Online Catalog

GCRIO Home ->arrow Library-> arrow 1997-> arrow Common Questions about Climate Change -> arrow How Reliable Are Predictions of Future Climate? Search
U.S. Global Change Research Information Office logo and link to home
Page last modified
24 May 2006

Common Questions about Climate Change
Published in 1997 by the United Nations Environment Programme - World Meteorological Organization




Britannica Internet Guide Selection
See: Information from Encyclopædia Britannica about global warming


More Questions and Answers from the Education page of the U.S. Global Change Research Program.







How Reliable Are Predictions of Future Climate?

Predictions of climate change are calculated by means of computer models that mathematically simulate the interactions of the land, sea, and air, which together determine the Earth's climate. Our confidence in these models rests largely on their basis in accepted physical laws, their ability to describe many aspects of current climate accurately, and their skill at reproducing some of the important features of past climates.

Climate models are based on a wealth of scientific observations and well established laws of physics, including the laws of gravity and fluid motion, and the conservation of energy, momentum, mass, and water. It is this reliance on basic physical laws that lends high confidence to the prediction that a buildup of greenhouse gases will eventually lead to a significant alteration in the Earth's climate.

A second important reason for having confidence in climate models is because of their ability to reproduce many of the observed features of the atmosphere and ocean. For the purposes of predicting the behavior of the atmosphere for only a few days ahead, an atmosphere-only model, with no simulation of the ocean, can be used. This is the method employed in making short-term weather forecasts, whose relative accuracy demonstrates the ability of this sort of model to reproduce some of the important details of the atmosphere's behavior.

While reliable weather forecasts can only be made for periods up to ten days, predictability for greater lengths of time can be obtained for averages of weather, i.e., the climate. For example, with regard to longer periods (several years or more), climate models in which both the oceans and the atmosphere are represented are able to simulate the main features of current climate and its variability, including the seasonal cycle of temperature, the formation and decay of the major monsoons, the seasonal shift of the major rain belts and storm tracks, the average daily temperature cycle, and the variations in outgoing radiation at high elevations in the atmosphere as measured by satellites. Similarly, many of the large-scale features observed in the ocean circulation have been reproduced by climate models.

It is possible for a model to simulate current climate well but still fail in its prediction of climate change. So another test of models is to compare their simulations of earlier climates to historical data, including the climate of the past century (Figure 7.1). These efforts have been hampered by our imprecise knowledge of a variety of factors, including how humans have changed the amounts of small particles in the atmosphere and variations in the energy output of the sun.

Figure 7.1

Calculated globally averaged surface air temperature is compared to observed values over the period 1860 to 1994.

Nevertheless, using estimates of some of these factors, climate models can reproduce many changes observed over the last century, including the global mean surface warming of 0.3 to 0.6°C (about 0.5 to 1°F), the reduction in temperature differences between day and night, the cooling in the atmosphere above about 14 km (about 9 miles), the increases in precipitation at high latitudes, the intensification of precipitation events in some continental areas, and a rise in sea level. Moreover, a climate model has correctly predicted broad features of the globally averaged surface cooling and subsequent recovery associated with the eruption of Mt. Pinatubo in 1991.

Climate models can also be used in attempts to reproduce the main features of prehistoric climates, but this effort has been limited by the scarcity and the indirect nature of the evidence available from sediment cores, tree rings, preserved pollen, and ice core data used to infer earlier climates. Even so, the models have reproduced some of the general features of reconstructed past climates, such as the enhanced North African monsoon 6000 to 9000 years ago, and the approximate level of cooling during the last ice age.

The major weakness of models is their reliance on approximations of some aspects of climate. It takes too much computer time, or is simply beyond the capacity of even supercomputers, to represent some of the key smaller-scale processes that affect climate. Even if adequate computers were available, scientists' understanding of the detailed physics of such processes is limited. So, some aspects of climate are approximated, based on a combination of physical laws, laboratory experiments, and direct observations of climate. For example, it is not possible to represent the details of the formation and dissipation of clouds. So, approximations are used. The approximation of cloud behavior is a major source of uncertainty in climate models.

In summary, the fact that models are based upon the known physical laws of nature and can reproduce many features of the current climate and some general aspects of past climates gives us increasing confidence in their reliability for projecting many large-scale features of future climate. However, there remains substantial uncertainty in the exact magnitude of projected globally averaged temperature rise caused by human activity, due to shortcomings in the current climate models, particularly in their representation of clouds. Furthermore, scientists have little confidence in the climate changes they project at the local level. Other uncertainties, not arising from specific limitations in the climate models, such as estimates of the rate of future greenhouse gas emissions, also restrict the ability to predict precisely how the climate will change in the future.

Go to "Are Recent Extreme Weather Events, Like the Large Number of Atlantic Hurricanes in 1995, Due to Global Warming?"

U.S. Global Change Research Information Office, Suite 250, 1717 Pennsylvania Ave, NW, Washington, DC 20006. Tel: +1 202 223 6262. Fax: +1 202 223 3065. Email: . Web: Webmaster: .
U.S. Climate Change Technology Program Intranet Logo and link to Home