Aren’t the computer models used to predict climate really simplistic?

Four differenet maps of the US with different resolution for climate model
The amount of detail in climate models has continued to increase in recent years, largely because of the calculation power provided by newer supercomputers. In the 1990s, high-resolution global climate models operated on the T42 resolution scheme (upper left). At this resolution, temperature, moisture, and other features were tracked in grid boxes that each spanned about 200 by 300 kilometers at midlatitudes (120 x 180 miles), an area roughly as large as West Virginia.

In more recent modeling that led up to the 2007 IPCC Working Group I report, the NCAR-based Community Climate System Model (CCSM) routinely operated at T85 resolution (upper right), with midlatitude grid points of about 100 by 150 km (60 x 90 miles)—the size of Connecticut.

Better resolution not only provides a more true-to-life depiction of atmospheric processes, but also allows for more realistic topography, which makes regional climate projections more accurate. For example, the highest Rocky Mountains appear as two coarse grid points at T42 but as a more diverse assortment of high peaks at T170 (lower left). Enhancements in computing power will help scientists explore the use of higher resolutions, such as T170 and T340 (lower right). Click here or on the image to enlarge. (©UCAR. Illustration courtesy Warren Washington, NCAR. This image is freely available for media and nonprofit use*)

Global climate models—the software packages that simulate the past, present, and future of our atmosphere—have grown in complexity and quality over the last 10 to 20 years, and the most sophisticated models agree on the big picture of climate change. This includes the rough amount of warming expected and the idea that poles will warm faster than lower latitudes. As models have continued to improve, increasing agreement on regional details has emerged, such as the likelihood of more precipitation in the northern subpolar latitudes and a northward expansion of the hot, dry subtropics around 30°N. Climate models are not perfect, but the main ingredients are well understood and tested, and scientists are making progress in areas that remain a challenge, such as the behavior of tropical oceans and the evolution of cloud patterns.

It’s important to note that models are not the only reason why scientists are concerned about climate change.  For more than a century—long before many recent advances in science, and long before computer models—we’ve known that increased greenhouse gases could produce a global temperature increase.  Observations of the last century of climate, including those from instruments and from the behavior of ice and plants, concur that the planet is warming.  As greenhouse gases continue to increase, it stands to reason that more warming can be expected.

Even the earliest models of the 1960s, which were quite crude by today’s standards, showed that a doubling of carbon dioxide in the atmosphere could increase global temperature by around 5°F (3°C). That projection remains close to the modern consensus, and temperatures over the last 30 years have risen at a rate consistent with this early estimate.

Far more information is available from today’s models, such as the NCAR-based Community Earth System Model, because they now include many more aspects of the Earth system, including ice sheets, vegetation, cloud areas, and soil moisture.

Research conducted for the 2007 IPCC Working Group 1 assessment compared the output from major models at research centers around the world. While these models are far from perfect, scientists are confident that they capture the key processes that drive climate. For example, models now replicate the ups and downs of 20th-century global temperature quite accurately.

As in other areas of science, rigorous testing and continual improvement are part and parcel of climate modeling. Researchers can test models against reality, identify and correct flaws, and compare their models with others. 


flow chart showing increasing complexity of climate models
The complexity of global climate models has increased enormously over the last four decades, as shown in this graphic. The most powerful models, such as the Community Earth System Model (developed by scientists at the Department of Energy and NCAR with colleagues at other organizations), now have the capability of simulating a broad range of atmospheric processes, such as the impact of marine ecosystems on the atmosphere. Click here or on the image to enlarge. (©UCAR. This image is freely available for media and nonprofit use*)
3-D view of column of air rising through layers of atmosphere
Computer models reach high into a virtual atmosphere and deep into the ocean. They simulate climate by dividing the world into into 3-dimensional grid boxes, measuring physical processes such as temperature at each grid point. Such models can be used to simulate changes in climate over years, decades, or even centuries. Click here or on the image to enlarge. (©UCAR. This image is freely available for media and nonprofit use*)
3-D view of Earth system, from ocean floor to troposphere
The NCAR-based Community Earth System Model (CESM) is one of the world’s most sophisticated models of global climate. Created by scientists at NCAR, the Department of Energy, and collaborators, this powerful model simulates the many processes in our climate system, ranging from clouds and atmospheric chemicals to ice to marine ecosystems. Click here or on the image to enlarge. (©UCAR. This image is freely available for media and nonprofit use*)