In order to be efficient, new astronomical methods, like high angular resolution and speckle interferometry require subarc second seeing conditions and long speckle boiling time. It is now well known that there is a close relationship between the seeing and the integrated vertical profile of atmospheric turbulence, from the focus plane of the telescope up to the stratosphere. The measurement of the spread function of a star image does not tell anything about the respective contributions of the inside of the dome, the boundary layer and the free atmosphere. Disturbances coming from the first two slabs can be avoided by : removing the heat sources, cooling the floor, building the observatory above the inversion layer, simulating the site in a wind tunnel... But there is no a priori knowledge about the free atmosphere turbulence. In the literature, such profiles are too scarce and sparse to foresee a clima-tology. In this paper, it will be shown that there is a strong experimental correlation between a good seeing and a low wind speed at the tropopause level. The study rely on already published sesonal variations of the seeing in La Silla, Chile, and Hawaii, and on atlas of climatology of the atmospheric circulation at 200 millibars level. It seems that subarc second seeing conditions require tropopause wind speeds lower than 20ms-1. A quick look at the wind behavior let us hope very good image quality above sites located in the northern part of Chile, near Peru. A fortunate consequence of our assumption is that slow tropopause wind selected sites will likely have a long speckle time life.