodels (the latter used to probe data sets) are the bridgebetween the data collection,conceptual understanding,and predictive models. For example,tracer transport mod-els use atmospheric obser vations to test rates of transportderived from general circulation models (GCMs) or assimi-lated meteorological fields. Inverse models use atmos-pheric and oceanic observations of concentrations to esti-mate surface fluxes,providing a test for hypotheses aboutthe time-space variability of fluxes.Inverse modeling thusforms a link between data,theory, and prediction,provid-ing constraints on today’s fluxes that should be consistentin models of the potential future.

Advanced Component Models and Scaling.Process and theoretical studies lead to new understandingand,in turn,improvement in model output,as mecha-nisms are incorporated into models. The impact ofprocesses on the whole system must be evaluated bycomparing the improved models to earlier formulationsand to observations,and by conducting sensitivity analy-ses to determine their importance.

Some new scientific results must be scaled in time ands 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 CO2 n e tex ch a n ge and rates of atmospheric mixing) can be under-stood by micro m e t e o ro l o gical and plant phy s i o l o gi c a lt h e o ry. H oweve r, as it occurs on finer spatial and tempo-ral scales than those re s o l ved in global models, a na p p ro a ch for ag gregating the effects of the re c t i fier has tobe deve l o p e d . S i m i l a r ly, the biochemical effects of CO2on photosynthesis have been understood for over ad e c a d e , but the impact on long-live d , whole plants, p l a n tc o m mu n i t i e s , and ecosystems is uncl e a r. To ach i eve theo b j e c t i ves listed for Goal, 4 a vigo rous pro gram linkingp rocess studies to modeling must be maintained for the


43

n ew science to be incorporated in local and globalp re d i c t i ve models.

Models for Decision Analysis. The great majority ofterrestrial systems are currently managed. Managementchoices have implications for carbon storage (e.g., choicessuch as no-till agriculture, fire suppression,length of forestharvest rotation,and fertilization). Carbon storage is onefactor to consider in making land management decisions,and reliable tools are needed for evaluating the carbonstorage consequences of various land management scenar-ios. There is an urgent need for local- or regional-scalemodels in agriculture, forestry, urban ecosystems,andmarine systems that can be used to evaluate the effects ofspecific interventions (e.g.,no-till agriculture,deep oceanCO2 injection) to enhance carbon storage,along withassessing associated environmental and economic conse-quences.These models will form a link between the basicscience of the carbon research community and the needsof stakeholders and decision makers.

Carbon System Models and Carbon-ClimateModels. The carbon system is intimately coupled to thephysical climate system. Indeed,the focus on carbonderives from the role of CO2 as a greenhouse gas. To proj-ect the large-scale consequences of changes to the carboncycle,it is necessary to link knowledge of the carbon sys-tem to that of the climate system. For example,ifenhanced concentrations of greenhouse gases disruptedoceanic circulation,the ocean carbon system could beaffected by changes in both physics and biology. Thesechanges would help shape the effect of any anthro-pogenic increase in CO2 emissions on atmospheric con-centrations,thus modulating the climate response.

Similarly, feedbacks could occur through terrestrial sys-tems,with terrestrial carbon exchange possibly modifiedby climate then affecting atmospheric CO2, in turn modi-fying the climate impact of any given anthropogenic emis-sion scenario. These feedbacks are presently uncertain,depending on the rate,magnitude,and spatial distributionof climatic changes. For example,if climate changes aregreatest in peatlands at high latitudes,CO2 emissionswould likely increase. More equable warming could resultin either increases or decreases of carbon storage in theterrestrial biosphere.

To analyze the potential effects of future industrial andland use emission scenarios,Earth system models arerequired that couple physical models of the atmosphere,ocean,and cryosphere to biogeochemical models of ter -restrial and marine ecosystems. Earth system models pre-dict partitioning of emissions of CO2 (and other green-house gases) between the mobile terrestrial,oceanic,andatmospheric reservoirs,and then predict climaticresponse and the ef fects on biophysical processes in theland and ocean.

Earth system models build on the knowledge of mecha-nisms gained in process studies and advanced componentmodeling studies,and on the evaluation of knowledgederived from diagnostic studies and historical and paleo-records. They will allow us to examine questions aboutthe relationships among policy choices,sources and sinks,and resulting concentrations. These models demand rigor-ous examination of knowledge, extensive computingresources,and sustained efforts of testing and refiningusing integrated knowledge from the Carbon CycleScience Program overall.

All of the above models must be painstakingly com-pared to observations, requiring strong commitment totesting and refining models. It can be very difficult to testlarge-scale models against real data sets,obtained in specif-ic locations and at specific times. A coordinated effort topromote rigorous comparisons of models and measure-ments is needed on the part of the agencies involved inthe CCSP.

Experiments

Experimental studies on the future of the terrestrialcarbon cycle should be used to probe controlling process-es,especially when the nature of the controls may beobscured by spurious correlations or long time constants.These studies should explore the behavior of key mecha-nisms to forcing outside the range of current ambient con-ditions,and evaluate responses to forcing from simultane-ous changes in multiple environmental and ecological fac-tors. In addition, well-designed experiments can be pow-erful tools for making results from carbon cycle researchaccessible to a broad range of stakeholders. Manipulativeexperiments are,however, complements to models,notfull alternatives. The processes controlling the future ter-restrial carbon cycle operate over a broad range of spatialand temporal scales. Many processes with the potential todominate future changes in terrestrial carbon storage—forexample, changes in the tree species that dominate forestsor changes in the frequency of wildfire—are beyondaccess through short-term,small-scale experiments. Evenif it were feasible to make an aggressive commitment ofresources to work at greater temporal and spatial scales,manipulative experiments lose their utility as the timescale of the experiment merges into the tempo of globalchanges,the far from fully replicatable grand experimentat the global scale.

Manipulative experiments can provide powerful accessto key uncertainties about the future trajectory of theglobal carbon cycle. However, many aspects of the terres-trial carbon cycle are very difficult to study with manipu-lative experiments. Consequently, the research targetsneed to be selected with careful attention to the likelyimportance of the focal processes,the efficiency with


44


45

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 atmosphericCO2 may bear the signature of climate effects on ecosystems and the oceans. Understanding the mechanisms thathave caused global terrestrial and ocean carbon exchange to vary over the period of observation provides a vital testfor the “scaling”of process level knowledge and a basis for predictive modeling.


46

which the experiments probe these processes,and thefeasibility of obtaining the critical information using otherapproaches. In addition, experiments must be designedfor effective integration with observational studies andmodels. The following approaches should be pursuedand enhanced.

Natural Experiments. Frequently, nature presents aready-made experiment,providing investigators an oppor -tunity to study ecosystem responses to a major environ-mental forcing. Natural experiments are often permanentor at least persistent features of the landscape,allowingunique access to responses with long time constants,usu-ally with little or no cost for maintaining the naturalmanipulation. Recent studies with ecosystems near natu-ral CO2 vents are a good example of the applicability ofthese features for understanding the future of the globalcarbon cycle (Raschi et al.1997). The Hawaiian lava flowswith a range of different ages and local climates representa parallel example for basic ecological research that is fun-damentally relevant to the carbon cycle (Chadwick et al.1999,Torn et al.1997).

Natural experiments are invaluable links between long-term observations and purposeful manipulations. Theyare natural system analogs of research designed to takeadvantage of recent patterns of land use change. Becausethese experiments provide unique opportunities toobserve long-term,and sometimes large-scale, responses tocontrolling factors,they should receive greater emphasis.Natural experiments that should be evaluated include notonly CO2 vents and gradients of substrate age,but alsosteep,topographic gradients of temperature and precipita-tion, geothermal features,and sites with unusual flora andfauna as a result of isolation. Natural experiments form acontinuum with gradient studies. Some of the most pow-erful uses of the natural experiment approach may lie incombining common gradients with much less commonnatural experiments.

Multifactor Manipulations. Results from globalchange experiments are often difficult to interpret inmore general terms. Ecosystems targeted by differentexperiments typically differ in a number of features,creat-ing a barrier to partitioning control for the differentresponses among the different features. Multifactormanipulations can play a critical role in enhancing thefoundation for generalization and reducing uncertaintiesabout future responses. The next generation of multifac-tor experiments should address two kinds of questions.First,how do ecosystem-scale responses to global changecomponents change when the global changes occur incombination? Experiments should consider interactionsamong elevated CO2, warming,and altered levels of pre-cipitation,nitrogen deposition,and ozone. Second,howdo the characteristics of ecosystems themselves affectresponse to global change? A new generation of experi-

ments should focus explicitly on the way that differencesin ecosystem characteristics influence responses to globalchanges. For example,matched manipulations on sub-strates of different age or with ecosystems of different bio-logical diversity can greatly enhance the ability to general -ize. Some of these experiments should focus on the inter-face between land use and atmospheric change,withmanipulations of CO2, temperature,and other environ-mental factors at sites of different ages since disturbance,with different management practices,and with differentlevels of anthropogenically stimulated biological invasion.Multifactor experiments should address a broad range ofmechanisms that may influence future carbon storage.These mechanisms range from direct responses of plantgrowth,to enhanced success of one or more plant ormicrobial species,to altered decomposition,to changes insensitivity to fire,pests,or pathogens. Integrating resultsfrom these multifactor experiments in models and generalinterpretations will present many challenges,but it ismuch wiser to confront the challenges than to ignorethe processes to which these experiments provideunique access.

Experiments with Model Ecosystems. The agendaof multifactor experiments is challenging,complex,andessential. Many of the processes that may play criticalroles in ecosystem responses have large spatial scales andlong response times. This circumstance places a high pri-ority on optimizing the trade-offs between experimentaltractability and relevance to the ecosystems with thelargest potential impact on the carbon cycle. As in manyfields of science,progress can be rapid if a sizeable frac-tion of the resources is invested in studies on experimen-tal models. These are systems chosen with an emphasison tractability and access to the key processes,and withmuch less emphasis on the model system’s quantitativecontribution to the global carbon cycle. For example,studies on annual grasslands can address multiple globalchange factors while allowing reasonable replication andlarge numbers of individuals per replicate. For other pur-poses,studies on artificial ecosystems in meso-cosm facili-ties may provide a useful balance of tractability and rele-vance. With sufficient resources and planning,usefulexperiments on model systems might operate on a largescale,addressing, for example,the relative success ofgrassland species and rapidly growing trees under a rangeof manipulations.

A re s e a rch pro gram with an emphasis on model systemswill re q u i re careful planning and coordination acro s sex p e ri m e n t s . Scaling interpretations from the models tothe systems with the most direct re l evance to the carbonc y cle may re q u i re ex p e riments delibera t e ly designed toa d d ress uncertainties in the scaling. In the longer term , weshould expect a continuing iteration among ex p e ri m e n t s ,o b s e rva t i o n s , and models. Evidence from one of thesea p p ro a ches will re c e i ve added value from extension acro s s

the others , and hypotheses suggested by one appro a ch willbe tested by evidence from all three appro a ch e s .

Goal 5: Evaluating Management Strategies


• Sy nt hesize r esul t s of global t er r est r ial car bonsur v ey s, f lux measur ement s, and ot her r elev antr esul t s, and use t his analy sis t o dev elop t he sci -ent i f ic basis f or management st r at egies t oenhance car bon sequest r at ion.

• Det er mine t he f easibi l i t y , env i r onment alimpact s, st abi l i t y , and ef f ect iv e t ime scale f orcapt ur e and disposal of indust r ial CO2 in t he deepocean and in geological r eser v oi r s.

• Def ine pot ent ial st r at egies f or max imizing car -bon st or age t hat simul t aneously enhance economicand r esour ce v alues in f or est s, soi ls, agr icul -t ur e, and w at er r esour ces.

• Ident i f y t he cr i t er ia t o ev aluat e t he v ulner abi l -i t y and st abi l i t y of sequest r at ion si t es t o cl imat echange and ot her per t ur bat ions.

• Document t he pot ent ial sust ainabi l i t y , l i f e-t imes, and int er annual / decadal v ar iabi l i t y of di f -f er ent managed sequest r at ion st r at egies.

• Pr ov ide est imat es of t he uncer t aint ies r elat edt o managed sequest r at ion t hat ar e r equi r ed f orpol icy discussions and decisions.

• Dev elop t he moni t or ing t echniques and st r at egyt o v er i f y t he ef f icacy and sust ainabi l i t y of car -bon sequest r at ion pr ogr ams.

• Cr eat e a consist ent dat abase of env i r onment al ,economic, and social per f or mance measur es f ort he pr oduct ion, ex ploi t at ion, and f at e of w ood andsoi l car bon “ f r om cr adle t o gr av e” ( f or w ood,f r om st and est abl ishment t hr ough f inal disposal ;f or soi l car bon, f r om f or mat ion t hr ough bur ial ) .

• Dev elop and par amet er ize an analy t ic f r ame-w or k f or pr edict ing and ev aluat ing managementand pol icy al t er nat iv es consider ing mat er ialt r ansf er , economics, societ al f act or s, w ast esand emissions, r aw mat er ial , labor and ener gyinput s, car bon sour ces, sinks, and f lux es, andout put f or mass car bon balance.

Human land use may be directed toward enhancing ter-restrial carbon sinks through a variety of managementoptions. Focusing on the United States,such optionsinclude manipulation of forestry practices (such as harvestdynamics,and pest and pathogen control) and agriculturalland management practices (such as tillage,crop rotations,and fertilizer use;see Lal et al.1998). To evaluate thepotential of management strategies to sequester carbon,there must be improved process understanding of therelationships between changes in land cover and biophysi-cal controls of carbon exchange. One way to gain suchunderstanding is the close coupling of historical informa-tion on land use with eddy-correlation experiments acrossa network of sites encompassing large temporal and spa-tial variability of fluxes and land uses (beginning with theAmeriFlux network). The land use information is neededto infer historical ecologies (e.g.,species composition,soilorganic matter dynamics) that help explain current car-bon fluxes. Where possible in this network, experimentaldesigns should allow for the observation of the ef fects ofalternative management strategies (e.g., fire suppression,age-selective forest harvesting). Efforts should be made toensure that new tower sites have highly variable land usehistories. This will broaden the base of knowledge fromwhich to project future fluxes. Further coupling of landuse and carbon flux information with biogeochemicalmodels,such as those participating in the Vegetation/Ecosystem Modeling and Analysis project (VEMAP),maybe used to predict carbon uptake by whole ecosystems.

Results from the U.S. Forest Service Forest InventoryAnalysis (FIA;Schroeder et al.1997),of periodic measure-ments of wood volume on thousands of plots of landclassed as “forest,” indicate lower values of net carbonaccumulation in forests (about 0.3 Gt C/yr) for NorthAmerica than are inferred from inverse analyses (0.9 to 1Gt C/yr) or tower flux data (~0.7 Gt C/yr). However, theFIA data set has not been optimized nor sufficiently ana-lyzed as yet to ensure that the estimated rates for net car-bon sequestration derived from it are accurate. The focushas been to define the wood volume potentially availablefor harvest. Thus, wood volumes on land reclassified asnonforest and wood harvested as forest products are bothremoved from the inventory, but a very large quantity ofthis wood may remain in the form of wood for long peri-ods. Properly accounting for such “removals”in the car-bon budget should increase corresponding estimates ofnet carbon sequestration by 50 to 100 percent. In theCCSP, we propose broadening the FIA to incorporate anexplicit focus on carbon,including upgrading the sam-pling scheme,increasing the number of parameters meas-ured,and increasing the frequency of measurements.

In deep ocean sequestration of CO2, industrial CO2recovered from fossil fuel power plants may be injecteddirectly into the deep ocean for disposal. This arrange-


47

ment would circumvent the slow natural transfer rate ofCO2 from the surface to the deep ocean. Laboratory stud-ies and in situ experiments show that liquid CO2 is denserthan seawater at water depths below about 3,000 meters.Additionally, rapid growth of CO2 hydrate crystals alongthe water-liquid CO2 interface has been observed. Thus,liquid CO2, when piped onto the seafloor, is likely to settlethere and be quickly transformed into solid hydrate forms.The liquid CO2 and the hydrate crystals dissolve graduallyinto the overlying seawater and mix into the surroundingdeep sea. The retention time of CO2 in the deep oceandepends on ocean transport and chemical processes suchas neutralization reactions with calcareous sediments.Once the water returns to the ocean surface,some CO2can escape back to the atmosphere. Hence,disposal inrelatively isolated regions of the ocean is preferredbecause of longer retention times. The effects of acidifiedseawaters on benthic ecosystems must be evaluatedbefore any large-scale disposal can be contemplated.Near-field as well as far-field distributions of CO2 around adisposal site must be investigated. The scientific investiga-tions outlined in the CCSP, such improvements in oceangeneral circulation models (GCMs),will be directly appli-cable in evaluating CO2 sequestration in the oceans.

Not all management strategies will be economicallyviable or socially acceptable. For example, over the longrun,a rising global demand for timber products may pro-vide a strong inducement to keep forest harvest cyclesshorter than the optimal time for sequestering the greatestamount of carbon. Collaboration among biogeochemicalmodelers,and economists and other social scientists willbe necessary to examine the economic efficiency andsocial desirability of various management strategies. Suchcollaboration will require information flows across a per-meable boundary between the initiatives proposed in thisplan and associated work in the human dimensionsresearch community.

Observing and Discovering

An essential complementary perspective to testingdefined hypotheses is suggested by the second central sci-entific question for the CCSP: “What will be the futureatmospheric CO2 loading resulting from both past andfuture emissions?” It is very difficult to frame this ques-tion within the context of a single hypothesis or even aframework of closely linked hypotheses. Scientists mustlook toward the future with considerable humility, bearingin mind the significant uncertainties involved in identify-ing the processes responsible for a carbon sink as large asthe hypothesized Northern Land Sink,or in trying to pre-dict how the ocean sink may develop over time. Someissues concerning future carbon cycling can be addressedthrough such hypotheses as the Increasing OceanInventory. It is also possible to formulate alternative

assumptions as competing hypotheses for carbon cyclingin the future,but none stands out as pervasively importantor useful in the sense that the Northern Land Sink andIncreasing Ocean Inventory hypotheses dominate currentstudies. Questions about future atmospheric CO2 levelshighlight the need for a comprehensive and balanced pro-gram of observations,manipulative experiments,and mod-els. These must contribute to the focused testing ofhypotheses while remaining comprehensive enough toentrain the new ideas and discoveries that will guidehypotheses for the future. Through testing currenthypotheses,unforeseen questions,as well as answers,willsurely emerge.

In particular, current attempts to project the futurerequire assumptions and speculation that extend wellbeyond the range of issues of any program focused onpresently testable hypotheses. A future could be hypothe-sized for atmospheric CO2 based on reasonable emissionsscenarios,consistent with the notion that future trendswill follow the pattern of past trends (in biological andocean uptake,emissions,etc.). However, this assumptionencompasses many untested hypotheses. For example,the IPCC scenarios for future CO2 levels are based onmodels that have been calibrated using a “fertilizationeffect”to account for the past and present “missing”car-bon sink. Yet this effect is just one of many proposedmechanisms for terrestrial carbon uptake,and large-scaleefforts to test it are just beginning. It is unknown howstable terrestrial uptake mechanisms are;the fertilizationeffect might be expected to continue for much longerthan some other contributing mechanisms,such as refor-estation in the United States and Europe. Likewise,theIPCC CO2 scenarios depend on ocean models that assumea steady climate. These assumptions seem increasinglyprecarious as more is learned about the interplay betweenoceanic and atmospheric dynamics,and as the prospect ofsignificant changes in the global climate system becomemore likely. Clearly, even as the scientific communityworks toward resolving uncertainties through focusedhypothesis testing,strong action is necessary to assurethat new ideas and discoveries continue to help solvequestions about the future behavior of the Earth system.

As this report has made clear, an essential componentof the research strategy is a sustained observational effort.This endeavor should provide improved,quantitative esti-mates of natural and anthropogenic carbon distributions,sources,and sinks,and their temporal and spatial variabili-ty at regional as well as global scales. Efforts to developmore focused studies of particular problems,or tests ofparticular hypotheses,should not obscure the need tocontinue and expand a program of comprehensive basicobservations,with parallel modeling to assimilate the data.Because these observations will be useful in assessing car-bon sequestration activities,policy needs must be treatedwith utmost attention during planning. The design and


48

implementation of such a program,in a cost-effectivemanner and in concert with the international researchcommunity, is a major task.

In addition to a program of comprehensive observa-tions,support is suggested for four types of studies out-side the realm of programs to test current hypotheses.Because all these study types are inherently innovative,they are defined by example here, rather than by descrip-tions that might prove too restrictive.

Studies That May Anticipate “Surprises.” Somestudies may help reveal unexpected interactions betweenthe global carbon cycle and the climate system. Recentstudies of paleoclimate have revealed the past occurrenceof surprising short-term instabilities in the Earth’s climatesystem. Concerns about instabilities in future climatemust be extended to the potential effects of these instabil-ities on carbon cycling.

An example of one such surprise was the finding ofcoupled atmosphere-ocean simulations of climate change,which predict large increases in the vertical stratificationof the ocean with warming in lower latitudes and freshen-ing in higher latitudes. These changes would lead to largereductions in ocean thermohaline circulation as well as invertical mixing in general (Manabe and Stouffer 1994).The impact of these changes on ocean biology would like-ly be extremely large. The effects on the air-sea balance ofcarbon are difficult to predict,,because of the counteract-ing effects of slowed circulation and a potentially moreefficient biological pump taking up a higher fraction ofsurface nutrients (Sarmiento et al.1998).

Climatic warming appears to promote carbon storagein mid-latitude forests,but possibly to promote release ofmuch larger stores of organic carbon from boreal forestsand peatlands through increased rates of fire and mineral-ization (Goulden et al.1998,Kasischke et al.1995)

Studies That Suggest Innovative Hypotheses orIdeas. Creative ideas often do not fit preconceivedhypotheses or priorities. Important new concepts mayarise from unexpected sources,often from scientific sub-disciplines not normally associated with carbon research.

A recent paper (Stallard 1998) uses evidence from geo-morphology and sedimentology to hypothesize that largeamounts of carbon may be buried on land as a result ofenhanced erosion and subsequent deposition accompa-nied by fertilization associated with cultivation. This studyhas focused attention on the need to account for the fateof large quantities of eroded soil carbon.

The hypothesis that iron might play an important rolein oceanic photosynthesis has a long history. However,the first trustworthy in vitro tests of the hypothesis had toawait the development of accurate techniques for meas-urements in seawater (Martin 1991). Recent in situ ironfertilization experiments provide dramatic support (Coale

et al.1996) for the bottle experiments. It has beenhypothesized that iron supply played an important role inthe changes of atmospheric CO2 content of the last iceages (Martin 1990). Iron also also appears to be a majorfactor in nitrogen fixation in the low-latitude surfaceocean (e.g.,Karl et al.1997,Michaels et al.1996a,Michaelset al.1996b). A major source of iron to the ocean is dustdelivered through the atmosphere. One notable questionthat has arisen,then,is whether changes in dusttransport that accompany climate change could af fectoceanic photosynthesis.

Studies of Past Geologic Variations in the CarbonCycle. It is widely recognized that the global carboncycle has varied in the geologic past. Analyses of gasessampled from ice cores have shown that atmospheric CO2and CH4 concentrations varied in concert with the pro-nounced fluctuations between glacial and interglacial cli-matic conditions over the past several hundred thousandyears (Barnola et al.1987,Stauffer et al.1988, Jouzel et al.1993). Larger past CO2 variations—comparable to thoseanticipated from future human activities—have beeninferred from models based on the geologic record of thelast 100 million years and longer (Berner 1994). Many ofthe processes involved in these changes are so slow thatthey have little relevance to time scales of usual humanconcern. However, the geologic record is a vital source ofinformation about the dynamic nature of carbon cyclingand its interactions with the global climate system over abroad spectrum of time scales. Geologic research can pro-vide important and unexpected insights that are directlyrelevant to the principal questions addressed in thisscience plan.

Measurements from the paleorecord suggest thatexchanges between reservoirs of the carbon system arenot static on millennial time scales. For example,analysesof an ice core from Taylor Dome,Antarctica,show thatatmospheric CO2 concentrations changed over time dur-ing the Holocene,with a steady increase from 260-280ppm over a period of 7000 years (Indermühle et al.1999).These changes may have been due to changes in theamount of terrestrial biomass and/or ocean CO2 storage.Numerical models of anthropogenic CO2 generally assumethat a steady state existed before human influence. Thisassumption appears to be reasonable for global modelsbecause the rate of CO2 change during the Holocene wasvery slow compared to the anthropogenic CO2 perturba-tion. However, the changes indicated by the ice core datamay be quite significant for studies of carbon budgets on aregional scale,because the inferred imbalances in the car-bon cycle are not likely to have been uniformly distrib-uted over the Earth’s surface.

Using precise correlations among ice and sedimentcores,paleoclimatologists have shown that the mostrecent glacial cycles were punctuated by large andextremely abrupt climate events (of years to decades)


49

(Blunier et al.1998,Dansgaard et al.1993). Many of theserapid events appear to correlate with abrupt changes inatmospheric methane (Brook et al.1993,Chappellaz et al.1993),and some may have been associated with modestchanges in atmospheric CO2 (Stauffer et al.1998). Thus,even on time scales of years to decades,there is geologicevidence of close interaction between climate and the car-bon cycle. The evidence suggests that atmosphericmethane may be a more sensitive indicator of certainchanges than atmospheric CO2.

Studies of the “Human Dimensions” of FutureCarbon Cycle Trends. Future terrestrial carbon budgetswill be strongly influenced by human land use. The bestattempts to forecast future trends may be confounded byuncertainties about future human activities that are influ-enced by factors beyond the usual scope of scientificinterests in carbon cycling.

Future rates of tropical deforestation will depend large-ly on the expansion of regional agricultural capacity. Mostof the developed countries,which are usually outside thetropics,will depend almost exclusively on technology-driv-en increases in productivity per unit land area to increasefuture capacity. In these countries,new changes in theamount of cropland in production are likely to be negligi-ble.The developing countries will depend on a mix ofincreases in productivity along with expansion of crop-land to increase future capacity. Excluding China,thesecountries have about 2.5 billion hectares of land on whichrain-fed crops could give reasonable yields,with approxi-mately 80 percent of it in tropical Africa and Latin America(FAO 1995). Much of the deforestation taking place in thetropics is the result of the disorderly, unrecorded expan-sion of cropland (FAO 1995). The FAO (1995) projectsthat deforestation from this unrecorded expansion ofcropland will not only likely continue,but will probablydo so at a rate exceeding that required to meet nationalagricultural capacity targets. Quantification of rates ofunrecorded conversion of tropical forest to cropland is acritical need for projecting future rates of deforestation.

Expected Results

The ultimate goal of the U.S.CCSP is to provide thescientific understanding required to predict futureconcentrations of atmospheric CO2 and evaluate alterna-tive scenarios for future emissions from fossil fuels,effectsof human land use,sequestration by carbon sinks,andresponses of carbon cycling to possible climate change.

These issues are addressed by Goals 4 and 5—improv-ing projections of atmospheric CO2 through complemen-tary methodologies and by incorporating regulating mech-anisms,and developing a scientific basis for carbonsequestration management strategies. Achieving these two

goals will permit us to carry out several critical tasks:

• Present a new generation of models, rigorously testedusing time-dependent,three-dimensional data sets,suit-able for predicting future changes in atmospheric CO2.

• Develop management and mitigation strategies thatoptimize carbon sequestration opportunities using ter-restrial ecosystems,primarily forest and agricultural sys-tems,in addition to evaluating the efficiency of oceancarbon sequestration.

Currently, estimates of terrestrial sequestration andoceanic uptake of CO2 vary significantly, depending onthe data used and the analytical approach (e.g., ForestInventory Analysis,direct flux measurements in majorecosystems,inverse model analysis of CO2 data from sur-face stations, changes over time of global CO2, O2,13CO2/12CO2, etc).

The proposed research program under Goals 1,2,and3,will reconcile these estimates to a precision sufficientfor policy decisions by providing new types of data andnew, stringent tests for models and assessments. The spe-cific deliverables from these three goals are as following.

Goal 1 is to establish accurate estimates of the magni-tude of the potential Northern Hemisphere terrestrial car-bon sink and the underlying mechanisms that regulate it.

• Define the existence and magnitude of past and pres-ent terrestrial carbon sinks.

• Elucidate the impacts of climate variations (e.g.,lengthof growing season,soil moisture,long-term temperaturechanges),of fertilization (CO2, NOx, etc.),and of landuse changes on atmospheric CO2.

• Document and constrain uncertainties related to thepotential Northern Hemisphere terrestrial carbon sink.

Goal 2 is to establish accurate estimates of the oceaniccarbon sink and the underlying mechanisms that regulateit. In the near-term,the focus will be on the NorthAtlantic and North Pacific oceans to best complement theterrestrial focus on the Northern Hemisphere and tropics.

• Incorporate better understanding of ocean processes incurrently undersampled regions,such as the SouthernOcean,into general circulation models (GCMs).

• D e t e rmine the ex i s t e n c e ,m ag n i t u d e , and intera n nu a lva ri ability of oceanic carbon sinks and sources on are gional scale through assimilation of new observa t i o n s .

• Explain observed changes in the ocean carbon sink dueto variations in circulation and biological and chemicalchanges using enhanced models tested rigorouslyagainst new observations.

• Incorporate improved oceanic CO2 flux estimates toimprove constraints on inverse models for estimating


50

terrestrial sinks.

Goal 3 is to establish accurate estimates of the impactsof historical and current land use patterns and trends onthe evolving carbon budget at local to continental scales.

• Document the existence,location,and magnitude ofcarbon sources and sinks resulting from historical andcurrent land use changes and land management prac-tices in critical regions.

• Provide improved assessments of the uncertaintiesrelated to land use,particularly for agriculture in NorthAmerica and the tropics.

The results of all the studies under goals 1 through 5will be communicated to the public and policy makers inpapers published in peer-reviewed journals.


51

Table 5.1 breaks out the Carbon Cycle Science Plan(CCSP) program elements and summarizes the researchcomponents and approximate costs. The elements arealso mapped onto the CCSP goals as described below.

Initial Funding Priorities

Most of the program elements outlined below servemore than one of the major long-term and five-year goals.Again,the program requires a coordinated,integratedapproach by the responsible agencies: the value of anintegrated program will greatly exceed the return fromuncoordinated program elements.

The individual program elements described in thisplan,however, are not at equal stages of readiness. Someprogram elements represent work that has been con-strained by limited resources in the past and could beginimmediately with a near-term infusion of funding. Newtechnology enterprises, for example,have suffered dispro-portionately in the recent past due to such resource con-straints. Since many of the program elements in the CCSPcall for focused technology development prior to large-scale implementation, we recommend that a high prioritybe placed on funding those technology developmentefforts that have been deferred due to insufficient fund-ing. More specifically, the following program elementsshould be considered as high priorities for initial funding:

• Both facility and technology development for theenhanced flux network,airborne sampling,and auto-mated and streamlined ocean sampling for long time-series and underway measurements

• A i r b o rne CO2 m o n i t o ring pro gra m s , both dispers e dwe e k ly measurements and in support of re gional studies

• An expanded and enhanced surface monitoring net-work for atmospheric CO2

• Improved forest inventories (with carbon measure-ments a key focus and an explicit goal) and develop-ment of new techniques for remote sensing of above-ground biomass

• Analysis of World Ocean Circulation Experiment/JointGlobal Ocean Flux Study (WOCE/JGOFS) data for CO2uptake by the oceans

• New ongoing program of air-sea carbon flux and oceaninventory measurements

• Continued ocean process studies and enhanced manip-ulation experiments

• Enhanced development of Earth system modeling toinclude interactive carbon and climate dynamics.

Mapping of Program Elements toCCSP Goals

The following paragraphs map the program elementsshown in Table 5.1 to the goals and objectives of theCarbon Cycle Science Plan.

Goal 1: Program Elements 1-4, 6-9, 11

The program elements (1) expanded terrestrial fluxnetwork,(2) airborne CO2 observation program,(3) globalCO2 monitoring network,and (4) land use inventories,taken together, address the first part of the CCSP Goal 1:to quantify the Northern Hemisphere terrestrial sink andmore generally, to quantify the global terrestrial sink forCO2. Program elements (1),(2),and (3) together providedirect long-term measurements defining sources and sinkson regional scales. The proposed airborne observations(2) are capable of providing integrated measures ofregional net exchange several times per month;the con-ceptual framework for interpreting these observationsneeds to be strengthened and tested through a series ofstrategically planned regional observation experiments(6).The expanded flux network (1) complements theother elements by determining monthly and annually aver-aged regional net uptake or release in typical ecosystems.The flux network will be able to define the systematic dif-ferences between flight days and non-flight days for theairborne profile measurements. Flux data can helpaccount for CO2 net exchange on days when analysis ofthe regional net exchange is not possible from atmo-spheric data,e.g.,during frontal passages or largeweather events.

Improved carbon and land use inventories (4) providea first-order check on the inferences from atmosphericmeasurements. By telling us where the carbon is going(or coming from),program element 4 also contributes sig-nificantly to the second part of Goal 1,to understand theunderlying mechanisms that regulate the NorthernHemisphere terrestrial sink,and more generally, global ter-restrial sinks and sources.The long-term terrestrial obser-vations (7),process studies and manipulations (8) provide

53

Chapter 5: Summary of Program Elementsand Resource Requirements

54

fundamental tools to develop new understanding of ter-restrial sinks and sources.

Estimates of the Northern Hemisphere terrestrial car-bon sink made from atmospheric CO2 observations arevery sensitive to the magnitude of the carbon sink in theNorth Atlantic and North Pacific. Oceanic observations(10) in the North Atlantic and Pacific Oceans thus provideimportant constraints on the Northern Hemisphere terres-trial carbon sink.

Modeling and synthesis (11) provide a large-scale checkon inferred Northern and global fluxes when combinedwith global network (3) and airborne (2) data,and allowtests of our understanding through simulations of past andpresent conditions.

Goal 2: Program Elements 2, 3, 6, 9-11

The elements (9) surveys,time series, remote sensing,(10) a quantitative understanding of air-sea exchangeprocesses,(2) airborne CO2 observation program,and (3)global CO2 monitoring network together address the firstpart of Goal 2: to quantify the oceanic uptake of CO2:global ocean measurements.These elements providedirect long-term measurements defining sources and sinkson the scale of major ocean regions. A critically importanttask is to successfully integrate expanded observations oftime series at key locations,observations of atmosphere-ocean exchange,periodic global ocean surveys, remotesensing of the oceans,large-scale airborne measurementsover the oceans,and atmospheric data from island sta-tions. The conceptual framework for interpreting theseobservations will be developed and tested in element (6),strategically planned regional observation experiments.

Ocean process studies and manipulations (10) tell uswhy the carbon is going (or coming from) major oceanregions and provide the basis for predicting long-termtrends. Program elements (10) and (6) thus contributesignificantly to the second part of Goal 2:to understandthe mechanisms of oceanic uptake of CO2.

Modeling and synthesis (11) provide large-scale checkson inferred oceanic and global fluxes,especially whenexercised to provide global constraints using data fromthe global surface and airborne networks (elements (2)and (3). Models allow tests of our understanding throughsimulations 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 landuse) and (4) (expand global ter restrial carbon and landuse inventories) are specifically designed to address Goal3: to determine the impact of historical and current land

use. The expanded flux network (1),long-term terrestrialobservations (7), and terrestrial process studies andmanipulations (8) will provide integral checks and con-straints on the interpretation of results from (4) and (5).Modeling and synthesis (11) will be the major tools bring-ing together the data and concepts developed by theseprogram elements.

Goal 4: Program Element 11 (integrating elements 1-10)

The program element modeling and synthesis (11) rep-resents the most comprehensive and integrating tool forGoal 4,projecting future atmospheric concentrations ofCO2. This goal is a major scientific undertaking,in whichthe models and analysis must be closely integrated with allother elements of the CCSP. To succeed, responsible agen-cies will need to develop a managerial framework with aunified vision of the program and with greatly enhancedmutual collaboration and strategic planning.

Goal 5: The Entire CCSP (all elements,integrated and coordinated)

Goal 5—developing the scientific basis for evaluatingmanagement decisions relating to CO2 in many criticalways represents the culmination of the entire CarbonCycle Science Plan. The ultimate measure of a successfulcarbon cycle research program will be found in its abilityto provide practical answers to both scientific andsocietal questions.

Chapter 5: Summary of Program Elements and Resource Requirements

55

Implementation Principles

Past experience with large-scale, multidisciplinary glob-al change research programs has provided invaluableinsights into principles for successful program implemen-tation. During its deliberations,the authors of this report(the Carbon and Climate Working Group) and sponsoringagencies of the U.S.Global Change Research Program(USGCRP) identified many aspects of the successfulTropical Ocean–Global Atmosphere (TOGA) Program aspossible models for the Carbon Cycle Science Plan. Manyof the principles and strategies described in this chapterhave their roots in the TOGA Program and are drawn fromthe experiences of Working Group members,sponsoringagency representatives,and community colleagues,includ-ing a 1996 National Research Council review of the TOGAProgram (NRC 1996). Working Group deliberations anddiscussions at the August 1998 Carbon Cycle SciencePlanning Workshop identified the following key principlesfor program implementation.

Shared Scientific Vision

This plan re p resents the critical fi rst step in any success-ful global ch a n ge re s e a rch pro gram—the development of as h a red pro gram vision focused on cl e a r ly defined pro bl e m swith identifi able delive rables of value to both the scientifi cc o m munity and the potential benefi c i a ries in the publ i cand pri vate sectors . D evelopment of this shared visionshould invo l ve broad community participation in an openp ro c e s s , with pro grammatic fl exibility to evo l ve continu a l lyas new scientific insights emerge ,n ew tech n o l o gies ared eve l o p e d , and new info rmation needs are identifi e d .

In this context,the Working Group felt it essential toprovide a clear statement of scientific priorities for afocused Carbon Cycle Science Program,while recognizingthe critical contributions of related national and interna-tional global change research programs. These programsalso address such issues as ecosystem dynamics,landuse/land cover change, climate variability and change,andatmospheric chemistry. The Carbon Cycle ScienceProgram proposed here describes the essential elementsof a program designed specifically to improve the quanti-tative characterization of past and present CO2 sourcesand sinks;to develop models to predict future sourcesand sinks;and to provide a scientific basis for evaluatingpotential carbon sequestration strategies and for measur-ing net emissions from major regions of the world.

Chapter 6: Program Implementation andCritical Partnerships

57

Shared Programmatic Responsibility

The National Research Council’s review of the TOGAProgram (NRC 1996) points to the importance of imple-menting this shared scientific vision through a dynamic,interactive partnership between the scientific communityand responsible federal agencies. In this partnership,responsibility for program direction,implementation,andreview is shared,and all partners commit to the program-matic discipline needed to implement the shared vision ofthe program. As seen in TOGA,commitment to a sharedscientific vision requires that,once a Carbon CycleScience Plan has been adopted,the federal agencies agreethat resources will be secured and allocated in accor-dance with the plan. The scientific community must alsoaccept an appropriate level of responsibility for settingpriorities,sustaining relevant existing commitments,andmaking a compelling case for new resources.

As was the case in TOGA,the programmatic partner-ship proposed here requires a strong mechanism for inter -agency coordination to take inventory of individualagency assets and ongoing programs. Agencies must alsoagree on appropriate agency roles and responsibilities forimplementing elements of the Carbon Cycle ScienceProgram,based on the agencies’individual capabilities andmissions. Resources can,and probably should, remain inindividual agencies,provided that participating agenciesagree to a single plan,a single committee for the pro-gram’s scientific oversight and review, a unified propos-al/program review process,and a single address for scien-tific community and public access to information aboutthe program. When successfully implemented,thisensures that funding decisions reflect scientific merit andprogrammatic relevance and are essentially “independent”of agency boundaries. The program management struc-ture proposed later in this chapter is based on this modelof interagency coordination.

Program Integration

Like the TOGA Program,the Carbon Cycle ScienceProgram necessitates an integrated approach,combiningsustained measurements and data analysis and synthesis;targeted process studies and field research campaigns;organized modeling and prediction ef forts;and informa-tion management. In addition to requiring good integra-tion among these programmatic activities,a successfulCarbon Cycle Science Program will require a

multidisciplinary approach. It is critical to integrate stud-ies of atmospheric,oceanic,and terrestrial components ofthe carbon cycle,together with research on the humandimensions of carbon cycle changes.

Scientific Guidance and Review

E x p e rience with nu m e rous global ch a n ge re s e a rch pro-grams has demonstrated the importance of establ i s h i n gclear pro c e d u res for scientific guidance and rev i ew of theCarbon Cycle Science Pro gram from the outset. In the caseof TO G A , an NRC panel (under the auspices of the ClimateR e s e a rch Committee of the Board on A t m o s p h e ric Sciencesand Climate) provided a source of ongoing scientific guid-ance and rev i ew throughout the pro gra m ’s life t i m e . T h eWo rking Group recommends separating the scientific guid-ance and rev i ew processes in a stru c t u re that draws fro mboth the TOGA ex p e rience and the World Ocean Cir-culation Experiment (WOCE) and the Joint Global OceanFlux Study (JGOFS) ex p e ri e n c e s . We recommend that anindependent Scientific Steering Committee be re s p o n s i bl efor scientific guidance. (The fo rmation and composition ofthis committee is addressed in the section on pro p o s e dm a n agement stru c t u re below.) In addition, the Wo rk i n gG roup recommends a periodic (e.g., eve ry three to fi veye a rs) ex t e rnal rev i ew of the Carbon Cycle ScienceP ro gra m . The Wo rking Group suggests that the NRCCommittee on Global Change Research is an appro p ri a t eb o dy to ove rsee the conduct of these periodic rev i ews ofissues related to scientific quality, re l eva n c e ,a c c o m p l i s h-m e n t s ,f u t u re plans, and pro gram implementation.

Links to International Programs

As the TOGA Pro gram demonstra t e d ,e s t ablishing stro n gties to the related scientific effo rts of other countries andfo rmal international re s e a rch pro grams (the World ClimateR e s e a rch Pro gram in the case of TOGA) can yield signifi-cant benefi t s ; it re p resents another key principle for suc-cessful implementation of the Carbon Cycle ScienceP ro gra m .The case for international collab o ration is eve ns t ro n ger for carbon cycle re s e a rch in light of the ongo i n ge ffo rts of the Intergove rnmental Panel on Climate Changeand current national and international deliberations re l a t e dto the United Nations Fra m ewo rk Convention on ClimateC h a n ge . E s t ablishing and sustaining effe c t i ve links withrelated international pro grams is described in further detailb e l ow under “ C ritical Pa rt n e rs h i p s .”

Access to Data and Communication ofResearch Results

A ch i eving the goals described in this plan re q u i res a cul-t u re of open ex ch a n ge of observations and associated datap ro d u c t s , re s e a rch fi n d i n g s , and model re s u l t s . In part ,t h i s

re q u i rement calls for a new level of collab o ration amongscientists engaged in modeling, m e a s u rement pro gra m s ,and field re s e a rch . In this contex t , the Wo rking Gro u pnotes that early and continuous investment in data-modeli n t e gration and synthesis is essential to providing a fullu n d e rstanding of past, p re s e n t , and future carbon cycl eb e h av i o r. A commitment to easy and open access alsoimplies greater responsibilities for info rmation manage-ment stra t e gy to ensure timely provision of data, re s e a rchfi n d i n g s , and model re s u l t s ; to assimilate/integrate observa-tions and data from diffe rent platfo rm s ,i n s t ru m e n t s ,a n dfield campaigns; to adhere to appro p riate standard s ,p ro t o-c o l s , and guidelines; and to provide metadata critical to thea n a lysis from nu m e rous individual scientists and part i c i p a t-ing institutions. The continuation of support for anyre s e a rch group under this plan will depend on the timelyand complete ava i l ability of data and models ge n e ra t e d .

Experience with other multidisciplinar y, global changeresearch programs like WOCE and TOGA highlights thetechnical challenges and opportunities associated withopen access and exchange within the scientific communi -ty. The Carbon Cycle Science Program described in thisplan,however, places an additional burden of responsibili -ty on the scientific community and sponsoring agencies:effective communication of research results to intendedusers such as businesses,communities,land managementagencies,and government officials. The U.S.GlobalChange Research Program uses the term “assessment”when describing such an organized effort to convey scien-tific results to potential beneficiaries in useful forms andto establish a continuing,interactive dialogue with thoseusers. Current national and international discussions relat-ing to the Framework Convention on Climate Changehighlight the importance of sustaining a dialogue that pro-vides government and private sector decision makers withreliable,quantitative information on the sources and sinksof carbon dioxide,both present and future. One elementof this dialogue will involve scientific support for formalassessment programs such as the work of theIntergovernmental Panel on Climate Change. In addition,participants in the Carbon Cycle Science Program should,from the beginning,plan an organized program of commu-nication,outreach,and education that supports the under-standing and dissemination of new scientific insights andresearch results to interested parties in and beyond thescientific community.

Proposed Program Management Structure

With these principles in mind,the Working Group rec-ommends establishment of a collaborative managementstructure for the Carbon Cycle Science Program as depict-ed in Figure 6.1.

This tri p a rtite management stru c t u re re flects the share dresponsibilities of the scientific community and the spon-


58

s o ring fe d e ral age n c i e s , and provides a mechanism to coor-dinate their contri b u t i o n s . While each component of them a n agement stru c t u re has pri m a ry responsibility for diffe r-ent elements of the pro gra m , success re q u i res that thesem a n agement components wo rk in unison, i n t e racting on ad ay - t o - d ay basis, ex ch a n ging info rmation fre e ly and openly,and jointly resolving critical issues and pro bl e m s . The pro-gram will stand only if all three “ l e g s ”a re balanced, s t ro n gand positioned to support their special re s p o n s i b i l i t i e s .

Responsibility for continuous scientific guidance wouldbe provided by a Scientific Steering Committee (SSC)comprised of an expert panel actively engaged in variousaspects of carbon cycle science. Individual members ofthe SSC would be selected by the sponsoring federal agen-cies in consultation with the scientific community andother interested parties. The NRC,through the Committeeon Global Change Research,could be a source of nomina-tions and/or review of a membership slate for the pro-posed SSC. Members of the steering committee shouldhave a vested interest in the success of the program,andtherefore,committee membership should likely includefunded participants in the Carbon Cycle Science Program.Membership should reflect both individualdisciplinary/programmatic expertise and the special chal-lenges posed by the integrated nature of the programdescribed in this plan. In addition,consideration shouldbe given to including individuals who represent potentialbeneficiaries of the program in government,the privatesector, and public interest groups.

Working with their colleagues in the sponsoring federalagencies,the SSC would have primary responsibility forensuring that the detailed implementation of the CarbonCycle Science Program follows directly from the scientific

goals and programmatic objectives described in the plan.The SSC would achieve this by continually evaluatingprogress; by reviewing priorities and revising plans asneeded to reflect new scientific insights,technology, andinformation needs; by evaluating proposals for majoradjustments to measurement, field research,and modelingprojects,ensuring a strong scientific review process forindividual projects;and by periodically submitting theCarbon and Climate Science Program to rigorous reviewby an outside body. The Working Group recommends thatthe sponsoring federal agencies look to the NRC’sCommittee on Global Change Research to organize peri-odic (e.g., every three to five years) outside reviews of theCarbon Cycle Science Program.

An Interagency Wo rking Group (IWG) comprised ofagencies contributing re s o u rces to support the CarbonC y cle Science Pro gram would have responsibility for inter-agency coordination and pro gram manage m e n t .The IWGwould have pri m a ry responsibility for identifying and sus-taining re l evant existing commitments and securing appro-p riate re s o u rces for new activities in support of the pro-gra m . As for the TOGA Pro gra m , the IWG for the CarbonC y cle Science Pro gram should be comprised of pro gra mm a n age rs with the responsibility and authority to commitre s o u rces in support of the pro gra m . The IWG shouldh ave access to the scientific ex p e rtise residing in individ-ual age n c i e s , and IWG members should have the re s p o n s i-bility of ensuring that their agency assets—fi s c a l ,h u m a n ,and capital—are brought to bear appro p ri a t e ly in supportof the pro gra m . In addition to their responsibilities fo rc o m munication within and among the participating scien-t i fic age n c i e s , IWG members will also be re s p o n s i ble fo re n s u ring that pro gram results are incorporated appro p ri-a t e ly into fe d e ral policy fo rmulation and mission age n c y

Chapter 6: Program Implementation and Critical Partnerships

59

Figure 6.1 Carbon Cycle Science Program management structure.

decision making. S i m i l a r ly, as public serva n t s ,m e m b e rs ofthe IWG have a special responsibility for commu n i c a t i n gthe results of the Carbon Cycle Science Pro gram to inter-ested parties in the public and pri vate sector.

One of the IWG’s most important initial tasks will beidentifying existing agency assets (human,programmatic,and fiscal) and agreeing on individual agency roles andresponsibilities in implementing the shared scientificvision described in this plan.Agency responsibilities forthe Carbon Cycle Science Program should properly reflectagency missions, expertise,and experience. By way ofexample,discussions during the August 1998 Workshop onCarbon Cycle Science highlighted the following potentialroles for individual agencies.The National ScienceFoundation would continue its tradition of support forcritical long-term field studies and laboratory experimentsto illuminate key processes for carbon cycle and Earth sys-tem modeling,and seek funding for exploratory researchby individual investigators representing a variety of view-points and approaches. Operational mission agenciessuch as the National Oceanic and AtmosphericAdministration, U.S.Geological Survey, and U.S.Department of Energy might play critical roles in support-ing sustained measurements and integrated,predictivemodeling projects. Resource management agencies suchas the Department of Interior, Department of Energy, andthe Department of Agriculture might assume primaryresponsibility for examining the roles of ter restrial reser-voirs for carbon dioxide and the impacts of human activi-ties on terrestrial ecosystems. The National Aeronauticsand Space Administration might provide the lead in appli-cations of space-based, remote-sensing techniques andtechnologies for carbon cycle science. None of theseresponsibilities would be exclusive,but some identifica-tion of appropriate,individual agency roles will be essen-tial to successful implementation of an integrated inter-agency program.

Once the sponsoring federal agencies and the scientificcommunity have adopted a shared scientific vision for theprogram and reached agreement on individual agencies’roles,the IWG would be responsible for overseeing thejoint implementation of the Carbon Cycle ScienceProgram on behalf of the participating federal agencies.We would like to emphasize that joint implementationdoes not necessarily require pooling of resources or thedesignation of a single “lead agency.” However, it doesrequire a joint commitment and adherence to the sharedscientific vision (as described in the Carbon Cycle SciencePlan) and to a primary source of scientific advice (the SSCdescribed above);to the implementation of a consistentand unified approach to proposal and program reviewprocesses;and to the provision of a clear point of contactfor and source of information on the Carbon CycleScience Program. In addition,the IWG would likely con-stitute the primary liaison between the carbon cycle

science community and government policy officials andresource management agencies in the federal government.

Managing the day-to-day implementation of this part-nership between the scientific community and the spon-soring federal agencies would be the responsibility of aCarbon Cycle Science Program Office. Responsibility forthis component of the program management structurecould,as in the case of TOGA, reside within a federalagency or, as in the WOCE program,be assigned through acompetitive process to a qualified scientific institutionoutside the government. As envisioned by this WorkingGroup,the function of the Carbon Cycle Science ProgramOffice should not require establishing a large bureaucracyor the expenditure of a significant level of resources.Some investment of resources (human and fiscal) will berequired,however, to ensure that the Program Office caneffectively provide the programmatic and institutional“glue”to sustain the critical interaction between the scien-tific community and federal agencies responsible forimplementation of the program. The Carbon CycleScience Program Office staff should have appropriate lev-els of scientific and programmatic expertise (e.g.,a Ph.D.or equivalent experience).

S p e c i fic responsibilities of the Pro gram Office wo u l di n clude providing a pro gramwide point of contact ands o u rce of info rmation on pro gram dire c t i o n ,a c t i v i t i e s ,s t a-t u s , and accomplishments; p roviding pro gramwide liaisonwith the National Research Council’s Committee on GlobalC h a n ge Research and other re l evant scientific org a n i z a-tions and re s e a rch pro gra m s ;s e rving as pri m a ry liaison fo rthe U. S .p ro gram with re l evant international scientific pro-gra m s ;s u p p o rting the day - t o - d ay aspects of pro gram imple-m e n t a t i o n , assisting in pro gram deve l o p m e n t , re s o u rc em a n age m e n t , and pro gram evaluation effo rt s ; and prov i d-ing secre t a riat services for both the SSC and the IWG.

Critical Partnerships

Successful implementation of the Carbon Cycle ScienceProgram described in this plan will require the creationand maintenance of a number of critical partnerships.First and foremost,the program should represent adynamic and innovative partnership between the scientif-ic community and the sponsoring agencies in the federalgovernment. The proposed management structuredescribed above reflects this partnership and a sharedresponsibility for ensuring scientific quality and program-matic relevance,setting priorities,adhering to plans,secur-ing required resources,and interpreting and disseminatingthe scientific and information products that emerge fromthe Carbon Cycle Science Program.

The Working Group believes that implementation ofthe program described in this plan would benefit signifi-cantly from efforts to build sustained partnerships


60

between federal laboratories and the extramural researchcommunity, leveraging the special capabilities and exper t-ise of those partners. For example,partnership in thedesign,development,testing,and deployment of measure-ment systems could help address a number of issues fac-ing the scientific community today. This partnershipcould ensure an appropriate mix of researchers aware ofthe scientific requirements,and engineers and techniciansfamiliar with system capabilities;and facilitate theexchange of ideas and experience among groups workingon the development of systems to meet similar needs. Inshort,it could create an environment of creative synergyrather than simply duplicate ef fort. The Carbon CycleScience Program envisioned in this plan provides a frame-work for creative new partnerships among various disci-plines and programmatic areas of expertise—such as indi-viduals involved in measurements,modeling,and processresearch—within the scientific community. Similarly, inno-vative partnerships among supporting federal agenciesshould bring individual agency assets and expertise tobear on a shared program in complementary ways thatbuild on existing capabilities,leverage limited resources,avoid duplication,and produce new opportunities forscientific progress.

The global nature of the carbon cycle and the intern a-tional context of current policy deliberations related toC O2 and other greenhouse gases highlight the critical needfor international collab o ration in carbon cycle inve s t i g a-t i o n s . As has been the case in the past, a strong U. S .p ro-gram can serve as a catalyst for similar pro grams in otherc o u n t ries and provide a focal point for international coor-d i n a t i o n . S u ch multinational collab o ration offe rs opport u-nities to leve rage re s o u rces and take adva n t age of compara-ble methodologies and joint pro j e c t s . E a r ly collab o ra t i o nwith Canada, for ex a m p l e , could produce significant bene-fits for proposed investigations of the North A m e rican ter-re s t rial sink. S i m i l a r ly, i nvestigations of ocean sources andsinks in the Pa c i fic will benefit from international collab o-ration in mu ch the same way that TO G A’s investigations ofEl Niño benefited from the contributions of part n e rsa round the Pa c i fic Basin and throughout the wo r l d .

As has also been seen in programs like TOGA,WOCE,and JGOFS,one particularly beneficial approach to inter-national partnerships involves the development andimplementation of strong U.S.contributions to establishedinternational global change research programs. TheCarbon Cycle Science Program should establish strong tiesto the World Climate Research Program (particularly theCLIVAR Program),the Global Energy Water CycleExperiment (GEWEX),the International Geosphere–Biosphere Programme (IGBP, particularly JGOFS,GAIM,and IGAC),and the International Human Dimensions ofGlobal Change Program (particularly the Land Use/LandCover Change project being implemented jointly with

IGBP).In addition,the sustained measurement compo-nents of the program described in this plan could providesubstantial contributions to emerging global observingprograms such as the Global Climate Observing System(GCOS),the Global Ocean Observing System (GOOS),andthe Global Terrestrial Observing System (GTOS).Conversely, these emerging multinational endeavors couldleverage resources (such as space- and ground-basedobservational platforms);develop and demonstrate newtechnologies;and provide a global,Earth system contextfor carbon cycle measurements emphasizing large-scaleecosystem dynamics and carbon cycle interactions withvariability and change in the climate system.

Fi n a l ly, the Wo rking Group believes that the CarbonC y cle Science Pro gram described here would benefit fro man early and sustained part n e rship with the pri vate sector.Some of the pro gra m ’s go a l s — n o t ably, p roviding a scientif-ic basis for evaluating potential carbon sequestration stra t e-gies and measuring net emissions at a re gional scale—should be of interest and value to businesses invo l ved inthe energy sector, re s o u rce use and management (e.g.,fo re s t ry ) , and agri c u l t u re , for ex a m p l e . P ri vate sectorex p e rtise and assets in tech n o l o gy R&D could accelera t ethe development and demonstration of critical new meas-u rement tech n o l o gi e s . Scientists interested in the design,d eve l o p m e n t , testing and deployment of measurement sys-tems face a number of obstacles today, and this Wo rk i n gG roup believes that innova t i ve part n e rships among fe d e ra ll ab o ra t o ri e s ,u n i ve rs i t i e s , and the pri vate sector (in the U. S .and ab road) would be mu t u a l ly benefi c i a l . I nvestments inthe development of new systems should be supplementedwith re s o u rces to make use of commerc i a l ly ava i l abl em e a s u rement systems in the fi e l d . Decisions to supportthe development of new systems should re p resent a com-mitment of the time and re s o u rces re q u i red to take newi n s t rumentation from concept to re a l i t y. E a r ly invo l ve m e n tof the pri vate sector in this process could help ensure asmooth transition from limited ex p e rimental prototype toc o m m e rc i a l ly ava i l able (and affo rd able) tech n o l o gy to sup-p o rt broad community needs.

Joint implementation of carbon cycle measurementprograms and collaborative development of integratedmodeling and assessment capabilities could result in earlyprogress,with direct benefits to the scientific community,public policy officials,and private sector interests directlyaffected by carbon cycle and climate policies. TheWorking Group was unable to engage in detailed discus-sion of the character of such a public-private partnershipin carbon cycle research during this initial planningphase. But we strongly encourage a thorough explorationof the challenges and opportunities of partnership withthe private sector as an early implementation task.

Chapter 6: Program Implementation and Critical Partnerships

61

Archer, D.,H.Kheshgi and E.Maier-Reimer, MultipleTimescales for the Neutralization of Fossil Fuel CO2,Geophys.Res.Lett., 24, 405-408,1997.

Archer, S., D. S.Schimel and E.A.Holland,Mechanisms ofshrubland expansion:Land use, climate or CO2,Climatic Change, 29, 91-99,1995.

Aumont, O, J. C.Orr, D. Jamous, P. Monfray, O. Marti,and G.Madec,A degradation approach to accelerate simula-tions to steady-state in a 3-D tracer transport model ofthe global ocean, Climate Dynamics, 14, 101-116,1998.

Baar, H. J.W. d., J.T. M.d. Jong, D. C.E.Bakker, B.M.Löscher,C.Veth, U. Bathmann and V. Smetacek,Importance ofiron for plankton blooms and carbon dioxide draw-down in the Southern Ocean, Nature, 373, 412-415,1995.

Bacastow. R.and C.D. Keeling,Atmospheric carbon diox-ide and radiocarbon in the natural carbon cycle. II,Changes from A.D. 1700 to 2070 as deduced from ageochemical reservoir. In: G.M.Woodwell and E.V.Pecan (eds.) Carbon and the Biosphere. U.S.Dept.ofCommerce (Springfield,Virginia),86-135,1973.

Bakwin, P. S., P. P.Tans,C.Zhao, W. Ussler and E.Quesnell,Measurements of carbon dioxide on a very tall tower,Tellus, 47B, 535-549,1995.

Bakwin, P.S., P.P.Tans, D. F. Hurst,and C.Zhao,Measurements of carbon dioxide on very tall towers:results of the NOAA/CMDL program. Tellus, 50B, 401-415,1998.

Baldocchi, D.,Valentini,R.,Running,S.,Oechel,W. and R.Dahlman, Strategies for measuring and modeling car-bon dioxide and water vapor fluxes over terrestrialecosystems, Global Change Biology 2, 159-169,1996.

Barnola, J. M., D. Raynaud,Y. S. Korotkevich and C.Lorius,Vostok ice core provides 160,000-year record of atmo-spheric CO2, Nature, 329, 408-414,1987.

Bates, N. R.,A. F. Michaels and A.H.Knap,Seasonal andinterannual variability of the oceanic carbon dioxidesystem at the U. S.JGOFS Bermuda Atlantic time-seriesstudy site, Deep-Sea Res., Part II, 1996.

Battle,M.,M.Bender,T. Sowers, P. P.Tans, J. H.Butler, J.W.Elkins, J.T. Conway, N. Zhang, P. Lang and A.D. Clarke,Atmospheric gas concentrations over the past centur ymeasured in air from firn at the South Pole, Nature,383, 2312-235,1996.

Bibliography

63

Behrenfeld,M. J. and P. G. Falkowski,A consumers guide tophytoplankton primary production models, Limnol.Oceanogr., 42, 1479-1491,1997.

Bender, M.L.,T. Sowers, J. N. Barnola and J. Chappellaz,Changes in the O2/N2 ratio of the atmosphere duringrecent decades reflected in the composition of air inthe firn at Vostok Station,Antarctica, Geophys.Res.Lett.,21, 189-192,1994.

Berner, R.A.,Geocarb II: A revised model of atmosphericCO2 over Phanerozoic time, Am. J. Sci., 294,56-91,1994.

Billings,W. D., J. O. Luken, D.A.Mortensen and K.M.Peterson,Arctic tundra:a source or sink for atmospher-ic carbon-dioxide in a changing environment,Oecologia, 53, 7-11,1982.

Blunier,T., J. Chappellaz, J. Schwander,A.Dallenback,B.Stauffer,T. F. Stocker, D. Raymond, J. Jouzel,H.B.Clausen,C. U. Hammer and S. J. Johnson,Asynchrony ofAntarctica and Greenland climate during the last gla-cial, Nature, 1-18,1998.

Bolin,B.Changes of land biota and their importance forthe carbon cycle, Science, 196, 613-615,1977.

Brewer, P. G.,C.Goyet and D. Dyrssen,Carbon dioxidetransport by ocean currents at 25-N latitude in theAtlantic Ocean, Science, 246, 477-479,1989.

Broecker,W. S.and T.-H. Peng,Interhemispheric transportof carbon dioxide by ocean circulation, Nature, 356,587-589,1992.

Brook,E. J.,T. Blunier and e.al.,Rapid variations in atmos-pheric methane concentration during the past 110,000years, Science, 273, 1087-1091,1993.

Cao,M.and F. I.Woodward,Dynamic response of terrestri-al ecosystem carbon cycling to global climate change,Nature, 393, 249-252,1998.

Chadwick, O.A.,L.A.Derry, P. M.Vitousek,B. J. Huebert andL. O. Hedin,Changing sources of nutrients during fourmillion years of ecosystem development, Nature, 397,491-497,1999.

Chappellaz, J.,T. Blunder, D. Raynaud, J. M.Barnola, J.Schwander and B.Stauffer, Synchronous changes inatmospheric CH4 and Greenland climate between 40and kyr BP, Nature, 366, 443-455,1993.

Chou,W.W.,Estimation of Net Regional Surface Fluxes ofCO2 and O3 Over Amazonia from Aircraft Data,1999.

Ciais, P., P. P.Tans,M.Trolier, J.W. C.White and R. J. Francey,Alarge Northern Hemisphere terrestrial CO2 sink indi-cated by the 13C/12C ratio of atmospheric, Science,269, 1098-1102,1995a.

Ciais, P., P. P.Tans, J.W. C.White,M.Trolier, R. J. Francey, J.A.Berry, D. R.Randall, P. J. Sellers, J. G. Collatz and D. S.Schimel, Partitioning of ocean and land uptake of CO2as inferred by _13C measurements from the NOAAClimate Monitoring and Diagnostics Laboratory GlobalAir Sampling, J. Geophys.Res., 100, 5051-5070,1995b.

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

Dai,A.and I.Y. Fung,Can climate variability contribute tothe “missing”CO2 sink?, Global Biogeochem.Cycles, 7,599-609,1993.

Dansgaard,W.,S. J. Johnsen, J. B.Clausen, D. Dahl-Jensen, N.S.Gundestrup,C. U. Hammer, C.S.Hvidberg, J. P.Steffensen,A.E.Sveinbjornsfottir, J. Jouzel and G. Bond,Evidence for general instability of past climate from a250-kyr ice-core record, Nature, 364, 218-220,1993.

DeGrandpre,M. D.,T. R.Hammar, S. P. Smith and F. L.Sayles,In situ measurements of seawater pCO2, Limnol.Oceanogr., 40, 969-975,1995.

Denman,K.,E.Hofmann and H.Marchant,Marine bioticresponses to environmental change and feedbacks toclimate,in Climate Change 1995, vol.edited by J.T.Houghton,L. G. M. Filho,B.A.Callander, N. Harris,A.Kattenberg and K.Maskell,483-516,CambridgeUniversity Press,New York,1996.

Denning,A.S.,M.Holzer, K.R.Gurney, M.Heimann,R.M.Law, P. J. Rayner, I.Y. Fung,S.-M. Fan,S.Taguchi, P.Friedlingstein,Y. Balkanski,M.Maiss and I.Levin,Three-dimensional transport and concentration of SF6:Amodel intercomparison study (TransCom 2), Tellus, inpress, 1999.

Desjardins,R.L., J. L.MacPherson,L.Mahrt, P. Schuepp,E.Patty, H.Neumann, D. Baldocchi,S.Wofsy, D. Fitzjarrald,H.McCaughey, and D.W. Joiner, Scaling up flux meas-urements for the boreal forest using aircraft-tower com-binations. J. Geophys.Res.102, 29125-29134,1997.

Dixon,R.K.,S.Brown,R.A.Houghton,A.M.Solomon,M.C.Trexler and J.Wisniewski,Carbon pools and flux ofglobal forest ecosystems, Science, 263, 185-190,1994.

Doney, S.C., D. M.Glover and R. G. Najjar,A new coupled,one-dimensional biological-physical model for theupper ocean:applications to the JGOFS BermudaAtlantic Time Series (BATS) Site, Deep Sea Res., 43, 591-624,1996.

Esbensen,S.K.,and Y. Kushnir,The heat budget of theglobal ocean:An atlas based on estimates from surfacemarine observations, Climatic Research InstituteReport 29, Oregon State University, Corvallis,OR.

Etheridge, D. M.,L. P. Steele,R.L.Langenfelds and R. J.Francey, Natural and anthropogenic changes in atmo-spheric CO2 over the last 1000 years from air inAntarctic ice and firn, J. Geophys.Res., 101, 4115-4128,1996.

Falkowski, P. G.,Evolution of the nitrogen cycle and itsinfluence on the biological sequestration of CO2 in theocean, Nature, 387, 272-274,1997.

Falkowski, P. G.,Ironing out what controls primary produc-tion in the nutrient rich waters of the open ocean,Global Change Biology, 1, 161-163,1995.

Falkowski, P. G.,R.T. Barber and V. Smetacek,Biogeochemical controls and feedbacks on ocean pri-mary production, Science, 281, 200-206,1998.

Fan,S.-M.,M.Gloor, J. Mahlman,S. Pacala, J. L.Sarmiento,T.Takahashi and P.Tans,A large terrestrial carbon sink inNorth America implied by atmospheric and oceanicCO2 data and models, Science, 282, 442-446,1998.

FAO,World Agriculture:Towards 2010 - An FAO Study, invol.edited by 14-23, John Wiley and Sons, West Sussex,1995.

Fasham,M. J. R.,Variations in the seasonal cycle of biologi-cal production in subarctic oceans: A model sensitivityanalysis, Deep Sea Res., 42, 1111-1150,1995.

Feely, R.A.,R.Wanninkhof, C.E.Cosca, P. M.Murphy, M. F.Lamb and M. D. Steckley, CO2 distributions in the equa-torial Pacific during the 1991-1992 ENSO event, DeepSea Res., 42, 365-386,1995.

Feely, R.A.,R.Wanninkhof, C.E.Coscas,M. J. McPhaden,R.H.Byrne, F. J. Millero, F. P. Chavez,T. Clayton, D. M.Campbell and P. P. Murphy,The effect of tropical insta-bility waves on CO2 species distributions along theequator in the eastern equatorial Pacific during the1992 ENSO event, Geophys.Res.Lett., 21, 277-280,1994.

Feely, R.A.,R.Wanninkhof, C.Goyet, D. E.Archer and T.Takahashi,Variability of CO2 Distributions and Sea-AirFluxes in the Central and Eastern Equatorial PacificDuring 1991-94 El Niño, Deep Sea Res., 33, 1996.


64

Foster, D. R.,Land-use history and transformations of theforest landscape of central New England,in Humans asComponents of Ecosystems: Subtle Human Effectsand the Ecology of Populated Areas, vol.edited by S.T.A.Pickett and M.McDonnell,91-110,Springer-Verlag,New York,1993.

Foster, D. R.,Land-use history and four hundred years ofvegetation change in New England,in Principles,Patterns and Processes of Land Use Change: SomeLegacies of the Columbian Encounter, vol.edited by B.L.Turner, 253-318, John Wiley and Sons, New York,1995.

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

Frew, N. M.,The role of organic films in air-sea gasexchange,in The Sea Surface and Global Change, vol.edited by P. S.Liss and R.A.Duce,121-171,CambridgeUniversity Press,Cambridge,1997.

Friederich, G. E., P. G. Brewer, R.Herlien and F. P. Chavez,Measurements of sea surface partial pressure of CO2from a moored buoy, Deep Sea Res., 42, 1175-1186,1995.

Friedlingstein, P.,I.Fung,E.Holland, J. John, G. Brassuer, D.Erickson and D. Schimel,On the contribution of CO2fertilization to the missing biospheric sink, GlobalBiogeochemical Cycles, 9, 541-556,1995.

Fung,I.,C.B. Field, J.A.Berry, M.V.Thompson, J.T.Randerson,C.M.Malmstrom, P.M.Vitousek, G.J. Collatz,P.J. Sellers, D.A.Randall,A.S.Denning, F. Badeck and J.John,Carbon-13 exchanges between the atmosphereand the terrestrial biosphere, Global Biogeochem.Cycles, 11, 507-533,1997.

Gent, P. R., J.Willebrand,T. J. McDougall and J. C.McWilliams,Parameterizing eddy-induced tracer trans-ports in ocean circulation models, J. Phys.Oceanogr.,25, 463-474,1995.

Goulden,M.L., J.W. Munger, S.M. Fan,B. C.Daube and S.C.Wofsy, Exchange of carbon dioxide by a deciduous for-est: response to interannual climate variability, Science,271, 1576-1578,1996.

Goulden,M.L.,S.C.Wofsy, J.W. Harden,S.E.Trumbore, P. M.Crill,S.T. Gower,T. Fries, B.C.Daube,S.-M. Fan, D. J.Sutton,A.Bazzaz and J.W. Munger, Sensitivity of borealforest carbon balance to soil thaw, Science, 279, 214-217,1998.

Goyet,C., D. R.Walt and P. G. Brewer, Development of afiber optic sensor for measurement of pCO2 in seawater: design criteria and sea trials, Deep Sea Res., 39,1015-1026,1992.

Gruber, N.,Anthropogenic CO2 in the Atlantic Ocean,Global Biogeochem.Cycles, 12, 165-192,1998.

Gruber, N., J. L.Sarmiento and T. F. Stocker,An improvedmethod for detecting anthropogenic CO2 in theoceans, Global Biogeochem.Cycles, 10, 809-837,1996.

Hansell, D.A.and C.A.Carlson,Net community productionof dissolved organic carbon. Global Biogeochem.Cycles, 12, 443-453,1998.

Harden, J.W.,E.T. Sundquist,R. F. Stallard and R.K.Mark,Dynamics of soil carbon during deglaciation of theLaurentide Ice Sheet, Science, 258, 1921-1924,1992.

Harrison, D. E.and D. S.Luther, Surface winds from tropicalPacific Islands - Climatological statistics, Journal ofClimate, 3, 251-271,1990.

Haywood, J. M.,R. J. Stouffer, R.T.Wetherald,S.Manabe andV. Ramaswamy,Transient response of a coupled modelto estimate changes in greehouse gas and sulfate con-centrations, Geophys.Res.Lett., 24, 1335-1338,1997.

Holfort, J.,K.M. Johnson,B.Schneider, G. Siedler and D.W.R.Wallace,The meridional transport of dissolved inor-ganic carbon in the South Atlantic Ocean, GlobalBiogeochem.Cycles, 12, 479-499,1998.

Holland,E.A.,B.H.Braswell, J.-F. Lamarque,A.Townsend, J.Sulzman, J.-F. Muller, F. Dentener, G. Brasseru,H.L.II, J. E.Penner and G.-J. Roelofs,Variations in the predicted spa-tial distribution of atmospheric nitrogen deposition andtheir impact on carbon uptake by terrestrial ecosys-tems, J. Geophys.Res., 102, 15,849-15,866,1997.

Houghton, J.T.,L. G. M. Filho,B.A.Callander, N. Harris,A.Kattenberg and K.Maskell, Climate Change 1995, vol.pp.,Cambridge University Press,Cambridge,1996.

Houghton,R.A.,Is Carbon accumulating in the northerntemperate zone?, Global Biogeochem.Cycles, 7, 611-617,1993.

H o u g h t o n ,R .A . ,R . D. B o o n e , J. R . Fru c i , J. E .H o bb i e , J. M .M e l i l l o ,C .A . Pa l m ,B . J. Pe t e rs o n , G. R .S h ave r, G. M .Wo o dwe l l ,B .M o o re , D. L .S kole and N. M ye rs ,The flux ofcarbon from terre s t rial ecosystems to the atmosphere in1980 due to ch a n ges in land use: ge o graphic distri b u-tion of the global fl u x , Te l l u s , 3 9 B , 1 2 2 - 1 3 9 ,1 9 8 7 .

Houghton,R.A.,E.A.Davidson and G. M.Woodwell,Missing sinks, feedbacks,and understanding the role ofterrestrial ecosystems in the global carbon balance,Global Biogeochem.Cycles, 12, 25-34,1998.

Houghton,R.A., J. E.Hobbie, J. M.Melillo,B.Moore,B. J.Peterson, G. R.Shaver and G. M.Woodwell,Changes inthe carbon content of terrestrial biota and soilsbetween 1860 and 1980: A net release of CO2 to theatmosphere, Ecological Monographs, 53(3), 235-262,1983.

Bibliography

65

Hurst, D.F.,P. S.Bakwin,R.C.Meyers,and J.W. Elkins,Behavior of trace gas mixing ratios on a very tall towerin North Carolina, J. Geophys.Res., 102, 8825-8835,1997.

IGBP Terrestrial Carbon Working Group,The terrestrial car-bon cycle:implications for the Kyoto Protocol, Science,280, 1393-1394,1998.

Indermühle,A., T. F. Stocker, F. Joos,H. Fischer, H. J. Smith,M.Wahlen,B.Deck, D. Mastroianni,T.Tschumi,T. Blunier, R.Meyer and B.Stauffer, Holocene carbon-cycle dynamicsbased on CO2 trapped in ice at Taylor Dome,Antarctica,Nature, 398, 121-126,1999.

Jähne,B.,K. O. Münnich,R.Bösinger,A.Dutzi,W. Huber andP. Libner, On parameters influencing air-water gasexchange, J. Geophys.Res., 92, 1937-1949,1987.

Jouzel, J., N. I.Barkov, J. M.Barnola,M.Bender, J. Chappellaz,G. Genthon,V. M. Kotlyakov,V. Lipenkov, C.Lorius, J. R.Petit, D. Raynaud, G. Raisbeck,C.Ritz,T. Sowers,M.Stievenard, F.Yiou and P.Yiou,Extending the Vostok ice-core record of paleoclimate to the penultimate glacialperiod, Nature, 364, 407-412,1993.

Kamykowski, D. and S.-J. Zentara,Predicting plant nutrientconcentrations from temperature and Sigma-T in theupper kilometer of the world ocean, Deep Sea Res., 33,89-105,1986.

Karl, D.,R.Letelier, L.Tupas, J. Dore, J. Christian and D.Hebel,The role of nitrogen fixation in biogeochemicalcycling in the subtropical North Pacific Ocean, Nature,388, 533-588,1997.

Karl, D.M.,A sea of change:Biogeochemical variability inthe North Pacific subtropical gyre, Ecosystems 2:181-214,1999.

Kasischke,E.S., N. L.Christensen,and B.J. Stock, Fire,Global Warming,and the carbon balance of boreal for-est. Ecol.Appl. 5, 437-451,1995.

Keeling,C. D.,R.B.Bacastow,A. F. Carter, S.C.Piper,T. P.Whorf, M.Heimann,W. G. Mook and H.Roeloffzen,Athree-dimensional model of atmospheric CO2 transportbased on observed winds: 1 Analysis of observationaldata,in Aspects of Climate Variability in the Pacificand the Western Americas AGU Monograph 55, vol.edited by D. H. Peterson,165-236,AGU,WashingtonD.C.,1989.

Keeling,C. D.,T. P.Whorf, M.Wahlen and J.V. d.Plicht,Interannual extremes in the rate of rise of atmosphericcarbon dioxide since 1980, Nature, 375, 666-670,1995.

Keeling,R. F. and T.-H. Peng,Transport of heat,CO2 and O2by the Atlantic’s thermohaline circulation, Philo-sophical Transactions of the Royal Society, 348,133-142,1995.

Keeling,R. F.,S.C.Piper and M.Heimann,Global and hemi-spheric CO2 sinks deduced from changes in atmospher-ic O2 concentration, Nature, 381, 218-221,1996.

Keeling,R. F. and S.R.Shertz,Seasonal and interannual vari-ations in atmospheric oxygen and implications for theglobal carbon cycle, Nature, 358, 723-727,1992.

Lal,R., J. M.Kimble,R. F. Follett and C.V. Cole, ThePotential of U. S.Cropland to Sequester Carbon andMitigate the Greenhouse Effect, vol.pp.,Ann ArborPress,Chelsea,Michigan,1998.

Landry, M.R.et al.,Iron and grazing constraints on primaryproduction in the central equatorial Pacific. An EqPacsynthesis, Limnol.Oceanogr., 42, 405-418,1997.

Law, R.M., P. J. Rayner,A.S.Denning, D. Erickson,I.Y. Fung,M.heimann,S.C.Piper, M.Ramonet,S.Taguchi, J.A.Taylor, C.M.Trudinger and I. G.Watterson,Variations inmodeled atmospheric transport of carbon dioxide andthe consequences for CO2 inversions, GlobalBiogeochem.Cycles, 10, 783-796,1996.

Lloyd, J.,B.Kruijt, D.Y. Hollinger, J. Grace,R. J. Francey, S.-C.Wong, F. M. Kelliher,A.C.Miranda,and E.-D. Schwartz,Vegetation effects on the isotopic composition ofatmospheric CO2 at local and regional scales:Theoretical aspects and a comparison between rain for-est in Amazonia and a boreal forest in Siberia,Australian J. Plant Physiology, 23, 371-399,1996.

Lloyd, J. and G. Farquahar, 13C Discrimination during CO2assimilation by the terrestrial biosphere. Oecologia, 99,201-215,1994.

Luz,B.,E.Barkan,M.Bender, M.Thiemens and K.Boering,Triple-isotope composition of atmospheric oxygen as atracer of biosphere productivity, Nature, 400, 547-550,1999.

Madec, G.,and M.Imbard,A global ocean mesh to over-come the North Pole singularity, Climate Dynamics,12, 381-388,1996.

Maier-Reimer, E.,Geochemical cycles in an ocean generalcirculation model:Preindustrial tracer distributions,Global Biogeochem.Cycles, 7, 645-677,1993.

Maier-Reimer, E.and K.Hasselmann,Transport and storageof CO2 in the ocean - an inorganic ocean-circulationcarbon cycle model, Clim.Dyn., 2, 63-90,1987.

Manabe,S.and R. J. Stouffer, Multiple century response of acoupled ocean-atmosphere model to an increase ofatmospheric carbon dioxide, Journal of Climate, 7,5-23,1994.


66

Marenco,A.,V.Thouret, P. Nédélec,H.Smit,M.Helten, D.Kley, F. Karcher, P. Simon,K.Law, J. Pyle, G. Poschmann,R.Von Wrede,C.Hume,and T. Cook,Measurement ofozone and water vapor by Airbus in-service aircraft:TheMOZAIC airborne program,An overview, J. Geophys.Res., 103, 631-625,1998.

Marland, G.,R. J.Andres,and T.A.Boden. Global, regional,and national CO2 emissions,In:Trends ‘93,T.A.Boden,D. P. Kaiser, R. J. Sepanski,and R.W. Stoss,(Eds.),publica-tion No.ORNL/CDIAC-65,Carbon Dioxide InformationCenter, Oak Ridge National Laboratory, Oak Ridge,TN,pp.505-581,1994.

Martin, J. H.,Glacial-interglacial CO2 change: the ironhypothesis, Paleoceanography, 5, 1-13,1990.

Martin, J. H.,Iron,Liebig’s Law and the greenhouse,Oceanography, 4, 52-55,1991.

Matear, R. J. and A.C.Hirst,Climate change feedback onthe future oceanic CO2 uptake, Tellus, in press.

McGowan, J.A., D. R.Cayan,and L.M.Dorman,Climateocean variability and ecosystem response in theNortheast Pacific, Science, 281, 210-217,1998.

Melillo, J. et al.,Land-use change in the Soviet Unionbetween 1850 and 1980:causes of a net release of CO2to the atmosphere, Tellus, 40B, 116-128,1988.

Merlivat,L.and P. Brault,CARIOCA buoy:Carbon dioxidemonitor, Sea Technology, 10, 23-30,1995.

Michaels,A. F., D. Olsen, J. Sarmiento, J.Ammerman,K.Fanning,R. Jahnke,A.H.Knap, F. Lipschultz and J.Prospero,Inputs,losses and transformations of nitrogenand phosphorus in the pelagic North Atlantic Ocean,Biogeochemistry, 35, 181-226,1996a.

Michaels,A. F., J. L.Sarmiento and J. Prospero,Excess nitrateand the rate of nitrogen fixation in the Sargasso Sea,EOS, 76, OS84,1996b.

Myneni,R.B.,C. D. Keeling,C. J.Tucker, G.Asrar and R.R.Nemani,Increased plant growth in the northern highlatitudes from 1981 to 1991, Nature, 386, 698-702,1997.

National Research Council, Global EnvironmentalChange:Research pathways for the next decade,National Academy Press,Washington, D.C.,1998.

National-Research-Council, Learning to Predict ClimateVariations Associated with El Niño and the SouthernOscillation:Legacies of the TOGA Program, vol.pp.,National Academy Press,Washington, D.C.,1996.

Nilsson,S.and W. Schopfhauser,The carbon-sequestrationpotential of a global afforestation program, ClimaticChange, 30, 267-293,1995.

Oechel,W. C.,S. J. Hastings, G.Vourltis,M. Jenkins, G.Riechers and N. Grulke,Recent Changes of ArcticTundra Ecosystems from a Net Carbon Dioxide Sink toa Source, Nature, 361, 520-523,1993.

Oeschger, H., U. Siegenthaler, U. Schotterer and A.Gugelmann,A box diffusion model to study the carbondioxide exchange in nature, Tellus, 27, 168-192,1975.

Potosnak,M. J.,S.C.Wofsy,A.S.Denning,T. J. Conway, andD. H.Barnes,Influence of biotic exchange and combus-tion sources on atmospheric CO2 concentrations inNew England from observations at a forest flux tower. J.Geophys.Res., in press,1999.

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

Randerson, J.T.,M.V.Thompson,T. J. Conway, I.Y. Fung andC.B. Field,The contribution of terrestrial sources andsinks to trends in the seasonal cycle of atmosphericcarbon dioxide, Global Biogeochem.Cycles, 11, 535-560,1997.

Raschi,A., F. Milietta,R.Tognetti and P. R. v. Gardigen (Ed.),Plant Responses to Elevated CO2: Evidence fromNatural Springs, vol.Cambridge University Press,Cambridge,1997.

Raynaud, D., J. Jouzel, J. M.Barnola, J. Chappellaz,R. J.Delmas and C.Lorius,The ice record of greenhousegases, Science, 259, 926-934,1993.

Rayner, P. J.,I. G. Enting,R. J. Francey and R.Langenfelds,Reconstructing the recent carbon cycle from atmo-spheric CO2, ∆13C and O2/N2 observations, Tellus, Inpress,1998.

Sabine,C.L.,R.M. Key, K.M. Johnson, F. J. Millero,A.Poisson, J. L.Sarmiento, D.W. R.Wallace,and C. D.Winn,Anthropogenic CO2 inventory of the Indian Ocean,Global Biogeochem.Cycles, 13, 179-198,1999.

Sarmiento, J. L.,T. M.C.Hughes,R. J. Stouffer and S.Manabe,Simulated response of the ocean carbon cycleto anthropogenic climate warming, Nature, 393, 245-249,1998.

Sarmiento, J. L.and C.Le Quéré,Oceanic carbon dioxideuptake in a model of century-scale global warming,Science, 274, 1346-1350,1996.

Sarmiento, J. L.,R.Murnane and C.L.Quéré,Air-Sea CO2transfer and the carbon budget of the North Atlantic,Philosophical Transactions of the Royal Society, 348,211-218,1995.

Sarmiento, J. L., J. C.Orr and U. Siegenthaler,A perturbationsimulation of CO2 uptake in an ocean general circula-tion model, J. Geophys.Res., 97, 3621-3646,1992.

Bibliography

67

Sarmiento, J. L.,and E.T. Sundquist,Revised budget for theoceanic uptake of anthropogenic carbon dioxide.Nature, 356, 589-593,1992.

Schimel, D. S.,VEMAP participants,and B.H.Braswell,Spatial variability in ecosystem processes at the conti-nental scale:models,data and the role of disturbance,Ecol.Monog., 67, 251-271,1997.

Schimel, D., D.Alves,I.Enting,M.Heimann, F. Joos, D.Raynaud and T.Wigley, CO2 and the Carbon Cycle,inClimate Change 1995, vol.edited by J.T. Houghton,L.G. M. Filho,B.A.Callander, N. Harris,A.Kattenberg andK.Maskell, 76-86,Cambridge University Press,Cambridge,1996.

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

Schroeder, P.,S.Brown, J. M.Mo,R.Birdsey and C.Cieszewski,Biomass estimation for temperate broadleafforests of the United Staes using inventory data, ForestScience, 42, 424-434,1997.

Smith,S.V. and R.W. Buddemeier, Global change and coralreef ecosystems, Ann.Rev. Ecol.Syst., 23, 89-118,1992.

Stallard,R. F.,Terrestrial sedimentation and the carboncycle: Coupling weathering and erosion to carbonburial, Global Biogeochem.Cycles, 12, 231-258,1998.

Stauffer, B.,T. Blunier,A.Dallenback,A.Indermuhle, J.Schwander,T. F. Stocker, J.Tschumi, J. Chappellaz, D.Raynaud,C. U. Hammer and H.B.Clausen,AtmosphericCO2 concentration and millennial-scale climate changeduring the last glacial period, Nature, 392, 59-62,1998.

Stauffer, B.,E.Lochbronner, H.Oeschger and J. Schwander,Methane concentration in the glacial atmosphere wasonly half that of the preindustrial Holocene, Nature,332, 812-814,1988.

Stephens,B.B.,R. F. Keeling,M.Heimann,K. D. Six,R.Murnane and K.Caldeira,Testing global ocean carboncycle models using mesurements of atmospheric O2and CO2 concentrations, Global Biogeochem.Cycles,12, 213-230,1998.

Takahashi,T.,R.A. Feely, R. F.Weiss,R.H. Wanninkhof, D.W.Chipman,S.C.Sutherland and T.T.Takahashi,Global air-sea flux of CO2: An estimate based on measurementsof sea-air pCO2 difference, Proc. Natl.Acad.Sci., 94,8992-8299,1997.

Ta k a h a s h i ,T. R . ,R .H .Wa n n i n k h o f, R .A . Fe e ly, R . F.We i s s , D.W. C h i p m a n , N. B a t e s , J. O l a f s s o n ,C .S ab i n e , and S. C .S u 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 pCO2 d i ffe r-e n c e .I n : The proceedings of the 2n d C O2 in the OceansS y m p o s i u m,T s u k u b a , Ja p a n , Ja nu a ry 18-23, 1 9 9 9 .

Tans, P. P., P. S.Bakwin and D.W. Guenther,A feasible globalcarbon cycle observing system:a plan to deciphertoday’s carbon cycle based on observations, GlobalChange Biology, 2, 309-318,1996.

Tans, P. P., J.A.Berry and R. F. Keeling,Oceanic 13C/12CObservations: A new Window on Ocean CO2 Uptake,Global Biogeochem.Cycles, 7, 353-368,1993.

Tans, P. P.,I.Y. Fung and T.Takahashi,Observational con-straints on the global atmospheric CO2 budget,Science, 247, 1431-1438,1990.

Taylor, N. K.Seasonal uptake of anthropogenic CO2 in anocean general circulation model, Tellus, 47B, 145-169,1995.

Toggweiler, J. R.,K.Dixon,and K.Bryan,Simulation ofradiocarbon in a coarse-resolution world ocean model,1.Steady state pre-bomb distributions, J. Geophys.Res.,94, 8217-8242,1989.

Torn,M.S.,S.E.Trumbore, O.A.Chadwick, P. M.Vitousekand D. M.Hendricks,Mineral control of soil organic car-bon storage and turnover, Nature, 389, 170-173,1997.

Turner, D. P., G. J. Koerper, M.E.Harmon and J. J. Lee,A car-bon budget for forests of the coterminous UnitedStates, Ecol.Appl., 5, 421-436,1995.

Wallace, D.W. R.,Monitoring global ocean inventories,OOSDP Background Rep.5,Ocean Observ. Syst.Dev.Panel,Texas A&M Univ.,1995.

Wanninkhof, R.,Relationship between wind speed and gasexchange over the ocean, J. Geophys.Res., 97, 7373-7383,1992.

Winn,C. D., F.T. Mackenzie,C. J. Carillo,C.L.Sabine and D.M.Karl,Air-sea carbon dioxide exchange in the NorthPacific Subtropical Gyre: Implications for the globalcarbon budget, Global Biogeochem.Cycles, 8, 157-164,1994.

Wofsy, S.C.,M.L.Goulden, J. M.Munger, S.M. Fan, P. S.Bakwin et al.,Net Exchange of CO2 in a mid-LatitudeForest, Science, 260, 1314-1317,1993.

Woodwell, G. M,R.H.Whittaker,W.A.Reiners, G. E.Likens,C.C.Delwiche,and D. B.Botkin,The biota and theworld carbon budget, Science, 199, 141-146,1978.

Zhang, J. R.and P. D. Quay,The total organic carbon exportrate based on C-13 and C-13 of DIC budgets in theequaotrial Pacific region, Deep-Sea Research Part II,

44, 2163-2190,1997


68

69

Acronyms and Abbreviations

ARS Agricultural Research Service (USDA)

BOREAS BOReal Ecosystem-Atmosphere Study

BR Bowen Ratio/Energy Balance (flux measurement towers)

CCSP Carbon Cycle Science Plan

CFCs chlorofluorocarbons

CLIVAR Climate Variability and Predictability Programme (WCRP)

DIC dissolved inorganic carbon

DOE Department of Energy

DOM dissolved organic matter

FAO Food and Agriculture Organization

FIA Forest Inventory Analysis (program of theU.S. Forest Service)

ENSO El Niño–Southern Oscillation

GAIM Global Analysis,Interpretation and Modelling (IGBP program)

GCOS Global Climate Observing System

GCTE Global Change and Terrestrial Ecosystems(IGBP program)

GEOSECS Geochemical Ocean Sections program

GEWEX/GCIP Global Energy and Water Cycle 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–BiosphereProgram

IPCC Intergovernmental Panel on Climate Change

JGOFS Joint Global Ocean Flux Study(IGBP program)

LTER Long-Term Ecological Research (studysites maintained by the National Science Foundation)

MERIS Medium Resolution Imaging Spectrometer

MODIS Moderate-Resolution Imaging Spectroradiometer

NAO North Atlantic Oscillation

NASA National Aeronautics and Space Administration

NBP net biome production

NEP net ecosystem production

NOAA National Oceanic and Atmospheric Administration

NRC National Research Council

NSF National Science Foundation

POM Particulate organic matter

TOC total organic carbon

TOGA Tropical Ocean–Global AtmosphereProgram

TRANSCOM Atmospheric transport model comparison study (GAIM project)

USDA U.S.Department of Agriculture

USGCRP U.S.Global Change Research Program

VEMAP Vegetation/Ecosystem Modeling and Analysis Project

WCRP World Climate Research Programme

WOCE World Ocean Circulation Experiment (WCRP program)

Documents

A U.S. CARBON CYCLE SCIENCE PLAN...porting government agencies. The nature of the problem demands it. Carbon dioxide is exchanged among three major active reservoirs,the ocean,land,and