WUWT editors’ note:

Watts Up With That? is committed to fostering open discourse on climate science and related topics. While we respect the authors’ perspective and their dedication to exploring climate dynamics, we find aspects of the CO₂ thermalization theory presented in this article to be inconsistent with well-established experimental and empirical evidence. As Richard Feynman famously stated, “It doesn’t matter how beautiful your theory is, it doesn’t matter how smart you are. If it doesn’t agree with experiment, it’s wrong.”

Extensive laboratory spectroscopy and direct atmospheric observations confirm that CO₂ plays a role in radiative heat transfer, and while water vapor is indeed the dominant greenhouse gas, the claim that CO₂’s effects are negligible does not align with measured data. That said, scientific inquiry thrives on scrutiny and debate, and we encourage readers to critically evaluate all perspectives in light of experimental validation and real-world measurements. Anthony has written primer on Carbon Dioxide Saturation in the Atmosphere also worth reading, as it describes how the logarithmic effect of CO₂ versus temperature will continue to lessen its impact even as atmospheric CO₂ concentrations increase.

By Andy May & Tom Shula

Fundamentally the entire man-made CO₂ global warming concept, boils down to the interaction of energy and matter in Earth’s atmosphere. The only reason that CO₂ and other greenhouse gases (GHGs) are special is that they absorb most of the radiation emitted by Earth’s surface. Water vapor absorbs across almost the entire emission spectrum and is, by far, the most significant absorber. The cloud-free atmosphere is mostly transparent to sunlight, so Earth’s surface absorbs most of the sunlight that makes it through the clouds. In response to this stimulation, it emits infrared radiation (IR).

Because the humid lower atmosphere is nearly opaque to most surface emitted radiation that is outside the atmospheric windows, surface emissions are absorbed by GHGs very close to the surface. According to Heinz Hug, at sea level, with a CO₂ concentration of 357 PPM and 2.6% water vapor, 99.94% of all surface radiation in the main CO₂ frequency band at about 15 μm is normally absorbed in the lower 10 meters of the atmosphere (Hug, 2012). Even at the edges of the deep CO₂ frequency band (see figure 1, as well as figures 4 & 5 here) where any increase in the CO₂ effect would be observed, 99.9% of the surface radiation is absorbed in the first 690 meters (Hug, 2000).

Heinz Hug goes on to say that is why climate change caused by CO₂ cannot be measured directly in the laboratory and can only be modeled. In our opinion, the effect of CO₂ is so small it will likely never be measured. In a similar fashion, any “back radiation” that makes it to the surface, outside atmospheric windows, is from the lower 10 meters of the atmosphere, the remaining emissions from the lower 10 meters of the atmosphere are captured by other greenhouse gases, almost always water vapor molecules.

Surface emissions in the frequencies that cannot be absorbed or emitted by GHGs, those in the so-called “atmospheric windows” are not captured, these are the frequencies utilized by IR thermometers and scanners, typically 7.5 to 14 micrometers as shown in figure 1. Water vapor is often a very weak absorber and emitter in portions of these windows. Carbon dioxide strongly absorbs and re-emits IR at two key frequencies: around 4.26 μm (microns) and 14.99 μm. The common vanadium oxide (VOx) based microbolometer long-wave infrared detectors cover wavelengths from 8-14 µm range. So, both CO₂ absorption bands are outside the range of the common hand-held infrared thermometer/bolometer.

The radiation seen when IR thermometers and scanners are pointed at the sky is surface radiation scattered by atmospheric particles and clouds. The radiation seen by IR thermometers and scanners cannot be emitted by greenhouse gases or clouds because neither GHGs nor clouds emit in frequencies that can be detected by the devices. As noted in van Wijngaarden and Happer (2025) scattered longwave IR originates only in water droplets or ice or other particulates, there is negligible scattering of IR by molecules, especially in the atmospheric windows.

When GHG molecules absorb radiative emissions from the surface or other GHGs they become excited and rise above their molecular ground state and then either dissipate the excess energy among their neighbors as kinetic energy through collisions, or emit the energy according to their specific frequency of emission (Hug, 2000). In the lower atmosphere, dissipation is much more common than emission, but when emission takes place, the emitted energy is quickly captured by nearby GHGs and they dissipate it to their neighbors. Radiant energy from the surface or other GHGs that is captured by a greenhouse gas molecule is held for a relatively long time, around a half second, before it is re-emitted. In this half second, the molecule will have around three billion collisions with other molecules at sea level (Siddles et al). Siddles et al. also report that the excited molecule is 50,000 times as likely to dissipate excess energy as emit it as energy at sea level. Radiative return to the ground state is insignificant in the lower atmosphere (Hug, 2000).

Dissipating the excess energy via collisions warms the neighborhood around excited GHG molecules, and is called thermalization. Thermalization increases the gas’s sensible heat and stimulates convection, these processes increase both evaporation and conduction of heat from the surface. Conduction directly transfers sensible heat from the surface to the air and evaporation carries away latent heat.

Now we reach a point where it gets confusing. The surface has emitted most of its excess thermal energy and stored the rest. What happens now? Most descriptions of the greenhouse effect emphasize heat transfer through the atmosphere via radiation and either ignore heat transported by convection or fudge an adjustment in the tropospheric lapse rate to “correct” for convection. If a vertical atmospheric temperature profile is created using a radiative transfer model it does not match observations. Thus, to create a reasonable atmospheric radiative heat transfer model, one must assume a temperature profile that approximates reality. A typical assumed profile can be seen in Wijngaarden and Happer (2020) as part of their figure 1.

In Manabe and Wetherald (1967) and in Manabe and Strickler (1964) they simply force the lapse rate to be below 6.5°C to accommodate the effect of convection. Convection decreases the lapse rate to about 6.5°C/km on average from about 9.8°C/km in the pure radiative equilibrium case as shown in figure 2 from Manabe and Strickler. The reduction is due to extra heat being retained in the climate system by convective processes. Radiative heat transfer is faster than convective cooling and the oceans and atmosphere (collectively the “climate system”) have a considerable heat capacity and store thermal energy for varying lengths of time. The radiative heat transfer assumptions in the conventional “consensus” greenhouse gas model of climate change do not match the real world, so the vertical temperature profile must be assumed, it cannot be modeled.

Convection

When the Sun elevates the surface temperature, conduction and evaporation cause the lower air to become less dense, and it begins to rise. Convection starts spontaneously. Convection carries heat, both latent and sensible, higher into the atmosphere where it is colder. The water vapor condenses in the cooler upper air, releasing its latent heat, and the resulting drier and denser air descends to evaporate more water and continue the circulation.

The uppermost boundary of the circulation is the tropopause at the top of the troposphere. At the tropopause, the air pressure and density are lower, and water vapor is nearly gone. The tropopause is well above the so-called “emission layer” (about 5 km on average, with a temperature of about 255K) where water condenses, and on average, sends most of its latent heat to space as emitted radiation. The latent heat release warms GHGs (mostly water vapor molecules) in the neighborhood and stimulates them, which induces emissions. In this atmospheric region, between the emission layer and the tropopause, water vapor largely disappears, convection subsides, and most emissions of OLR (outgoing longwave radiation) to space occur. In this region, thermalization is harder to achieve due to lower atmospheric density and low humidity, and emitted radiation goes farther. At some altitude within the region, and below it for some frequencies, emitted radiation can escape to space.

Thermalization as described above, can work in reverse. Molecules warmed by latent heat that is released by condensing water vapor or upward convection of warm air can collide with GHGs and cause them to become excited and emit radiation. This is especially true of water vapor which is more easily excited by collisions than CO₂. This is another reason why nearly all emissions to space are from water vapor.

Koll & Cronin

Koll & Cronin (2018) show that for typical terrestrial temperatures, the magnitude of total outgoing longwave radiation (OLR) is a linear function of near surface temperature. This is consistent with Newton’s law of cooling.

Most of the energy lost to space comes from water vapor emissions, emissions by other GHGs are insignificant. Koll and Cronin go through a very tortured analysis of their data in order to continue calling water vapor a “feedback” to CO₂, but their data shows that water vapor is in the driver’s seat and the other GHGs have little effect on Earth’s cooling rate. All GHGs can absorb energy emitted from the surface, but nearly all the energy (except in deserts and at the poles in winter) is absorbed by water vapor. There are a lot more water vapor molecules than molecules of the other GHGs in the troposphere, so water vapor both absorbs and emits nearly all the radiation.

In a radiative world, one might assume that OLR would be consistent with the Stefan-Boltzmann equation (σT⁴, red line in figure 3), however Koll and Cronin’s data show this is not the case. The red line shows the outgoing IR radiation calculated assuming that Earth’s surface was cooling via radiation, as in Manabe’s CO₂ hypothesis. Newton’s law of cooling predicts that surface temperature will be linear with OLR if the surface cools via convection. The only condition is that the fluid properties should not change much.

What Shula & Ott propose is that the radiation emitted by the surface and the radiation observed from a satellite are decoupled from one another by the conversion of surface radiation to sensible heat by GHGs very near the surface. The added sensible heat is what drives the convection. Convection transports thermal energy upward, and in the critical region between around 2 and 7 km spontaneous radiation emissions, mostly from water vapor, are radiated to space. It is not surprising that the previously mentioned “emission layer,” at 5 km deduced from satellite OLR observations, with a temperature of about 255K (-17.5°C) lies in the middle of this region. Between 2 and 7 km is where upwardly convected water vapor condenses or freezes out of the air, releasing its latent heat, and forms clouds. The extra heat stimulates other water molecules (and a few other minor GHGs) causing them to emit radiation, much of which makes it to space. Hermann Harde modeled water vapor emissions as seen from 12.5 km and figure 4 shows the spectrum from his model (Harde 2013).

Water vapor dominates atmospheric OLR emissions because it can emit across nearly the entire IR spectrum. Water vapor is also more easily stimulated to emit radiation than other GHG molecules (Harde, 2013).

Discussion

The argument presented in most descriptions of the radiative greenhouse effect is one-dimensional and relies on average temperature profiles and solar irradiance. In order to use these one-dimensional models in a three-dimensional global climate model, modelers invoke a hypothetical local radiative equilibrium. Local thermodynamic equilibrium (LTE) is a mathematical abstraction and tool used in climate models. It means that within an “air parcel” of arbitrary size all of the molecules are in thermodynamic equilibrium. The air parcels are not in thermodynamic equilibrium with one another. Air parcels move heat and mass between each other, but not internally. The size and definition of a parcel is determined by the computer modeler, and is usually too large to be realistic. Clearly, regionally, over large areas, near the surface, the atmosphere is never in equilibrium and convection is persistent. If a parcel is large enough to contain a tornado, it is obviously not in thermodynamic equilibrium.

In modern general circulation climate models (GCMs or ESMs, short for Earth System Models in AR6) the cells in the model (“air parcels”) are one degree latitude by one degree longitude or more than 10,000 square km at the equator. These cells can easily contain a large thunderstorm containing multiple tornados. Even higher resolution regional models are no better than 100 sq. km (AR6, WGI, page 1140). By way of comparison the average diameter of a thunderstorm is 24 km, this is an area of about 450 sq. km.

Earth, as a whole, is a dynamical system with diurnal and seasonal cycles and is never in equilibrium. Its radiative energy input and output never matches or is in equilibrium anywhere on Earth’s surface, except in very small volumes over very short periods of time. The whole greenhouse effect concept assumes that energy-in and energy-out approximately balance over the whole planet (Manabe and Wetherald, 1967) and anything left over, the “energy imbalance,” is what warms or cools the planet on average (Trenberth, et al. 2014).

If the planet had a constant input and radiative heat transfer were the actual cooling mechanism, this could be true. In that scenario, perturbing the model by increasing the CO₂ concentration to create a “radiative forcing” would result in a different equilibrium temperature. However, even if the surface cooled with radiative emissions and the process were not short circuited by convection, the radiative forcing of CO₂ is approximately logarithmic with its atmospheric concentration. This is due to the distribution of its absorption coefficients in the large CO₂ 15μm wavelength band (Romps, et al., 2022). Put another way, the radiative impact of going from 50 ppm to 100ppm is the same as going from 400 ppm to 800 ppm.

However, the data we’ve shown suggests that radiative heat transfer only occurs at the top and bottom of the atmosphere, in between convection rules and convection is very complex with a lot of constantly changing associated heat, or more precisely thermal energy, storage capacity. Standard radiative models use simplifying assumptions to account for the average changes to the vertical temperature distribution caused by convection. These assumptions can work to make reasonable one-dimensional models, but do not work in our three-dimensional rotating real world. In reality, the vertical temperature profile, and the lapse rate, change constantly and from place to place.

Convection is not just a train that transports heat from the surface to the TOA at a constant rate everywhere. Its pathway and efficiency are constantly changing, which causes our weather. Plus, it has a very powerful energy storage cell at the bottom, the world ocean. As circulation changes, the amount of energy stored in the ocean changes. The amount is trivial to the ocean, with its immense heat capacity, so its temperature normally does not change significantly, except in the shallow mixed layer. But when atmospheric and ocean circulation changes, and become more or less efficient, the atmospheric temperature changes dramatically due to its smaller heat capacity and density. Everyone seems to ignore the considerable heat storage in the climate system and the storage time factor. Energy residence time makes a difference, and it does change with time. Earth’s surface contains more heat (aka thermal energy) than the surface of Venus, yet the surface temperature on Venus is 464°C, because Venus has no water or oceans.

The impact of energy storage in the climate system can be seen in long-term temperature records, such as the Vostok record compiled by Petit, et al. In figure 5 we can see that the entry into a warm interglacial climate state is very rapid as these are caused by increased insolation onto the critical northern continents. However, the descent into the next glacial is very slow since draining the heat stored in the oceans is a very slow process. All this needs to be incorporated into climate models for them to make more sense. On shorter time scales, the effect of changing ocean storage on our climate can be seen in the ENSO cycle (see figure 2.4 here), the Atlantic Multidecadal Oscillation (AMO, see figure 6 here), and in the Pacific Decadal Oscillation (PDO, see figure 4.8 here). Also see the discussion of the AMO and global average surface temperature around figure 2 of May & Crok 2024.

Reality is more complex than we can explain today, and we haven’t even touched on the impact of variations in cloudiness (van Wijngaarden & Happer, 2025).

A bibliography for this post can be downloaded here.

Much of this post is a result of discussions with Markus Ott. Contributions to the post were also made by Will Happer and Anthony Watts.

A related paper by Tom and Markus can be downloaded here.

[H/T Watts Up With That]