Turbulence-driven scale dependence of probability distributions for global MODIS-AQUA cloud properties

Data from the MODerate resolution Infrared Spectrometer are used to characterize changes in the probability distribution functions (PDFs), mean values and standard deviations of cloud properties relevant to climate uncertainties as a function of the pixel resolution scale. Those properties are cloud fraction, cloud particle effective radius, and cloud liquid water path. Global PDFs are characterized as a function of the spatial averaging. Analytical functions were identified that fit best to the observed PDFs and their scale dependence is compared to predictions for a homogeneous turbulent system that displays intermittency. It is confirmed that the mean value and standard deviation change with scale. For the range of scales considered here the global means change within 3% for the cloud fraction, 7% for liquid water path, and changes very little (~0.2%) for the cloud particle effective radius. We indeed found the remarkable result that while the closest analytical expressions to the observed PDFs differ for all of the variables, their PDFs are reasonably well described by lognormal distributions when the mean is normalized by the standard deviation inside each averaging domain. This suggests a possible approach for both subgrid modeling and unified stochastic models of these variables at all scales. The observed scale-dependence also highlights the relevance of answering the question of what is the adequate spatial scale to analyze and model climate feedbacks.