As Dr. Roy Spencer points out in his book,
"The most obvious way for warming to be caused naturally is for small, natural fluctuations in the circulation patterns of the atmosphere and ocean to result in a 1% or 2% decrease in global cloud cover. Clouds are the Earth’s sunshade, and if cloud cover changes for any reason, you have global warming — or global cooling."
According to the authors of this new paper, current global climate models "predict a robust increase of 0.5-1 K in EIS over the next century, resulting in a 2.3-4.5% increase in [mixed layer model] cloudiness."
EIS or estimated inversion strength has been shown by observations to be correlated with cloudiness, as demonstrated by the 2nd graph below from the University of Washington, indicating a 1 K increase in EIS results in an approximate 4-5% increase in low cloud cover [CF or cloud fraction]. Thus, a combination of observational data and modelling indicate clouds have a strong net negative feedback upon global warming that is "not captured" by current climate models.
Journal of Climate 2012 ; e-View
PETER M. CALDWELL,* YUNYAN ZHANG, and STEPHEN A. KLEIN
Large-scale conditions over subtropical marine stratocumulus areas are extracted from global climate models (GCMs) participating in Phase 3 of the Coupled Model Intercomparison Project (CMIP3) and used to drive an atmospheric mixed layer model (MLM) for current and future climate scenarios. Cloud fraction is computed as the fraction of days where GCM forcings produce a cloudy equilibrium MLM state. This model is a good predictor of cloud fraction and its temporal variations on timescales longer than 1 week but overpredicts liquid water path and entrainment.
GCM cloud fraction compares poorly with observations of mean state, variability, and correlation with estimated inversion strength (EIS). MLM cloud fraction driven by these same GCMs, however, agrees well with observations, suggesting that poor GCM low cloud fraction is due to deficiencies in cloud parameterizations rather than large-scale conditions. However, replacing the various GCM cloud parameterizations with a single physics package (the MLM) does not reduce inter-model spread in low-cloud feedback because the MLM is more sensitive than the GCMs to existent inter-model variations in large-scale forcing. This suggests that improving GCM low cloud physics will not by itself reduce inter-model spread in predicted stratocumulus cloud feedback.
Differences in EIS and EIS change between GCMs are found to be a good predictor of current-climate MLM cloud amount and future cloud change. CMIP3 GCMs predict a robust increase of 0.5-1 K in EIS over the next century, resulting in a 2.3-4.5% increase in MLM cloudiness. If EIS increases are real, subtropical stratocumulus may damp global warming in a way not captured by the GCMs studied.
From the University of Washington Dept. of Atmospheric Sciences: