Lisa Moore is a scientist in the Climate and Air Program.
Of everything in climate science, what seems to spark the most skeptical questions is our use of computer models. In this post and another to follow, I'll talk about exactly what these models do, and how they contribute to our understanding of global warming.
I'll start with why we use models in the first place. We want to project what will happen to our climate in the future – will it be warmer? How much warmer? Will it change in different ways in different places? Climate models use our knowledge of how the climate system works to calculate what different emissions scenarios mean for the future. Here's how the models are built.
General Circulation Models, or GCMs, are computer programs that use basic physics and chemistry to calculate properties of the atmosphere and ocean, such as temperature, precipitation and sea level.
GCMs represent the Earth’s ocean, biosphere, and atmosphere in three-dimensional cells. The models stack ocean and atmospheric layers in grids across the Earth’s surface. An example schematic of atmospheric cells is shown below. The models use differential equations to calculate transfers of radiation, heat, carbon, water, and other substances between cells.
A Coarse Example of Atmospheric Cells
Figure from the Australian Bureau of Meteorology
How we use a GCM depends on its complexity. The simplest models take less than a minute to complete a 100-year simulation. A more complex, high-resolution model can simulate complex phenomena such as regional weather extremes or atmospheric dust transport, but using these kinds of GCMs requires substantially more computing power and time. Documentation and data from all of the GCMs used in the latest IPCC report are available here.
To run a GCM, we need to give it input. This information includes solar radiation, volcanic emissions and human-produced emissions of greenhouse gases. We have this information for the 20th century, but of course we don’t know what emissions will be in the future. Scientists deal with that uncertainty by using a range of emissions scenarios.
Some emissions scenarios are “fossil fuel intensive”, with emissions continuing to increase (for example, Scenario “A2” shown below in red). Others assume that greenhouse gas emissions decrease at various rates over time. For example, emissions cuts in Scenario “B1” (blue) are steeper than those in “A1B” (green). As shown in the graph below, using different scenarios results in different predictions for global temperature change.
Each scenario has a thick line with some shading on either side of it. The line represents the average of several model runs. The shading shows the range of temperature predictions from all the models.
All of these models were given exactly the same inputs — so why don’t they give the same results? Some of this has to do with the different levels of complexity from model to model. Some of the spread also comes from uncertainties in various factors such as the carbon cycle (the rate at which soils and plants store or release carbon), climate sensitivity (the amount of warming for a given amount of greenhouse gas) and cloud formation (which can lead to warming or cooling). Each model deals with these factors slightly differently, which leads to variations in their predictions.
These uncertainties lead to the most common question about climate models — How accurate are they? — which is the topic of part two. Stay tuned!