During the last forty years vertical exchange in the atmospheric surface layer has been parameterized with the aid of the Monin-Obukhov similarity theory. Currently it is understood that the concept of local flux-gradient correspondence underlying that theory and most traditional turbulence closures breaks down in convective conditions. Physical essence of the problem is as follows. In strong convection large-scale semi-organised coherent structures embrace the entire convective boundary layer

more »

... ~1 km in height). They generate pronounced largescale (2-3 km in the horizontal) flow patterns close to the surface, which play an important role in the horizontal dispersion of atmospheric pollutants. The largescale structures also yield local velocity shears and consequently the sheargenerated turbulence, which crucially affects heat/mass transfer. On the contrary, local shears, oriented in the opposite directions, only slightly affect the mean momentum transfer (hence the term "inactive turbulence"). Only first steps have been made in analysing these effects theoretically. In the present paper a review of the problem is given, and a new theoretical model is proposed capable of reproducing the above essential features of convection. The model is verified using data from recent large-eddy simulation of convective boundary layers together with the reference atmospheric data. Recommendations are given as to how to proceed in practical parameterization of the surface fluxes and horizontal velocity variances in pollution dispersion models.