Limitations of Air Pollution Episodes Forecast due to Boundary-Layer Parameterisations Implemented in Mesoscale Meteorological Models

Abstract

Dispersion models require information on the turbulence characteristics in the planetary boundary layer (PBL). This information is most often extracted from either meteorological measurements or from numerical (prognostic or diagnostic) models, and the requested turbulence parameters are then estimated using a PBL pre-processor. Traditionally, Monin-Obukhov (M-O) similarity theory is applied when estimating the surface turbulent fluxes and the various vertical profiles of averaged quantities in the surface layer of the PBL (Beljaars and Holtslag 1991; Hanna and Chang 1992; Zilitinkevich et al. 2002b). In this similarity approach several simplifying assumptions are made, among which the requirement of quasi-stationary and horizontally homogeneous flow, and constant (independent of height) turbulent fluxes are the most crucial (Arya 1988). In urban areas and in complex terrain these assumptions are obviously not fulfilled. The theory is particularly questionable in very stable conditions, i.e. under conditions typically prevailing during pollution episodes in winter. In very stable conditions turbulence tends to be sporadic, and wave-turbulence interaction becomes increasingly important as well as drainage effects due to even small terrain slopes (Hogstrom 1996). Moreover, observational data suggest that developed turbulence can exist in the stably stratified surface layer at much larger Richardson numbers than the