A comparative study of observation- and LES-derived turbulence

One of the most important statistical variables representing turbulence structure is the probability density function (PDF). A comprehensive knowledge of the PDF and its rela-tionship to the underlying in physical processes is important for both the understanding of the turbulence structure and the parameterization of these processes. Usually, an evaluation of LES results involves comparisons between observation- and model-derived turbulence fluxes and variances. However, a good comparison in these variables does not necessarily guarantee a consistency in the turbulence structure of the model simulations and observa-tions. Since a PDF includes additional information and may reveal critical features of the turbulence structure and physical processes, we analyzed observation- and LES-derived PDFs for stratocumulus clouds. We derived the PDFs from both LES simulations and DYCOMS II field turbulence data; evaluated the LES results and studied the entrainment dynamics. We have found that some of 2D-joint PDFs of vertical velocity and scalars (such as moisture) clearly demonstrate some Gaussian features; but the 2D-joint PDF of vertical velocity and liquid water potential temperature close to the inversion is clearly non-Gaussian, showing strong impacts of the entrainment and radiative cooling. This fea-ture is qualitatively similar for the PDFs derived from both the LES and the turbulence data. Both observation- and LES-derived w PDFs show larger values for negative w near the cloud top, a sign of negative skewness. One consistent difference between observed and modeled PDFs is that the former usually show broader distributions than the latter, in-dicating larger variability in the observations. It is particularly true for moisture near the cloud top. This could be due to the different entrainment effects in the simulations and ob-servations.