Raw Ingredients for Evaluating and Improving Turbulence Parameterizations

Poster PDF

Description

Turbulence is a process that redistributes water vapor and other atmospheric gases, sensible heat, and momentum in the atmosphere. It is a stochastic process, and is best represented by statistics of various moments. It has been shown that the ARM Raman lidar has the accuracy and signal-to-noise level to measure the second- and third-order moments of the water vapor mixing ratio distribution in the convective boundary layer (CBL), from which profiles of variance and skewness of the water vapor (which serves as a passive tracer of atmospheric motion in cloud-free scenes) can be derived. We have identified 300 afternoon cases observed by the SGP Raman lidar over a 6-year period wherein the CBL was quasi-stationary; i.e., it had reached its maximum depth and the turbulent eddies could be assumed to be non-changing with time. From these cases, a “climatology” of variance and skewness profiles of water vapor in the CBL was derived. These cases demonstrated that the gradient of water vapor across the top of the CBL is strongly correlated with the magnitude of the water vapor variance at the top of the CBL, as has been suggested by analysis of large eddy simulation (LES) model output. Furthermore, while turbulence at the top of the CBL typically leads to drying of the CBL due to the mixing of drier free tropospheric air into the CBL, this dataset showed that about 10% of the time the CBL actually moistened, and that the moistening/drying did not depend on the magnitude of the latent heat flux at the surface. We have also analyzed aerosol backscatter profiles observed by a high-spectral-resolution lidar (HSRL) collected over 17 days wherein the CBL was stationary. Due to the very high signal-to noise level in the HSRL, the fourth moment (and hence kurtosis) could be derived in the CBL. This study demonstrates that turbulence at the top of the CBL is never Gaussian distributed, and that there is a strong relationship between skewness and kurtosis there. We have developed new similarity relationships that link turbulent moments and larger-scale variables that might provide a basis for new parameterizations of turbulence. These high-resolution advanced lidar observations (as well as the vertical wind moments from Doppler lidars) are the raw ingredients needed to evaluate these similarity relationships, the explicit representation of turbulence in LES models, and the turbulence parameterization schemes used in cloud resolving and general circulation mod