Science Framework and Implementation

On this page: Preface | TOC Executive Summary

Editors:
Josep G. Canadell, Robert Dickinson, Kathy Hibbard, Michael Raupach and Oran Young

Prepared by the Scientific Steering Committee of the Global Carbon Project and other contributors:
Michael Apps, Alain Chedin, Chen-Tung Arthur Chen, Peter Cox, Robert Dickinson, Ellen R.M. Druffel, Christopher Field, Patricia Romero Lankao, Louis Lebel, Anand Patwardhan, Michael Raupach, Monika Rhein, Christopher Sabine, Riccardo Valentini, Yoshiki Yamagata, Oran Young.

Please, cite this document as:
Global Carbon Project (2003) Science framework and Implementation. Edited by J.G Canadell, R Dickson, K Hibbard, M Raupach, O Young. Earth System Science Partnership (IGBP, IHDP, WCRP, DIVERSITAS) Report No. 1; Global Carbon Project Report No. 1, 69pp, Canberra.

---

Preface

We are pleased to launch the Earth Systems Science Partnership (ESSP) report series with the publication of the Science Framework and Implematation Strategy of the Global Carbon Project. This report marks the beginning of a new era in international global change research, as well as a significant departure from the usual way of treating the carbon cycle.

The ESSP, comprising four global change programmes - the International Programme of Biodiversity Science (DIVERSITAS); the International Geosphere-Biosphere Programme (IGBP); the International Human Dimensions Programme on Global Environment Change (IHDP); and the World Climate Research Programme (WCRP) - has been formed for the integrated study of the Earth System, the changes that are occurring to this system, and the implications of these changed for global sustainability. The Global Carbon Project, along with other ESSP projects on food systems, water resources and human health, are designed to make the links between the fundamental research on global change and earth systems carried out in the programmes themselves and issues of vital concern for people.

Carbon cycle research is often carried out in isolation from research on energy systems and normally focusses only on the biophysical patterns and processes of carbon sources and sinks. The Global Carbon Project represents a significant advance beyond the status quo in several important ways. First, the problem is conceptualised from the outset as one involving fully integrated human and natural components; the emphasis is on the carbon-climate-human system (fossil-fuel based on energy systems + biophysical carbon cycle + physical climate) and not simply on the biophysical carbon cycle alone. Secondly, the development of new methodologies for analysing and modelling the integrated carbon cycle is a central feature of the project. Thirdly, the project provides an internally consistent framework for the coordination and integration of the many national and regional carbon cycle research programmes that are being established around the world. Fourthly, the project addresses questions of direct policy relevance, such as the management strategies and sustainable regional development pathways required to achieve stabilisation of carbon dioxide in the atmosphere. Finally, the Global Carbon Project goes beyond the traditional set of stakeholders for a global change research project by seeking to engage the industrial and energy sectors as well as the economic development and resource management sectors in the development regions of the world.

We believe that this document will help to encourage, promote and shape carbon cycle research around the world for at least the next decade. Furthermore, we believe that it will provide the framework of a substantially enhanced knowledge base for dealing more effectively wit the challenge of transforming energy systems and managing the global carbon cycle.

Micheal Loreau Chair, DIVERSITAS	Anne Laringauderie Executive Director, DIVERSITAS
Guy Brasseur Chair, IGBP	Will Steffen Executive Director, IGBP
Coleen Vogel Chair, IHDP	Barbara Gobel Executive Director, IHDP
Peter Lemke Chair, WCRP	David Carson Director, WCRP

---

---

[TOC] |

Executive Summary

The Changing Carbon Cycle

The carbon cycle is central to the Earth system, being inextricably coupled with climate, the water cycle, nutrient cycles and the production of biomass by photosynthesis on land and in the oceans. A proper understanding of the global carbon cycle is critical for understanding the environmental history of our planet and its human inhabitants, and for predicting and guiding their joint future.

Human intervention in the global carbon cycle has been occurring for thousands of years. However, only over the last two centuries have anthropogenic carbon fluxes become comparable in magnitude with the major natural fluxes in the global carbon cycle, and only in the last years of the 20th century have humans widely recognised the threat of adverse consequences and begun to respond collectively. This development adds a new feedback into the global carbon cycle that will have a profound influence on the future of the Earth system, as humankind begins to grapple with the challenge of managing its planetary environment.

The Global Carbon Project

The challenge to the scientific community is to monitor (quantify), understand (attribute) and predict the evolution of the carbon cycle in the context of the whole Earth system, including its feedbacks with human components. This demands new scientific approaches and syntheses that cross disciplinary and geographic boundaries, and place particular emphasis on the carbon cycle as an integral part of the coupled carbon–climate–human system.

Three international global environmental change research programmes have come together to bring a coordinated programme into reality: the International Geopshere–Biosphere Programme (IGBP), the International Human Dimensions Programme on Global Environmental Change (IHDP), and the World Climate Research Programme (WCRP). The result is the Global Carbon Project (GCP). The present document outlines the project’s framework for research and its implementation strategy. The document is addressed to the large research and agency communities, including multiple disciplines of natural and social sciences, and policy makers.

Science Themes

The goal of the GCP is to develop comprehensive, policy-relevant understanding of the global carbon cycle, encompassing its natural and human dimension and their interactions. This will be accomplished by determining and explaining three themes:

1. Patterns and variability: What are the current geographical and temporal distributions of the major pools and fluxes in the global carbon cycle?
2. Processes and interactions: What are the control and feedback mechanisms — both anthropogenic and non-anthropogenic — that determine the dynamics of the carbon cycle?
3. Carbon Management: What are the likely dynamics of the carbon–climate-human system into the future, and what points of intervention and windows of opportunity exist for human societies to manage this system?

Implementation strategy

The GCP will implement its research agenda through collaborative efforts with national and international carbon programmes and funding agencies, and by leading a limited number of difficult and highly interdisciplinary new research initiatives that are feasible within a 3–5 year framework. The implementation strategy is organised around the three science themes.

Theme 1: Patterns and variability

Quantify current geographical and temporal distributions of the major carbon pools and fluxes through compiling new sectorial and regional budgets and developing model–data fusion.

• Major carbon stocks and fluxes. Provide a coordinated international effort to complement and strengthen regional and national carbon cycle programmes by fostering common protocols, sharing data, promoting rapid transfer of information on new applications and techniques, and leveraging resources in joint projects.
• Model–data fusion. Develop and implement methods for assimilating atmospheric, ocean and terrestrial data into carbon–climate–human system models, with particular emphasis on the application of multiple constraints (from the simultaneous use of atmospheric, oceanic and terrestrial data and models) to the problem of determining patterns and variability in the carbon cycle.
• Comprehensive national, regional and sectoral carbon budgets. Promote the harmonisation of existing approaches to national, regional and basin-scale carbon budgets to ensure comparability amongst regions.

Theme 2: Processes and feedbacks

Promote new research and synthesis to increase understanding of the controls on natural and human-driven sources and sinks of carbon, and the spatially explicit links between causes and effects, with particular emphasis on understanding the interactions among mechanisms and feedbacks among components of the coupled carbon–climate–human system.

• Mechanisms and feedbacks controlling carbon stocks and fluxes. Promote research and synthesis to identify the source and sink mechanisms, their relative importance and their interactive effects. Explore how the processes of the carbon system work, both individually and collectively.
• Emergent properties of the coupled carbon–climate system. Investigate additional system properties that emerge when the perturbed carbon cycle is included as an interactive element in the full carbon–climate system; in particular, investigate whether thresholds, instabilities and surprises could emerge from this full-system coupling.
• Emergent properties of the coupled carbon–climate–human system. Initiate cross-disciplinary research on the coupling of models (quantitative or qualitative) of the physical, biochemical and human components of the carbon cycle, and highlight novel behaviours that emerge when all these subsystems are coupled. Stimulate the development of more detailed predictive tools and conceptual frameworks.

Theme 3: Carbon Management

Identify and quantify points for intervention and windows of opportunity in the carbon cycle to steer the evolution of the coupled carbon–climate–human system.

• Points of intervention and options for mitigation. Identify and assess specific points of intervention at which the future evolution of the carbon cycle might be influenced, and critically assess the achievable mitigation potential of the options, once sustainable development concerns are considered (i.e., triple bottom line: economy, society, and environment).
• Carbon management in the context of the whole Earth system. Develop a framework to assess the best mix of mitigation options in a full-system analysis framework, design dynamic portfolios of carbon mitigation options for specific regions, and analysis/design appropriate institutions for carbon management.
• Carbon consequences of regional development pathways. Undertake a comparative analysis of a network of regional case studies to understand:
  • The consequences of different pathways of regional development on carbon stocks and fluxes.
  • The critical processes and interactions in development that result in pathways with widely differing carbon consequences.
  • The trade offs and synergies between changes in carbon stocks and fluxes with other ecosystem services, especially the provision of food, water and clean air, and the maintenance of biodiversity.

Synthesis and communication

The GCP will deliver high-level syntheses of information on the carbon cycle aimed at the research and assessment communities. Written products and web-based resources will be developed for policy makers, educators and general public. Specific products for multidisciplinary audiences will be developed to foster a common understanding and language.

Capacity building

The GCP will develop a number of capacity building activities associated with the main research themes. This will promote the development of a new generation of young and senior scientists trained in the highly interdisciplinary topics of the carbon cycle.

Products

The products of a 10-year research programme are envisioned as being:

• Improved knowledge of the coupled carbon–climate–human system with increased capacity to quantify, attribute and predict.
• A systemic framework, implemented in a suite of linked models, of the coupled biophysical and human interactions controlling the carbon cycle.
• Improved coordination between the research, monitoring and assessment communities, leading to a capability for rapid assessments and responses to trends in the carbon cycle.
• Improved outputs from national and international research and monitoring programmes, through better coordination, linkage and information exchange.
• Outreach and communication products, including synthesis of research in journal issues and books; international coordination platform through the GCP project and affiliated offices, electronically available resources (e.g., data, graphics and presentation material), quality websites including a carbon portal, educational resources (e.g., posters and leaflets) and opportunities for higher education through the various research activities.

Stakeholders

Major stakeholders of the GCP are the scientific, assessment and policy communities dealing with:

• Quantifying and predicting carbon budgets from local to global scales.
• Policies to reduce net greenhouses gas emissions.
• Development of, and compliance with, international conventions.
• Regional development aimed at meeting environmental, economical and social goals.

Connections with national and international programmes

Because of the integrative nature of the project, there will be a need to build upon many existing projects and to work with communities whose spheres of interest intersect (but do not necessarily coincide) with that of the GCP. In particular, the GCP will work with:

• Research communities coordinated through IGBP, IHDP, WCRP and other members of the Integrated Global Observing Strategy Partnership (IGOS-P).
• National and regional carbon cycle programmes.
• Assessment and policy communities dealing with the consequences of changes in the carbon cycle, vulnerability and the links to water resources, food systems and biodiversity.

The GCP Mandate

• To develop a research framework for integrating the biogeochemical, biophysical and human components of the global carbon cycle, recognising the need for work across disciplines, and temporal and geographical boundaries.
• To provide a global platform for coordinating international and national carbon programmes to improve the design of observation and research networks, data standards, information transfer, and timing of campaigns and process-based experiments, and the development of model–data fusion techniques.
• To strengthen the carbon-related research programmes of nations, regions, and international programmes such as IGBP, IHDP, WCRP, DIVERSITAS and the observation community, through better coordination, articulation of goals and development of conceptual frameworks.
• To foster research on the carbon cycle in regions that are poorly understood but have the potential to play important roles in the global carbon cycle.
• To synthesise and communicate new understanding of the carbon–climate–human system to the broad research and policy communities.

---

[TOC] |

1. Introduction

This document outlines the research framework of the Global Carbon Project (GCP), a research project on the global carbon cycle developed jointly by the International Geopshere–Biosphere Programme (IGBP), the International Human Dimensions Programme on Global Environmental Change (IHDP), and the World Climate Research Programme (WCRP). The GCP is also one of the first projects established under the Earth System Science Partnership (ESSP) sponsored by IGBP, IHDP, WCRP and DIVERSITAS. This document is, therefore, addressed to the large research and agency communities, including multiple disciplines of natural and social sciences, and policy makers.

The document is organised in three major sections. Section 1 (Introduction) gives an overview of the project, the motivation, vision and the main strategic elements. Section 2 (Science themes) outlines the three science themes of the GCP, which together provide a comprehensive picture of the global carbon cycle and its interactions with climate and human activities. For each of these themes, there are subsections on the relevant knowledge base, present research areas, uncertainties and research priorities. Section 3 (Implementation strategy) outlines the initial activities that the GCP, in coordination with a number of other projects and programmes, will execute over the next 3–5 years, and includes a vision that extends to the full life time of the project (about 10 years). At the end of the document, there are a number of appendixes containing information on national and international programmes and networks relevant to global research of the carbon cycle.

The carbon challenge

The carbon cycle is central to the Earth system, being inextricably coupled with climate, the water cycle, nutrient cycles and the production of biomass by photosynthesis on land and in the oceans. This production sustains the entire animal kingdom, including humans through their dependence on food and fibre. Hence, a proper understanding of the global carbon cycle is critical for understanding the environmental history of our planet and its human inhabitants, and for predicting and guiding their joint future. The Vostok ice core record (VIEW Figure 1 [DOWNLOAD ZIP, 521Kb]) illustrates the limits and patterns of natural variability of atmospheric carbon dioxide (CO₂) and the correlation of atmospheric CO₂ and methane (CH₄) concentrations to inferred temperature over the last 420,000 years. From about one-half million years ago until about 200 years ago, the climate system has operated within a relatively constrained range of temperature and concentrations of atmospheric CO₂ and CH₄. In the pre-industrial world, atmospheric CO₂ concentrations oscillated in roughly 100,000-year cycles between 180 and 280 parts per million by volume (ppmv), as the CO₂ climate system pulsed between glacial and interglacial states. The ice core record clearly illustrates that atmospheric composition and climate (especially temperature) are closely linked.

Comparison of the Vostok record with contemporary measurements of atmospheric CO₂ concentration reveals that the Earth’s system has dramatically left this regular domain of glacial–interglacial cycling (VIEW Figure 2 [DOWNLOAD ZIP, 215Kb]). Atmospheric CO₂ concentrations are now nearly 100 ppmv higher than at the interglacial maximum, and the rate of increase has been at least 10, and possibly 100, times faster than at any other time in the past 420,000 years. Concentrations of other greenhouse gases, including CH₄ and N₂O, are increasing at comparable rates. These increases are unquestionably due to human activities, and are already having consequences for climate. For example, a temperature record for the past millennium indicates that the contemporary climate system is now responding to changing greenhouse gas concentrations in the atmosphere. Far greater changes are predicted over a time scale of centuries, with a confidence that has increased substantially between the second and third assessments of the Intergovernmental Panel on Climate Change (IPCC 1996; 2001a,b,c). These changes indicate that the Earth system has moved well outside the range in which the carbon cycle operated over the past half million years. Change has been unidirectional and of unprecedented rate; that is, humans have pushed the Earth system into uncharted territory.

The role of humans in the carbon cycle is not new. Human activities have influenced it for thousands of years through agriculture, forestry, trade and energy use in industry and transport. However, only over the past two or three centuries have these activities become sufficiently widespread and far reaching to match the great forces of the natural world. Moreover, human societies and institutions (social, cultural, political and economic) are not unidirectional drivers of change: they are impacted upon by changes to the carbon cycle and climate, and respond to these impacts in ways that have the potential to feed back on the carbon cycle itself (Young 2002; VIEW Figure 3 [DOWNLOAD ZIP, 451Kb]). One example is the attempt to manage greenhouse gas emissions as part of the global atmospheric commons.

Efforts to identify the location and magnitude of carbon exchanges between atmosphere, land and ocean illustrate the complex interactions between the natural and human aspects of the system, and the difficulty of separating them. The locations of current terrestrial CO₂ sinks (i.e., areas of land that take up CO₂ from the atmosphere) may be largely due to historic patterns of land-use change, and their magnitudes the result of physiological response to repeated disruption. Patterns of oceanic CO₂ sinks may also be modified by atmospheric transport of iron-laden dust from continents, which, in turn, is influenced by land-use and climatic variability. Areas where humans might manipulate the carbon cycle include enhancing sequestration of carbon in terrestrial ecosystems and the oceans, and minimising the massive emissions from fossil fuel combustion. 

Research is focusing on monitoring and understanding these patterns and processes in the global carbon cycle, and their environmental impacts. Different research communities are using a variety of resources and methods. For example, satellite data, air sampling networks and inverse numerical methods (‘top-down’ approaches) allow the strength and location of the global and continental-scale carbon sources and sinks to be determined. Surface monitoring and process studies (‘bottom-up’ approaches) provide estimates of land–atmosphere and ocean–atmosphere carbon fluxes at finer spatial scales, and allow examination of the mechanisms that control fluxes at these regional and ecosystem scales (VIEW Figure 4 [DOWNLOAD ZIP, 435Kb]). An understanding of the natural dynamics and the potential for mitigation in the carbon cycle will ultimately allow pathways for decarbonisation to be developed that can be implemented through policy instruments and international regimes.

The Vision

The central vision of the GCP is to develop comprehensive, policy-relevant understanding of the global carbon cycle, encompassing its natural and human dimension and their interactions.

Achieving this vision will require coordination by the international scientific community across all relevant disciplines and regions, and application of a large number of available resources and techniques. At present, no single international research programme provides this framework. The GCP was created to fill this gap and provide overall coordination to address highly interdisciplinary and complex problems of the carbon–climate–human system.

The GCP will to take an interdisciplinary approach to understanding the natural unperturbed carbon cycle, the perturbed carbon–climate–human system, and the feedbacks between societies’ responses to a perceived or real threat and the dynamics of the natural system (VIEW Figure 5 [DOWNLOAD ZIP, 436Kb]).

Through a series of workshops from 1999 to 2003, the scientific community identified three broad science themes for carbon cycle research. These themes define the scientific scope of the GCP and contribute to the development of a strong capacity for detection, attribution and prediction. The prediction component focuses strongly on where and how humans can intervene in the future dynamics of the perturbed carbon cycle. Each of the themes is described by an overarching question, as follows:

1. Patterns and variability: What are the current geographical and temporal distributions of the major pools and fluxes in the global carbon cycle?
2. Processes and interactions: What are the control and feedback mechanisms — both anthropogenic and non-anthropogenic — that determine the dynamics of the carbon cycle?
3. Carbon management: What are the likely dynamics of the carbon–climate-human system into the future, and what points of intervention and windows of opportunity exist for human societies to manage this system?

Mandate and approach

To implement the GCP vision and cover the three major themes, the GCP will be driven by the following scientific mandate:

• To develop a research framework for integrating the biogeochemical, biophysical and human components of the global carbon cycle, recognising the need for work across disciplines, and temporal and geographical boundaries.
• To provide a global platform for coordinating international and national carbon programmes to improve the design of observation and research networks, data standards, information transfer, timing of campaigns and process-based experiments, and the development of model–data fusion techniques.
• To strengthen the carbon-related research programmes of nations, regions, and international programmes such as IGBP, IHDP, WCRP, DIVERSITAS and the observation community, through better coordination, articulation of goals and development of conceptual frameworks.
• To foster research on the carbon cycle in regions that are poorly understood but have the potential to play important roles in the global carbon cycle.
• To synthesise and communicate new understanding of the carbon–climate–human system to the broad research and policy communities.

Approach: The GCP will implement its research agenda in two ways. First, the more disciplinary-oriented research on the carbon cycle is already implemented through a number of projects under the auspices of the GCP’s sponsoring programmes (Appendix A1), and sub-global research efforts are implemented through many national/regional carbon programmes (Appendix A2). The GCP will enhance and add value to this research by facilitating collaboration towards a higher-level integration, supporting the GCP’s mandate of putting together the broader picture of the global carbon cycle. Secondly, the GCP will initiate and lead a limited number of new research initiatives that are feasible within a 3-5 year framework on difficult and highly interdisciplinary problems of the carbon cycle.

Scientific guidance: The work of the GCP is guided by a scientific steering committee (SSC) made up of scientists covering the main interdisciplinary areas of the GCP science framework. The SSC will also consider recommendations made by its sponsor programmes and their projects.

Governance and time frame: The GCP answers to a committee made up of the chairs and directors of its three sponsoring programmes. A time frame of 10 years is envisioned for the GCP, beginning in 2002. A mid-term review by the three sponsoring programmes will assess how well the project has met its near-term objectives, monitor progress towards the longer term goals and suggest modifications needed to enhance the effectiveness of the project.

Institutional linkages: In a broader context, research on the carbon cycle is an essential component of many activities addressing the environmental science of the whole Earth system and the sustainable development agenda at an international level. The GCP will establish formal and informal partnerships to work with a number of observation, assessment and policy bodies:

• An integrated strategy for observing the global carbon cycle (Integrated Global Carbon Observation, IGCO) is under active development within the Integrated Global Observation Strategy Partnership (IGOS-P), with contributions by the global observing systems (Global Ocean Observing System (GOOS), Global Terrestrial Observing System (GTOS), Global Climate Observing System (GCOS)) and the GCP (Appendix B).
• The global carbon cycle is at the centre stage of policy development for climate mitigation, sustainable development and the provision of ecosystem services, both at national and international levels. There is a need to connect, through appropriate assessment bodies, with international and national policy communities.
• Assessment of scientific research on the carbon cycle, and its interpretation for the policy community, is carried out by the IPCC as requested by the Subsidiary Body for Scientific and Technological Advice (SBSTA) of the United Nations Framework Convention on Climate Change (UNFCCC), the Millennium Ecosystem Assessment (MA) and other assessment programmes.

---

[TOC] |

2. Science themes

The science framework of the GCP is organised around three themes: patterns and variability; processes and interactions; and carbon management. This section describes, for each theme, the knowledge base (what we already know from past work), current or planned research, and the main areas where important knowledge is lacking, described here as areas of uncertainty. The definition of areas of uncertainty leads finally into a number of questions that define research priorities for each theme. It is notable, however, that many of the research questions bridge across the three themes.

---

[TOC] |

Science Themes
Theme 1: Patterns and variability

Motivation

The basic structure of the carbon cycle is determined by the flows of carbon between major pools, including carbon in the atmosphere (mainly as CO₂); in the oceans (surface, intermediate waters, deep waters and marine sediments); in terrestrial ecosystems (vegetation, litter and soil); in rivers and estuaries; and in fossil carbon, which is being remobilised by human activities. Both the flows of carbon among these pools and their carbon content have a rich spatial and temporal structure reflecting natural dynamics and human activity (VIEW Figure 6 [DOWNLOAD Figure 6a ZIP, 159Kb][DOWNLOAD Figure 6b ZIP, 779Kb][DOWNLOAD Figure 6c ZIP, 613Kb][DOWNLOAD Figure 6d ZIP, 714Kb]).

An understanding of the patterns and variability in this structure is crucial for defining the basic anatomy of the carbon cycle, providing diagnostic insight into the driving processes and underpinning reconstructions of past and predictions for the future — especially a future subject to anthropogenic perturbations outside the range experienced by the Earth system in recorded history.

Knowledge base

Present understanding of the patterns and variability of global carbon fluxes is based on:

• Global observations, including the atmospheric concentrations of CO₂ and other gases, satellite observations, and in situ terrestrial and oceanic measurements.
• Modelling of atmospheric and oceanic dynamics and biogeochemical processes.
• Mass balance principles.

Together, these provide strong evidence to support the following points (IPCC 2001a, Field and Raupach 2003; CDIAC 2003):

1. Global fossil fuel emissions have been rising since pre-industrial time and were 5.2 petagrams of carbon (PgC) in 1980 and 6.3 PgC in 2002, with the vast majority occurring in the northern hemisphere.
2. Atmospheric carbon dioxide (CO₂), methane (CH₄), and nitrous oxide (N₂O) have increased by 31%, 150%, and 16% since 1750, respectively.
3. About half of the CO₂ emitted to the atmosphere by fossil fuel sources is taken up by a combination of terrestrial and oceanic sinks.
4. Observed distributions of atmospheric CO₂ and the oxygen/nitrogen ratio (O₂/N₂), together with atmospheric model inversion studies, suggest that the terrestrial sink occurs predominantly in the northern mid-latitudes.
5. Land-use change results in significant emissions of atmospheric CO₂ in tropical latitudes, whereas land-management change is responsible for a significant carbon sink in northern mid-latitudes.
6. For the last few decades, observed changes in atmospheric CO₂ concentrations have varied widely (VIEW Figure 7 [DOWNLOAD ZIP, 429Kb]); the implied rate of carbon accumulation in the atmosphere varies between years by nearly as much as average annual fossil fuel emissions.
7. The interannual variability in carbon exchange with the atmosphere is dominated by terrestrial ecosystems rather than the ocean.
8. Imports and exports of cereals, wood and paper products accounted for about 0.72 PgC yr-1 of ‘embodied’ carbon trade in 2000, affecting regional sinks (production), sources (consumption) and temporary storage (e.g., furniture) (VIEW Figure 8 [DOWNLOAD ZIP, 439Kb]).
9. The net global air–sea flux is 2.2 PgC (-19% to +22%) into the ocean for the reference year 1995; ocean models and observations suggest that the interannual variability of the global ocean CO₂ flux is around 0.5 PgC yr-1, with the largest interannual variability apparently occurring in the equatorial Pacific Ocean.
10. The broad pattern of oceanic sources and sinks of atmospheric CO₂ are known: tropical waters generally act as sources and higher latitude waters act as sinks; the strongest oceanic CO₂ sink is the North Atlantic Ocean and the strongest source is in the equatorial Pacific Ocean.
11. Lateral fluxes in rivers and trade/commerce are important in explaining patterns and distribution of the carbon sources and sinks; carbon exports from rivers to the coastal ocean are higher than 1 PgC yr-1.

Current research

The above conclusions are largely based on observation and modelling. This section describes continuing work on carbon cycle patterns and variability using these two approaches, including observations of human interactions with the carbon cycle and strategies for combining observations with models.

Global monitoring

Long-term monitoring is an essential research tool for detecting, attributing and predicting the spatial and temporal patterns in the global carbon cycle. Major time series have become touchstones for the science of the carbon cycle and the Earth system (IPCC 2001a). Examples include the multidecadal records of atmospheric composition (notably CO₂ concentrations) from baseline observing stations at Mauna Loa, Cape Grim and elsewhere (e.g., Keeling and Whorf 2000); and the 420,000-year Vostok ice core record shown in VIEW Figure 1 [DOWNLOAD ZIP, 521Kb] (Petit et al 1999). Spatial data are also critical, for example, the global net terrestrial primary production (NPP) inferred from a number of biogeochemical models (VIEW Figure 6a [DOWNLOAD ZIP, 159Kb]).

The global observation tools necessary to understand the Earth system (including the global carbon cycle and human impacts on it) are being assembled in a cooperative global observing strategy, the Integrated Global Observing Strategy Partnership (IGOS-P). The principle behind IGOS-P is to develop a strategy for coupling major Earth and space-based systems for global environmental observations of the land, oceans and atmosphere.

As part of IGOS-P, a strategy for international global carbon cycle observations over the next decade is being developed through an IGCO theme, in close collaboration with the GCP (Appendix B). This strategy will:

• Integrate remote and in situ observations.
• Link ocean, terrestrial and atmospheric observing strategies.
• Involve close collaboration with the international carbon cycle research and assessment communities.

Towards these goals, a Terrestrial Carbon Observation (TCO) component of the GTOS component has already been developed to provide information on the spatial and temporal distribution of carbon sources and sinks in terrestrial ecosystems, using data obtained through ground and satellite-based observations.

The new products from the Moderate Resolution Imaging Spectroradiometer (MODIS) and other satellites will provide an important dynamic long-term record of the terrestrial and ocean metabolism. This record will include a number of consistent, calibrated and near-real-time measures of major components of the global carbon cycle including global net primary productivity (NPP) at 1 × 1 km resolution every eight days (VIEW Figure 9 [DOWNLOAD Figure 9a ZIP, 1.51Mb][DOWNLOAD Figure 9b ZIP, 1.33Mb]).

Atmospheric observations

Numerous countries currently sponsor measurements of atmospheric trace gas concentrations, in most cases as part of research programmes. These data have made pivotal contributions to the awareness and understanding of climate change. The atmosphere is an excellent filter of spatially and temporally varying surface fluxes, integrating short-term fluctuations while retaining the large-scale signal (Tans et al 1990). The distribution of CO2 in the atmosphere and its time evolution can thus be used to quantify surface fluxes.

Regional carbon budgets are currently calculated from CO₂ measurements at about 100 sites, supplemented by a few tall towers and aircraft programmes, using atmospheric inversion methods. Among the most significant impacts to date of network observations (and their interpretation by inversion methods) has been the discovery of major CO₂ net sinks in the northern hemisphere, both terrestrial and oceanic (IPCC 2001a, Gurney et al 2002, Rödenbeck et al 2003) However, retrieval of the space–time patterns of surface fluxes is highly uncertain. Without the use of additional constraints, it is hardly possible to resolve sources or sinks within longitudinal zones or between oceans and continents, even in the most densely sampled regions, the northern mid-latitudes. Even when such constraints are available from local process-oriented studies (e.g., Wofsy et al 1993), it is difficult to connect this understanding to global CO₂ patterns (Braswell et al 1997). Without a comprehensive spatial coverage of CO₂ measurements, uncertainties cannot be localised unequivocally to transport model or data error, or inversion procedures.

To overcome accuracy and consistency problems in these measurements, GLOBALVIEW-CO₂ was established as a cooperative atmospheric data integration project. It presently involves approximately 24 organisations from 14 countries (VIEW Figure 10 [DOWNLOAD ZIP, 548Kb]). An internally consistent 21-year global time series has been compiled. In addition to CO₂, the observing system includes measurements of ¹³C and ¹⁸O in CO₂, CH₄, CO, the O₂/N₂ ratio, and many other species. Measurements of ¹³C and O₂/N₂ provide information on the partition of net carbon fluxes into the atmosphere between fossil fuel emissions, land–atmosphere exchange and ocean–atmosphere exchange. Measurements of ¹⁸O are used to estimate gross primary production, as opposed to net ecosystem exchange. The CH₄ and CO measurements are used to estimate the contribution of combustion, in addition to the significance of CH₄ as a greenhouse gas. In addition, GLOBALHUBS has outlined a plan for global intercalibration of CO₂ concentrations and isotopes.

Three significant developing contributions to atmospheric observation are as follows:

• Continental and opportunistic measurements of atmospheric composition will extend the network of observations not only for CO₂ but also for other gases mentioned above. Existing atmospheric observing networks focus largely on measurements in the remote marine boundary layer, to avoid contamination by local sources and sinks. These data are invaluable in providing a baseline. However, there is a need for additional measurements over the continents. These are more complicated, due to strong variability in space and time caused by surface heterogeneity and diurnal cycling of the atmospheric boundary layer between convective and stable states, which affects the mixing of CO₂. Developments in sampling strategies are likely to progressively overcome these difficulties. Such measurements are commencing, using a combination of flask-sampled and continuous data from Fluxnet sites (see below), commercial and specially deployed aircraft, and Ships of Opportunity (SOOP). For continuous CO₂ measurements, a key technological development is the recent availability of lightweight, low-maintenance CO₂ sensors with precision comparable to present continuously attended baseline instrumentation.
• Methods for network optimisation will improve the next generation of upgrades to existing sampling networks. These rely on the use of data-assimilation methods as a primary technique to optimise network design.
• The measurement of CO₂ from space will have major impacts in filling the present sparse and uneven ground-based atmospheric sampling network on land, at sea and in the atmosphere, which, as noted above, severely limits the atmospheric-inverse approach (Rayner and O'Brien 2001).

Satellite observations of atmospheric CO₂

Remote sensing of the earth’s surface and atmosphere by space-borne instruments will improve all aspects of carbon cycle research. Two new infrared instruments for operational meteorological soundings are currently being developed for the measurement of CO₂ from space: the Atmospheric Infrared Sounder (AIRS), launched on board the Earth Observing System (EOS) satellite EOS-Aqua in March 2002; and the Infrared Atmospheric Sounder Interferometer (IASI), on board the first Meteorological Operational Polar Satellite (METOP) in 2005. Both instruments will measure most of the infrared spectrum at high spectral resolution and will be accompanied by the Advanced Microwave Sounding Unit (AMSU), a microwave sounder that can be used synergistically with either AIRS or IASI. The significance of this is that AMSU detects only the atmospheric temperature, while AIRS and IASI are also sensitive to CO₂ concentration. It is anticipated that additional properties of CO₂ will be retrieved from these sensors (Chedin et al 2003a).

A proof of concept study has been completed with existing instruments such as the Television Infrared Observational Satellite-Next (TIROS-N) Operational Vertical Sounder (TOVS), flown on board the United States National Oceanic and Atmospheric Administration (NOAA) polar meteorological satellites since 1978. Despite the quite limited spectral resolution of these space-based radiometers, clear signatures of the seasonal cycles and trends in CO₂ and other greenhouse gases (N₂O and CO) may be extracted from TOVS measurements and interpreted in terms of seasonal and annual variations of their atmospheric concentrations (Chedin et al 2002, 2003b).

Also important for retrieving CO₂ concentrations from space is the Scanning Imaging Absorption Spectrometer for Atmospheric Cartography (SCIAMACHY) launched on the Envisat platform in 2002 (Bovensmann et al 1999). This instrument will provide high-resolution spectra of the sunlight reflected by the Earth, including the absorption bands that are being considered for retrieving the greenhouse gases CO₂, CH₄, N₂O, H₂O and CO. The estimated total column precision is about 1% for CO₂, CH₄ and H₂O and about 10% for CO and N₂O (Buchwitz et al 2000). The horizontal resolution of the SCIAMACHY nadir measurements is typically 30 km × 120 km for relevant gases (30 km × 240 km at high latitudes). A similar passive differential absorption technique has also been recently proposed for the CARBOSAT (European Space Agency mission dedicated to monitoring the carbon cycle) and Orbiting Carbon Observatory (OCO) instruments (with greatly improved spatial and spectral resolutions).

The key assignment for each of these missions is a set of column CO₂ measurements of individual precision better than 1% (< 3ppmv). Simulations show that satellite measurements improve measurement of the carbon fluxes by a factor of up to 10 as compared to the network of surface stations. The greater coverage in time and space provided by the satellite data will improve existing estimates even though the precision of individual measurements may be an order of magnitude lower than those estimated from the air sampling network (Rayner and O'Brien 2001).

Terrestrial observations

Traditionally, the exploitation of biomass resources has been the primary reason for terrestrial carbon observations, motivating many countries to establish inventories (Cannell et al 1999; Houghton 2003) and monitoring networks to support the sustainable use of forests, croplands and grasslands. In parallel, national research programmes have initiated long-term ecophysiological observations at numerous sites, and increasingly use remotely sensed observations of land cover.

Currently, there are a number of existing internationally coordinated networks relevant to terrestrial carbon observation (data providers), including both ground networks of global scope and satellite-based observations (Appendix C). Among the ground-based networks, the Fluxnet programme coordinates a global network of over 200 sites, at which tower-based eddy covariance methods provide continuous measurements of the land–atmosphere exchanges of CO₂, water vapour, heat and other entities (VIEW Figure 11 [DOWNLOAD ZIP, 415Kb]). At many of these sites, complementary measurements are made of carbon stocks and fluxes in vegetation, litter and soil pools, and other ecophysiological variables. Flux tower data with scaling techniques have been already applied successfully to calculate continent-wide fluxes (Papale and Valentini 2003) and are yielding important insights on the controls of seasonal and interannual dynamics. Fluxnet is also becoming an important validation tool for the new MODIS products (e.g., net primary productivity) which will be generated every eight days.

The International Long-term Ecological Research Network (ILTER) provides a far more extensive network of lower technology ecophysiological observations, together with measurements of ecological changes. Harmonisation of regional ground observation programmes is being addressed by GTOS, as part of the GT-Net programme (Appendix C). Some research programmes are also addressing the harmonisation of data collected nationally; for example, a comparison of national forest inventories in North America and Asia (Goodale et al 2001), and a comparison of datasets from a number of countries on soil organic matter (Smith et al 2001).

Data users are agencies and programmes requiring information on the carbon cycle in terrestrial ecosystems (Appendix C). Data requirements differ in coverage (global, continental and national), type of product and the user group. For certain activities, national agencies require consistent information beyond their territories.

In addition to the above acquisition and product generation programmes, a number of projects have been undertaken that contribute to the development of systematic global observing capabilities, such as the Global Observations of Forest Cover (GOFC) project; the World Fire Web, providing data and information about biomass burning; the GTOS net primary production (NPP) project, providing data to support NPP estimation; and the IGBP NPP-intercomparison project, contributing to the improvement of algorithms for ecosystem productivity (Cramer and Field 1999).

Several major emerging trends in the observation of terrestrial carbon pools and fluxes are likely to accelerate in the next few years:

• Increased attention will be given to methods for combining measurements at multiple scales, such as eddy covariance, ecophysiological and process-level data, and remotely sensed data (see Current research: Scale interactions, in Theme 2).
• A closely related direction will be the synthesis of observations and models, through inversion, data assimilation and multiple-constraint approaches applied to a combination of terrestrial models and observations (see Current research: Synthesis of observations and models, in Theme 1).
• The use of isotopes and other tracers (¹³C, ¹⁴C, ¹⁸O, ¹⁵N, ²H, ³H) will provide additional measurement possibilities and constraints on models.
• There will be an increasing diversity of terrestrial observations, as nations implement carbon monitoring programmes for determining stocks and fluxes in the mandated categories for greenhouse gas emission estimations under the Kyoto Protocol.

Ocean observations

Traditional oceanographic surveys are a necessary element of any sampling strategy, providing continuity with historical data and the capability for full water column sampling, high accuracy and precision laboratory measurements, and detailed process studies. A continuing global survey programme is under way, to be coordinated by the International Ocean Carbon Coordination Project (IOCCP), a pilot project of the GCP and the SCOR-IOC Advisory Panel on Ocean CO₂. The IOCCP will work in collaboration with CLIVAR, which is making plans to re-occupy some of the World Ocean Circulation Experiment (WOCE) hydrographic lines.

Higher resolution spatial data is available for some in situ surface measurements, in particular the sea surface CO₂ partial pressure (pCO₂) required for air–sea carbon flux estimates. Shipboard underway pCO₂ systems are commonly used on oceanographic research cruises (a recent example being the WOCE-Joint Global Ocean Flux Study (JGOFS) hydrographic survey), as well as a growing effort with Ships of Opportunity (SOOP). The quantity of such measurements will increase in the future and will need better coordination to optimise the basin-scale and global coverage.

Because of their expense and logistic requirements, large-scale shipboard surveys are conducted only infrequently. Such temporally limited measurements offer a picture of the approximate average state of the ocean but do not resolve well the variability on seasonal, interannual and decadal time scales. To resolve these temporal patterns, long-term time-series measurements of carbon and other biogeochemical variables at fixed locations are crucial. The best-known open-ocean time series at present are the more than decade-old United States JGOFS stations near Hawaii (Hawaii Ocean Time-series programme — HOT) and Bermuda (Bermuda Atlantic Time-series Study — BATS). The HOT and BATS monthly data include carbonate system parameters and other traditional biogeochemical data, such as primary productivity, chlorophyll, nutrients, and near-surface sediment traps. They have led to a number of key discoveries, including the demonstration of increasing surface dissolved inorganic carbon (DIC) concentrations and the importance of nitrogen fixation in the subtropical Pacific Ocean. To be most effective, these time-series sites should be thoroughly integrated into the hydrographic and SOOP survey programmes, including measurements from moorings and drifters. The time-series data can provide the temporal context for the spatial surveys and vice versa.

A number of satellite data sets have direct applicability to the ocean carbon system (Appendix C). The most obvious are ocean colour data, which were collected beginning with Coastal Zone Colour Scanner Data (CZCS) (1979–1986) and have been greatly expanded in recent years (Ocean Colour and Temperature Scanner, OCTS, 1996–1997; Polarization and Directionality in the Earth Reflectances, POLDER; Sea-viewing Wide Field-of-view Sensor, SeaWiFS, late 1997 to present). Relevant physical data sets include sea surface altimetry (TOPEX, a United States/French mission to track sea-level height with radar altimeters/Poseidon, a satellite research programme; and the European Remote Sensing programme) for mesoscale variability and physical circulation, sea surface temperature (Advanced Very High Resolution Radiometer (AVHRR) and other platforms), and surface wind speed (National Aeronautics Space Agency Scatterometer, NSCAT; QuickScat). New developments may make it possible to measure salinity from satellites.

Major emerging directions for ocean observations include the following:

• Continuing expansion of ocean observations, in temporal and spatial density, and in the range of chemical and biological parameters measured. Satellite data on sea surface temperature (SST), winds and ocean colour will continue to provide critical information on large-scale patterns and variability of upper ocean physics and biology. For those quantities that cannot be resolved from space, in situ autonomous measurement/sampling technologies are being developed. Particularly promising directions for in situ chemical sampling include new autonomous sensors (e.g., pCO₂, DIC, nutrients, particulate inorganic carbon, particulate organic carbon (POC), bio-optics) and ocean platforms (e.g., moorings, drifters, profiling floats, gliders and autonomous underwater vehicles).
• Enhanced methods for the interpretation of ocean observations will provide additional information on regional interannual variability in air–sea fluxes. Such information is now emerging from repeat observations of surface water pCO₂. Estimates of changes in ocean carbon inventories and transports are needed to contribute to basin-scale carbon budgets for the ocean.
• The development of a comprehensive ocean carbon observing system can be advanced through improved organisation and coordination. This will involve (1) identifying and supporting those programme elements that are currently in operation (such as time-series stations, hydrographic sections and SOOP lines) or in the planning stages; (2) convening and encouraging international meetings of expert groups to refine observing system requirements for scientific and operational monitoring goals; and (3) developing cooperative relationships with other physical, chemical and biological ocean field efforts, with special emphasis on CLIVAR and GOOS. Projects are presently underway in both the North Atlantic and North Pacific to continuously monitor properties on the basin-scale from SOOP lines. However, these programmes need long-term support to build and maintain available datasets.
• The development of ocean carbon assimilation and inverse models is advancing rapidly (as for atmospheric and terrestrial observations). The inclusion of enough process-level information will be critical to address spatial–temporal patterns and detection–attribution of controls of fluxes. As the observational programmes mature, they will provide an unprecedented data stream that can be quickly fed into data-driven models. These models can help provide the time and space scale interpolation to evaluate global fluxes and inventories of carbon.

Observations of human interactions with the global carbon cycle

The human components of the global carbon cycle include emissions, sinks, lateral flows, commodity production and consumption. These human-induced carbon fluxes interact with a range of other human variables including population, wealth, energy systems, technological pathways, and environmental values and constraints (Dietz and Rosa 1997). Such interactions occur both through perceptions of the consequences of human-induced changes in the carbon cycle, and through other major factors such as economic and social drivers, and water and food supplies.

A range of existing systems provide relevant data on these human-mediated carbon fluxes. These systems include inventories of national emissions, forestry and land-use; national carbon accounting systems, regional environmental reporting, and data on trade and commodity production. The challenge is clearly to integrate these disparate and often indirect data sources.

In the area of observations of human interactions with the carbon cycle, major emerging directions include:

• The differing roles of countries, regions and sectors in the carbon cycle. For example, the vast majority of fossil fuel emissions occur in the northern hemisphere, while land use change dominates carbon emissions in tropical latitudes. International corporations are key contributors of data and analyses of such trends (Mason 1997).
• Increasingly refined assessment of regional impacts and vulnerability to climate and carbon-cycle changes. Although there are already different documents on this issue, a major challenge will be to explore the ways in which different regions, sectors, ecosystems and social groups will be confronted by, and/or be able to manage, changes in the carbon cycle (O'Brien and Leichenko 2000).

Synthesis of observations and models

Only a few of the observations described above give direct information on the fluxes and stocks that constitute the global carbon cycle, and none offer an adequate direct picture of spatial and temporal patterns. It is therefore necessary to infer these indirectly. Numerous methods have been developed for this, all based on the synthesis of information from both observations and models. The term “model-data fusion” is sometimes used as an umbrella descriptor for these activities. The general principle is to find an optimal match between observations and model by varying one or more than one property of the model. All applications of this principle involve two basic choices, the first being the model property (or properties) to be varied. There are four broad options: parameters (notionally constant quantities entering the model equations); boundary conditions in space; initial conditions in time; or the model state variables themselves. The second choice is the search method for finding the optimum values of these properties, for which there are many options depending on the formulation and complexity of the problem. In all cases the optimising process should provide three kinds of output: the optimal values of the varied quantities, uncertainty statements about these values, and an assessment of whether the model fits the data, given prior specified uncertainties on the data.

There are several well-established pathways among this suite of possibilities. Three are summarised briefly below.

Atmospheric and Oceanic Inverse Methods: Global atmospheric inversions use observations of atmospheric composition from global flask and baseline networks, together with global atmospheric transport models, to infer spatially averaged net fluxes of CO₂ and other entities between the surface and the atmosphere (Enting 2002; Gurney et al. 2002). The principle is to seek the source-sink distribution of a passive tracer, typically CO₂, which, together with a transport model, provides maximum consistency with global concentration measurements. Thus, in this case, the model property being varied is a boundary condition. The term “inverse” refers to the search method, which in essence is to run the transport model backwards.

Atmospheric inversions provide constraints on total carbon sources and sinks, but do not offer information on the processes responsible. Currently, their spatial resolution is extremely coarse. They can partition between the tropics and northern and southern hemisphere extratropical regions, and between land and ocean exchanges, but they do not provide a tropical carbon balance and cannot satisfactorily resolve longitudinal patterns ((Schimel et al. 2001). Their regional resolution is highly limited by lack of data, particularly in the tropics and the interiors of continents. There is an ongoing effort to use vertical profiles to help fill this gap. Inversions also depend on the choice of atmospheric transport model, especially on scales of ocean basins, continents or smaller.

Atmospheric inversion methods have also been applied regionally, using mesoscale models (Gloor et al. 2001) and atmospheric boundary-layer budget approaches, (e.g., (Lloyd et al. 1996)). Plume studies of forest fires and urban areas have also been used to obtain otherwise unavailable information on gaseous sources, through species and isotopic measurements. Applications at the scale of vegetation canopies have been used to partition sources and sinks between vegetation and soil (e.g., (Raupach 2001)).

Ocean inversions use similar principles to infer ocean–atmosphere CO₂ exchanges, using ocean pCO₂ and other data. Their data requirements are broadly similar to those of atmospheric inversions. In particular, the accuracy and density of measurements is a major issue, and results are sensitive to the ocean transport model employed.

Parameter estimation: In this case the model properties being varied are parameters which are poorly constrained by process understanding. For biogeochemical and carbon cycle models, these may include quantities such as quantum yields, light use efficiencies, temperature controls on respiration or pool turnover times (Barret 2002). It is almost always necessary to choose such parameters so that the model best fits sets of test data. There are many techniques for finding the best (“optimum”) parameters, ranging from simple graphical fits (such as choosing the slope of a line to give best fit) to advanced search procedures for finding multiple parameters simultaneously.

An emerging direction is the simultaneous use of multiple kinds of data in parameter estimation (“multiple constraints”). Many different kinds of data – atmospheric composition, remote sensing, in-situ measurements of pools or fluxes – are available. Different kinds of data constrain different processes in a model. For example, atmospheric concentration measurements and eddy fluxes constrain net CO₂ exchanges (Net Ecosystem Exchange) while remote sensing provides indirect constraints on gross exchanges (Gross Primary Production) through indices such as the Normalised Difference Vegetation Index (NDVI). Thus, different model parameters are constrained by different kinds of data, and the simultaneous use of several kinds of data is needed to constrain a comprehensive model adequately. Several preliminary applications of this approach have already been made, including the combined use of atmospheric concentrations and surface data at continental scale (Wang & Barret 2002), investigations of the combination of atmospheric composition, remotely sensed data and eddy flux data at global scale (Kaminski et al. 2002), use of genetic algorithms to constrain terrestrial ecosystem models of the global carbon cycle with ecological data at continental scale (Kaminski et al. 2002) and applications at the scale of vegetation canopies (Styles et al. 2002).

The multiple-constraint approach relies on access to multiple sources of constraining data with vastly different spatial, temporal and process resolutions, thus producing more constrained predictions. The approach potentially offers a means for discriminating between important and less important avenues for research to improve the process representations in the carbon-cycle model, because the inverse techniques currently used yield uncertainties on estimated parameters. A reduction in these uncertainties constitutes an increase in the information content of the overall prediction of the model. Potential data sources can be assessed for the reduction in uncertainty they provide for model parameters. Importantly, this approach requires the uncertainty characteristics of the data but does not require actual data to be available, thus allowing preliminary testing of experimental designs.

Data assimilation methods: Data assimilation involves adjusting the (time-varying) model state variables themselves as the model is integrated forward in time. This may also be done by sequential adjustment of initial conditions, as in four-dimensional variational data assimilation (4DVAR) methods now being made operational in weather forecasting. Here, time series of global data are used to force a dynamic model into optimal conformity with the data at a given time, while respecting the conservation requirements on the various fields represented in the model, such as conservation of mass (Chen & Lamb 2000, Park & Zupanski 2003). The application of these methods to carbon cycle modelling is still in the future.

Major emerging directions in the synthesis of data and models include:

• Development of other reasonably passive tracers that will offer increased understanding of the carbon cycle (noting that some currently available carbon cycle-related tracers are still not used), and continuing improvements in measurement density, calibration and interpretation, particularly for these additional tracers, but also for atmospheric CO₂ and ocean pCO₂.
• Improved data coverage that will allow downscaling to regional estimates, although regional estimates will also require improved knowledge of the global background (roughly — improvement anywhere hinges on improvement everywhere).
• Three promising technologies to collect more data relevant to regional inversions: (1) continuous measurements, to allow synoptic variations in transport to provide regional source signatures; (2) potential global coverage of CO₂ column integrals from space; and (3) potentially disposable light-weight sensors for use in low-maintenance environments — all of these technologies can interpolate gaps in the current network but must be well linked to it and will also require a major international data management effort to cope with the associated expansion in data flow.
• Inversions regionally and in ‘campaign mode’ (i.e., snapshots in time in a more constrained area) that can provide information on processes (e.g., via atmospheric plume studies from fires or urban areas and regional ocean transport studies).
• Application of multiple-constraint approaches to the modelling of combined physical, biological and biogeochemical processes.
• Development of carbon cycle process models, for use in multiple-constraint studies, focused on modelling at an appropriate level of parameterisation (noting that most fully process-based models are over-parameterised for use in this way).
• Developments in practical nonlinear search methods.
• More rigorous testing for model inconsistencies by the use of subsets of multiple data sources.
• Further development of uncertainty analyses, particularly in the context of nonlinear inversions.

Areas of uncertainty and research priorities

Despite progress over the last decade, substantial uncertainties remain.

• Existing global models and observations are unable to determine carbon sources or sinks with acceptable accuracy at regional, continental or interannual time scales, largely because of the sparse observing network. For example, the partition of the northern hemisphere terrestrial sink between North America and Eurasia remains unclear.
• There is no systematic and convincing agreement between ‘top-down’ and ‘bottom-up’ approaches to determining the spatial patterns of major fluxes in the carbon cycle. Budgets at regional, continental or basin scales are not consistent with the global analysis, with major uncertainties in key regions such as the Southern Ocean and terrestrial tropics. In addition, estimates of some critical fluxes, such as those associated with land-use change, are only obtainable by bottom-up methods and remain highly uncertain in the global context. Recent evidence suggests a much larger role of lateral transport and the coastal zones for regional carbon budgets than previously thought.
• The temporal patterns of the major carbon fluxes, and their consequences for stocks, are poorly understood at time scales greater than a few years. It is unclear which major stocks, whether resolved regionally or biogeochemically, contribute to the long-term variability in atmospheric CO₂ evident in the Vostok ice core record or in shorter records.
• Global estimates of oceanic flux patterns must currently be obtained from data collected over several decades, during which time the spatial pattern of fluxes has changed. This leads to considerable uncertainty in the estimates for any given year, or even decade. Results are dependent on the assumptions made to interpolate both in time and space between the often sparse measurements.
• There are uncertainties in the spatial and temporal distributions of human-induced fluxes, and the influence of human decision processes. Examples are the fluxes associated with land clearing and anthropogenic terrestrial sinks, and fossil fuel emissions (IPCC 2000a,b).

The above assessment prompts the following research priorities for Theme 1: What are the current geographical and temporal distributions of the major pools and fluxes in the global carbon cycle?

1. What are the spatial patterns of carbon fluxes and stocks at large scales (continents, ocean basins)?

• Determine the carbon budget of the terrestrial tropics, and particularly constrain carbon emissions due to land-use change.
• Determine the longitudinal distribution of the northern hemisphere terrestrial sink between North America and Eurasia, and within Eurasia between Europe and Asia.
• Determine the spatial patterns and magnitudes of ocean carbon sources, sinks and stocks, particularly in the Southern Ocean.
• Determine the fluxes and stocks of carbon associated with water flow from land to terrestrial water bodies to the coastal zone, and exchanges between the coastal zone and open oceans.
• Determine the consequences for the global carbon budget of the uplift, transport and deposition of sediments by both water and wind.
• Determine the role of non-CO₂ gases (e.g., methane and volatile organic compounds) in the global carbon budget.

2. How do regional and subregional patterns in carbon fluxes interact with the global-scale carbon cycle?

• Determine the space–time dynamics of the biological and solubility pumps in the oceans, and their relative roles in regional and global ocean carbon balances.
• Determine current trends in the carbon budgets of key terrestrial biomes (tropical, savannah, mid-latitude, boreal, tundra) which are changing as a result of global-scale changes in the coupled carbon–climate system.
• Develop methodologies for using regional and subregional carbon budgets to constrain the global budget, and vice versa.

3. What are the seasonal- to decadal-scale temporal variations in the fluxes and stockss making up the carbon budget at global and regional scales, and what are the causes of these variations?

• Determine the relative roles of the oceans, the terrestrial biosphere and fluctuations in human emissions.

4. What are the space–time patterns of human influences on the carbon cycle, including emissions from fossil fuel burning and land-use practices?

• Quantify the carbon fluxes and stocks associated with critical regions and sectors of human influence. Here regions include both rural and urban areas (especially megacities), and sectors include both industrial and agricultural activities.
• Resolve discrepancies in measurements of the historical and current rates and patterns of land-use and land-cover change.
• Determine the role of human activities in the terrestrial tropics, especially carbon sources due to land use change.

5. What are the social impacts of changes in the carbon cycle?

• Analyse the social and regional patterns of vulnerability and adaptation to the changes in the carbon cycle.

---

[TOC] |

Science Themes
Theme 2: Processes and interactions

Motivation

The behaviour of the fluxes and stocks that make up the carbon cycle is governed by a set of processes. These include:

• Physical processes in the atmosphere, oceans and terrestrial hydrosphere.
• Biological and ecophysiological processes in the oceans and on land.
• Biogeochemical transformations.
• A range of natural and anthropogenic disturbances to terrestrial ecosystems, such as fire, agriculture and clearing.The processes associated with the release of fossil carbon by humans (i.e., energy systems).

Some of these emerge as crucial controls on the global carbon cycle and its responses to anthropogenic forcing (e.g., terrestrial sink saturation, the stability of the thermohaline circulation, and the behaviour of the oceanic biological pump).

An understanding of these processes is needed at the level of their basic mechanisms and also of factors that emerge when they act in combination. Such knowledge is central to understand current and future dynamics of the carbon cycle and includes the recognition and interpretation of past and present interactions and feedbacks among key processes and mechanisms. Process understanding is needed for the development of diagnostic and prognostic tools (Activities 1.2, 2.2 and 2.3, below) that will eventually integrate the dynamics of biophysical systems and human behaviour. These tools will allow the exploration of critical system thresholds (e.g., vulnerabilities) beyond which it may be unwise to proceed (Activity 2.3) and the identification of mitigation options and their contribution to stabilising atmospheric CO₂.

Knowledge base

There is already a wealth of understanding of many of the processes governing the carbon cycle, especially at the level of detailed mechanisms. This understanding has been obtained through field observations, laboratory studies, manipulative experiments in the field and process modelling. Carbon-cycle processes that are well understood include the following (Walker et al 1999, IPCC 2001a, Field and Raupach 2003):

1. Net ocean–atmosphere exchanges of carbon are largely controlled by physical processes involving ventilation of thermocline and deeper waters (the solubility pump), with additional influences by biological processes that redistribute carbon from surface to deep waters (the biological pump).
2. A key biological control of the biological pump is the large phytoplankton cells that are responsible for much of the export of particulate carbon to the deep ocean (VIEW Figure 12 [DOWNLOAD ZIP, 332Kb]). Apart from the effect of large-scale changes in ocean circulation on the biological pump, future oceanic uptake of carbon from the atmosphere is expected to increase as atmospheric CO₂ levels rise.
3. A suite of feedback processes control the coupled energy, water and carbon exchanges between terrestrial surfaces and the atmosphere, causing the response of these fluxes to perturbations (such as land-cover transitions or changes) to be significantly scale dependent. Significant feedbacks in this context include plant physiological responses to atmospheric temperature, humidity, and soil temperature and moisture.
4. The current northern hemisphere carbon sink is due to multiple processes including, for example, forest regrowth, CO₂ and nitrogen fertilisation, climate change, soil erosion and accumulation in freshwater bodies. The relative importance of these processes is not well known (VIEW Figure 13 [DOWNLOAD ZIP, 100Kb]).
5. The strength of the terrestrial sink may eventually level off and then decline, despite some potential to increase due to anticipated atmospheric and climate change over the next decades.
6. Under sufficiently elevated atmospheric CO₂ concentrations (> 550 ppm), in conjunction with commonly limiting environmental conditions such as water and nutrients, photosynthetic assimilation by terrestrial plants quickly becomes physiologically saturated.
7. Deliberate land-surface modifications will constitute a large forcing of both physical climate and the carbon cycle in the medium-term future (a few decades). The climate response to this forcing may feedback onto land management practices.
8. At large (regional to continental) scales, several factors determine the magnitude and direction of CO₂ exchange between terrestrial ecosystems and the atmosphere:
   • Extreme climate events such as drought, large shifts in seasonal temperatures, or changes in radiation, induced by large-scale perturbations in levels of atmospheric aerosols (e.g., volcanic eruptions).
   • Changes in the frequency of fire, clearing and other large-area disturbances that lead to large, short-term carbon losses followed by a long, slow recovery of carbon stocks; global NPP by land plants is about 57 PgC yr-1 of which approximately 5–10% returns to the atmosphere through combustion (e.g., fuel, wildfires).
   • Changes in distributions and biome boundaries of plant species, induced by environmental forcing or changes in land use, which affect carbon storage and turnover over large areas (e.g., the conversion of evergreen to deciduous forest, forest to grassland or grassland to woodland) (Archer 1995, Hibbard et al 2001).
   • Losses of biodiversity and invasions by exotic species that may effect the use, efficiency and retention of carbon, nutrients (particularly nitrogen) and water (Schulze et al 2000).
9. Instabilities and multiple equilibrium states can potentially occur in the coupled biophysical and biogeochemical system, consisting of the physical climate, hydrological, carbon and nutrient cycles. These can arise primarily due to the nonlinearities and feedbacks involved in the exchanges of energy and matter between atmosphere, land, oceans and ice. Suggested examples include changes to the El