ON THE OBSERVED MORPHOLOGY OF GRAVITY-WAVE AND EQUATORIAL-WAVE VARIANCE IN THE STRATOSPHERE

A first systematic attempt is made to explain the variance of horizont al velocities u'2 + upsilon'2 and relative temperature fluctuations T' 2 produced by gravity waves and equatorial waves in the 20-60 km heigh t range of the atmosphere. Single-wave and spectral theories of the wa ve field are applied to derive simplified quantities which parameteriz e both dissipative and nondissipative effects controlling wave varianc es. The major simplification is the omission of variations in source s trengths and background winds, which can modify wave variances. The su ccess or otherwise of the resulting simulations gives some measure of the relative importance of the retained terms compared to the neglecte d terms. The simplifications produce terms independent of individual w ave parameters, making them valid over the entire wave field, and the neglect of background winds enables their computation from background temperatures alone. This approach accurately models both the phase and depth of the observed annual variation of u'2 + upsilon'2 and T'2 at high latitudes, and its attenuation on moving equatorward. The simulat ed annual cycle arises principally from seasonal variations in the den sity stratification of the atmosphere below 60 km, although dependence on background Brunt-Vaisala frequency N accounts for the deeper annua l T'2 variation. The observed decrease in T'2 above 40 km is also cons istent with the decrease in N at upper heights. An observed Gaussian d istribution about the equator of variance at large vertical wavelength s agrees with equatorial-wave theory. Several prominent features are n ot simulated, and so are probably due to processes omitted during init ial simplification of the theory. The semiannual and quasi-biennial va riation of equatorial-wave variances are consistent with their perceiv ed role in driving similar oscillations of the background atmosphere i n these regions. The nonsimulated peak in upper-stratospheric variance s during July-September at sites in the United States coincides with w eakening westward flow. It is argued that this will reduce vertical wa venumbers, and so increase variances in order to conserve the vertical flux of wave action. The hypothesis can explain the confinement of th e peak to July-September, and its attenuation on moving to higher lati tude. However, it cannot explain a similar feature observed over Wales (52-degrees-N), nor the absence of a peak in data from a Japanese roc ket station. Appreciable differences in annual-mean variances among si tes in and around North America indicate geographical differences in s ource intensities. The accurate modelling of seasonal variations at th ese sites suggests a dominant source that is geographically variable b ut temporally constant, consistent with topographic forcing and suppor ting previous lower-atmosphere studies over the United States.