What are the main sources of uncertainty in equilibrium climate sensitivity, and why did CMIP6 models run hotter than CM

Equilibrium climate sensitivity (ECS)—the long-term warming expected if atmospheric CO2 concentrations were to double—remains a central focus of climate science. Its uncertainty is primarily driven by the complexity of how different components of the Earth system interact.

Main Sources of Uncertainty in ECS

The range in ECS estimates stems from how climate models represent "feedback loops," which can either amplify or dampen the initial warming caused by greenhouse gases.

1. Cloud Feedbacks: This is the largest source of uncertainty. Models struggle to accurately represent the formation, distribution, and optical properties of clouds—particularly low-level stratocumulus clouds. Whether these clouds dissipate or become more reflective as the planet warms significantly alters the net energy balance.
2. Aerosol-Cloud Interactions: Aerosols (tiny particles like sulfates or soot) can reflect sunlight and serve as nuclei for cloud droplets. Understanding how these aerosols interact with clouds—and how they have offset warming in the past—is difficult to parameterize at the fine scales required for global models.
3. Convection and Small-Scale Processes: Climate models operate on grids that are often too coarse to resolve individual thunderstorms or turbulent air masses. These processes must be "parameterized" (approximated), and different mathematical approaches to these approximations lead to different climate sensitivities.

Why CMIP6 Models Run "Hotter"

The sixth phase of the Coupled Model Intercomparison Project (CMIP6) yielded an average ECS higher than that of CMIP5, with several models exceeding the range considered likely by observational evidence. This shift is generally attributed to the following:

• Improved Physical Realism: In an effort to better represent microphysical processes, developers made the models more sensitive to small-scale cloud dynamics. Specifically, many CMIP6 models adopted new, more sophisticated schemes for cloud droplet formation and ice-phase transitions.
• Stronger Positive Feedbacks: Some of these new parameterizations inadvertently created a stronger positive feedback loop than exists in the real climate. For example, some models simulate a faster reduction in cloud cover in response to warming, which lets more solar radiation reach the surface, thereby increasing sensitivity.
• Calibration Choices: When tuning models to match historical data, developers face a "constrained optimization" problem. Because they are trying to fit historical warming while simultaneously representing more complex physics, some models end up with higher ECS values that still remain consistent with historical temperature trends, primarily because they also include higher aerosol forcing (which provides a "cooling" effect that masks the high sensitivity).

In short, the higher sensitivity in CMIP6 is largely seen as a byproduct of trying to capture more complex, high-resolution physical processes that, in some cases, result in cloud responses that are more aggressive than what the paleoclimate record and current observational constraints suggest.