High-altitude wind speeds have been adopted as a parameter for astronomical site selection based on the relationship found at the Paranal and Cerro Pachón sites between the average velocity of the turbulence (V0) and winds at the 200-mbar pressure level (V200). Although this relationship has not been checked at any other site in the world and a connection between image quality and V200 has not been proven anywhere, high-altitude wind speed (V200) is a parameter for checking the suitability of sites for adaptive optics and surveying potential sites for extremely large telescopes. We present comprehensive and reliable statistics of high-altitude wind speeds and the tropospheric flows at the location of five important astronomical observatories. We have used the National Center for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) Reanalysis data base to collect wind data at nine tropospheric pressure levels for the five selected sites. For comparison and validation of the data from the climate diagnostic model, we have also obtained wind profiles from radiosonde stations. The degrees of correlation found indicate a high level of significance between NCEP/NCAR Reanalysis and balloon data bases, pointing to NCEP/NCAR Reanalysis as a useful data base for site characterization. Statistical analysis exclusively of high-altitude winds points to La Palma as the most suitable site for adaptive optics, with a mean value of 22.13 m s-1 at the 200-mbar pressure level. La Silla is at the bottom of the ranking, with the largest average value 200 mbar wind speed (33.35 m s-1). We have found a clear annual periodicity of high-altitude winds for the five sites under study. We have also explored the connection of high- to low-altitude atmospheric winds as a first approach of the linear relationship between the average velocity of the turbulence and high-altitude winds. We may conclude that high- and low-altitude winds show good linear relationships at the five selected sites. The highest correlation coefficients correspond to Paranal and San Pedro Mártir, while La Palma and La Silla show similar high- to low-altitude wind connection. Mauna Kea shows the smallest degree of correlation, which suggests a weaker linear relationship. Our results support the idea of high-altitude winds as a parameter for ranking astronomical sites in terms of their suitability for adaptive optics, although we have no evidence for adopting the same linear coefficient at different sites. The final value of this linear coefficient at a particular site could drastically change the interpretation of high-altitude wind speeds as a direct parameter for site characterization.