Abstract
The problem of scale-selection of Kelvin waves in the stratosphere by forcing from tropospheric heating is analyzed using a simple linear model. The effect of vertical wind shear is excluded because the phase speed of the waves is fast relative to the range of the mean zonal wind in the vicinity of the tropopause at which level the upward energy flux due to forcing is evaluated. Results of this analysis modify Holton's (1973) theory in that 1) the forcing is most efficient for the longest zonal wavelength even if the heat sources are distributed randomly, and 2) the most favored vertical wavelength of the excited waves is about twice the vertical scale of beating. The calculated vertical wavelengths exceed slightly those observed and the discrepancies are discussed.
