In climate simulation, hot models are ones that predict that global warming will produce significantly higher temperatures than is considered plausible. The problem was first seen in the sixth phase of the Coupled Model Intercomparison Project, where a fifth of the models included began to 'run hot'. According to these models, average global temperatures in the worst-case scenario would rise by more than 5°C above preindustrial levels by 2100, with a "catastrophic" impact on human society. The tendency of models to run hot is typically measured by their estimate of the climate sensitivity, which is how much the earth is expected to warm following a doubling of . Hot models show a climate sensitivity of more than 5°C, whereas empirical observations indicate that the likely range is between 2.3 and 4.7°C. Models with a high climate sensitivity are also known to be poor at reproducing known historical climate trends, such as warming over the 20th century or cooling during the last ice age. Hot models may be the result of problems simulating clouds.<ref name="Voosen 2022" />
|