Observations were taken at Mount Zugspitze, involving 4-year-long monitoring with a web camera, in-situ measurements on both sides of the mountain, as well as a week-long campaign with measurements from a variety of instruments. The occurrence of banner clouds at this mountain turns out to be essentially independent of the wind speed; it shows a pronounced diurnal cycle as well as a seasonal cycle. Reasons are suggested and discussed. In-situ measurements indicate that the windward and the leeward side are characterized by different wind regimes, although there is mean uplift on both sides; in addition, the leeward air is both moister and warmer than the windward air.
Numerical simulations were carried through with the EULAG-model using idealized orography. It turns out that there is a strong windward-leeward asymmetry in the Lagrangian uplift when air flows over and around a pyramid-shaped obstacle. This asymmetry (and, hence, the likelihood for banner cloud formation) increases with height and steepness of the orography. Latent heat release leads to the observed unstable stratification at the top of the banner cloud, but it is of secondary importance for the banner cloud as a whole. The large leeward uplift is associated with boundary layer separation and a rather complex lee-vortex geometry. It follows that banner clouds can be entirely due to orographic dynamics and that moisture asymmetries are not essential. The asymmetry in uplift is lost and even reversed when the orography becomes more ridge-shaped with the flow becoming increasingly two-dimensional. In that case "flow around the mountain" is replaced by "flow over the mountain", and gravity waves start to play a more dominant role. We also simulated flow conditions which are more conducive to cap clouds than to banner clouds. Both cloud types can be distinguished from each other with the help of trajectories: banner clouds primarily consist of air parcels which go around the mountain and experience most of their uplift on the leeward side.