Prospects of global climate change resulting from human activities present a crucial and urgent scientific problem, one with large social and economic implications. The problem has generated much debate; in particular, the possibility of global warming as a response to the increased atmospheric concentration of greenhouse gases has become a subject of widespread public attention.
The American Meteorological Society (AMS) applauds the concern of the public over the matter, as well as the increased attention of governments generally and our own Congress and Administration in particular. To help set an objective and expert framework for continuing discussions and actions, the AMS is issuing this statement for the information of all concerned.
At present, observations suggest, but are insufficient to prove, that atmospheric warming caused by human activities has already occurred. Climate models, which are simplified mathematical representations of the complex climate system, are the principal means available to predict or project the magnitude, timing, and spatial distribution of global climate change. General Circulation Models (GCMs), the most comprehensive physically-based models of the climate system, currently provide the best estimates of future global warming. These suggest that global warming will occur, though the magnitude and timing of this warming are debatable. However, even the most comprehensive present day models now in existence have significant limitations. These limitations, which are expected to diminish as model improvements occur, could be masking elements which would produce greater warming or could be causing over-estimates of the warming to be anticipated.
General Circulation Model studies have emphasized the importance of the greenhouse gases, such as carbon dioxide, methane, and chloroflourocarbons (CFCs); however, these are not the only human influences on the climate system. For example, increased concentrations of atmospheric aerosols and land-use changes may have additional effects. The full extent of large- scale human impacts is yet to be assessed with confidence.
Observations do reveal unmistakable changes in the composition of the atmosphere that can be traced --without ambiguity-- to human activities. Particularly notable is the increased concentration of greenhouse gases which interact strongly with the earth's thermal radiation. The increase in concentration of carbon dioxide results from combustion of fossil fuels and from tropical deforestation. The major sources of CFCs are refrigeration, industrial uses, and aerosol propellants. Methane sources are less certain, but agricultural activities are a significant contributor to the observed increase.
The theory of how these greenhouse gases directly influence the earth's energy balance is not controversial. If no other factors counter their influence, increases in their concentration will lead to global warming. However, the magnitude and timing of climate change are controversial because of uncertainties in our knowledge of all factors which may influence future climates and of the interactive processes which may act to modify the direct influence of the greenhouse gases.
Since the industrial revolution, the concentration of carbon dioxide in the atmosphere has increased more than 20 percent. Detailed studies of the worldwide record of temperature suggest an increase during the last 100 years. Temperatures are estimated to have increased 0.5°C (0.9°F) on a globally averaged basis over this period. Further, some of the warmest years of the century have occurred during the last decade. For two reasons, however, the evidence is insufficient to state conclusively that human-induced global warming has occurred.
First, both the number and distribution of observations over the last century are insufficient to decipher the exact nature and character of temperature changes which have taken place. The network of observations has grown in response to local needs for accurate weather forecasting, not always in accord with a design to achieve a long-term, consistent, global record of climate change. Weather stations in growing urban areas, for example, are often influenced by the urban "heat island" effect. Changes in instruments have occurred over the last century. Perhaps most importantly, the oceanic and polar regions are poorly represented. Substantial effort has been applied to remove biases in the observations record; analyses of the resulting data provide evidence for warming during the more recent segment of the record, but they are insufficient to determine the mechanisms which would explain the temperature changes.
Secondly, the climate system is characterized by substantial natural variability over different time scales. Many factors influence the climate system, resulting in short- and long- term fluctuations through many complex interactions. Differences from year to year, and from decade to decade are typical. Thus, any extrapolation of the temperature record from a period of decades to a longer term trend, or the imputation of such a trend to a single mechanism, will be the subject of debate and difference in opinion. The length of the record and our confidence in the observations are insufficient to conclude positively that the apparent warming over the last century has exceeded what might be expected from natural variability in the climate.
However, the evidence must be considered in the context of human activities. The years of record warmth in the last decade and the apparent trend over the last century, taken in conjunction with the measured increase in radiatively active greenhouse gases, are consistent with an hypothesis that the products of human activity have reached a level where they can influence climate. If this hypothesis is correct, increased frequency of warm years over the next decade should reduce the uncertainty as to whether the record is solely the product of natural variability, or whether it includes a substantial component of human-induced climate change.
The most comprehensive climate models to date suggest that the continued increase in the concentration of carbon dioxide would cause a globally-averaged temperature increase in several °C over the next 50 to 100 years. Such a warming would be larger than any sustained global warming experienced in the past 100,000 years. For perspective, this projected climate change is of the same magnitude (but in the opposite direction) as the last ice age when continental ice caps penetrated well into Europe and North America (reconstructed as a 3-5°C globally-averaged temperature decrease relative to today). The difference is that the global warming may occur over the next few decades, whereas the ice-age changes occurred over thousands of years! The rate and magnitude of the model-predicted global warming, with the potential for even larger regional changes in some areas, could have enormous impact.
Beyond the limitations already mentioned, today's climate models provide little or no useful and consistent information on regional distributions of climate change. We can only be confident that some geographical areas would experience temperature increases larger than the globally-averaged warming, while other areas would be less affected. Additionally, the models are not now capable of predicting changes in average precipitation on a regional or otherwise localized basis.
The climate system is exceedingly complex, consisting of diverse processes and interactions involving the atmosphere, the oceans, the land surface, the biosphere, and snow and ice. Every feature of this complex system cannot be incorporated into a model, certainly not with precision based on a complete understanding; consequently, the primary task in modeling is to simplify the complex climate system while incorporating the important feedbacks and interactions necessary to simulate climate and climate changes. However, uncertainties are introduced by this need to simplify the climate system to produce a tractable model.
For perspective, this projected climate change is of the same magnitude (but in the opposite direction) as the last ice age when continental ice caps penetrated well into Europe and North America... The difference is that the global warming may occur over the next few decades, whereas the ice-age changes occurred over thousands of years!
First, comprehensive climate models, because they require extensive computational capabilities, have limited spatial resolution, now typically on the order of 5° of latitude and longitude. Processes on scales smaller than a model's resolution tend to be poorly represented. Coarse resolution introduces uncertainty. Consequently, some of the possible climate changes of societal importance cannot be simulated. Changes in regional climate, some of which are bound to be greater that the projected global changes, are a matter of speculation.
Secondly, the focus of climate models has been primarily the atmosphere, secondarily the oceans. Incorporation of the land surface, the biosphere, biogeochemical interactions, and polar ice has been rudimentary. Yet the response to a change in climate forcing involves all the interactive parts, and these factors may serve either to amplify or to dampen the earth system response. Substantial research is required to couple all the earth system components, particularly ocean-atmosphere interactions, biogeochemical interactions, the role of the hydrologic cycle, ecosystem response, and human responses to a changing environment. Inadequate computation resources and insufficient knowledge of the processes which govern system interactions are the primary impediments to improved predictive system models.
Thirdly, there are serious weaknesses in our current understanding of many important atmospheric processes. For example, one of the most serious problems is assessing the impact of clouds --in specifying how they produce precipitation and thereby release heat, which is an important driver of atmospheric circulation, and how their water and ice particles interact with and change the incoming and outgoing radiation. Improvement of the way models simulate cloud processes is critical. The radiative response of clouds could either enhance or moderate the effects of increased greenhouse gases. Comparisons among different GCMs suggest that much of the variation among them in temperature estimates, and particularly the significant regional differences in precipitation and soil moisture, may result from differences in model representations of cloud processes, none of which presently can be regarded as adequate or well tested.
The scientific debate on these issues and efforts in model improvement are likely to continue for some time before high confidence can be placed in model predictions of climate change. Nevertheless, to address the serious practical issues associated with global climate change, numerical models provide the best available estimate of human impact on the changing climate; they indicate a reasonable probability for large climate changes over the impending decades.
The nature of the current uncertainties defines the priorities for research needed in the atmospheric sciences if we are to better understand the likely future of our planet's climate. The primary needs are: