Boundary-Layer Meteorology, 2010, Vol.136(2), pp.253-284
High-resolution water vapour measurements made by the Atmospheric Radiation Measurement (ARM) Raman lidar operated at the Southern Great Plains Climate Research Facility site near Lamont, Oklahoma, U.S.A. are presented. Using a 2-h measurement period for the convective boundary layer (CBL) on 13 September 2005, with temporal and spatial resolutions of 10 s and 75 m, respectively, spectral and autocovariance analyses of water vapour mixing ratio time series are performed. It is demonstrated that the major part of the inertial subrange was detected and that the integral scale was significantly larger than the time resolution. Consequently, the major part of the turbulent fluctuations was resolved. Different methods to retrieve noise error profiles yield consistent results and compare well with noise profiles estimated using Poisson statistics of the Raman lidar signals. Integral scale, mixing-ratio variance, skewness, and kurtosis profiles were determined including error bars with respect to statistical and sampling errors. The integral scale ranges between 70 and 130 s at the top of the CBL. Within the CBL, up to the third order, noise errors are significantly smaller than sampling errors and the absolute values of turbulent variables, respectively. The mixing-ratio variance profile rises monotonically from ≈0.07 to ≈3.7 g 2 kg −2 in the entrainment zone. The skewness is nearly zero up to 0.6 z / z i , becomes −1 around 0.7–0.8 z / z i , crosses zero at about 0.95 z / z i , and reaches about 1.7 at 1.1 z / z i (here, z is the height and z i is the CBL depth). The noise errors are too large to derive fourth-order moments with sufficient accuracy. Consequently, to the best of our knowledge, the ARM Raman lidar is the first water vapour Raman lidar with demonstrated capability to retrieve profiles of turbulent variables up to the third order during daytime throughout the atmospheric CBL.
Boundary-layer turbulence ; Convective boundary layer ; Raman lidar ; Turbulence statistics
View full text in Springer