Structure of temperature fluctuations in turbulent convective boundary layers

The Large-Eddy Simulation technique is exploited to investigate statistics of temperature fluctuations, Delta(r)theta, in Atmospheric Boundary Layers ( ABLs) with different degrees of convection. We found statistical characterizations for both strong and weak fluctuations. In terms of probability density functions ( pdfs) of Delta(r)theta, weak and strong fluctuations reflect themselves in different rescaling properties of pdf cores and tails, respectively. For the cores, the observed rescaling is P(Delta(r)theta) =r(-alpha)(P) over tilde(Delta(r)theta/r(alpha)); while for the tails, data are compatible with P(Delta(r)theta) alpha r(zeta infinity). Such two rescaling properties are equivalent to saying <vertical bar Delta(r)theta vertical bar(p)> similar to r(zeta p), with zeta(p) = alpha p for small p's and zeta(p) = zeta(infinity) = constant for large p's. Both alpha and zeta(infinity) turn out to be z-independent within the mixed layer and, more importantly, they do not appreciably vary by changing the degree of convection in the ABL. We also address the question related to the geometrical structure of temperature jumps contributing to large vertical bar Delta(r)theta vertical bar. Finally, the possible relevance of our results to the long-standing problem of subgrid scale parameterizations is discussed.