Assessing the warming biases in CMIP6 models: the roles of fast response and cumulative effects to external forcings

The “hot model” problem is a widely recognized concern, highlighting the need to assess whether a climate model exhibits a warm bias before utilizing its simulations. Traditionally, climate sensitivity indicators such as the Transient Climate Response or the Equilibrium Climate Sensitivity have been used for the assessment, which, however, requires substantial computational resources and suffers from high uncertainty. Here we propose a novel method based on the scaling behavior of the climate system to objectively evaluate warm biases in climate models. The method relies on two indices, $$a$$ and $$H$$, which measure the fast responses of global mean surface temperatures to external forcings and their cumulative effects, respectively. By comparing the ($$a,H$$) values from climate models with those derived from observations, one can readily identify whether a model tends to be too warm or too cold. Detailed analysis indicates that the overestimated cumulative effects of temperature responses to external forcings are a primary driver of warming biases in CMIP6 models. Since the two indices can be derived directly from historical observations and simulations, they together provide an efficient framework for model evaluation, improvement, and calibration.