Distinct Sensitivities of Extratropical Low Cloud Regimes to Cloud Controlling Factors Across Multiple Timescale

Extratropical marine low cloud feedback remains a primary source of uncertainty in estimating Equilibrium Climate Sensitivity (ECS). Recent studies have proposed using the observed seasonal cycle of low clouds as an "emergent constraint" to reduce this uncertainty. However, the validity of such constraints relies on the assumption that the physical mechanisms governing seasonal variations are consistent with those driving long-term climate feedback (timescale invariance). In this study, we investigate the dominant physical processes driving extratropical low cloud variability by decomposing clouds into three regimes based on cloud-top height: Multi-layer clouds (MLC), Upper boundary-layer clouds (UBC), and Lower boundary-layer clouds (LBC). Using CloudSat-CALIPSO observations and CMIP6 model outputs (CESM2 and MPI-ESM), we quantify the sensitivity of these regimes to key Cloud Controlling Factors (CCFs)—including Sea Surface Temperature (SST), Estimated Inversion Strength (EIS), relative humidity at 700 hPa, and vertical velocity—across seasonal and interannual timescales. We find that while MLC and UBC exhibit consistent responses across timescales, the response of Lower Boundary-layer Clouds (LBC) to surface temperature is inconsistent between seasonal and interannual timescales. This breakdown is mechanistically driven by the differing responses of tropospheric stability (EIS) to surface warming on different timescales: seasonally, EIS is controlled by dynamic subsidence, whereas interannually, it is modulated by the thermodynamic response to widespread ocean warming. Consequently, the seasonal cycle of LBC is not a robust proxy for its long-term feedback. Furthermore, a decomposition of the cloud feedback spread reveals that inter-model uncertainty arises primarily from the cloud sensitivity term (d LCF/d CCF) rather than large-scale state changes (dCCF/dTs). Specifically, the divergent model simulations of cloud sensitivities to dynamic (vertical velocity) and thermodynamic (relative humidity) drivers are identified as the dominant sources of spread. These results highlight the necessity of improving physical parameterizations of cloud-turbulence interactions to reduce uncertainty in climate projections.