CO2 as the Primary Driver of Climate Change During the Phanerozoic

I saw a 24 minute talk by R.B. Alley a while back that did a marvelous job of explaining why geologists have overwhelmingly concluded that greenhouse gases (and in particular CO2) are the primary driver of climate changes on geologic time scales. This presentation was given prior to the publication of Judd et al 2024,[1] so I thought it might be fun to show how his argument would be enhanced even more with the more recent data we now have about Phanerozoic temperature and CO2. But let me set the stage.

GMST is set by a balance between incoming absorb solar radiation (ASR) and outgoing longwave radiation (OLR). ASR is affected by changes in how much sunlight reaches the earth (solar variability), where and when sunlight reaches the earth (orbital cycles) and how much is reflected vs absorbed (albedo). Outside influences can also at least theoretically play a role in affecting process on earth (galactic cosmic rays) that could change how much incoming solar is absorbed. OLR is affected by the greenhouse effect (greenhouse gases and high altitude clouds). What geologists have discovered is that Earth's temperature history does not make sense if you try to explain it only with variability in ASR. We're far enough away from the Sun that it cannot heat the planet to the temperatures we experience, even if nearly all the solar energy the Earth receives is absorbed. Earth's temperature history only makes sense when allowing for changes on the OLR side of the energy balance equation. And this is not all that difficult to show. 

ASR: Solar, Orbital and Albedo-Related Influences

Solar Variability

We know that the Sun was only about 70% as bright as it is today when the Earth was formed. This is the basic physics of solar evolution we've understood since Carl Sagan, who did the relevant calculations. The Sun fuses hydrogen into helium, and as it does the density of the solar core increases, and so fusion must occur at a faster rate to balance gravity. Consequently, the Sun has been getting brighter with time, though it's only significant on geologic time scales. TSI can be accurately calculated for any time in Earth's geologic history.[3] 

However, the Earth is so far away from the Sun that even at the Sun's current brightness (1361 W/m^2) our planet should be a giant snowball. With a current value for Earth's albedo of 0.3, the Earth's effective temperature is currently about 255 K (-18°C), but the Sun's brightness was only about 972 W/m^2 when Earth was formed, so the Earth's effective temperature (assuming the same albedo) would be 235K (-38°C) then. So if the Earth was governed solely by changes in TSI, we should have always been a giant snowball in space. 

Orbital Forcings

Orbital forcings are small, and their total effect can be quantified by a variability of about 0.5 W/m^2.[2] These are undetectable on geologic time scales (the longest cycle is 100,000 years) unless the earth is experiencing ice age conditions with varying amounts of polar ice sheets. These forcings are significant for glacial cycles during the Quaternary, but these are too small to show up in the graph below and would not affect the overall trend in TSI or GMST over the last 4.5 billion years.

Albedo

The Earth today reflects about 30% of the sunlight it receives due largely to clouds and ice, but even accounting for a large variability in Earth's albedo, Earth's temperature history on geologic time scales should have been in constant snowball earth conditions, with temperatures increasing by about 20°C over the last 4.5 billion years and variability from albedo adding or subtracting a couple °C throughout Earth's history. Below I show TSI on Earth for the last 4.5 billion years along with the range for Earth's effective temperature assuming a range for albedo between 0.28 and 0.32.[3] If TSI and albedo fully determined Earth's temperature, then GMST would vary between the two blue dotted lines on the graph below.

But this does not resemble Earth's actual temperature history at all. Throughout most of Earth's history, really since we had a solid crust and oceans, the Earth has been warm enough to have liquid water at the surface. In fact, snowball earth episodes make up a very small fraction of Earth's temperature history (and none of these episodes occur during the Phanerozoic), and even "ice ages" with polar ice caps take up much less geologic time than time without polar ice caps. The Earth has normally been much hotter than today even when the Sun was much less bright than today. Now, it's possible that Earth's albedo has varied by more than what I show above, but as R.B. Alley has pointed out, even if albedo was as low than the albedo of the moon (0.11), we would not be able to explain why it is that the Earth has had liquid water at the surface for almost all of Earth's geologic history. If the Earth's temperature were governed solely by the Sun and albedo, the prevailing climate conditions seen on Earth for the last 4.5 billion years would be impossible. 

Cosmic Rays

But what about galactic cosmic rays? Galactic cosmic rays ionize the atmosphere, and this affects cloud condensation nuclei, and this can affect cloud cover, which in turn affects albedo. So maybe cosmic rays can make radical changes in Earth's albedo that may play a role in explaining earth's temperature history. Well this can be tested. Below is a graph Alley shows of a record from the GRIP ice core. A little over 38,000 years ago, the Earth's magnetic field was reduced to near zero, which is seen in the 10-Beryllium spike at the bottom of the graph below. This means that the Earth's surface was virtually unprotected from cosmic rays. And yet even with cosmic rays spilling into the Earth's atmosphere at much higher rates, the effect on temperature in the ice core record is undetectable.

And in fact, numerous studies have investigated the impact of cosmic rays (bibliography here) and they have been unable to detect any trend in cosmic rays that can explain any significant fraction of current warming. This does not mean that cosmic rays do nothing at all. Rather, it means that whatever effect they do have, it's swamped by other factors and so is not useful as an explanation for Earth's temperature history on geologic time scales (or for the last 200 years). Galactic cosmic rays are not a significant factor in explaining Earth's climate history. So while we can clearly see that the Earth's temperature history on geologic time scales is affected by changes in incoming absorbed solar radiation, we can conclusively rule out that these explain all of the variability in Earth's climate. The earth has been much too warm throughout its history for these explanations alone to work.

OLR: Greenhouse Gases

If we include the greenhouse effect, however, we can explain why it is that the Earth has had liquid water at the surface for most of geologic history. With Judd et al 2024, we now have pretty good evidence for both CO2 and GMST for the last 485 million years. CO2 cycles in an out of the climate system on geologic time scales through volcanism, which supply atmospheric CO2, and chemical weathering of rocks is slowly working to remove it as CO2 and H2O interacts with CaSiO3 to remove CO2 from the atmosphere. Living organisms also play a role here, since calcified organisms take up CO2 in the process of forming CaCO3 shells. When these organisms die, they sink to the bottom and their shells get subducted under subduction zones, where volcanoes may return the carbon to the atmosphere as CO2. Tectonic forces change the rates at which these processes occur, as does temperature. Chemical weathering occurs at faster rates when climate is warmer, and so this is a stabilizing influence on climate (on time scales of a half million years). Chemical weathering increases when climate is warm, reducing CO2 concentrations, and it decreases when temperatures are cool, allowing CO2 to accumulate in the atmosphere.

Nevertheless, perturbations of the carbon cycle by long-term and large scale volcanic eruptions (as in the eruption of the Siberian Traps that largely caused the Great Dying) can disrupt the climate system faster than these stabilizing mechanisms can restore it, and so atmospheric CO2 levels have changed by large amounts, with CO2 levels reaching 2000 ppm or even 3200 ppm during some times in the Phanerozoic. Below I show Judd's reconstruction of CO2 concentrations for the last 485 million years. This is the most accurate proxy record we have to date, though there are studies showing variability in CO2 on much shorter time scales that are not represented well on this graph.

The fascinating aspect of Judd's paper is that, even ignoring solar evolution for the last 485 million years, CO2 explains a great deal of Earth's temperature variability, where 2xCO2 correlates with ~7.7°C on geologic time scales with an r^2 of 0.52. CO2 is not the only factor controlling GMST, but it's a very important one.

During the Cenozoic, when data poverty issues are much less of a problem, CO2 concentrations predict global temperatures even better. Here the slope of the regression is a little higher at ~8.2°C for 2xCO2 (x-axis is not logarithmic below) and an r^2 of 0.94.

Perhaps more importantly, because the overall trend in CO2 has been towards a slow decrease in CO2 concentrations, the combination of solar, CO2, and geographic (essentially the positions of continental land masses) forcings explains why we do not see an upward trend in temperatures during the Phanerozoic, as would be expected if solar evolution was the dominant driver of Earth's temperature history. The stabilizing influence from varying rates of chemical weathering appear to be working. Earth's temperature has varied a lot during the Phanerozoic, due to perturbations of the carbon cycle, but there has been no overall trend towards warming on geologic time scales.

Below I show just Judd's GMST reconstruction with ln(rCO2), since the relationship between CO2 and temperature is logarithmic. Clearly the correlation is quite good. And yes I know that correlation is not causation, but there is good work on causation elsewhere (bibliography here). Now, there are some weak spots during the Mesozoic. GMST appears to go very high almost 100 million years ago without a corresponding increase in CO2. Judd refers to this as the Mesozoic Conundrum. This is potentially explained by data poverty issues. It may also be that there are other forcings at work during that time that are not yet accounted for.

However, what is absolutely clear from the geologic record that the Earth's temperature history makes no sense apart from the greenhouse effect. Without a greenhouse effect, the Earth would be a giant snowball for virtually all of its history, and the warmest snowball earth conditions would occur during Earth's recent history. And, w know is not the case. With the greenhouse effect, and in particular CO2, we can explain why Earth has had liquid water at its surface for most of hits history and even what caused snowball earth episodes when they did occur. 

These tectonic and erosional processes perturb the carbon cycle and drive long-term warming and cooling during the Phanerozoic, and the Phanerozoic appears to be dominated by these processes in which CO2 "leads" temperature. However, this does not explain every feature of the climate system. Bolide impacts cause sudden disturbances to the climate system that have little to do with CO2, at least initially - as in the asteroid impact killed the non avian dinosaurs. And the glacial cycles of the Quaternary were triggered by orbital cycles, not by any perturbation of the carbon cycle. These cycles can trigger warming in the Antarctic, and as Antarctic Temperatures (AT) rise, CO2 outgases from the warmer oceans (since CO2 is less soluble in warmer water). This added CO2 amplifies the warming signal and causes global warming.[4] In this case, CO2 follows AT warming but leads global warming. Here, CO2 acts as a feedback in response to Antarctic warming and make a localized warming event become a global warming event. As a result, CO2 forcings in the preindustrial Quaternary are about 6x greater than the forcings from orbital forcings alone[2]. So even in these cases, you can't get observed temperature variability without the impact of CO2.

Conclusion

Of course there is still geologic work to be done, like with the apparent disconnect between CO2 and GMST about 100 million years ago. There are other greenhouse gases at work beyond CO2. There are also time periods with data poverty that may cause limitations for the above reconstruction. But what should be obvious is that the evidence is clear that greenhouse gases, and especially CO2, play the dominant role in explaining changes in Earth's climate on geologic time scales.

Human carbon emissions are causing a perturbation of the carbon cycle that is happening at much greater rates than has been detected in geologic history, including the PETM and the eruption of the Siberian Traps that caused end-Permian extinction. Human emissions have totaled about 750 GtC. The carbon emissions that caused PETM were about 4x larger, but they occurred over tens of thousands of years. Most of our emissions have occurred in the last 150 years. Geologically speaking, the Earth is used to these kinds of perturbations of the carbon cycle. Give the planet a half million years, and chemical weathering will stabilize this perturbation of the carbon cycle, and the Earth will be just fine. But the effect on human civilizations over the next few hundred years will be substantial and expensive. In fact, in terms of the rate of change, we're unwittingly conducting an experiment on Earth's climate that, to the best of our geological knowledge, has never happened before, but the best analogues to it in climate history (PETM) indicate that human civilization is in for a rough ride.

References:

[1] Emily J. Judd et al., A 485-million-year history of Earth’s surface temperature. Science 385,eadk3705 (2024).DOI:10.1126/science.adk3705

[2] Friedrich et al, "Nonlinear climate sensitivity and its implications for future greenhouse warming," Sci. Adv. 2.11 (2016): e1501923.
https://www.researchgate.net/publication/309791338_Nonlinear_climate_sensitivity_and_its_implications_for_future_greenhouse_warming

[3] The Sun increases in brightness predictably over time, and so TSI at any time in geologic history can be calculated as following:

TSI = TSIp/[1+0.4(1-T/To)], where

    TSI = the solar constant at T
    TSIp = the solar constant at present (1361 W/m^2)
    T = time in the past from the age of the earth
    To = the age of the earth (4567 million years)

To calculate effective temperature from TSI, I used the following formula:

Teff = [TSI*(1-α)/(4εσ)]^0.25, where

    α = Albedo

    ε = Surface emissivity

    σ - Stefan Boltzmann Constant

[4] Parrenin, F. et al. “Synchronous Change of Atmospheric CO2 and Antarctic Temperature During the Last Deglacial Warming.” Science 339, 1060 (2013). DOI: 10.1126/science.1226368
https://pdfs.semanticscholar.org/d61d/0fbcb5828af1d434d1bd0282ed36e0f00d2a.pdf