A Physically-Based Cloud Classification Scheme Resolves Extratropical Low Cloud Feedback Discrepancies

The Equilibrium Climate Sensitivity (ECS) of global climate models remains poorly constrained, partly due to conflicting emergent constraints on extratropical low-cloud feedback. Jiang et al. (2023) linked a winter-maximum low-cloud seasonal cycle to higher ECS, while Furtado et al. (2023) associated a summer-maximum with strong negative feedback. This study reconciles this dichotomy by introducing a novel, physically-based framework that classifies "low clouds" into three types based on cloud-top height: Multi-Layer (MLC), Upper Boundary-layer (UBC), and Lower Boundary-layer Clouds (LBC). We show that the apparent contradiction arises from conflating these distinct regimes, which exhibit opposing seasonal cycles: MLC and UBC (tied to storm tracks) peak in winter, whereas LBC (boundary-layer driven) peaks in summer. The winter-maximum emerges in observations and high-ECS models due to dominant MLC, while the summer-maximum in low-ECS models is an artifact of systematically underestimated UBC, which allows the LBC signal to dominate. Furthermore, passive satellite data (e.g., ISCCP) artificially reinforce this model bias by missing wintertime low clouds under upper-level layers. Our work demonstrates that progress in constraining ECS requires a paradigm shift from analyzing total low-cloud fraction to evaluating model fidelity in simulating the distinct physics and mix of these cloud regimes.