Why do you think that the chart makers present two different temperature scenarios?

Models help us to work through complicated problems and understand complex systems. They also allow us to test theories and solutions. From models as simple as toy cars and kitchens to complex representations such as flight simulators and virtual globes, we use models throughout our lives to explore and understand how things work.

Climate Models, and How They Work

This image shows the concept used in climate models. Each of the thousands of 3-dimensional grid cells can be represented by mathematical equations that describe the materials in it and the way energy moves through it. The advanced equations are based on the fundamental laws of physics, fluid motion, and chemistry. To "run" a model, scientists specify the climate forcing (for instance, setting variables to represent the amount of greenhouse gases in the atmosphere) and have powerful computers solve the equations in each cell. Results from each grid cell are passed to neighboring cells, and the equations are solved again. Repeating the process through many time steps represents the passage of time. Image source: NOAA.

Climate models are based on well-documented physical processes to simulate the transfer of energy and materials through the climate system. Climate models, also known as general circulation models or GCMs, use mathematical equations to characterize how energy and matter interact in different parts of the ocean, atmosphere, land. Building and running a climate model is complex process of identifying and quantifying Earth system processes, representing them with mathematical equations, setting variables to represent initial conditions and subsequent changes in climate forcing, and repeatedly solving the equations using powerful supercomputers.

Check out The Very, Very Simple Climate Model »

Climate Model Resolution

Climate models separate Earth’s surface into a three-dimensional grid of cells. The results of processes modeled in each cell are passed to neighboring cells to model the exchange of matter and energy over time. Grid cell size defines the resolution of the model: the smaller the size of the grid cells, the higher the level of detail in the model. More detailed models have more grid cells, so they need more computing power.

See an animation showing different grid sizes »

Explore practical information for those interested in participating in  global climate model experiments »

Climate models also include the element of time, called a time step. Time steps can be in minutes, hours, days, or years.  Like grid cell size, the smaller the time step, the more detailed the results will be. However, this higher temporal resolution requires additional computing power.

How are Climate Models Tested?

Once a climate model is set up, it can be tested via a process known as “hind-casting.”  This process runs the model from the present time backwards into the past. The model results are then compared with observed climate and weather conditions to see how well they match. This testing allows scientists to check the accuracy of the models and, if needed, revise its equations.  Science teams around the world test and compare their model outputs to observations and results from other models.   

Using Scenarios to Predict Future Climate

Once a climate model can perform well in hind-casting tests, its results for simulating future climate are also assumed to be valid. To project climate into the future, the climate forcing is set to change according to a possible future scenario. Scenarios are possible stories about how quickly human population will grow, how land will be used, how economies will evolve, and the atmospheric conditions (and therefore, climate forcing) that would result for each storyline. 

In 2000, the Intergovernmental Panel on Climate Change (IPCC) issued its Special Report on Emissions Scenarios (SRES), describing four scenario families to describe a range of possible future conditions. Referred to by letter-number combinations such as A1, A2, B1, and B2, each scenario was based on a complex relationship between the socioeconomic forces driving greenhouse gas and aerosol emissions and the levels to which those emissions would climb during the 21st century. The SRES scenarios have been in use for more than a decade, so many climate model results describe their inputs using the letter-number combinations.

In 2013, climate scientists agreed upon a new set of scenarios that focused on the level of greenhouse gases in the atmosphere in 2100. Collectively, these scenarios are known as Representative Concentration Pathways or RCPs. Each RCP indicates the amount of climate forcing, expressed in Watts per square meter, that would result from greenhouse gases in the atmosphere in 2100. The rate and trajectory of the forcing is the pathway. Like their predecessors, these values are used in setting up climate models.

Learn more about RCPs »

Results of Current Climate Models

Around the world, different teams of scientists have built and run models to project future climate conditions under various scenarios for the next century. So the groups can make a fair comparison of their results, they run the same experiment. Because each climate model is slightly different, the results show a a range of projections. Though yearly values projected for temperature and precipitation differ among the models, the trend and magnitude of change is fairly consistent. 

Global climate model results from groups around the world project that global temperature will continue to increase. They also show that human decisions and behavior we choose today will determine how dramatically climate will change in the future.

Check the World Climate Research Programme's experimental design for the most recent phase of the Coupled Modeled Intercomparison Project »

How are Climate Models Different from Weather Prediction Models?

Unlike weather forecasts, which describe a detailed picture of the expected daily sequence of conditions starting from the present, climate models are probabilistic, indicating areas with higher chances to be warmer or cooler and wetter or drier than usual. Climate models are based on global patterns in the ocean and atmosphere, and records of the types of weather that occurred under similar patterns in the past.

View maps showing short-term climate forecasts »

Scientists have been making projections of future global warming using climate models of increasing complexity for the past four decades.

These models, driven by atmospheric physics and biogeochemistry, play an important role in our understanding of the Earth’s climate and how it will likely change in the future.

Carbon Brief has collected prominent climate model projections since 1973 to see how well they project both past and future global temperatures, as shown in the animation below. (Click the play button to start.)

A post shared by Carbon Brief (@carbonbrief) on Oct 5, 2017 at 10:09am PDT

While some models projected less warming than we’ve experienced and some projected more, all showed surface temperature increases between 1970 and 2016 that were not too far off from what actually occurred, particularly when differences in assumed future emissions are taken into account.

How have past climate models fared?

While climate model projections of the past benefit from knowledge of atmospheric greenhouse gas concentrations, volcanic eruptions and other radiative forcings affecting the Earth’s climate, casting forward into the future is understandably more uncertain. Climate models can be evaluated both on their ability to hindcast past temperatures and forecast future ones.

Hindcasts – testing models against past temperatures – are useful because they can control for radiative forcings. Forecasts are useful because models cannot be implicitly tuned to be similar to observations. Climate models are not fit to historical temperatures, but modellers do have some knowledge of observations that can inform their choice of model parameterisations, such as cloud physics and aerosol effects.

In the examples below, climate model projections published between 1973 and 2013 are compared with observed temperatures from five different organizations. The models used in the projections vary in complexity, from simple energy balance models to fully-coupled Earth System Models.

(Note, these model/observation comparisons use a baseline period of 1970-1990 to align observations and models during the early years of the analysis, which shows how temperatures have evolved over time more clearly.)

Sawyer, 1973

One of the first projections of future warming came from John Sawyer at the UK’s Met Office in 1973. In a paper published in Nature in 1973, he hypothesised that the world would warm 0.6C between 1969 and 2000, and that atmospheric CO2 would increase by 25%. Sawyer argued for a climate sensitivity – how much long-term warming will occur per doubling of atmospheric CO2 levels – of 2.4C, which is not too far off the best estimate of 3C used by the Intergovernmental Panel on Climate Change (IPCC) today.

Unlike the other projections examined in this article, Sawyer did not provide an estimated warming for each year, just an expected 2000 value. His warming estimate of 0.6C was nearly spot on – the observed warming over that period was between 0.51C and 0.56C. He overestimated the year 2000’s atmospheric CO2 concentrations, however, assuming that they would be 375-400ppm – compared to the actual value of 370ppm.

Broecker, 1975

The first available projection of future temperatures due to global warming appeared in an article in Science in 1975 published by Columbia University scientist Prof Wally Broecker. Broecker used a simple energy balance model to estimate what would happen to the Earth’s temperature if atmospheric CO2 continued to increase rapidly after 1975. Broecker’s projected warming was reasonably close to observations for a few decades, but recently has been considerably higher.

This is mostly due to Broecker overestimating how CO2 emissions and atmospheric concentrations would increase after his article was published. He was fairly accurate up to 2000, predicting 373ppm of CO2 – compared to actual Mauna Loa observations of 370ppm. In 2016, however, he estimated that CO2 would be 424ppm, whereas only 404 pm has been observed.

Broecker also did not take other greenhouse gases into account in his model. However, as the warming impact from methane, nitrous oxide and halocarbons has been largely cancelled out by the overall cooling influence of aerosols since 1970, this does not make that large a difference (though estimates of aerosol forcings have large uncertainties).

As with Sawyer, Broecker used an equilibrium climate sensitivity of 2.4C per doubling of CO2. Broecker assumed that the Earth instantly warms up to match atmospheric CO2, while modern models account for the lag between how quickly the atmosphere and oceans warm up. (The slower heat uptake by the oceans is often referred to as the “thermal inertia” of the climate system.)

You can see his projection (black line) compared to observed temperature rise (coloured lines) in the chart below.

Broecker made his projection at a time when scientists widely thought that the observations showed a modest cooling of the Earth. He began his article by presciently stating that “a strong case can be made that the present cooling trend will, within a decade or so, give way to a pronounced warming induced by carbon dioxide”.

Hansen et al, 1981

NASA’s Dr James Hansen and colleagues published a paper in 1981 that also used a simple energy balance model to project future warming, but accounted for thermal inertia due to ocean heat uptake. They assumed a climate sensitivity of 2.8C per doubling CO2, but also looked at a range of 1.4-5.6C per doubling.

* SAR trend differences are calculated over the period from 1990-2016, as estimates prior to 1990 are not readily available.
# Differences in parenthesis based on blended model land/ocean fields

Climate models published since 1973 have generally been quite skillful in projecting future warming. While some were too low and some too high, they all show outcomes reasonably close to what has actually occurred, especially when discrepancies between predicted and actual CO2 concentrations and other climate forcings are taken into account.

Models are far from perfect and will continue to be improved over time. They also show a fairly large range of future warming that cannot easily be narrowed using just the changes in climate that we have observed.

Nevertheless, the close match between projected and observed warming since 1970 suggests that estimates of future warming may prove similarly accurate.

Methodological note

Environmental scientist Dana Nuccitelli helpfully provided a list of past model/observation comparisons, available here. The PlotDigitizer software was used to obtain values from older figures when data was not otherwise available. CMIP3 and CMIP5 model data was obtained from KNMI Climate Explorer.

Comments

View Comments () Close Comments