Just How Good Were the Early Climate Models?

From the Cornwall Alliance for the Stewardship of Creation

by David R. Legates

An article by Nadir Jeevanjee, a Research Physical Scientist at the National Oceanic and Atmospheric Administration (NOAA), recently published in The Conversationand reprinted by Space.Com, suggests that climate models are being given a bad rap. It cites a recent Department of Energy report as using the complexity of climate models as the primary reason why these models cannot be trusted. The article protests, “But the history of climate science tells a different story” because early climate models were instrumental in shaping our understanding of the Earth’s climate.

Okay, let’s revisit history—because those who can’t remember history correctly are destined to get it wrong.

First, we need to get one fact correct. The recent Department of Energy report did not simply cite the complexity of climate models as the primary reason they cannot be trusted. It goes into detail as to why current climate models cannot be trusted: namely, they cannot reproduce current conditions.

For example, models are not capable of determining the equilibrium climate sensitivity to increasing carbon dioxide concentrations. As shown in this graph from the DOE report, they tend to “run hot,” or overstate surface warming.

Global surface air temperature trends (°C/decade), 1979–2024, from various CMIP6 climate models (red, 30-model average in orange); and the average of three thermometer datasets (HadCRUT5, NOAA Global Temp, and Berkeley 1 deg.) and two reanalysis datasets (ERA5 and NCEP/NCAR R1) in blue. Data source: https://climexp.knmi.nl/start.cgi.

They also overstate warming of the tropical troposphere, and their simulation of stratospheric cooling is inconsistent.

With regards to the tropical troposphere, the excessive warming was noted as a problem in the first United States National Climate Assessment and has been noted in every Intergovernmental Panel on Climate Change report since. This problem has become more extreme over time—i.e., simulations from later models, despite decades of “improvements” costing billions of dollars, deviate farther from observations than earlier models—and its spatial extent now encompasses the entire globe.

But the article asserts that early climate model forecasts about global warming were correct decades before the forecasts could be evaluated. It concludes, “it is this track record of success that gives us confidence in interpreting the changes we’re seeing now, as well as predicting changes to come.” It then lists five forecasts the early models made that underscore how well even those early models could reproduce Earth’s climate.

Now, let’s give credit where credit is due, but let’s not reward models for getting the basics of climate correct. In other words, models should not be praised for figuring out that the Poles are cold and the Equator is warm. They should not be given credit for simulating the Hadley Cell circulation, or demonstrating that oceans have a moderating effect on climate, for examples. These basic premises of climate are obvious and must have been a product of the simulation; otherwise, the models would have been discarded before they were even published.

So, let’s test the article’s evaluation of the models.

Forecast number 1: early models correctly simulated surface global warming from increased carbon dioxide. Yes, but this was obvious long before the first models were created. Carbon dioxide is a greenhouse gas, and thus the more of it added to the atmosphere, the higher surface temperature should rise. Way back in 1896, Svante Arrhennius (1859–1927) showed that an increase in carbon dioxide would increase global temperatures.

The question always has been, and still is, how much should the temperature rise from, say, a doubling of carbon dioxide?

Early models argued that the temperature rise would be between 2.0 and 3.5 degrees Celsius. These numbers are actually closer to reality than estimates provided by later model simulations, but they are still higher than data-driven estimates using historical and paleoclimate data.

Verdict? Early models would have been discarded if they had simulated cooling or no effect. That they simulated warming from increased greenhouse gases is no proof of their reliability.

Forecast number 2: the models predicted stratospheric cooling from adding carbon dioxide to the atmosphere. This should also have been relatively obvious. If outgoing terrestrial radiation is being intercepted by more carbon dioxide at the surface, and some of it is radiated back to the surface, then less will reach the stratosphere and, consequently, the stratosphere should cool. Again, the question that demands an answer is, “How much cooling?”—and even current climate models cannot simulate this correctly or consistently.

Forecast number 3: Arctic amplification. The Conversation suggests that it was surprising that the early models accurately predicted that rising carbon dioxide concentrations would warm the Arctic more than the Tropics. But that was no surprise at all. Given a warming planet, the Arctic will warm faster than the Equator due to six factors.

1. Colder air warms more than warmer air from the same amount of energy input due to the derivative of the Stefan-Boltzmann Radiation Law.
2. Moist tropical air has a higher specific heat than dry polar air. (Specific heat is the amount of energy required to raise the temperature of one gram of a substance by one degree Celsius.) This is because water vapor has a higher specific heat than dry air and warm, tropical air has more moisture than cold, polar air.
3. The change in albedo—or surface reflectance—is greater in the Arctic because of the melting of highly reflective ice and snow, uncovering darker soils and tundra.
4. Sea ice provides a layer of insulation between the unfrozen water beneath and the potentially much colder air above. Warming reduces the sea-ice coverage, which allows energy in the relatively warmer water to warm the air above it.
5. The lack of convection at the Poles keeps the warming closer to the surface. Unlike the tropics, where the atmosphere becomes unstable due to surface heating and rising air, the polar regions are not heated sufficiently to create rising air motions, and the creation of the Polar High inhibits vertical motion. Thus, the warmer air stays near the ground.
6. Finally, the evaporation of water stores energy as latent heat—that is, energy in the phase transition of water from liquid to gas—which then is transported poleward by global circulation. This latent energy is stored in equatorial regions and released as condensation occurs in higher latitudes, thereby transporting energy polewards.

Therefore, a greater warming of the Arctic was obvious to climatologists even before the first climate model was created. That models simulate it is no big feat.

Forecast number 4: land-ocean contrast. The article notes that the coupled atmosphere-ocean model “led to a slew of insights, including the observation that land generally warms more than ocean, by a factor of about 1.5.” While this may sound impressive to the uninitiated, climatologists have long known the moderating effect of the oceans, again due to their high specific heat.

Consider San Diego and Dallas, at about the same latitude of 33 degrees north. The annual temperature range in San Diego is about 8 degrees Celsius, while that in Dallas is about 22 degrees—a factor of almost three. Extending any transect to both coasts indicates that the annual temperature range—and the diurnal temperature range, for that matter—increases with distance from the coast.

To anyone who understands the Earth’s climate, the central Plains have a much greater temperature range than either coast, for example. So, why would this have been a novel finding back in 1979? Any competent climatologist would have known this, so it must have been a basic component of any climate model worth its salt at the time.

Forecast number 5: delayed warming of the Southern Ocean around Antarctica. This issue arises from the supposedly novel fact that the Arctic has a greater warming signal than that of the Southern Ocean.

But why is that surprising? The change in albedo is not as great in Antarctica, since there is little land area exposed by melting snow and ice, while there is much in the Arctic. Sea ice dynamics differ considerably in the Southern Hemisphere, and the loss of sea ice isn’t as dramatic to warming Antarctica. Moreover, the Southern Ocean is bounded by a hemisphere of ocean, which has a moderating effect on the climate as mentioned earlier. Thus, important contributions to Arctic warming are not present in the Southern Ocean, and so we would have expected the Southern Ocean to be different from the Arctic.

The take-home message? If early climate models missed any of these supposed climate novelties, climate modeling wouldn’t have gotten off the ground. What early climate modeling showed was that the basics of climate could be replicated with simple mathematical approximations. But that’s the basics. The details are that climate is still very complex, and while the general patterns can be simulated accurately, specific details as to how climate varies spatially and how it is likely to change under various scenarios are still a far-off challenge.

The article concludes: “Climate models have their limitations, of course. For instance, they cannot predict regional climate change as well as people would like. But the fact that climate science, like any field, has significant unknowns should not blind us to what we do know.”

I agree, but I would also note that we were not blind to these facts before the advent of climate models. Syukuro Manabe deserves credit for demonstrating that the basics of climate can be reproduced with a rudimentary computer program run on early computers. But the fact remains that current model simulations still deviate significantly from real-world observations. And let’s not lose sight of the fact that early climatologists understood quite a bit about how the Earth’s climate works. Yes, there were climatologists before the climate modelers, and they knew an awful lot about the Earth’s climate.

David R. Legates, Ph.D. (Climatology), is retired Professor of Climatology at the University of Delaware and Director of Research and Education for the Cornwall Alliance for the Stewardship of Creation.