Investigations of the Earth and the environment it provides for life have progressed over the course of many decades to the point where it is now possible to address the system as a whole. Present understanding is based on the observations that:
1) while the planetary environment provides the general geophysical conditions for the sustenance of the biosphere, life itself alters these conditions in ways that both respond to, and buffer, changes;
2) major changes in the biosphere have occurred in conjunction with climatic changes on a variety of timescales from decades to hundreds of millennia and longer;
3) human activities have perturbed the system very rapidly and pushed it out of equilibrium in ways that have never happened before. While each aspect of the system individually has not been pushed beyond natural precedents, the present combination of atmospheric chemistry, ocean and atmospheric temperature, ice extent, and land cover has never occurred naturally in the past;
4) it is possible to develop models of each aspect of the Earth System in some detail, but it is more difficult to combine these models in the form of an integrated system model. Differences in scaling, parameterizations, and boundary conditions lead to complications in coupling existing subsystem models.
Investigation of the dynamic behavior of the Earth System is motivated by our limited knowledge about the consequences of large-scale perturbations by human activities such as fossil-fuel combustion or the fragmentation of terrestrial vegetation cover. There are key outstanding questions regarding behavior of the system that cannot be addressed with current models of individual system components such as the terrestrial biosphere or atmospheric circulation. A holistic system-level approach must be taken. One of the most notable of these questions is: "Is the system resilient with respect to anthropogenic disturbances, or could it be driven towards qualitatively new modes of planetary operation?" Another critical issue is the anticipated alterations of human behavior in response to perceived impacts of global change. This bears on international emissions and land use policies, and is presently exceedingly difficult to predict based on numerical models of socio-economic systems.
The present state-of-the-art is based on a conceptualization of the Earth System as three coupled parts: Physical climate, the Biosphere, and Human society. Understanding of each of these is being developed "independently" by the appropriate scientific communities (e.g. World Climate Research Program (WCRP), International Geosphere Biosphere Program (IGBP), International Human Dimensions Program (IHDP), respectively). However, we are already approaching the point at which it is no longer possible to advance without directly coupling each of these to the others. In earlier phases, with only a crude understanding of climate and the biosphere, it was sufficient (and necessary for practical reasons) for atmospheric physicists, for instance, to consider the ocean and land surfaces as fixed boundary conditions. As atmospheric general circulation models (AGCMs) evolved, it became clear that more is to be gained now by coupling with land surface and ocean circulation models than there is to maintaining fixed boundary conditions. However, it is not readily apparent a priori at what point the suite of subsystem models to be coupled are sufficiently robust for coupling and relaxation of prescribed boundary conditions to actually REDUCE uncertainty of the model results relative to the results of the decoupled models individually. This is an important area of investigation for GAIM and the scientific community at large.
Setting the stage for integration and synthesis
Cross-cutting theme: The Carbon Cycle
Carbon is the basis of biogeochemical cycles, with a pivotal role in the Earth System. As such, considerable effort has been expended by GAIM and others in exploring the processes and data necessary for a better understanding of the global carbon cycle. As the linkages between the processes that control the various aspects of the carbon cycle are identified, it has become clear that the behavior of the global biogeochemical system cannot be understood unless carbon is treated on a global, cross-disciplinary basis. Thus terrestrial, marine, atmospheric, and petrologic cycling and storage of carbon must be reconciled in a unified attack on the global biogeochemical system.
Model Intercomparison
One of the most significant contributions to date made by GAIM is the refinement of techniques and tools for model intercomparison, using the carbon cycle as a "test case". Whereas individual scientists or modelling groups can and do develop numerical models of various aspects of the Earth System, the value of the results of isolated models is greatly enhanced by comparison with other models. The discrepancies in model results between different approaches to the same problem provide critical insights into model shortcomings, and pave the way for model refinement and improvement. Further, the differences in sensitivity of different models to variations in input parameters and boundary conditions shed light on the differing conceptualizations utilized in model formulation. Elucidation of these differences yields insights regarding the actual Earth System processes being modeled. In its early stages, GAIM supported the development of three major sub-system level model intercomparison projects, abbreviated as TransCom, OCMIP, and EMDI in the atmosphere, ocean, and terrestrial realms, respectively. Descriptions of these projects can be found on the GAIM web site at http://gaim.unh.edu. In the next few years, and the tools developed through these activities will be applied to the interpretation and assessment of broader Earth System models.
Integration of earth system science
Based on the progress made at the subsystem level by GAIM, the IGBP Core Projects and others, GAIM has initiated its efforts toward subsystem model integration. This entails working with the broader research community to ensure that subsystem models are developed in a way that facilitates model coupling by matching boundary conditions and fluxes, temporal and spatial resolution, and establishing common numerical protocols.
The Waikiki Principles: Rules for an Evolving GAIM
During a meeting of the GAIM Task Force in Hawaii (Feb., 2000), the integration challenge was intensively discussed and identified as the central research issue of the next decade of global change science. A clear-cut picture emerged that may be summarized in the following three “Waikiki Principles”.
1) GAIM is to explore and promote cognitive opportunities arising from the appropriate combination of core project results and tools. This means, in particular, to play the role of a trans-project topics scout and a feasibility assessor.
2) GAIM is to advance the integration of wisdom inside and outside IGBP. This means, on the one hand, to make available the best integrative methodologies and, on the other hand, to include the systems and problems primarily investigated by the sister programmes WCRP and IHDP [and now, also DIVERSITAS].
3) GAIM is to implement Earth System analysis by organizing the construction, evaluation and maintenance of a hierarchy of Earth System models. This means, in particular, to help generate models of different degrees of complexity and to employ the resulting complementary ensemble for conducting virtual planetary experiments with respect to past, present, and future global changes.
Analysis of the outstanding problems to be solved in Earth System Science: The "Hilbertian Questions"
In response to the challenge of the first of the Waikiki Principles, GAIM developed a set of basic questions as a challenge for Earth System analysis in the 21st century. These questions are not limited in scope to those that can be answered by individual research projects, programs, or even communities. Rather, they help to define the overall context of global change science regardless of present ability to address the issues articulated therein. The meaning and implications of the questions are further explored in the GAIM Newsletter (Research GAIM, Summer, 2002). The total number of questions in these groups is 23. This is the number of challenges that David Hilbert listed in 1900 for 20th century mathematics, and so this could be viewed as a "Hilbertian" approach to Earth System analysis in the 21st century. These questions cannot in general be addressed within a single discipline, but rather will require interdisciplinary communication and collaboration throughout the scientific community, and with a wide range of researchers in the social sciences and humanities.
Analytic Questions:
1. What are the vital organs of the ecosphere in view of operation and evolution?
2. What are the major dynamical patterns, teleconnections and feedback loops in the planetary machinery?
3. What are the critical elements (thresholds, bottlenecks, switches) in the Earth System?
4. What are the characteristic regimes and time-scales of natural planetary variability?
5. What are the anthropogenic disturbance regimes and teleperturbations that matter at the Earth-System level?
6. Which are the vital ecosphere organs and critical planetary elements that can actually be transformed by human action?
7. Which are the most vulnerable regions under global change?
8. How are abrupt and extreme events processed through nature-society interactions?
Methodological Questions:
9. What are the principles for constructing “macroscopes”, i.e., representations of the Earth System that aggregate away the details while retaining all systems-order items?
10. What levels of complexity and resolution have to be achieved in Earth System modelling?
11. Is it possible to describe the Earth System as a composition of weakly coupled organs and regions, and to reconstruct the planetary machinery from these parts?
12. Is there a consistent global strategy for generating, processing and integrating relevant Earth System data sets?
13. What are the best techniques for analyzing and possibly predicting irregular events?
14. What are the most appropriate methodologies for integrating natural-science and social-science knowledge?
Normative Questions:
15. What are the general criteria and principles for distinguishing non-sustainable and sustainable futures?
16. What is the carrying capacity of the Earth as determined by humanitarian standards?
17. What are the accessible but intolerable domains in the co-evolution space of nature and humanity?
18. What kind of nature do modern societies want?
19. What are the equity principles that should govern global environmental management?
Strategic Questions:
20. What is the optimal mix of adaptation and mitigation measures to respond to global change?
21. What is the optimal decomposition of the planetary surface into nature reserves and managed areas?
22. What are the options and caveats for technological fixes like geoengineering and genetic modification?
23. What is the structure of an effective and efficient system of global environment & development institutions?
Integrated earth system modelling
Investigating the dynamic behavior and complexities of the Earth System remains a “grand challenge” for the scientific community. It is motivated by our limited knowledge about the consequences of large-scale perturbations of the Earth System by human activities such as fossil-fuel combustion or the fragmentation of terrestrial vegetation cover. During the past decades, marked progress has been achieved in modelling the separate elements of the geosphere and the biosphere, focusing on atmospheric and ocean circulation, and on land vegetation and ice-sheet dynamics. These developments have stimulated preliminary attempts to put all the separate pieces together on the basis of physical and biogeochemical processes. It has been the rule rather than the exception that surprising behavior has emerged when these components are coupled.
System-level models are being developed at three levels. The simplest models are those of the "Daisy-world" class, in which there is strong coupling and global scope, but very little detail geographically, or in the processes through which the various parts of the Earth System interact with each other. These simple models do not require international, multi-institutional coordination, and are thus not treated by GAIM explicitly.
At the other conceptual extreme are three-dimensional comprehensive models with coupled atmospheric and oceanic circulation in two-way interaction with marine and terrestrial ecosystems, human activity and policy, as well as the response of anthropogenic perturbations (emissions, land use) to perceived negative impacts of predicted global change. This highly non-linear problem can be presently addressed only for short time slices, and there are no models to date that incorporate robust formulations for societal responses to predicted global changes in the context of quantitative alterations of "business as usual" socioeconomic scenarios. There are also scaling and run-time problems presently being addressed, and there is much work to be done before practical and reliable prognostic models of this type emerge. However, for certain time slices, the spatially explicit treatment of the system afforded by such full-complexity models provides critical insights into the linkages between distinct biomes, synoptic systems, and human perturbations such as specific land use and emissions. In addition, these models can already be applied to specific limited applications.
It is becoming clear that one of the most fruitful approaches is on middle-ground, i.e., the development of Earth System Models of Intermediate Complexity (EMICs). These models capture the critical linkages between the various Earth subsystems, while being simple enough to run for long model times on present computers. With these intermediate models, it is possible to match spatial, temporal, and computational scales between subsystems, and also to scrutinize processes that occur at major boundaries such as the ocean-atmosphere, land-atmosphere, and land-ocean. In addition, with these models, it is possible to include socio-economic aspects, so that future anthropogenic global change drivers can be predicted. A great advantage of modelling at an intermediate level of complexity is that they can readily be constructed in a "balanced" manner, with equivalent contributions from marine, terrestrial, and atmospheric components. This alleviates both the mismatched scaling problems inherent in full-complexity models, as well as the problem of impractical run times on existing computers. Preliminary EMICs have already produced valuable results, such as the "prediction" of the rapid collapse of the Saharan ecosystem at 5.4 kybp (Claussen et al., 1999), and the mode switching of North Atlantic thermohaline circulation (Fig 1).
Earth System models can be constructed in two fundamental ways - by coupling existing subsystem models, and by developing integrated system models "from scratch". Both approaches have advantages and disadvantages. The insights from both will shed light on model sensitivities and weaknesses, and will point to the most fruitful directions for model refinement. As such, GAIM uses a two-pronged approach. The first is EMICs, which show great promise for highlighting the critical feedbacks and interactions that control the stability and sensitivity of the Earth System to anthropogenic perturbations. The second approach to system-level integration is coupling of full-complexity subsystem models. This approach introduces a different set of challenges in terms of space and timescale compatibilities, boundary condition specification, abandonment of flux corrections, and other issues inherent in the existing subsystem models. With the convergence of these two approaches in addition to support from further developments at the sub-system level, GAIM is working toward its ultimate goal of development of a workable suite of prognostic Earth System models.
Each of the above approaches is valuable in its own right and our proven techniques for model intercomparisons will help modelers of each "complexity" evaluate and refine their models. The greatest overall insights may come, in fact, from a comparison of the performance of each "complexity" relative to the others. Do full-complexity models provide qualitatively different results than EMICs due to their explicit spatial treatment? Are differences merely quantitative, and if so, what is the point of diminishing returns in terms of prognostic reliability for critical aspects of the Earth System (e.g. ice sheet collapse, ocean circulation modes, global precipitation patterns, major biome shifts, etc.). These questions are central to GAIM efforts for the next five years and will represent some of the first major prognostic results of integrated IGBP research.
We envision the emergence of a suite of coupled models, each with consistent coupling and interactions between model components, but each based on a different style of formulation. The parallel development of multiple Earth System models in parallel has several advantages. The first is that because no single model accounts for all processes and interactions in the Earth System, each model will necessarily result in slight differences in inter-component fluxes and sensitivities. This will set the stage for Earth System model intercomparison which will highlight the relative importance of the various processes, interactions, and feedbacks between subsystems modeled by each of the integrated models. This should ultimately lead to modified integrated models which correctly account for the interactions to which the Earth System is most sensitive, while becoming unburdened from those to which it is demonstrably insensitive.
Humans as an interactive component of the Earth System
A critical issue that must ultimately be addressed in any prognostic model is the anticipated alteration of human behavior in response to perceived impacts of global change. This bears on international emissions and land use policies, and is presently exceedingly difficult to predict based on numerical models of socio-economic systems. As people become aware of anthropogenic perturbation of the various ecosystems and of its consequences regarding the operation of the Earth System itself, they are likely to alter their behavior in ways that will affect land use and emissions, the two main anthropogenic forcing factors. As such, the problem moves from a simple scientific cause and effect problem, to a highly non-linear problem in which multiple players seek to optimize personal, regional or national economies while considering issues of "preservation of the commons" to a highly variable degree. If human behavior is altered by the knowledge generated by prognostic Earth System models, then the models themselves become invalidated unless they can be "self-inclusive." This phenomenon is comparable to one in quantum physics in which the observation of an entity causes a change in the nature of the entity itself. As such, there is an inherent “Heisenberg Uncertainty” in the human dimensions of Earth System science.
Let us consider international environmental policy for example: Developing countries may aspire to the same per capita resource exploitation opportunities afforded to the industrialized world in previous centuries. Given present population and projected growth, this would lead to unprecedented levels of deforestation, disruption of terrestrial ecosystems globally but most severely in tropical regions, extinction of a large number of plant and animal species, and changes in atmospheric chemistry and fresh water utilization at levels that would greatly accelerate global change processes. However, if the "players" are aware that such changes will have consequences for crop viability, sea level, and other issues important at the local level, they may call for changes in behavior, with each agreeing to make concessions only after concessions have been made by others. The human dimensions of the problem of determining the two-way interactions between ecosystems, climate, and human activities is thus one of the most complex problems facing modelers of the Earth System, but one that must be addressed before reliable predictions regarding future biospheric conditions and viability can be made.
Earth-System Models of Intermediate Complexity (EMIC)
Earth System models need to be globally comprehensive because the fluxes within the system are global (e.g., the hydrological cycle): changes in one region may well be caused by changes in a distant region. A currently open question is:
•How much spatial (regional) resolution is required to appropriately capture processes with global significance?
This question faces models at all levels of complexity, and must be addressed using as many approaches as are available. Earth System models probably need not capture all aspects of interaction between system components at the regional scale. The challenge is to determine the minimum model complexity necessary to account for the critical causes, effects, and interactive non-linearities at all spatial scales that affect global feedbacks.
In order to explore the relationships between system components that may control the behavior of the system as a whole, it is necessary to develop models that capture the critical feedbacks and interactions that drive the system to one mode of operation or another. It is also critical to be able to run such models for long model times to validate models, ensure model stability, and to bring to light long-term changes in system behavior. Current full-complexity models cannot operate over sufficiently long model times (thousands of years) for the complete Earth System - at least not within reasonable computer run times (days to weeks). Consequently it is necessary to develop EMICs that can run quickly on present machines and that can run for thousands of model years while capturing the critical interactions of components within the Earth System.
EMICs can be characterized in the following way: They describe most of the processes implicit in comprehensive models, albeit in a more reduced, i.e. a more highly parameterized form. They explicitly simulate the interactions among several components of the climate system including biogeochemical cycles. On the other hand, EMIC’s are simple enough to allow for long-term climate simulations for the Holocene or even glacial cycles. Similar to those of comprehensive models, but in contrast to conceptual models, the degrees of freedom of an EMIC exceed the number of adjustable parameters by several orders of magnitude. Tentatively, we may define an EMIC in terms of a three-dimensional vector: Integration, i.e. number of components of the Earth System explicitly described in the model, number of processes explicitly described, and detail of description of processes (Fig. 2).
Currently, there are several preliminary EMICs in operation including 2-dimensional zonally averaged models (e.g., Gallée et al., 1991), 2.5-dimensional models with a simple energy balance (e.g., Marchal et. al., 1998; Stocker et al., 1992), or with a statistical-dynamical atmospheric module (e.g. Petoukhov et al., 2000), and reduced-form comprehensive models (e.g., Opsteegh et al., 1998). Some preliminary EMICs have already been used for a number of paleostudies, because they provide the unique opportunity for transient, long-term ensemble simulations (e.g., Claussen et al., 1999) (Fig. 2). This is in contrast to so-called time slice simulations in which the climate system is implicitly assumed to be in equilibrium with external forcings (rarely a realistic assumption). EMICs have been used in preliminary explorations of the climate system's behavior under various scenarios of greenhouse gas emissions to assess the potential of abrupt changes in the system (e.g., Stocker and Schmittner, 1997; Rahmstorf and Ganopolski, 1999). However, these are all at a relatively early stage of development, and we are only beginning to compare and assess EMIC model results so that they may be subsequently refined and improved (Claussen et al 2002).
Full-Form Earth System Models: Coupled Carbon-Cycle Climate Model Intercomparison Project (C4MIP)
Full from models treat each part of the Earth system in the greatest possible detail, and include all known physical and biogeochemical processes. Although considerable work is underway, and some full-form models have shown great promise in their ability to handle the myriad interactions within the Earth system at high spatial and temporal resolution, it would be unrealistic at present to expect them to perform as working simulations of the Earth System. However, for short time slices and under certain conditions, such comprehensive models can be practical, and only this type of model can answer certain key questions about the Earth System and our understanding of the key processes that drive responses of the system to anthropogenic perturbations at the level of resolution required for regional issues or short-term drivers/impacts.
C4MIP introduces terrestrial and oceanic carbon cycle modules into coupled atmosphere-ocean-land climate models, with CO2 as a prognostic variable, to investigate the co-evolution of climate and CO2 given emission scenarios. The excitement lies in the identification and investigation of interactions in a climate space beyond known experience.
The goal of C4MIP is to evaluate the sensitivity of the coupled carbon-climate system to anthropogenic perturbations. The project focuses on CO2 emissions and concentration and the response of the Earth System to CO2 forcing, given a fixed scenario for future emissions (e.g. Rayner and Law, 1995). This "experiment" uses an increase in atmospheric CO2 concentration of 1%/yr. While this may be a modest increase relative to "business as usual" scenarios, it provides a useful baseline for this initial development and application of a full-complexity models.
The procedure is to solve simultaneously a coupled family of equations for different specifications of external source/sinks of CO2 and other greenhouse gases:
f CO2 / f t = Fba - Fab + Foa + External Sources/Sinks = f (climate)
f (climate) / f t = f (CO2, other GHGs)
where F is flux, b = terrestrial biosphere, o = ocean, a = atmosphere.
The project involves a control for the pre-industrial era with no external sources/sinks of CO2, and a forward integration from the pre-industrial to beyond AD2000 for a specified emission scenario for CO2 and the other greenhouse gases.
In the contemporary carbon budget, the fossil fuel source is ~5% of the one-way gross terrestrial or oceanic flux. Small annual carbon flux imbalances or errors, like air-sea heat and freshwater flux errors, if sustained over a long-enough time, may lead to significant climatic migrations. Surprises may emerge from non-linearities in terrestrial and oceanic carbon dynamics or from the climate system. Hence C4MIP could be viewed as an exploration of the non-linearities inherent in the Earth System.
One should treat C4MIP as a grand challenge to our understanding of the carbon cycle as well as of carbon-climate interactions. It should be the stimulus to take existing models to another level. While glimpses of realism should be expected, a simulation of the full operation of Earth system should not. Much is already being learned in the project during the process of model development, intercomparison, and refinement.
A look ahead
The IGBP is moving into a new phase of existence, with the conclusion and synthesis of most of the original Core Projects and an eye to the future based on the knowledge gained over the past decade or so. As IGBP builds on this knowledge in its second phase, integration of the knowledge gained will be necessary to make progress toward understanding the operation of the Earth System as a whole. GAIM's goal is to "promote the development of a suite of Earth System models" and this overall goal will not change as IGBP moves into its next phase. However, the types of activities that are conducted toward that goal will be changing at a most fundamental level reflecting new capabilities in addressing the Earth System. Whereas GAIM previously focused on modelling of specific subsystems (e.g. ocean, land, atmosphere) in the context of the carbon cycle, it has recently been turning its attention to Earth System level issues at an accelerating pace. This is both in response to abilities provided by advances in disciplinary research, and as a proactive step toward identifying the issues that may be important at the system level but remain inadequately explored within and beyond IGBP. As such, GAIM is attempting to act as a "lighthouse" to ensure that the research community is not missing any key issues that will prove to be important at the system level. The Hilbertian Questions are a first step in this "Analysis" role. In addition, GAIM's "Integration" role will involve working closely with the IGBP Core Projects and beyond to promote and conduct research activities that lie beyond the scientific scope of individual (or pairs) or Core Projects, but require input from across a wide range of IGBP and ESSP scientists. One such project in the planning stages is the construction of an Earth System Atlas in collaboration with IGBP, WCRP, IHDP and DIVERSITAS. The Atlas will serve as a medium through which the global change research results and data generated in the last decade or so can be disseminated to the broader community, with version of accompanying text directed to scientists, the general public and policy community, and school children. In addition to the educational and outreach aspects of the atlas, it will serve the scientific community by highlighting gaps in our understanding and/or observational knowledge of the Earth system. GAIM's "Modelling" activities will continue to expand into treatment of the broader Earth System, as enabled by the Analysis and Integration activities.
The GAIM Task Force
The GAIM Task Force is composed of scientists from all over the world who share the view of the Earth as a single system, but come from a wide variety of backgrounds and "disciplines." The Task Force is under the oversight of the Scientific Committee of the IGBP (SC-IGBP). Nominees are reviewed and selected by the Officers of the IGBP and serve at the behest of the International Council of Scientific Unions (ICSU). The GAIM Task Force for 2002 is comprised of the following scientists:
John Schellnhuber (Chair)
Eric Barron
Richard Betts
Martin Claussen
Wolfgang Cramer
Brad de Young
A. Scott Denning
Paul Falkowski
Pierre Friedlingstein
Inez Fung
Pavel Kabat
Maria Kanakidou,
Rik Leemans
Tim Lenton
Natalie Mahowald
John Mitchell
Ian Noble
Carlos Nobre
Wandera Ogana
Jim Orr
Colin Prentice
Peter Rayner
Arild Underdal
The IGBP/GAIM Office is under the direction of Dork Sahagian at the University of New Hampshire, and has been supported by NSF, NOAA, DOE and EPA.
References
Claussen, M., V. Brovkin, A. Ganopolski, C. Kubatzki, V. Petoukhov, and S. Rahmstorf, A new model for climate system analysis: Outline of the model and application to palaeoclimate simulations, Env. Modeling and Assessment, 4, 209-216, 1999.
Claussen, M., L. Mysak, A. Weaver, M. Crucifix, T. Fichefet, M. Loutre, S. Weber, J. Alcamo, V. Alexeev, A. Berger, R. Calov, A. Ganopolski, H. Goosse, G. Lohman, F.L. Mokhov, V. Petoukhov, P. Stone, and Z. Wang, Earth system models of intermediate complexity: closing the gap in the spectrum of climate system models., Clim. Dynam., 18, 579-586, 2002.
Gallee, H., J. Vanypersele, T. Fichefet, C. Tricot, and others, Simulation of the last glacial cycle by a coupled, sectorially averaged climate-ice sheet model 1. The climate model, J. Geophys. Res., 96, 13139-13161, 1991.
Opsteegh, J., R. Harsma, F. Selten, and A. Kattenberg, ECBILT: A dynamic alternative to mixed boundary conditions in ocean models, Tellus Ser. A, 50, 348-367, 1998.
Petoukhov, V., A. Ganopolski, V. Brovkin, M. Claussen, A. Eliseev, C. Kubatzki, and S. Rahmstorf, CLIMBER-2: A climate system model of intermediate complexity. Part 1: Model description and performance for present climate, Clim. Dyn., 16, 1-17, 2000.
Rahmtorf, S., and A. Ganopolski, Simple theoretical model may explain apparent climate instability, J. Clim., 12, 1349-1352, 1999.
Rahmstorf, S.: Warum das Eiszeitklima Kapriolen schlug, Spektrum der Wissenschaft, pp. 12-14, September 2001.
Stocker, T., and A. Schmittner, Influence of CO2 emission rates on the stability of the thermohaline circulation, Nature, 388, 862-865, 1997.
Stocker, T., D. Wright, and L. Mysak, A zonally averaged, coupled ocean atmosphere model for paleoclimate, J. Clim., 5, 773-797, 1992.
|
is Published by 
Figure 1: Three modes of thermohaline circulation during the last ice age. Top: Heinrich event. Center: Stable cold conditions that dominated glacial times. Bottom: Dansgaard-Oeschger event (modified after Rahmstorf, 2001).