Earth’s Climate Sensitivity
Halfway up the Ringberg, overlooking the Tegernsee, lies a castle used by the Max Planck Society as a conference venue. It is a contemplative place where scientists have long come to retreat, to think, and debate. One topic that has been discussed here on several occasions is the extent of human-made global warming. In March 2015, 33 scientists gathered at Schloss Ringberg searching for an answer. They had come at the invitation of Bjorn Stevens, director at the Max Planck Institute for Meteorology (MPI-M) and Sandrine Bony of the Centre National de la Recherche Scientifique (CNRS) who were leading a new initiative of the World Climate Research Programme.
In a stable climate, the power of the visible radiation that Earth receives from the sun is balanced by the power of the thermal radiation that Earth returns to space. Currently, this balance is disrupted by anthropogenic emissions of carbon dioxide (CO₂): CO2 inhibits Earth’s ability to radiate thermal energy to space. To overcome this hinderance, and thus restore the radiation balance, Earth’s surface must warm. The extent of this warming depends on both the level of emissions and the planet’s response to them. Climate sensitivity is a measure of the latter and determined in the form of various metrics (see table).
| Equilibrium Climate Sensitivity (ECS) | Long-term temperature increase after an instantaneous doubling of CO2 upon reaching equilibrium |
| Effective Climate Sensitivity (EffCS) | Approximate value derived from observational data or models, based on the energy imbalance for a specific period (e.g., recent decades) that is shorter than that required to reach equilibrium. May change with the spatial warming pattern. |
| Transient Climate Sensitivity (TCR) | Temperature increase (°C) at the time of CO2 doubling in an idealized scenario with 1% CO2 growth per year (doubling occurs after approximately 70 years). The value is significantly lower than ECS or EffCS. |
| Transient climate response to cumulative emissions (TCRE) | Temperature increase per 1,000 gigatons of cumulative CO2 emissions. |
Many scientists use equilibrium climate sensitivity (ECS), by virtue of its simplicity. ECS is defined as the answer to a thought experiment phrased as follows: By how much would the global mean temperature need to increase to balance a doubling of the carbon dioxide content of the atmosphere, assuming that the great ice-sheets and other commensurately slow processes remained unperturbed?
The ECS value is derived from the ratio between the radiative forcing and the climate system’s response, often called feedback. Given a known radiative forcing, such as that from CO₂, the feedback must be determined to calculate the temperature increase or ECS. If the feedback is strongly negative, a slight warming is sufficient to restore balance. Conversely, if the feedback is weak, the temperature increase is larger. If it was positive, the system would be unstable, and the planet would warm uncontrollably.
If the Earth were an ideal body without an atmosphere, the feedback would be easy to calculate (black body radiation). In reality, the Earth’s response is modulated by the peculiarities of the atmosphere. For example, water vapor, the planet’s reflectivity (albedo), and clouds all play a role. Additionally, warming isn’t uniform; rather, a specific warming pattern develops. This also influences climate sensitivity, an effect that Bjorn Stevens and others refer to as the “pattern effect” (see box).
Pattern Effect
The pattern effect refers to the dependence of climate sensitivity on the geographical pattern of warming. This can change over time, leading to changes in climate sensitivity. Therefore, observations regarding climate sensitivity can only be interpreted correctly if the pattern effect is understood. The department headed by MPI-M Director Sarah Kang is therefore investigating the temperature pattern in the tropical Pacific and why it is not captured by conventional models.
How climate sensitivity is determined
The first famous attempt to determine ECS dates back to the end of the 19th century, when Svante Arrhenius presented first estimates of the temperature increase in response to an increase in CO2. In an influential paper published in 1967, Syukuro Manabe and Richard Wetherald calculated ECS using a simple model of the tropical atmosphere. Later, ECS was determined using three-dimensional atmospheric circulation models. In 1979, a group chaired by meteorologist Jules Charney prepared a report for the U.S. National Academy of Sciences on the effect of carbon dioxide on the climate. The report gave an ECS of 3°C with an uncertainty of 1.5°C. Over the next 40 years, further studies estimated the ECS using observations from the recent past, reconstructions from the distant past, and complex Earth system models. However, this did not initially change the range specified in the Charney report.
In March 2015, 33 scientists gathered at Schloss Ringberg to discuss ways to further narrow down the range of 1.5°C to 4.5°C. During the meeting, they shared their preliminary findings in presentations, asked each other critical questions, discussed possible solutions, and organized their thoughts on a blackboard. They also took a wholly new approach to the problem. Rather than proposing new values for the ECS, the scientists systematically re-examined existing ones, starting with the extreme values and using models, observations, and theory to assess their plausibility. Rather than asking what evidence allowed, they asked what it excluded. In a 2016 study, Bjorn Stevens and his co-authors presented this new approach. Which values for the ECS seem plausible when considering multiple lines of evidence? This method allowed them to reject very low and very high extreme values. Building upon and systematizing this approach, Steven Sherwood brought together a wide range of expertise to perform a more formal assessment, which was published in 2020. Shortly thereafter, the Intergovernmental Panel on Climate Change (IPCC) incorporated the results of this study into its Sixth Assessment Report, published in 2021, giving a range of 2.5°C to 4°C. This was the first robust reduction in uncertainty in estimates of the ECS since the original Charney report.
Open questions
The meeting at Schloss Ringberg marked an important turning point. Narrowing down the likely range of ECS was a significant achievement for climate research. However, the fact that the range is still quite large highlights how the seemingly simple concept of climate sensitivity is actually a complex problem. Many aspects of this problem are being addressed at the MPI-M. Researchers are investigating the planet’s response in the absence of clouds, as well as cloud feedbacks, which continue to cause large uncertainty. Additionally, they are investigating the temperature pattern in the tropical Pacific, which is relevant to the pattern effect. The tools used to study these issues include simple, idealized models, Earth system models, theoretical applications of radiant energy transfer, and observations.
For example, Lukas Kluft and Sally Dacie (now at the Potsdam Institute for Climate Impact Research) jointly developed a one-dimensional model for investigating the radiation balance. This model is similar to that of Manabe and Wetherald, but it uses modern methods. They named the model “Konrad” after the two essential processes in it: convection (Konvektion in German) and radiation. Using this model, Kluft and others were able to confirm the results of Manabe and Wetherald’s seminal study. However, Konrad also made it possible to resolve more details, including a better representation of convective adjustment of the vertical temperature profile, and a more physical, spectral treatment of the radiant energy transfer. “By looking at the ECS at different wavelengths, you can attribute changes in the ECS to different physical processes—for example, related to water vapor,” explains Kluft.
The role of water vapor
Water vapor is a powerful greenhouse gas that has a significant impact on how efficiently the Earth radiates heat into space. Kluft compares this to a pond: suspended particles cloud the water, obscuring the view of the bottom. Similarly, water vapor shifts the effective “view from space” away from the surface—however, by differing amounts depending on the wavelength. The bottom is only visible at a few wavelengths, and at many wavelengths, water vapor shifts the view from the warm surface to the higher, colder layers of air. In other words, the Earth can radiate heat more efficiently where the atmosphere is dry. In humid air, radiation into space is impeded. Thus, water vapor has a significant influence on climate sensitivity that can only really be understood by looking across the spectrum of thermal radiation, something Stevens and Kluft have called the colorful view of climate sensitivity.
To benefit from this colorful view, theoretical estimates of climate sensitivity must make an assumption about how water vapor changes as a result of warming. One plausible assumption, which the study by Manabe and Wetherald was based on, is that relative humidity remains constant. This means that temperature limits how much water vapor the atmosphere can absorb—the warmer it is, the more it can absorb. In principle, the assumption of constant relative humidity is very robust, especially compared to the alternative—the constancy of the total amount of water vapor. This is demonstrated by measurements in the tropics. MPI-M researchers Helene Gloeckner, Bjorn Stevens, and others evaluated data from the ORCESTRA measurement campaign, which took place in the tropical Atlantic in 2024 and was coordinated by the MPI-M. A comparison with the GATE expedition, which had been conducted 50 years earlier at the same location, showed that the temperature had risen as expected and the increase in water vapor in the atmosphere was also confirmed.
“This shows that our theoretical assumptions are generally valid,“ says Gloeckner. Nevertheless, reality is somewhat more complicated. For instance, different assumptions about the initial vertical distribution of relative humidity yield different values for climate sensitivity, as another study using the Konrad model by Stella Bourdin, Lukas Kluft, and Bjorn Stevens showed. Furthermore, when theoretical estimates are checked against observations, small deviations become apparent. Helene Gloeckner, Lukas Kluft, Hauke Schmidt, and Bjorn Stevens calculated the clear-sky long-wave feedback parameter using reanalysis data, which combine observations as assimilated by weather models. The scientists concluded that the calculated average value for the clear-sky long-wave feedback parameter is 10 to 20 percent lower (i.e., climate sensitivity is higher) than predicted for an atmosphere with constant relative humidity. The scientists emphasize the need for long-term monitoring of humidity variations in the atmosphere to determine the accuracy of the assumption of constant relative humidity.
The great unknown: clouds
Even for a cloudless sky, correctly calculating the feedback is challenging, although scientists have made significant progress. But things become even more complex when clouds come into play.
“Clouds remain the great unknown, a seemingly never-ending frontier, in climate research,” says Hauke Schmidt, group leader at MPI-M. They are complicated by their ability to do many things. They can either strengthen or weaken the feedback, and they can modify both the solar and terrestrial streams of energy. They behave differently at day versus night, in winter versus summer, over land versus ocean. Low-lying clouds, in particular, reflect incoming solar radiation and thus cool the Earth. This is especially effective over dark seas, for cumulus clouds in the trade wind region, for example. Conversely, clouds also prevent thermal radiation from escaping into space, which has a warming effect, and this effect is stronger the greater the temperature difference between the cloud and the surface. Hence it can dominate the effect of clouds on solar radiation in the case of high clouds, or for all clouds at night. Changes in the extent or brightness (i.e., albedo) of clouds, as well as their temperature, could therefore influence the cloud feedback and thus climate sensitivity. On top of this, changes in the underlying surface can confuse interpretations of the effect of cloud changes.
To isolate the aspect of changing cloud temperature, Konrad was used once again at MPI-M. Lukas Kluft and others examined the radiative effect of clouds at constant albedo. Mid-level and high clouds were assigned to higher altitudes to largely maintain their temperature. Although the presence of clouds, even at unchanged temperature, leads to more warming, the higher altitude also dampens the greenhouse effect of carbon dioxide. These two effects partly cancel each other out, so that the overall climate sensitivity remains nearly unchanged.
“The real uncertainty is how cloud cover will change,“ says Kluft. For a long time, there was concern, based on experiments with climate models using a statistical representation of cloud effects, that surface warming would cause the ubiquitous low cumulus clouds in the trade wind region to disappear. More specifically, it was hypothesized that warming would cause more dry air from above to mix into the lower cumulus cloud layer, drying out the clouds and causing them to reflect less radiation. Researchers tested this and other hypotheses in January and February of 2020 during the EUREC4A field campaign, led by the MPI-M and the CNRS at Sorbonne University in Paris. They used research vessels, aircraft, and autonomous and remote-controlled platforms, anchored by long term measurements at the Barbados Cloud Observatory, which is operated by MPI-M and the Caribbean Institute for Meteorology and Hydrology. Based on this data, a research team led by Raphaela Vogel from the University of Hamburg was able to refute the “mixing-desiccation” hypothesis. The team, which included researchers at the MPI-M, emphasized the importance of intermediate (or meso-) scale circulations, i.e., circulation extending up to 200 kilometers, in supporting cloud formation. Conventional climate models are oblivious to these, and hence such factors cannot be included in their statistical representation of clouds. This leads to a wrong response of trade wind clouds to global warming.
Nevertheless, this key finding of the EUREC4A campaign is only partial relief. It doesn’t say that shallow clouds won’t go away, it just says that if they go away it won’t be for the reasons that climate models purport. Furthermore, other cloud regimes that the models also don’t accurately represent may be a source of change. In a 2024 study led by Helge Gössling of the Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research in Bremerhaven the strong temperature increase in 2023 was attributed to reduced cloud cover, particularly in the mid-latitudes of the northern hemisphere and the tropics. Bjorn Stevens speculates that this might be related to changes in atmospheric circulation. But changes in the storm-tracks, paths in the extra-tropics followed by the familiar winter storm systems, might also be a factor.
One of the aims of the ORCESTRA measurement campaign was to measure and understand deep convection and mesoscale circulation in the tropical rain belt. Another measurement campaign, “STACCATO,” planned for 2029, will focus on subtropical stratocumulus clouds. In addition to Stevens, Franziska Glassmeier will also be involved in the campaign. She heads the independent Lise Meitner research group “Multiscale Cloud Physics” at MPI-M, which investigates the self-organization of clouds on different spatial scales.
Complex models
Since clouds are complex and the relevant processes extend from the microscale to the planetary scale, conventional climate models are forced to decouple them from the circulation systems within which they arise and to estimate their coverages statistically. To do so requires a great number of simplifying assumptions, which are then incorporated into the models in the form of parameterizations. State-of-the-art, high-resolution climate models offer the possibility of approaching these questions more physically. Their computational grid can be much finer than that of conventional models, allowing the circulation systems in which clouds form to be resolved. “By reducing the grid size of our model to less than one kilometer, we hope to learn more about how clouds are changing under global warming,” says Bjorn Stevens. In an earlier project called HD(CP)2 Stevens and colleagues were able to show that at scales of about 150 meters, many trade-wind cloud regimes could be well represented. The WarmWorld project, led by the MPI-M and funded by the German Federal Ministry of Research, Technology and Space, builds on this finding and is working toward global simulations at this scale. According to Stevens, continued warming will likely reveal the fate of clouds and climate sensitivity, provided that important observation systems are maintained and further developed. However, to successfully interpret these observations will still require theoretical and modeling advances.
The world is changing, more profoundly so than many researchers anticipated. These changes are particularly pronounced in the building blocks of climate sensitivity and are calling researchers back to Schloss Ringberg in March 2026. From 16-20 March, Bjorn Stevens expects 42 guests who will hope to structure hypotheses constructed to explain the growing radiation imbalance at the top of the atmosphere and devise new research programs through which they hope to test their ideas. They will follow on the EUCLIDYAN initiative led by Sarah Kang that will occupy the castle in the beginning of the same month, during which they aim to construct a research program for another aspect, namely understanding puzzling dynamical changes that might also be at play and complicate interpretations of models and measurements.
Original publications
Bourdin, S., Kluft, L., & Stevens, B. (2021). Dependence of climate sensitivity on the given distribution of relative humidity. Geophysical Research Letters, DOI: 10.1029/2021GL092462
Gloeckner, H. M., Kluft, L., Schmidt, H., and Stevens, B. (2025). Estimates of the global clear-sky longwave radiative feedback strength from reanalysis data. Geophysical Research Letters,52, e2024GL113495. DOI: 10.1029/2024GL113495
Kluft, L., Dacie, S., Buehler, S. A., Schmidt, H., and Stevens, B. (2019): Re-Examining the First Climate Models: Climate Sensitivity of a Modern Radiative–Convective Equilibrium Model. J. Climate, 32, 8111–8125, DOI: 10.1175/JCLI-D-18-0774.1
Kluft, L., Stevens, B., Brath, M., and Buehler, S. A. A conceptual framework for understanding longwave cloud effects on climate sensitivity (2025). Atmospheric Chemistry and Physics, 25, 9075–9084. DOI: 10.5194/acp-25-9075-2025
Stevens, B., Abe-Ouchi, A., Bony, S., Hegerl, G., Schmidt, G., Sherwood, S., and Webb, M. (2015) Ringberg15: Earth's Climate Sensitivities. 23-27 March, Schloss Ringberg, Germany. WCRP Report No. 11/2015.
Stevens, B., Sherwood, S.C., Bony, S., and Webb, M. J. (2016). Prospects for narrowing bounds on Earth’s equilibrium climate sensitivity, Earth’s Future, 4, 512–522, DOI: 10.1002/2016EF000376.
Stevens, B. and Kluft, L. (2023): A colorful look at climate sensitivity. Atmos. Chem. Phys., 23, 14673–14689, DOI: 10.5194/acp-23-14673-2023
Contact
Prof. Dr. Bjorn Stevens
Max Planck Institute for Meteorology
bjorn.stevens@mpimet.mpg.de
Dr. Hauke Schmidt
Max Planck Institute for Meteorology
hauke.schmidt@mpimet.mpg.de
Helene Glöckner
Max Planck Institute for Meteorology
helene.gloeckner@mpimet.mpg.de
Dr. Lukas Kluft
Max Planck Institute for Meteorology
lukas.kluft@mpimet.mpg.de