Mesoscale vertical velocities generated by stress changes in the boundary layer: linear theory *

coastal urban areas when onshore flow occurs. However, it may also be of significance in inland urban areas, either because of the contrast with surrounding agricultural rural areas or when there is a contrast between prairie and wooded areas. Under supportive synoptic conditions the developed vertical velocities may trigger convective clouds. The features of the induced vertical velocity may also be of importance in dispersing pollutants. In this paper we approach the problem of the vertical velocity which arises because of horizontal inhomogeneities in the surface shearing stress in the atmospheric planetary boundary layer. This study is an extension in more general terms of a previous paper (Dalu et al. 1988), where we reported on the waves generated by a change in surface roughness (CSR). Hunt and Simpson (1982) provided an excellent review of the studies, reported by that time, concerning flow perturbations induced by a roughness change. Furthermore, Belcher et al. (1990) and Hunt et al. (1991) presented solutions concerning the flow perturbation induced by roughness changes within and around the stress layer. Our work generally agrees with these results, however it is less detailed in the structure of the perturbation within the stress layer because we are more concerned with the flow perturbation in the free atmosphere above. Additional studies reporting on the impact of a sudden change in the surface roughness on the horizontal flow are given by Pendergrass and Arya (1984). Claussen (1987) computed, using a model simulation, the vertical velocity due to a CSR. However, the computed vertical velocities were, in general, very sensitive to the horizontal grid resolution, which must often be reduced to several hundred meters in order to appropriately resolve the related vertical velocity. Using a very coarse horizontal resolution, Vukovich and Dunn (1978), in their numerical model simulation of the St. Louis urban area, suggested that the surface roughness has only a small effect on the circulation for the wind speeds used in their study. Alestalo and Savijarvi (1985), using a hydrostatic two-dimensional model with a grid interval of 4 km simulated the airflow in the Baltic shore region of Finland Introduction