If we can’t predict the weather a week out, why should we believe what climate models say about the next 100 years?
Return to Climate Question & Answers
Weather and climate are not the same thing. Weather and climate models both call on the laws of physics to predict atmospheric phenomena by combining weather observations with mathematical formulas called algorithms. But they have different resolutions and time scales, and climate models add other features, such as solar variation and ocean circulation, that affect long-term climate but not day-to-day weather.
Weather models project day-to-day weather features forward; eventually, small-scale chaotic features grow to dominate the weather model, making it impossible to predict local weather features beyond about 8 to 10 days. In contrast, climate models look at the next century by using what we know about Earth’s orbit, the patterns of incoming sunlight and how it is absorbed by gases in the atmosphere, and other large-scale, long-term processes.
The chaotic features that make short-term forecasts impossible beyond a few days aren’t an issue with climate models, because they are not focusing on day-to-day weather but instead on the ebb and flow of seasonal, annual, and interannual features. These are driven by the climate variables above, which can be extrapolated for many decades based on what we know about Earth’s orbit and how greenhouse gases are increasing.
If it’s January 1 in Fairbanks, Alaska, we can’t predict the daily high on February 1, because weather patterns are too chaotic to predict a month ahead. However, we can say with confidence that July 1 will be warmer than January 1, because the factors that shape seasonal climate are well known. Similarly, we can say that Fairbanks in July is likely to run warmer in 2050 than it does now, due to the continued growth of greenhouse gases and their effects on global and regional climate.
Climate models have been tested and proven in simulations of 20th-century conditions, which they reproduce quite well. These “hindcasts” have even helped researchers identify anomalies in the observational record, such as overestimates of Northern Hemisphere cooling in the 1940s due to changes in ocean observing methods. The success of the 20th-century simulations strengthens the case for using climate models to project 21st-century climate.
More information: “General Circulation Models of Climate.” From: The Discovery of Global Warming, by Spencer Weart. The American Institute of Physics