January -March Issue


Predicting Global Climate Change: New Tools, New Insights

new climate-change model developed at MIT provides scientists and policymakers with an unprecedented ability to analyze potential impacts of human activities and proposed policies on the global climate system. The model simulates economic growth and associated emissions, flows of greenhouse gases into and out of land and oceans, chemical reactions in the atmosphere, climate dynamics, and changes in natural terrestrial ecosystems. It also includes previously ignored interactions among those processes--interactions that can have a substantial impact on predicted outcomes. For example, most models assume that emissions of methane and nitrous oxide from wetlands and soils are constant. But MIT simulations show that predicted rises in temperatures will cause natural emissions of those greenhouse gases to increase by 15%-60%. Other results suggest that when carbon dioxide levels become very high, the oceans may be a less effective "sink" for that gas than is often predicted. By simplifying certain components, the researchers made the model "computationally efficient" without compromising its ability to represent key climate phenomena. Performing long-term simulations and repeated runs with changed assumptions is therefore economically feasible. A series of runs using a range of plausible assumptions regarding human emissions, ocean circulation, and cloud cover suggests that predictions of temperature change are even more uncertain than generally believed. The researchers are now using the model to clarify key sources of uncertainty and to examine the effectiveness and costs of proposed policy options.

Policies to address global climate change are the subject of much national and international debate. How serious is the climate-change problem? What level of response is appropriate? And what types of policies would be effective in alleviating any potential threat? Answering those questions is difficult because connections among human behavior, emissions, and natural systems are so complex. Emissions of greenhouse gases depend on economic, technological, and political forces. How those emissions then affect climate depends on complicated, interacting phenomena in the atmosphere, the oceans, and land ecosystems. And any change in climate in turn influences all aspects of the system. Until we better understand this elaborate global system, we cannot reliably predict the impacts of continuing human emissions, nor can we predict the effectiveness of specific policies being considered.

Much climate-change research focuses on only one discipline and one component of the global system. In contrast, MIT's Integrated Global System Model (IGSM) both describes all the components and integrates them. This inclusive approach is made possible by the interdisciplinary nature of the large research team involved. The team includes experts in economic growth, technological change, and emissions; climate chemistry and physics; and the biology of ecosystems.

The specialists are brought together by the MIT Joint Program on the Science and Policy of Global Change, which draws on two MIT centers-the Center for Energy and Environmental Policy Research (CEEPR) and the Center for Global Change Science (CGCS). Added breadth is provided by participants from the Ecosystems Center at the Marine Biological Laboratory (Woods Hole, Massachusetts), other MIT departments, and industrial groups. Leading the effort are Ronald G. Prinn, TEPCO Professor of Atmospheric Chemistry and director of the CGCS, and Henry D. Jacoby, William F. Pounds Professor of Management and former director of the CEEPR.

In developing the IGSM, the MIT team had three goals. The model should permit researchers to examine the science of global climate change and the interactions among the many components involved. It should help clarify the effects and costs of proposed policies such as emissions quotas and taxes. And it should help investigators figure out what makes predictions of global climate change--both its magnitude and its timing--so uncertain. Many of the processes involved are not well understood, so modeling them requires making assumptions and approximations. Such uncertainty propagates through the integrated model in complicated ways, ultimately contributing to the uncertainty of the final result. The IGSM should enable scientists to identify and explore key sources of the uncertainty that now hinders both scientific understanding and policy assessment.

To perform all those functions, the IGSM must simulate a huge number of physical and chemical phenomena and track many individual chemical species, accounting for their behaviors and their impacts on climate. Moreover, its calculations must be resolved geographically. Amounts and types of emissions vary with latitude and are expected to change over time, shifting from Europe and North America to China and southern Asia during the next century. Temperatures and other climatic conditions and the nature of the earth's surface also vary with location, influencing both emissions and chemical reactions in the atmosphere. Because of such regional differences, the impacts of climate change and the effects of a given policy will differ dramatically from place to place.

After five years' work, the MIT team has developed a model that begins to provide the needed level of completeness and detail. The diagram below outlines the key models (in boxes) and information flows (in ovals) of the IGSM.

The Anthropogenic Emissions Prediction and Policy Analysis Model (at the top of the diagram) forecasts future anthropogenic emissions of carbon dioxide (CO₂), sulfur oxides (SO_X), nitrous oxide (N₂O), and other key gases. The model covers the period 1985 to 2100 in five-year steps and recognizes twelve regions of the world, linked by trade. For each region it tracks population change, technological change, economic growth, and the energy intensity of economic activity. Based on those forces, it predicts the magnitude, composition, and location of future anthropogenic emissions.

The Natural Emissions Model (at the left) predicts emissions from natural sources based on estimates of future temperatures, rainfall, and other climatic variables and predictions of the condition of land ecosystems, supplied by the Terrestrial Ecosystems Model (at the bottom). It recognizes different types of soils and wetlands and their geographical locations and determines the natural uptake and emission of N₂O from soils and methane (CH₄) from wetlands.

Predictions of anthropogenic and natural emissions (resolved by latitude) feed directly into the Coupled Atmospheric Chemistry and Climate Model (at the center). The climate portion of this integrated model simulates global climate dynamics including critical physical processes such as radiation and convection and the effects on those processes of all significant greenhouse gases and a variety of aerosols (suspended particulate matter, the most important being sulfate aerosols resulting from SO_X emissions). It also includes the crucial role of ocean circulations, which can remove both heat and CO₂ from the atmosphere. It differentiates between climate dynamics over land and ocean at each latitude. An interactive atmospheric chemistry submodel includes more than 50 chemical reactions and recognizes 25 chemical species. The passage of CO₂ into and out of the ocean is also calculated. Based on temperatures and other outputs of the climate model, the chemistry submodel determines the formation and destruction of important greenhouse gases and aerosols and feeds the predictions back into the climate model.

Finally, the Terrestrial Ecosystems Model predicts how changes in climate and atmospheric composition will affect the state of terrestrial ecosystems--one measure of the potential impact of global climate change. The model, developed by collaborators at the Marine Biological Laboratory, simulates fundamental biogeochemical processes in 18 terrestrial ecosystems around the world and makes monthly estimates of key carbon and nitrogen fluxes. The predicted uptake of CO₂ by land ecosystems feeds back into the chemistry submodel, and predicted soil compositions become inputs for the Natural Emissions Model.

The IGSM thus simulates all the relevant processes and feedbacks among them. And it can do so quickly--an important feature for scientific studies and policy assessments, which may require repeated runs using different variables and simulating extended periods of time. A traditional model may take months even on a supercomputer to simulate climate change over a century. The IGSM incorporates carefully chosen simplifications that make it much faster. For example, in initial tests the coupled chemistry/climate model--which considers latitude and altitude but not longitude--performed simulations twenty times faster than did more complicated models that also consider longitude. Yet the climate change patterns predicted by the two types of models were similar. And despite the simplifications, the coupled chemistry/climate model reproduces the major features of today's climate reasonably well, including seasonal variations.

Already the researchers have identified some feedbacks that have important impacts but are not addressed by other models designed for policy studies.

• Climate feedbacks on natural emissions: Results show that land ecosystems play an important role in taking up carbon and emitting CH₄ and N₂O. Further, interactions between the coupled models show that those fluxes are significantly altered by changes in climate and in atmospheric CO₂. For example, predicted increases in temperatures and precipitation will substantially increase natural emissions of CH₄ and N₂O. Indeed, the increase in natural emissions of N₂O may be as large as the increase in human emissions of that gas. Assuming that natural emissions are constant--as do most models now used in policy discussions--is therefore wrong and leads to significant underestimates of the future roles of those gases.
  
  
• Removal of sulfate aerosol: Among the emissions from coal-burning power plants are sulfate aerosols that remain suspended in the atmosphere. While aerosols are troublesome pollutants, they play a positive role in terms of climate change: they reflect incoming sunlight back to space, thereby cooling the earth. Model results suggest that aerosols may not be as abundant in the future as is generally assumed, as SO_X control schemes are put in place and as more aerosols may be washed out of the atmosphere by climate-change-induced increases in future precipitation.
  
  
• Ocean uptake of carbon: Oceans are known to be an important sink for CO₂, but here again traditional climate models generally omit an important feature: the effect of the ocean's pH on CO₂ uptake. Rather than assuming a given pH that remains constant, the ocean submodel in the IGSM explicitly predicts the ocean's pH. As levels of CO₂ in the ocean increase, the water becomes more acidic; and as acidity increases, the water's capacity for holding CO₂ decreases. Total carbon uptake by the ocean thus decreases. This more realistic approach yields significantly different oceanic carbon uptake rates over the next century than do simpler schemes.
  
  

The researchers have also begun examining sources of uncertainty by performing a series of sensitivity studies. For example, they have tested the effects of using different values for input parameters influencing three key components of climate change: anthropogenic emissions, ocean circulation, and climate sensitivity. All three are potentially important but not well understood. The level of anthropogenic emissions depends on hard-to-predict factors such as population growth and technological change. The behavior of the ocean is not well understood, yet the rate at which surface waters circulate to the ocean depths is critical. The more rapid the circulation, the more quickly heat and CO₂ will be removed from the atmosphere, slowing warming. The third parameter--climate sensitivity--indicates how much warming will occur for given levels of CO₂ and other greenhouse gases in the atmosphere. One of the most uncertain factors affecting climate sensitivity is cloud cover. Clouds reflect incoming sunlight, so their presence lowers the temperature. But they also redirect outgoing infrared radiation back to the earth's surface, thereby raising the temperature. The net impact is highly uncertain.

The figure below shows the effect on temperature predictions of using different "reasonable" assumptions about the underlying input parameters. The middle curve--the reference case--represents the outcome using assumptions consistent with estimates from the Intergovernmental Panel on Climate Change (IPCC). The higher curve reflects a carbon-intense world, a slowly overturning ocean, and high positive feedback because of water vapor and cloud behavior. The lower curve reflects a cleaner world, a faster ocean overturn, and lower cloud feedback. The amount of temperature change predicted for the year 2100 varies from a little over 1°C to about 5°C--at the high end, enough to raise sea level significantly, to shift agricultural regions, and to cause other ecological and economic changes. The IPCC recently provided a range of estimates of temperature predictions from 1°C to 3.5°C by 2100. But that summary was based on a compilation of results from a variety of climate change models. If the results from a single model--the IGSM--can vary by that much and more, the IPCC range is apt to be too narrow.

The researchers are now continuing to refine the IGSM. They are improving its ability to track factors such as energy prices, resource depletion, and levels of stratospheric ozone (another greenhouse gas). They are developing a model that can predict future distributions of potential croplands under various climate-change scenarios. They are modifying the submodel for terrestrial ecosystems so it can operate simultaneously and interactively with the chemistry/climate model (rather than iteratively, as it has in the past). And they are examining the costs and impacts of proposed policy options such as restricting emissions and stabilizing atmospheric concentrations of CO₂ at particular levels. New work focuses on the potential impact on global warming of proposed restrictions on SO_X emissions aimed at reducing local and regional air pollution. Coal-burning electric utilities may use scrubbers to remove SO_X from their stack gases yet leave levels of CO₂ emissions unchanged. In the absence of the cooling aerosols, any potential global warming will progress more quickly. Using the IGSM, researchers are quantifying such effects so that policymakers can assess the trade-offs involved when considering potentially conflicting local, regional, and global policies.

This research was supported through the MIT Joint Program on the Science and Policy of Global Change by the US Department of Energy, the National Science Foundation, the US National Oceanic and Atmospheric Administration, the US Environmental Protection Agency, the Royal Norwegian Ministries of Industry and Energy and Foreign Affairs, and a group of corporate sponsors from the United States, Europe, and Japan. Further information can be found in the following references.