Climate Model Hindcasting

Loading data...

Model assumptions

Compare the predictions of global temperature made by several different climate models to the actual obserations.

Choose a model from the list, and compare its temperature predictions against what actually happened.
How did it do?

See the for more information about the model you've chosen.

The dashed line marks the year that the model was published. The model's prediction after this time is called a "forecast". The model's "prediction" for earlier times is called a "hindcast".

The "forecast" label is from the model's point of view. Looking back on these predictions today, they are really all "hindcasts" now, so we can see how good their predictions really were!

You may notice that this learning resource does not include major recent climate model projections. That's just because they're so new, there hasn't been enough observation time to compare them to. See the for more details.

All models make assumptions. The most important assumptions for these models are shown in the controls below the graph. You can adjust these assumptions to see how much difference they make to the model's prediction.

This learning resource was created by KCVS, in collaboration with NCSE, and draws, with permission, on the data and discussions in these articles:

Hausfather, Z. (2017) Analysis: How well have climate models projected global warming? CarbonBrief website, accessed August 2021.

Hausfather, Z., Drake, H. F., Abbott, T., & Schmidt, G. A. (2020). Evaluating the performance of past climate model projections. Geophysical Research Letters, 47, e2019GL085378. https://doi.org/10.1029/2019GL085378

Resolution: ° by °.none (1D model)

When a model is chosen, you'll find more information about it here.

In this very readable article, Sawyer went over some of the evidence that humans are increasing the CO2 concentrations by burning fossil fuels. He discussed what happens to CO2 in the atmosphere and the ocean, and how this feeds back into other Earth system processes. Sawyer also acknowledged several of the factors which made (and still make) the warming effect harder to predict, and estimated how much impact those uncertainties could have on the prediction.

Sawyer stated that the CO2 concentration was expected to increase by 25% by the year 2000. This implied Earth's temperature would increase by 0.6°C by the year 2000, assuming a climate sensitivity of 2.4°C.

This projection used a radiative energy balance model, which compared the rate at which Earth absorbs energy from the sun to the rate at which it is emitted into space. The difference was used to estimate the rate at which the Earth's temperature will rise. CO2 and other greenhouse gases absorb some of the outgoing radiation and re-emit it in all directions, including back to Earth, reducing the amount of energy escaping into space. (In Sawyer's model, the only greenhouse gases included were CO2 and water vapour.)

Sawyer, J.S. (1972). Man-made Carbon Dioxide and the "Greenhouse" Effect. Nature, 239, 23–26. https://doi.org/10.1038/239023a0

In addition to presenting a projection of the Earth's temperature in response to rising CO2 concentrations, this article compared the CO2 effects to known natural climate cycles observed in ice core temperature records. At the time of publication, the Earth's temperature had been cooling, which fit into the natural cycle pattern. Broecker's model predicted that the CO2 effect would overwhelm the natural cycles within a decade or so.

This temperature prediction begain with estimates of the amount of fossil fuels that would be burned. CO2 emissions were calculated from there, with some assumptions about what fraction of emitted CO2 remains in the atmosphere. An energy balance model was then used to translate CO2 levels into changes in global temperature.

Broecker noted that the main uncertainties in his model projections were the climate sensitivity parameter and the actual rate of CO2 emission. He also observed that the effects of aerosols on global temperature were not yet well understood. However, Broecker emphasized that, in spite of these uncertainties, climate change due to increasing global temperatures was clearly coming.

This projection used a radiative energy balance model, which compared the rate at which Earth absorbs energy from the sun to the rate at which it is emitted into space. The difference was used to estimate the rate at which the Earth's temperature will rise. CO2 and other greenhouse gases absorb some of the outgoing radiation and re-emit it in all directions, including back to Earth, reducing the amount of energy escaping into space. (In Broecker's model, the only greenhouse gas included was CO2.)

Broecker, W. (1975). Climatic Change: Are We on the Brink of a Pronounced Global Warming? Science, 189, 460–463. https://doi.org/10.1126/science.189.4201.460

Prediction scenarios:
  • 1: fast growth
    (454 ppm CO2 in 2017)
  • 2: slow growth
    (412 ppm CO2 in 2017)
  • 3: no growth
    (336 ppm CO2 in 2017)
We are showing scenario 1 ("fast growth") by default, because its temperature predictions were the closest to observations.

The Earth's temperature was not noticeably higher in 1981 than it was at the time of Broecker's model in 1975. Hansen et al. began their paper by noting this was "the major difficulty" in accepting the greenhouse effect and its consequences. They describe in detail which aspects of climate they were including in their model and how they were representing them. They also explored many of their assumptions, showing how their predictions could change if they made different reasonable choices.

Hansen et al. used hindcasting to show how well their model worked, as well as to constrain some of the important input values.

This model calculated temperature as a function of altitude (a "one-dimensional radiative-convective model"), which allowed the scientists to more accurately model greenhouse gas concentrations, clouds, and aerosols. This advance allowed them to account for how long it takes to heat up the ocean.

The paper cited estimates of climate sensitivity of 2–3.5°C.

Hansen, J., Johnson, D., Lacis, A., Lebedeff, S., Lee, P., Rind, D., Russel, G. (1981). Climate Impact of Increasing Atmospheric Carbon Dioxide. Science, 213, 957–966. https://doi.org/10.1126/science.213.4511.957

Prediction scenarios:
  • A: exponential growth
    (408 ppm CO2 in 2017)
  • B: linear growth
    (402 ppm CO2 in 2017)
  • C: net zero by 2000
    (368 ppm CO2 in 2017)
We are showing scenario B ("linear growth") by default, because its temperature projections were the closest to observations.

This was one of the first truly three-dimensional climate models, so it could track the interaction of land and ocean, simulate clouds, and more. This also meant it could predict how climate change would affect different parts of the world.

Understanding of the ocean was a major source of uncertainty for this model. Hansen et al. described their simplified ocean as a "surprise free" representation, and emphasized the need to understand it better to improve future models.

Hansen, J., Fung, I., Lacis, A., Rind, D., Lebedeff, S., Ruedy, R., Russell, G., and Stone, P. (1988). Global climate changes as forecast by Goddard Institute for Space Studies three-dimensional model. Journal of Geophysical Research, 93, 9341-9364, https://doi.org/10.1029/JD093iD08p09341

Prediction scenarios:
  • A: Business as usual
    (420 ppm CO2 in 2017)
  • B & C: Lower-carbon fuels
    (256 ppm CO2 in 2017)
  • D: Tight controls
    (230 ppm CO2 in 2017)
(B and C assume the same CO2 emissions until about 2050.)
We are showing scenario A ("Business as usual") by default, because its CO2 projections were the closest to observations.

The First Assessment Report from the Intergovernmental Panel on Climate Change (IPCC) was released in 1990. The IPCC made the climate projections in this report in two (very large) stages: First, they used a variety of general circulation models from groups around the world to estimate the climate sensitivity and other parmaeters. They fed those results into a simpler energy balance model to make predictions about temperature and other aspects of climate change. The models used for the first stage were complex enough that it would have been almost impossible to explore a range of scenarios using them directly with the computer processing power available at the time.

IPCC (1990). Climate change: The IPCC scientific assessment. Report prepared by Working Group I. In J. T. Houghton, G. J. Jenkins, & J. J. Ephraums (Eds.), Intergovernmental Panel on Climate Change (p. 365). Cambridge, UK and New York, NY: Cambridge University Press. Available at the IPCC website.

Prediction scenarios:
  • IS92a: Business as usual
    (410 ppm CO2 in 2017)
  • IS92b: Some emission reduction
    (406 ppm CO2 in 2017)
  • IS92c and IS92d: Global population decline, slow economic growth.
    (399 ppm CO2 in 2017)
  • IS92e: Rapid economic growth
    (420 ppm CO2 in 2017)
  • IS92f: Rapid population growth, modest economic growth
    (415 ppm CO2 in 2017)
We are showing IS92a ("Business as usual") by default, because it was what the IPCC report used as its own default.

The IPCC's projections in their Second Assessment Report used a similar procedure to their first report: detailed models from around the world were used to study the climate, and the IPCC combined their results using a simpler energy balance model to make its projections. Again, there was not enough computing power available to apply all of the models directly across a range of scenarios. The models included better treatment of anthropogenic aerosols, improved carbon cycle simulation, and more.

The scenarios in this report described emission of a variety of greenhouse gases, aerosols, and more, and they covered many different possible assumptions about government policies, demographic changes, and human behaviour. (We are representing them here as different CO2 projections for simplicity.) The IPCC emphasized that these are "representative" scenarios to show what is likely to happen to the climate, and what can be done about it. There were many other possible assumptions and combinations that could be made. These would change the details of the projections, but would lead to the same conclusions: Our behaviour has serious consequences, but there was (and is) still time to change.

The different scenarios all lead to similar predictions for 2017; the differences between them are more obvious by mid-century or so.

Scenarios: IPCC (1992). In J.T. Houghton, B.A. Callander, S.K. Varney (Eds.), Climate Change 1992: The Supplementary Report to the IPCC Scientific Assessment. Cambridge, UK and New York, NY: Cambridge University Press. ISBN: 0-521-43829-2. Available at the IPCC website.

Report: IPCC (1996). In J. T. Houghton, L. G. Meira Filho, B. A. Callander, N. Harris, A. Kattenberg, & K. Maskell (Eds.), Climate change 1995: The science of climate change, Contribution of Working Group I to the Second Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge, UK and New York, NY: Cambridge University Press. ISBN 0‐521‐56433‐6. Available at the IPCC website.

Prediction scenarios:
The Third and Fourth Assessment Reports used a set of 40 scenarios called SRES, which were based on a range of possible global social and economic behaviours. All scenarios gave almost identical near-term CO2 projections, which is why they're not listed individually here.
We are showing SRES A2 by default, because its projection data was readily available, but the choice makes little difference.

For the Third Assessment Report, the IPCC was able to use the results of almost 20 atmosphere-ocean GCMs from around the world, as well as other models for specific aspects of the climate. Many of these models included a wider range of greenhouse gases, improved input parameters, and more detailed studies of model uncertainties.

The IPCC's projections in their Third Assessment Report followed the same procedure as ealier: detailed models from around the world were used to study the climate, and the IPCC combined their results using a simpler energy balance model to make its projections. Again, there was not enough computing power available to apply all of the models directly across a range of scenarios. The time limit was even tighter this time, because the SRES scenarios had been only recently published when the IPCC wanted to produce their next Assessment Report.

The new "SRES" scenarios used in this report covered a wide range of possible global social and economic behaviours. All forty or so scenarios gave nearly the same predictions for the first couple of decades. Longer-term projections, on the other hand, depended much more strongly on the scenario.

Scenarios: IPCC (2000). Nebojsa Nakicenovic and Rob Swart (Eds.), Special Report on Emission Scenarios. Cambridge, UK: Cambridge University Press. Available at the IPCC website.

Report: IPCC (2001). In J. T. Houghton, Y. Ding, D. J. Griggs, M. Noguer, P. J. van der Linden, X. Dai, K. Maskell, & C. A. Johnson (Eds.), Climate change 2001: The scientific basis, Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge, UK and New York, NY: Cambridge University Press. ISBN 0‐521‐80767‐0. Available at the IPCC website.

Prediction scenarios:
The Third and Fourth Assessment Reports used a set of 40 scenarios called SRES, which were based on a range of possible global social and economic behaviours. All scenarios gave almost identical near-term CO2 projections, which is why they're not listed individually here.
We are showing SRES A1B by default, because its projection data was readily available, but the choice makes little difference.

Major model improvements available for the IPCC's Fourth Assessment Report included more accurate and detailed atmospheric simulations and overall finer model resolution. In addition, a number of the models used had a combination of coarser and finer resolution. Important events such as cyclones, as well as Earth system processes such as vegetation growth and spread, often require very fine resolution to properly simulate and understand, but the computing power required to run the entire GCM at such fine resolution would be enormous. "Coupling" models at different scales will likely be an important technique in climate modeling well into the future.

Important Earth system processes were included in the models of the Fourth Assessment Report which had not been available earlier. Highlights include the biogeochemistry of carbon cycles, simulation of land ice dynamics, and atmospheric chemistry.

Scenarios: IPCC (2000). Nebojsa Nakicenovic and Rob Swart (Eds.), Special Report on Emission Scenarios. Cambridge, UK: Cambridge University Press. Available at the IPCC website.

Report: IPCC (2007). In S. Solomon, D. Qin, M. Manning, Z. Chen, M. Marquis, K. B. Averyt, M. Tignor, & H. L. Miller (Eds.), Climate change 2007: The physical science basis, Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge, UK and New York, NY: Cambridge University Press. ISBN 978‐0‐521‐88009‐1. Available at the IPCC website.

What is "hindcasting"?

One important way to check how well a model did is to have it "predict" things that we already know. This is also an important part of the process of making a model. Climate model predictions are also called "forecasts", so this sort of check is sometimes called "hindcasting".

In order to evaluate something, we need to decide what it's for so we know what to focus on. Climate models like these are predicting how global patterns will change over several decades, depending on our behaviour and other factors. In the process, many climate models predict changes from year to year (and even from day to day!). These "short term" changes are an important part of how the model works, but modelers are really only interested in the "long term" results: the averages and trends over decades. This works because those small, short-term variations naturally "average out" over longer times.

So we need to squint a bit when we're evaluating these models. A model doesn't need to get all of the wiggles right (and many of them give smooth lines!). Instead, we look at things like what the slope would be if we drew a smooth line through all the wiggles, for example, or how much warmer the Earth is in the 2000s than it was in the 1970s.

This is also why more recent models are not included in this learning resource. In order to evaluate a model's predictions like this, we need a decade or more of observations to compare to, since that's the time scale the models are really designed for.

It's also important to remember that how the climate changes depends on our own behaviour, since we're the ones putting most of that CO2 and other greenhouse gases into the air. With some models, researchers were able to make predictions under several different scenarios, representing different assumptions about our future behaviour. At other times, researchers did not have the resources for several predictions, so they had to based their assumptions on what seemed to be our most likely path. Always think of climate model predictions as "if we do this, then that will happen."

In this learning resource, you can see roughly how much these assumptions matter to the predictions. Changing the slider for CO2 concentration will bend the forecast curve up or down appropriately. This usually won't be exactly the same as you'd see if you actually re-ran the original models, but it's quite a good approximation.

How do we measure the Earth's temperature?

When we talk about "the Earth's temperature" we're referring to the mean global temperature at the Earth's surface. Land surface air temperatures are measured by thermometers at weather stations around the world. Sea surface temperatures are measured by thermometers in buoys scattered around the oceans. All of these measurements are combined into a single mean surface temperature for the globe. Different sources (listed below) have different methods for combining the measurements, but the results are all quite similar.

On land, we measure the temperature of the air, but on the ocean we measure the temperature of the water. Until recently, climate models would use air temperatures over both land and sea to calculate their mean, which introduced a slight bias in the output since the ocean tends to be cooler on average. This only makes a difference of around 0.01°C, too small to matter for the model comparisons we're making here.

Annual global mean surface temperature data were from NASA GISTemp, NOAA GlobalTemp, Hadley/UEA HadCRUT4, Berkeley Earth, and Cowtan and Way, as provided in the supplemental material of Hausfather et al. (2020). The "Observed" line in the graph shows the mean of the five measurements. For any given year, the sources generally agreed to within 0.01°C.