Click for next page ( 17


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 16
2 Integrated Modeling of the Earth System OVERVIEW The possibility of major changes in the global environment presents the scientific research community with a difficult task: to devise ways of ana- lyzing the causes of and projecting the course of these shifts as they are occurring. Purely observational approaches are inadequate for providing the needed predictive or anticipatory information because response times of many terrestrial ecosystems are slow and there is a great deal of variability from place to place. Furthermore, many important processes cannot be measured directly over large areas, such as those processes that occur in soils. We need models to express our understanding of the complex sub- systems of the earth and how they interact with and respond to and control changes in the physical-climate and biogeochemical systems. By the year 2000, a fully coupled, dynamical model of the earth system (Figure 2.1) could be a reality. Such models would significantly improve capabilities for projecting changes in the earth system on a decadal time scale. The focus of this chapter is on the efforts required to achieve this goal. For instance, it is necessary to begin now to develop models that are more completely coupled albeit still partial-than those that are currently available. Even though these prototypes may themselves not be successful, This chapter was prepared by the working groups on Integrated Earth System Models established under the Committee on Global Change. Members of the group on Terrestrial-Atmosphere Modeling were Berrien Moore III, University of New Hampshire, Chair; John Aber, University of New Hampshire; Guy Brasseur, Na- tional Center for Atmospheric Research; Robert Dickinson, National Center for At- mospheric Research; William Emanuel, Oak Ridge National Laboratory; Jerry Melillo, 16

OCR for page 16
INTEGRATED MODELING OF THE EARTH SYSTEM cn I ~ i_ \ Biogeochemical Cycles 17 ~ E ~ cat a, c On 0 ~0 ~ ._ cn ~ a, c' ~ ~It -it 1 ' 1~' ~ | Tropospheric Chemistry 4 Pollutants | Atmosphenc Physics/Dynamics := ;! a!; IT Ti 1 ~ _ Terrestrial Or ean Dynamics Energy/Moisture ~ Marine l l Terrestrial Biogeochemistry I I Ecosystems | Global Moisture 1 ( Soil . ~ ~ ' ~' ~- Land _ Use Human Activities FIGURE 2.1 Status of earth system science in the year 2000 (ESSC, 1988~. they will teach us what is needed to realize our goal of a fully coupled, dynamical earth system model with a multidecadal scale of analysis. How- ever, it should not be overlooked that much of the real science is in the simple models and empirical observations that guide our understanding and give us a framework for interpreting (and creating) the more complex mod- els that evolve later. The early linking of complex models and the subse- quent addition of existing approaches should be balanced by efforts to cre- ate new, insightful simple models. Such insights provide the basis for qualitative improvements in model structures. It should be recognized at the outset that the muliidecadal temporal scale places important constraints and demands upon the character of earth sys- tem models (Bolin et al., 1986; ESSC, 1988; NRC, 1988~. For instance, the Marine Biological Laboratory; David Schimel, Colorado State University; Piers Sellers, University of Maryland; and Herman Shugart, University of Virginia. Members of the group on Ocean-Atmosphere Modeling were Berrien Moore III, University of New Hampshire, Chair; Mark Abbott, Oregon State University; Curt Covey, Lawrence Livermore National Laboratory; Nick Graham, Scripps Institution of Oceanography; Dale Haidvogel, Johns Hopkins University; Eileen Hoffman, Old Dominion University; Christopher Mooers, University of New Hampshire; James O'Brien, Florida State University; Albert Semtner, Naval Postgraduate School; and Leonard Walstad, Oregon State University. Members of the group on Atmospheric Physics-Atmospheric Chemistry were Berrien Moore III, University of New Hampshire, Chair; Guy Brasseur and Robert Dickinson, National Center for Atmospheric Research; Bill Gross, NASA Langley Research Center; and Chris Morris, University of New Hampshire.

OCR for page 16
18 RESEARCH STRATEGIES FOR THE USGCRP temporal scale demands inclusion of the biosphere and coupling across critical interfaces: terrestrial ecosystems and the atmosphere, the chemistry of the atmosphere and the physics of the atmosphere, and the oceans and the atmosphere. Advances at these interfaces are essential for progress. Differences in characteristic rates of change and fundamental processes of different components of the system will impose subsystem-specific de- mands and requirements on component models (Rosswall et al., 1988~. Ecological systems will most likely rest upon functional groups rather than species; understanding biogeochemical fluxes will require process-level models, but initial implementation at global scales will certainly require extensive pa- rameterization. Similarly, the nonlinear chaotic dynamics of the fluid sub- systems the oceans and atmosphere will continue to require a careful, step-by-step buildup in complexity; the simplistic thinking that must go into all initial modeling advances will tend to be eventually superseded by computationally intensive three-dimensional approaches. This is, in fact, occurring in many of the geophysical and biological-biogeochemical sci ences. The most complex models to date are the atmospheric and oceanic gen- eral circulation models (GCMs). These have structures largely determined by the need to solve the Navier-Stokes fluid equations, but they are rich in other physical processes as well. The atmospheric models and their climate role are especially strongly governed by water processes; however, it is precisely these aspects, including questions of scale and parameterization, that are among the least satisfactory of the models. Resolution is a problem in that the spatial scales of many of the impor- tant atmospheric water structures are poorly resolved by existing models. For example, many of the cloud systems that are most important for atmo- spheric radiation have vertical scales of less than the thickness of the layers in most existing GCMs. The horizontal structure of precipitating systems suffers not only from inadequate resolution but also from severe difficulties with the currently available numerical schemes that were designed prima- rily for effectiveness (minimal computational demands) in treating the model hydrodynamics. One obvious defect of these schemes is the tendency of truncated spectral series to give negative mixing ratios for water in high latitudes, a consequence of the failure of the series to represent properly the fields in going from relatively large mixing ratios to relatively small ones. The same difficulty can be encountered for any model tracer. For example, it was difficult to get models to treat global smoke fields properly in nuclear winter computations. The hope is that the new semi-Lagrangian schemes will cure these numerical difficulties. Another question in the treatment of water vapor in various atmospheric GCMs is whether vertical transport in the models resembles the process in nature, again because much of the real vertical transport occurs on scales that are small in comparison with that of the model. The subgrid-scale moist

OCR for page 16
INTEGRATED MODELING OF THE EARTH SYSTEM 19 convection parameterizations in the models are still fairly crude and have not improved much in the last decade, although considerable effort is now going into them (Anthes, 1983~. Adding the important chemical constituents and the reactions to an atmo- spheric GCM causes the issues of scale and computational challenges to become daunting. Many of the important chemical reactions are concentra- tion dependent and hence grid-scale dependent, and important processes often occur in the boundary layer, which generally is not well enough resolved. Further, the addition of atmospheric chemistry to a GCM places greater demands upon the terrestrial and oceanic boundary conditions and dynamic simulations (Lenschow and Hicks, 1989; NBC, 1984; Schimel et al., 1989~. In considering coupling atmospheric GCMs to terrestrial models, where the coupling transfers not only energy and water but also important gases, such as carbon monoxide, methane, and carbon dioxide for the carbon cycle, temporal- and spatial-scale issues again emerge. The macrobalance of ter- restrial carbon stocks, which determine the net flux of carbon dioxide, are difficult to derive by integrating across the short time scales at which en- ergy, water, and carbon dioxide and oxygen are actually exchanged because of the high degree of variability that these processes exhibit. Longer time step integrations have generally been more successful. On the other hand, the flux of methane and other short-lived species cannot be treated by simple mass balance and crudely time-averaged responses. Ecological changes, such as successional sequences of tree species, are not well treated on time steps that are appropriate for considering photon input and: water exchange or even trace gas fluxes and require some intermediate parameterization or model. The relatively simple coupling issue of land hydrology and atmosphere remains elusive, and yet it is quite important. The exchange of many re- duced gases (e.g., methane) depends on soil moisture conditions, and en- ergy fluxes are influenced by water balances. Modeling sensitivity studies have shown that if evapotranspiration were turned off over continental-scale areas, summer precipitation would be severely reduced and temperatures would be as much as 10 K higher than with normal fluxes. They also show that over tall vegetation the integrated resistance to transpiration implied by the stomata will have a major effect on Bowen ratios over the diurnal cycle. Since the rates of sensible heat exchange over the diurnal cycle determine the height reached by the planetary boundary layer as well as diurnal variations of precipitation in tropical and summer conditions, it is evident that it is important to include the role of vegetation in simulations of the hydrologi- cal cycle. Better field data are helping to establish the parameters needed for linking plant physiology to surface evapotranspiration, but considerable further effort is needed before the appropriate submodels can be applied with confidence over a wide range of vegetation cover (e.g., Dickinson, 1984; Eagleson, 1986; Sellers et al., 1986~.

OCR for page 16
20 RESEARCH STRATEGIES FOR THE USGCRP The coupling between the ocean and the atmosphere is central to the question of climate change. Atmospheric GCMs with prescribed oceans, long the mainstay of three-dimensional climate modeling, are inherently incapable of simulating the actual time-evolving response of the climate system to increasing greenhouse gases because this response involves heat uptake by the oceans. This is particularly clear when one realizes that the heat capacity of the atmosphere is roughly equivalent to that of the upper 3 m of the ocean. While it is true that the ocean may, partially, act in a passive manner, studies of the E1 Nino/Southern Oscillation (ENSO) show that the ocean-atmosphere system responds in a coupled fashion on interannual time scales, and paleo-oceanographic investigations suggest that aspects of longer-term climate change are associated with changes in the ocean's ther- mohaline circulation. The capability to predict these changes in circulation and heat exchange is necessary to describe the future evolution of global climate (e.g., Bryan et al., 1982; Cess and Goldenberg, 1981; IPCC, 1990; Sarmiento et al., 1988~. Fortunately, exciting and encouraging progress is being made in cou- pling key aspects of the major subsystems. Results from linking atmo- spheric and oceanic GCMs have already been reported in the literature and have shown significantly different behavior from that of simulations in un- coupled modes. Similarly, interactive simulations between atmosphere and land vegetation have been reported, and these have also exhibited new dynamical characteristics. The inclusion of biology in oceanic GCMs has begun, al- though the models are still simplistic and do not yet include climatic feedback in a coupled system. Representations of terrestrial biology are also preliminary and again without critical biogeochemical feedbacks. Finally, progress is being made toward model structures and data sets that will allow implemen- tation of atmospheric-oceanic-terrestrial models that include key biological- biogeochemical feedbacks. For the near term, developments in modeling the earth system should continue to focus on linking previously unlinked components, adding spe- cific subsystems to existing models (e.g., coupling oceanic and atmospheric GCMs or adding a marine biospheric model to an oceanic GCM), or im- proving existing linked treatments. In this spirit, the committee has arranged the following discussion around three interface models: 1. Atmosphere-terrestrial subsystem. 2. Physical-chemical interactions in the atmosphere. 3. Atmosphere-ocean subsystem including interactions with the biosphere. These three subsystems obviously overlap and do not include all interfaces. Further consideration is required on the issue of the role of the cryosphere and its coupling on multidecadal time scales (see OIES, 1989~. In the following sections the committee presents a brief general discus

OCR for page 16
INTEGRATED MODELING OF THE EARTH SYSTEM 21 sion of the current status of models at these three interfaces, including for each a focused report on recommended initiatives and themes. The two final sections deal with the cross-cutting issues of model tests and infra- structure. In order to provide perspective on the remainder of the chapter, the following considerations for each of the three interface models are pro- vided: For models that couple the terrestrial ecosystems and the atmosphere: The coupling must address questions such as how will a changing climate affect terrestrial carbon dioxide uptake and storage; how will . evapotransp~rat~on change; how will the distribution of vegetation and its seasonal pattern change; what are the effects on climate of changing pat- terns of vegetation, including large-scale deforestation; and what is the ef- fect of changing chemical conditions on terrestrial vegetation and trace gas exchange? The primary research issue in understanding the role of terrestrial ecosystems in global change is that of analyzing how processes with vastly differing rates of change, from photosynthesis to community change, are coupled to each other and to the atmosphere. Modeling these interactions requires coupling successional models to biogeochemical models to physiological models. Of these, only the physi- ological models can currently describe the exchange of water and energy between the vegetation and the atmosphere at fine time scales. . Terrestrial models should focus on linked models addressing plant community change, biogeochemistry, and physiology and~biophysics. Mod- els of the physics of the atmosphere couple directly to terrestrial physiology models; biogeochemical models serve as a bridge between physiology and community change as well as coupling to the chemistry of the atmosphere. The coupling must address how changes in the global environment, including the effects of land use and chemical stress, affect terrestrial eco- systems and how ecosystem changes affect the global system. Formidable problems of scale and parameterization are raised in three- and four-dimensional simulations of biology and atmospheric chemistry be- cause of nonlinear concentration-dependent phenomena. For models that couple physics and chemistry in the atmosphere: The coupling must address questions such as what is the spatial-tem- poral distribution of carbon monoxide, methane, and tropospheric ozone and how might it change; what is the effect of changing climatic or chemi- cal conditions on the aerosol-initiated stratospheric ozone depletion in the Arctic and the Antarctic; how might the exchange of water vapor between the troposphere and the stratosphere change in a changing climate; and what is the vertical transport of trace species by cloud convection and how might it change?

OCR for page 16
22 RESEARCH STRATEGIES FOR THE USGCRP Progress in the modeling of the coupled chemical-physical atmospheric system requires a better knowledge of surface sources of trace gases and their dependence on climatic conditions; chemical processes and reaction channels, both in the gas and in the aqueous phase, and their dependence on atmospheric conditions; and transport processes by advection and convec- tion, including the development of high-resolution transport models coupled to atmospheric GCMs with detailed representation of physical processes including cloud formation and associated transport, boundary layer trans- port, and troposphere-s~atosphere exchange. This progress is dependent on the acquisition of global data sets for validation of these treatments. Future progress will be dependent both on available computational resources and on progress in developing our understanding of fundamental physical and chemical processes and the nature of their coupling. For models that couple the ocean and the atmosphere: The coupling must address questions such as how will changing cli- mate affect oceanic carbon dioxide uptake and storage; how will oceanic heat storage and transport change; how will the amount and distribution of primary production change; how will the marine hydrological cycle change; and how will a changing ocean affect a changing climate? Critical issues include widely differing temporal and spatial scales, inclusion of biological and biogeochemical dynamics, and sparse data. Par- ticularly important and difficult tasks are the scaling of the biological-bio- geochemical components from local-regional domains to basin-global do- mains, formation of the upper mixed-layer physics, and inclusion of possible biological feedbacks on mixed-layer dynamics. Progress in the development of coupled oceanic-atmospheric models including biological-biogeochemical dynamics is limited, in part, by an in- adequate theoretical or observational understanding of certain key processes and a corresponding and continuing uncertainty as to how best to incorpo- rate or parameterize them in oceanic GCMs. . The set of field programs (JGOFS, WOCE9 the Coupled Oceans At- mosphere Research Experiment (COARE) organized under TOGA, and the Global Ocean Ecosystem Dynamics (GLOBEC)) required to acquire the data needed to advance our knowledge of fundamental oceanic processes is already well defined. These programs also offer valuable opportunities for simultaneous observational efforts, and these should be encouraged. The development of fully coupled models should be encouraged along two parallel paths: the first devoted to developing basin- and global-scale models with increasing levels of coupling, and the second leading to a series of regional fine-scale models that could provide boundary conditions and parameterization tests for the larger-scale models. Several overarching issues exist regarding approaches to and testing of models and the infrastructure necessary for their development:

OCR for page 16
INTEGRATED MODELING OF THE EARTH SYSTEM . 23 Validation is extremely difficult; models should be subjected to natu- rally occurring perturbation tests that exercise the coupling. In addition, large-scale phenomena offer a valuable opportunity for focusing model de- velopments and testing model dynamics. Studies of these large-scale pro- cesses will serve not only as diagnostic tests but also as prognostic tools. It is urgent that testing of models and model combinations begin as soon as possible. Experiments with global models will initially use simple representations, but the lessons learned and data bases developed will be critical to future improvements. Prototype global experiments will be espe- cially important to exploring feedbacks between the production of long- lived trace gas species and climate. Two important themes are important in early testing of partial earth system models: the global carbon cycle (carbon dioxide, methane, and carbon monoxide) and the transient response to a changing greenhouse forc- ing. The former exercises the chemistry and biology, whereas the latter stresses the physics and biology. The obvious next step is coupling these . . two themes. . The importance of experience gained through prototype modeling ex- periments, including failure, should not be underestimated. Careful analy- sis of failures can provide valuable information. Earth system modeling should serve as a focus and catalyst for inter- disciplinary science. No one institution or group of investigators has more than a fraction of the interdisciplinary talent necessary for the development of an earth system model focused on multidecadal time scales. Thus sev- eral teams and talented individuals should be supported, who with some coordination could help perform the incremental steps toward the integrated earth system model. Some of these groups may act primarily as synthesiz- ers, their principal interest being in linking component pieces, while in other groups the interest would be in component development. Also needed in an overall modeling strategy are centralized facilities and associated staff to serve the common needs of the various teams and individuals and focus on issues of synthesis, continuity, documentation, and extensive numerical experiments. ATMOSPHERE-TERRESTRIAL SUBSYSTEM The primary research issue for coupling atmosphere-terrestrial models is understanding how processes with vastly differing rates of change, from photosynthesis to community change, are coupled. Representing this cou- pling in models is the central challenge to modeling the terrestrial biosphere as part of the earth system (e.g., Allen and Wyleto, 1984; Huston et al., 1988; King et al., 1990; Moore etal., 1989b, Smith et al., 1989~. Terrestrial ecosystems participate in climate and in the biogeochemical cycles on several temporal scales. The metabolic processes that are respon

OCR for page 16
24 RESEARCH STRATEGIES FOR THE USGCRP sible for plant growth and maintenance, and the microbial turnover associ- ated with dead organic matter decomposition, move carbon and water through rapid as well as intermediate time scale circuits in plants and soil. More- over, this cycle includes key controls over biogenic trace gas production. Some of the carbon fixed by photosynthesis is incorporated into plant tissue and is delayed from returning to the atmosphere until it is oxidized by decomposition or fire. This slower carbon loop through the terrestrial com- ponent of the carbon cycle, which is matched by cycles of nutrients required by plants and decomposers, affects the increasing Rend in atmospheric car- bon dioxide concentration and imposes a seasonal cycle on that trend (Fig- ure 2.2~. The structure of terrestrial ecosystems, which responds on even longer time scales, is the integrated response to the intermediate time scale carbon machinery. The loop is closed back to the climate system since it is the structure of ecosystems, including species composition, that sets the terrestrial boundary condition in the climate system from the standpoint of surface roughness, albedo, and, to a great extent, latent heat exchange. These separate temporal scales contain explicit feedback loops that may modify the system dynamics. Consider again the coupling of long-term climatic change with vegetation change. Climatic change will drive vegeta- tion dynamics, but as the vegetation changes in amount or structure, this 355 350 Q Q z o E z C) z o Cal 8 345 340 335 330 325 320 315 310 _ ~,.1., I ,l,,,,,l,,,,,l,,,,,l,,,,,l, l , lL,.,,l,, ,l l l l l l l , ,,,l,.,,,,,,,,,l,,,,,,, ~ 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 90 YEAR FIGURE 2.2 Concentration of atmospheric carbon dioxide in parts per million of dry air (ppm) versus time for the years 1958 to 1989 at Mauna Loa Observatory, Hawaii. The dots indicate monthly average concentration. (From C.D. Keeling et al. (1989). Copyright (3 1989 by the American Geophysical Union.)

OCR for page 16
INTEGRATED MODELING OF THE EARTH SYSTEM 25 will feed back to the atmosphere through changing water, energy, and gas exchange. Biogeochemical cycling will also change, altering the exchange of trace gas species. The long-term change in climate, driven by chemical forcing functions (carbon dioxide and methane) will drive long-term eco- system change. Modeling these interactions requires coupling successional models to biogeochemical models to physiological models that describe the exchange of water and energy between the vegetation and the atmosphere at fine time scales. There does not appear to be any obvious way to allow direct reciprocal coupling of GCM-type models of the atmosphere, which inherently run with fine time constants, to ecosystem or successional mod- els, which have coarse temporal resolution, without the interposition of a physiological model. This is equally true for biogeochemical models of the exchange of carbon dioxide and trace species. This cross-time-scale coupling is important and sets the focus for the modeling strategy. A Modeling Strategy: Prognosis for Progress Intuitively, we might develop a global model of terrestrial ecosystem dynamics by combining descriptions of each of the physical, chemical, and biological processes involved in the system. In such a scheme, longer-term vegetation changes would be derived by integrating the responses of rapidly responding parts of the model. But we cannot simply integrate models that describe the rapid processes of carbon dioxide diffusion, photosynthesis, fluid transport, respiration, and transpiration in cells and leaves in order to estimate productivity of whole plants, let alone entire ecosystems. The nature of the spatial averaging implied in the selection of parameters and processes to consider is difficult because of nonlinearities, which means that the choice of scale influences the calculation of averages (see Rosswall et al., 1988~. To progress in the development of terrestrial ecosystem models, we choose processes to treat in different models based on the phenomenological scales involved. As is common in physical models, terms in fundamental equa- tions can be included or ignored depending on the temporal and spatial scales of interest (e.g., ignoring gravitational effects in quantum physics and including Coriolis effects in large-scale fluid motion). Careful organi- zation of a suite of models, each describing processes that operate at differ- ent rates, is crucial to the practical development of terrestrial ecosystem models for use in earth system models of global change. Based on current model structures, atmosphere-biosphere interactions can be captured with simulations operating with three characteristic time con- stants (Figure 2.3~. The first level represents rapid (seconds to days) bio- physical interactions between the climate and the biosphere (Figures 2.4a and b). The dynamics at this level result from changes in water, radiation,

OCR for page 16
26 C L 1 M A T 1 C D R 1 V E R S RESEARCH STRATEGIES FOR THE USGCRP | TIME STEP 1 , SECONDS - DAYS TIME STEP DAYS - WEEKS TIME STEP 1 ANNUAL H2O EVAPOTRANS PI RATI ON ENERGY / WATER / CO2 LAI (SEASONAL) FOLIAR C / N (SEASONAL) HYDROLOGY / SOIL CHEMISTRY / TRACE GASES DECOMPOSITION / MINERALIZATION / UPTAKE LAl COTS) NPP TOTAL) DECOMPOSITION / MINERALIZATION / UPTAKE NET CARBON EXCHANGE / NET ECOSYSTEM PRODUCTION FIGURE 2.3 Three different time steps at which existing models of terrestrial ecosystems use climatic information to modify rates of ecosystem function. LEVEL 1 A 4) C H2O B HZO ~ ~ ~1! CO2 3- TOTAL PRIMARY r , PRODUCTION _ I TEMPERATURE - _ l | l _ WATER, LIGHT FAST TIME 1 _: 1 DECAY POOL FIGURE 2.4 Two diagrammatic representations of models converting short-time- step environmental data (minutes to hours) into balances of energy, water, and carbon. For these models, ecosystem structure, including leaf display and canopy structure, are fixed. Nutrient fluxes other than emission and consumption of trace gases are not dealt with.

OCR for page 16
56 RESEARCH STRATEGIES FOR THE USGCRP measurements will be required to define the subsurface variability (satel- lites do not detect many key processes, for example, new production and vertical fluxes) and to provide a baseline for satellite measurements. Cur- rently, an important concern is whether or not there will be an ocean color satellite in orbit during the major field campaigns (i.e., WOCE and JGOFS). This represents a potential critical gap in being able to link biological and physical phenomena. Data assimilation is an emerging reality in physical oceanographic mod- eling and observational studies and may be used advantageously by biologi- cal and biogeochemical oceanographers. This methodology aids in the in- terpolation of physical observations by adding dynamical constraints. While the quantity of physical data required to describe oceanic phenomena may be reduced by the use of data assimilative models, it is more likely that field estimates will be improved as a consequence of data assimilation. The first attempts are now being made to develop the techniques neces- sary to assimilate ocean color measurements into regional physical-biologi- cal models. This is a promising direction for the development of models that ultimately will have predictive capability for biological distributions in the ocean. One aspect that makes data assimilation into physical-biological models challenging is that updating one ecosystem component (e.g., phytoplankton from ocean color) requires that all other ecosystem components be adjusted so that they are in equilibrium with the updated field. More specifically, assimilative models will require estimates of the error fields of both the assimilated data sets and the processes that are being parameterized. This will allow quantitative estimates of the confidence in the forecast (or hindcast) fields being produced. For example, in assimilating ocean color data into a multicomponent ecosystem model, one needs to have an estimate of the errors in the satellite data in time and space (i.e., particularly those associ- ated with gap filling) as well as an estimate on the effect of zooplankton grazing within the model. Such error estimation will require synergy be- tween the modeling effort and the field programs. In spite of these difficulties, the initial attempts at assimilating ocean color data into physical-biological models have shown that the accuracy of the model is improved, but that the improvement of the model diminishes after a short time. The implication that data are needed at frequent intervals for assimilation into physical-biological models is a potential area of re- search that could be an important aspect of developing models to address problems of carbon dioxide uptake by the ocean. Specifically, given the inherent nonlinearity of biological processes as well as their occasional "switching circuit" behavior, present data assimilation techniques are inad- equate. Research into techniques involving nonlinear and nondifferentiable forms is needed.

OCR for page 16
INTEGRATED MODELING OF THE EARTH SYSTEM 57 From a more classical approach, biological models in which the flow field is set as a boundary condition have been in use for about 15 years for specific regional studies. Consequently, the dynamics and limitations in- herent in these models are beginning to be understood. If physical-biologi- cal models are to be developed to address the larger question of carbon dioxide uptake by the ocean, then the question arises as to how to extend the knowledge gained from regional, physically forced biological modeling studies as well as from geographically restricted fully coupled models to models developed for basin or global domains. While, in principle, it seems straightforward to simply increase the model domain, in practice, this is not so. At least three issues must be addressed: (1) how to link the biological dynamics to biogeochemical changes important for global carbon studies, (2) how much of the complexity that characterizes coastal biological sys- tems needs to be transferred to larger-scale domains, and (3) how to match the space and time scale requirements of coastal processes with those of larger-scale systems. These three issues represent fundamental problems that must be addressed if coupled oceanic-atmospheric-biogeochemical models are to be developed to investigate carbon uptake by the world oceans. Biogeochemical models link biology and chemistry at the level of nutri- ents and carbon dioxide and are generally based on the forcing of nitrate or phosphate fields. The full effect of the biology on the chemistry is gener- ally not included, nor is the full effect of the chemistry on the biology. Yet it is likely in a changing climate system that these processes may be impor- tant. We need to better understand the sensitivity of the climate system to changing biogeochemical systems, and, if found to be relevant, biogeochemical systems must be included in our climate models. Including these relation- ships is expected to be computationally expensive, and yet ignoring the interaction of the physical-chemical-biological systems may lead to poor predictions of climatic change. Local to regional three-dimensional, coupled oceanic-atmospheric-bio- logical models that use circulation fields obtained from sophisticated re- gional primitive equation circulation models to produce "predictions" of biological distributions are currently under development. The development of this type of model should be given particular encouragement since this development is essential for the realization of fully coupled global carbon cycle models. In particular, these regional models can be used to test and validate parameterizations and generalizations that are used in larger-scale models. The development of realistic fine-scale regional models is also desirable in that the output from these models can be used to specify the boundary conditions for the basin- and global-scale models. Perhaps the appropriate direction for modeling is along two parallel paths: one to develop basin- and global-scale models with increasing levels of coupling and the second to develop a series of regional fine-scale models that could provide

OCR for page 16
58 RESEARCH STRATEGIES FOR THE USGCRP boundary conditions and parameterizations tests for the larger-scale models. Each regional model could, thereby, include the complexity and dynamics appropriate for simulating the processes in a specific region, and at the same time the necessity for maintaining reasonable and consistent interfaces with the basin- and global-scale models would give the entire modeling effort an overall framework. Finally, much can be learned from extant models despite their limita- tions. Careful analysis of the sensitivity of the systems, which are approxi- mations to the climate and biogeochemical systems, will indicate the emphasis needed in observational and modeling studies. The importance of experience gained through modeling experiments, including failure, should not be un- derestimated. Failure, when carefully analyzed in the refereed literature, can be valuable to the scientific community as a whole. Summary The strategy is to acquire data through field experiments (e.g., TOGA/ COARE, WOCE, JGOFS; see chapter 7) designed to develop an under- standing of processes and a description of phenomena. Models may aid in the design and execution of these experiments as well as in the analysis and interpretation of the measurements. New understanding of key processes will be used to improve models and reduce or improve parameterizations. These model enhancements are expected to lead to the ability to describe biogeochemical and physical phenomena. This enhanced modeling ability should lead to improved climate estimates, including error bounds from which rational decisions may be made. Hence the development of the fully coupled models should be encouraged along two parallel paths: one to develop basin- and global-scale models with increasing levels of coupling and the second to develop a series of regional fine-scale models that could provide boundary conditions and parameterization tests for the larger-scale models. Each regional model could include the complexity and dynamics appropriate for simulating the processes in a specific region, and the neces- sity for maintaining interfaces with the larger-scale models would provide an overall consistent structure for model development. In seeking to develop models of the coupled atmosphere-ocean-marine biosphere and biogeochemical system, it is important, as mentioned earlier, to recognize the value of "great failure." Linking atmospheric-oceanic- biospheric models, even though costly in terms of human and computer resources, should begin sooner rather than later. These early, relatively primitive attempts will shed light on the difficult issues of scale, both spatial and temporal, and the associated questions concerning the degree to which vari- ous process complexities or details are required. This clarification may be of particular importance as we scale the biological-biogeochemical compo

OCR for page 16
INTEGRATED MODELING OF THE EARTH SYSTEM 59 nents from local-regional domains to basin-global domains, or as we seek to better define and formulate the upper mixed-layer physics and possible bio- logical feedback, such as shading due to phytoplankton blooms. At the least, such attempts will encourage the creation of needed infra- structure and will provide a basis for assessing better the required resources. Part of the infrastructure enhancement would be the establishment of mod- eling teams. Obviously, several parallel efforts will be needed. Validation of these models is both difficult and critical. The first step is to ensure that they reproduce major climate phenomena (e.g., the spring bloom and E1 Nino). Testing (not validating) the "interfacing" models as well as the earth system models that link them can be addressed in the U.S. Global Change Research Program. CRITICAL MODEL TESTS The earth system modeling program should include three interface mod- els, as well as models of the fully coupled system. This approach allows for the rapid development of science and its inclusion into the less computationally demanding (although still challenging) interface models. Also, certain av- enues of validation are open to the interface models that will be difficult to use for a full earth system model. The required abilities of the models and the critical tests needed before they can be used with confidence are dis- cussed below. As concepts are developed and tested in the interface models, they should be included into an evolving earth system model that will form the basis for long-term prediction. The Challenge and Critical Tests All of the models described below must be able to simulate system re- sponse to the forcing induced, for example, by a carbon-dioxide-equivalent doubling in the atmosphere. That is, all of the models must be able to simulate the transient response of oceans, ecosystems, chemistry, or physi- cal atmosphere to a change in physical climate induced by a greenhouse gas ~ . forcing. Other drivers and critical feedbacks (e.g., land surface albedo, clouds, and oceanic heat transport) should be included when developing physical climate scenarios for use as forcing functions. The forcing functions given to the interface models will evolve as tested concepts from the interface models are incorporated into the earth system model, presumably modifying its predictions of whole-system response to changing greenhouse forcing. Thus a continual interplay of interface and earth system models is required, allowing for cyclic validation, failure, and modification.

OCR for page 16
60 RESEARCH STRATEGIES FOR THE USGCRP Finally, validation is impossible in the classical con~ol-experiment mode. We have no other earth system to serve as a control, not to mention the difficulties imposed by the multidecadal time frame and policy-relevant aspects of this science. As a step in appraising these models, a series of critical tests is described below that exploit various data sets, including satellite data (e.g., Earth System Science Committee, 1988) and paleo-records (see chapter 3~. These "tests" of these interface models provide an evalua- tion of the models' capabilities prior to either their use in a predictive mode or their inclusion in an earth system model. The Interface Models Atmosphere-Terrestrial Subsystem The challenge for an interface model of the atmosphere-land biosphere is to predict responses to changes in such phenomena as water and energy exchange (and more generally the hydrological cycle per se), trace gas biogeochemistry, primary productivity and ecosystem carbon storage, and vegetation composition and structure due to the changing macroclimatic forcing andlor atmospheric chemical composition. Critical tests of this model prior to its use in the predictive mode will be to reproduce current patterns of biogenic trace gas and carbon exchange, using past and current climate as drivers; reproduce key aspects of coupling between the paleoclimate and pa- leoecological records within regions of interest; simulate contemporary spatial and seasonal patterns of vegetation properties, including primary productivity worldwide, using satellite indi- ces as validation data; capture patterns of ecological change along anthropogenically induced chemical gradients in the land component of this interface model; and simulate surface fluxes of radiation, especially solar, and including spectral surface albedos in the atmospheric component of this interface model. Adequate simulation of amounts and spatial and temporal distribu- tion of precipitation must also be addressed. Physical-Chemical Interactions in the Atmosphere For this interface model of the physical atmosphere and the chemical atmosphere, the challenge is to predict the change in chemical climate through

OCR for page 16
INTEGRATED MODELING OF THE EARTH SYSTEM 61 a macroclimatic transient forcing. This will, of course, require either simu- lation or forcing functions for the biospheric sources, which could be de- rived from the atmospheric-terrestrial-biospheric model. Critical tests will include simulation of contemporary variations in global trace gas fields, especially of"inte- grator species" such as methyl chloroform and carbon monoxide; methane and carbon dioxide concentrations and isotope ratios; large-scale tropospheric ozone features such as are observed in the tropics; high-latitude stratospheric ozone; and exchange of water vapor between troposphere and stratosphere. Atmosphere-Ocean Subsystem For the interface model of the atmosphere-ocean subsystem, the chal- lenge is to predict the responses of water and energy exchange, carbon dioxide exchange and carbon storage, pattern of the spring bloom, and shifts in ecosystem composition and resultant shifts in oceanic mixed- layer chemistry (e.g., alkalinity). The critical tests of such models are whether they can capture the key aspects of such large-scale phenomena as E1 Nino/Southern Oscillation (ENSO), North Atlantic spring bloom, cross-shelf exchange, poleward heat flux, and biogeochemistry of aeolian deposits. INFRASTRUCTURE . . . . .. . . . .. The global change modeling effort, particularly on these longer time scales, encompasses a class of scientific problems far broader than those cnaractenzed by the physical climate system alone. The biogeochemical system (see Figure 2.1) merits an equal emphasis. An overall strategy that favors diversity is required, in that no one institution or group of investiga- tors has more than a fraction of the interdisciplinary talent necessary for the complete task. Research teams in a range of sizes should be supported. Larger groups are needed for an overall integration role; smaller groups (5 to 10 people) would achieve the incremental steps (i.e., the linkages be- tween the interface models) toward integrated earth system models. Indi

OCR for page 16
62 i RESEARCH STRATEGIES FOR THE USGCRP vidual investigators will obviously also make important contributions along these same lines. Some of these groups may act primarily as synthesizers whose principal interest would be in linking component pieces; others would develop the components. The larger groups would most likely be associated with various central- zed facilities, which would also serve the common needs of the various teams for computational resources and linkages with large-scale models. Examples of such needs include the preparation and analysis of large-scale observational data sets (such as those that will be developed in preparation for the NASA Earth Observing System (EOS), not to mention the essential data set that EOS will provide following launch); operation of the large, computationally intensive GCMs; documentation and maintenance of baseline codes and protocols for information exchange used by the community; and diagnostics of model output. These larger facilities would provide the physical locations for the most capable supercomputers. One of the more intriguing advances in computer technology is in the area of massively parallel architectures. It appears that many of our current models may be recast to operate in a parallel mode. Centralized facilities could devote resources to this longer-term investment that would be diffi- cult for smaller modeling teams to provide. Given the rapid improvements in CPU power, especially in the worksta- tion class and with parallel architectures, the gap between hardware and software is increasing. Many of the model codes that we use are many years old, and it is difficult to find the funds or the researchers required to convert such codes to take advantage of new hardware. In addition, many of the codes are unwieldy and poorly documented. Again, this is an area to which larger, central teams could commit resources. Specifically, more effort needs to be placed on software development than simply applying existing codes in faster machines. To aid in this process, all modeling teams, large and small, should be encouraged to take advantage of various debugging and software develop- ment tools. For example, code profilers that aid in parallelizing or vector- ing codes should be used. Object-oriented methods that allow codes to be reused or reconfigured more easily should be incorporated into new models. One of the limitations of such efforts is that such tools are often cumber- some and difficult to learn. Thus it is essential that new partnerships be formed between the various hardware and software vendors and the scien- tific users so that appropriate tools can be developed. Lastly, the realm of visualization is becoming increasingly important in handling the volume and increasing dimensionality of the data sets. Visual- izat~on tools also need to be made more accessible. In addition to facilitat- ing the analysis of the model output, such tools can play an important role in testing and debugging by allowing the modeler to see every time step of

OCR for page 16
INTEGRATED MODELING OF THE EARTH SYSTEM 63 the model, rather than relying on summary data sets. Visualization also requires close coupling between He model and a data base system to Rack the model output. Clearly, these centers and the smaller modeling groups and individual investigators must be mutually supporting. Smaller groups must be pro- vided with the capability to run process experiments on full GCMs and full earth system models at larger centers; moreover, they must have on-site computer support, including workstations, advanced graphics, geographical information systems, and ma~nfrarnes and, most importantly, the technical staff to allow a full application of the on-site computer facilities as well as the off-site supercomputers. Correspondingly, the centers must be able to incorporate advances in subsystem and interface representations formulated by the smaller modeling groups. Such a strategy must necessarily involve multiagency support over many years. REFERENCES Aber, J.D., J.M. Melillo, and C.A. Federer. 1982. Predicting the effects of rotation length, harvest intensity, and fertilization on fiber yield from northern hardwood forests in New England. Forest Sci. 28~1~:31-45. Allen, T.F.H., and E.P. Wyleto. 1984. A hierarchical model for the complexity of plant communities. J. Theor. Biol. 101:529-540. Anthes, R.A. 1983. Regional models of the atmosphere in middle latitudes (a review). Mon. Weal Rev. 111: 1306-1335. Bass, A. 1980. Modeling long-range transport and diffusion. Pp. 193-215 in Conference Papers, Second Joint Conference on Applications of Air Pollution Meteorology, New Orleans, La. Bolin, B., and R.B. Cook (eds.~. 1983. SCOPE 21: The Major Biogeochemical Cycles and Their Interactions. John Wiley and Sons, Chichester, England. Bolin, B., A. Bjorkstrom, K. Holmen, and B. Moore. 1983. The simultaneous use of tracers for ocean circulation studies. Tellus 35B:206-236. Bolin, B., B.R. Doos, J. Jager, and R. Warrick (eds.~. 1986. SCOPE 29: The Greenhouse Effect, Climatic Change and Ecosystems. John Wiley and Sons, Chichester, England. Botkin, D.B., J.F. Janak, and J.R. Wallis. 1972a. Some ecological consequences of a computer model of forest growth. J. Ecol. 60:849-873. Botkin, D.B., J.F. Janak, and J.R. Wallis. 1972b. Rationale, limitations, and as- sumptions of a northeastern forest growth simulator. IBM J. Res. Develop. 16: 101-116. Broecker, W.S., and T.-H. Peng. 1982. Tracers in the Sea. Lamont-Doherty Geological Observatory, Columbia University, New York, N.Y. Bryan, K., F.G. Komro, S. Manabe, and M.J. Spelman. 1982. Transient climate response to increasing atmospheric carbon dioxide. Science 215:56-58. Cess, R.D., and S.D. Goldenberg. 1981. The effect of ocean heat capacity on global warming due to increasing carbon dioxide. J. Geophys. Res. 86:498-602.

OCR for page 16
64 RESEARCH STRATEGIES FOR THE USGCRP Demerjian, K.L. 1976. Photochemical Diffusion Models of Air Quality Simulation: Current Status. Assessing Transportation-Related Impacts. Special Report 167. Transportation Research Board, National Research Council, Washington, D.C., pp. 21-33. Dickinson, R.E. 1984. Modeling evapotranspiration for three-dimensional global climate models. Pp. 58-72 in J.E. Hansen and T. Takahashi (eds.), Climate Processes and Climate Sensitivity. Geophysical Monograph 29. American Geophysical Union, Washington, D.C. Eagleson, P.S. 1986. The emergence of global-scale hydrology. Water Resour. Res. 22:6S-14S. Earth System Sciences Committee. 1988. Earth System Science: A Closer View. National Aeronautics and Space Administration, Washington, D.C. Emanuel, W.R., G.G. Killough, W.M. Post, and H.H. Shugart. 1984. Modeling terrestrial ecosystems and the global carbon cycle with shifts in carbon storage capacity by land use change. Ecology 65:970-983. Farquhar, G.D., and S. van Caemmer. 1982. Modeling of photosynthetic response to environmental conditions. In O.L. Lange, P.S. Nobel, C.B. Osmond, and H. Ziegler (eds.), Physicological Plant Ecology, Encycl. Plant Physiol. (NS) 12B:549-587. Springer, Berlin. Hanks, R.J. 1985. Soil water modeling. In M.G. Anderson and T.P. Burt teds.), Hydrological Forecasting. John Wiley and Sons, Chichester, England. Houghton, R.A., J.E. Hobbie, J.M. Melillo, B. Moore, B.J. Peterson, G.R. Shaver, and G.M. Woodwell. 1983. Changes in the carbon content of terrestrial biota and soils between 1860 and 1980: A net release of carbon to the atmosphere. Ecological Monographs 53:235-262. Huston, M., D.L. DeAngelis, and W.M. Post. 1988. New computer models unify ecological theory. BioScience 38:682-691. Intergovernmental Panel on Climate Change. 1990. Scientific Assessment of Cli- mate Change. Draft report. World Meteorological Organization and United Nations Environment Program. King, A.W., W.R. Emanuel, and R.V. O'Neill. 1990. Linking mechanistic models of tree physiology with models of forest dynamics: Problems of temporal scale. Forest Growth: Process Modeling of Responses to Environmental Stress. Timber Press, Auburn University, Auburn, Ala. Lenschow, D.G., and B.B. Hicks (eds.~. 1989. Global Tropospheric Chemistry: Chemical Fluxes in the Global Atmosphere. Report of the Workshop on Measurements of Surface Exchange and Flux Divergence of Chemical Species in the Global Atmosphere. National Center for Atmospheric Research, Boul- der, Colo. 107 pp. Logan, J.A., M.J. Prather, S.C. Wofsy, and M.B. McElroy. 1981. Tropospheric chemistry: A global perspective. J. Geophys. Res. 86:7210-7254. Maier-Reimer, E., and K. Hasselman. 1987. Transport and storage in the ocean- an inorganic ocean-circulation carbon storage cycle model. Climate Dynamics 2:63-90. Manabe, S., and R.T. Wetherald. 1975. The effects of doubling the CO2 concentration on the climate of a general circulation model. J. Atmos. Sci. 32:3-15.

OCR for page 16
INTEGRATED MODELING OF THE EARTH SYSTEM 65 Markar, M.S., and R.G. Mein. 1987. Modeling of evapotranspiration from homogeneous soils. Water Resour. Res. 23:2001-2007. Melillo, J.M., J.D. Aber, and J.F. Muratore. 1982. Nitrogen and lignin control of hardwood leaf litter decomposition dynamics. Ecology 63:621-626. Moore, B., B. Bolin, A. Bjorkstrom, K. Holmen, and C. Ringo. 1989a. Ocean carbon models and inverse methods. Pp. 409-449 in D.L.T. Anderson and J. Willegrand feds.), Oceanic Circulation Models: Combining Data and Dynam- ics. Kluwer Academic Publishers. Moore, B., M.P. Gildea, C.J. Vorosmarty, D.L. Skole, J.M. Melillo, B.J. Peterson, E.B. Rastetter, and P.A. Steudler. 1989b. Biogeochemical cycles. Pp. 113- 141 in M.B. Rambler, L. Margulis, and Rene Pester (eds.), Global Ecology: Towards a Science of the Biosphere. Academic Press, Inc. National Research Council. 1984. Global Tropospheric Chemistry. National Academy Press, Washington, D.C. National Research Council. 1988. Toward an Understanding of Global Change: Initial Priorities for U.S. Contributions to the Intemational Geosphere-Biosphere Program. National Academy Press, Washington, D.C. Oeschger, H., U. Siegenthaler, and A. Gugelman. 1975. A box diffusion model to study the carbon dioxide exchange in nature. Tellus 27:168-192. Office for Interdisciplinary Earth Studies. 1985. Opportunities for Research at the Atmosphere/Biosphere Interface. OIES, Boulder, Colo. Office for Interdisciplinary Earth Studies. 1989. Arctic Interactior~s. OIES, Boulder, Colo. Parton, W.J., J.W.B. Stewart, and C.V. Cole. 1988. Dynamics of C, N. P. and S in grassland soils: A model. Biogeochemistry 5:109-131. Pastor, J., and W.M. Post. 1985. Development of a linked forest productivity-soil carbon and nitrogen model. ORNL/TM-9519. Oak Ridge National Laboratory, Oak Ridge, Tenn. 162 pp. Pastor, J., and W.M. Post. 1986. Influence of climate, soil moisture, and succession on forest carbon and nitrogen cycles. Biogeochemistry 2:3-27. Pastor, J., and W.M. Post. 1988. Response of northern forests to CO2-induced cli- mate change. Nature 344:55-58. Riley, G.A. 1951. Oxygen, phosphate, and nitrate in the Atlantic Ocean. Bulletin of the gingham Oceanographic Collection, XIII, Article 1. Yale University, New Haven, Conn. Rosenzweig, C., and R. Dickinson (eds.~. 1986. Climate-Vegetation Interactions. Report DIES-2. Office for Interdisciplinary Earth Studies, Boulder, Colo. Rosswall, T., R.G. Woodmansee, and P.G. Risser (eds.~. 1988. SCOPE 35: Scales and Global Change: Spatial and Temporal Variability in Biospheric and Geospheric Processes. John Wiley and Sons, Chichester, England. Running, S.W., and J.C. Coughlan. 1988. A general model of forest ecosystem processes for regional applications. I. Hydrological balance, canopy gas exchange, and primary production processes. Ecological Modeling 42:125- 154. Sarmiento, J.L., J.R. Toggweiler, and R. NaJuar. 1988. Ocean carbon-cycle dynamics and atmospheric pCO2. Philos. Trans. R. Soc. London A325:3-21.

OCR for page 16
66 RESEARCH STRATEGIES FOR THE USGCRP Schimel, D.S., M.A. Stillwell, and R.G. Woodmansee. 1985. Biogeochemistry of C, N. and P in a soil catena of the shortgrass steppe. Ecology 66:276-282. Schimel, D.S., M.O. Andre ae, D. Powler, I.E. Galbally, R.C. Harriss, D. Ojima, H. Rodhe, T. Rosswall, B.H. Svensson, and G.A. Zavarzin. 1989. Priorities for an international research program on trace gas exchange. Pp. 321-331 in M.O. Andreae and D.S. Schimel (ads.), Exchange of Trace Gas Between Terrestrial Ecosystems and the Atmosphere. John Wiley and Sons, Chichester, England. Schlesinger, M.E. 1983. Simulating CO2-induced climate change with mathemati- cal climate models: Capabilities, limitations, and prospects. In Proceedings of the Carbon Dioxide Research Conference: Carbon Dioxide, Science and Consensus (CONF-820970), U.S. Department of Energy, Washington, D.C. Sellers, P.J., Y. Mintz, Y.C. Sud, and A. Dalcher. 1986. A simple biosphere model (SIB) for use within general circulation models. J. Atmos. Sci. 43:505-531. Shugart, H.H. 1984. A Theory of Forest Dynamics. Springer-Verlag, New York. Shugart, H.H., and D.C. West. 1980. Forest succession models. BioScience 30:308- 313. Shugart, H.H., M.Y. Antonovsky, P.G. Jarvis, and A.P. Sandford. 1986. CO2, cli- matic change and forest ecosystems. Pp. 475-521 in B. Bolin, B.R. Doos, J. Jager, and R. Warrick (eds.), SCOPE 29: The Greenhouse Effect, Climatic Change and Ecosystems. John Wiley and Sons, Chichester, England. Smith, T.M., H.H. Shugart, D.L. Urban, W.K. Lauenroth, D.P. Coffin, and T.B. Kirchner. 1989. Modeling vegetation across biomes: Forest-grassland tran- sition. Pp. 290-241 in E. Sjogren (ed.), Forests of the World: Diversity and Dynamics (abstracts). Studies in Plant Ecology 18:47-49. Solomon, A.M. 1986. Transient response of forests to CO2-induced climate change: Simulation modeling experiments in eastern Nor~ America. Oecologia 68:567- 579. Solomon, A.M., M.L. Tharp, D.C. West, G.E. Taylor, J.M. Webb, and J.C. Trimble. 1984. Response of unmanaged forests to CO2-induced climate change: Avail- able information, initial tests, and data requirements. U.S. Department of En- ergy, Washington, D.C. Toggweiler, J.R., K. Dixon, and K. Bryan. 1989. Simulations of radiocarbon in a coarse-resolution world ocean model. 1. Steady state prebomb distributions. J. Geophys. Res. 94:8217-8242. Vorosmarty, C.V., B. Moore, A.L. Grace, M.P. Gildea, J.M. blelillo, B.J. Peterson, E.B. Rastetter, and P.A. Steudler. 1989. Continental scale models of water balance and fluvial transport: An application to South America. Global Biogeochemical Cycles 3~3~:241-265. World Climate Research Program. 1990. Scientific Plan for GEWEX. May 1990 Draft Report. World Meteorological Organization, Geneva. Wunsch, C. 1978. The general circulation of the Nor~ Atlantic west of 50W degrees determined from inverse methods. Rev. Geophys. Space Phys. 16:583- 620.