Data obtatmed from the medlun-frequency radar at Wuhan (30°N, 114°E} from Ⅱ to 25 February and frum 28 February to 13 March have been used to study the mean wind and tidal ogcillations at mesopause (8O-98km) region over Wuhan to 13 Minter The observed zonal and meridional mean winds show obvious temporal variation and usually decrease widl the mcreasing heighls in most observational time. the zonal mean wind is eastward A Aynamic spectral analysis of disturbance amplutudc illustratcs a promincnt peak at a period of 24h al all hcighls and alines! uninterrupted, and there are occasionally shghlly stong senlldiurnal and weak lerdiurnal tidal disturbarces, indicating that in Ihe mid-lalitude regions. the diurnal tide is Ihe dominant tidal disturbance at tnesopausc in winter. The temporal variation of tidal ampbtudes and their wave kinematical cncrgies iffustrates that there may bc sigmificant resonant interaction among the diurnal, semidiurnal and terdiurnal rides A quantitative compariosn with Ihe global-scale wave model (GSWM) demonslrates that the observed diuulal and semidiurnal tidal amplitudes are less 3nd larger than lhe results of GSWM-02. respectively Tbe essentially linear GSWM carmot predicl Ihe decrease ot diurnal tidal anplitude at higher heigbts. which muy be due to several possible nonlinear processes. such at wave-Wave interaction aad wave break. Although the height vanations of tidal phases are Consistenly well with the prediction of GSWM. there are evident pbase differenees between our observation and tile model. The profiles of the diunanl tidal phases show nbvious decreasc trond with Ihe increasiong height, indicalmg the observed diurnal tide is propagating upward The vertical wavelength of Ihe observed diurnal tide is estimated to be about 37km. and die eorresponding downward phase velocity is about 15km-1h.

We examine the evolution of three-gravdy-wave interactions, with one initaly much stronger Ihun the other two, in a dissipative atmosphere, The characteristic time is used as an essential lime ca e for depicting the interactions. In the absense of viscosity, the solutions of the lineatized interaction equations, viz., the paramettic instability, give a good description for the evolution of secondary waves up to one-hail ol the characteristic time Beyond this time, the linear solutions are invalid; the evolution must be obtained by solving the complete interaction equations For given initial wave amplitudes, the characteristic time increases with the increasing wavelength of the primary wave; thus the criterion for the validity of resonant interaction approximation may be better satisfied The viscous dissipation makes the characteristoc time larger, and when severe enough a porlion oftbe seuondary waves, particularly at high wavenumbers, may fail to interacl with the primary wave effectively Additionally. the maxima of the energy densities for the secondary wavve are smaller than those in the absence of viscosity. In the ease of a significalt frequeney mismatch, only a fraction of the primary wave energy lakes part in the hlterchange Copyright.

By using a two-dimentional fug-implicit-continuous-Euleri&n scheme, numerical sLm ulation for nonlinear propagation of Gaussian gravity-wave packets in a compressible and isothermal a~osphere are carried out. The numerical analyses show that for an inkiaUy given upgomg gravity-wave packet whose disturbance velocity is much less than ambient wind velocity, although there exists nonlinear interaction, during the propagation, the whole wave packet and the wave-associated energy keep moving upward, while the wave front keeps moving downward, Wave-associated perturbation velocity increases with the increasing height, and the mean flow shows obvious enhancement when the wave packet passes. After a long time propagation (several periods), wave-associated perturbation and energy can still concentrate in a limited region that is comparable in size to that given initially, The propagation path of wave energy coincides well with the ray path predicted by the linear gravity wave theory, but the magnitude of wave energy propagation velocity is evidently smaller than the group velocity derived from the linear gravity wave theory. This indicates that once gravity waves are generated, they propagate almost freely along their rays, and the nonlinear effect will only lower the propagation velocity of the wave- associated energy While gravity-wave packets propagate in a nonlsothermal atmosphere, the nonlinar propagation paths of wave energy depart clearly from the ray paths derived from the Fncer gravity wave theory under the WKB approximation, which indicates that the linear gravity-wave theory under the UKB approximation can not predict the nonlinear propagation of gravity-wave packet in a nonisothennal atmosphere.

Observations with the SOUSV VHF Radar in the mesolpause region during summer 1989 show lay- ers of strong echo ictensity and wind speeds increasing with height. Occasionally, quasi-manochromatic internal gravity waves occur in the bottom side of the echo layer and vanish in the height intervaI around the echo intensity maximum which tends to increase withh increasing wave amplitude The data anaiysis can be performed using a Boussieesq and a WKB approximation, It shows that the horizontal phase trace velocity of the observed gravity waves is equal to the background wind velocity at the height of the echo intensity maximum. The observations thus reprsent nonlinear gravity wave-critical level encounters pruducing strong turbulence in the neutral gas. But the, estimated euergy disspation rates are too small to produce neutral gas turbulence at scales equal to haft the radar wave!ength yieiding an additional indication for yet unidentified mechanism generating small-scale structure in the electron gas.

Tbe spatial and temporal evohltion of gravity wave packet interactions is studied numerically. It is shown that Igrougb tile resonant parametric excitation an upgoing gravity wave packet can cause the growth of two secondary waves from noise level up to a significanl inlensity in several bours The prbnary wave packet is apparently debrmed as it decays. Tile energy transfer among tile intenlcling waves is no longer reversible since their amplitudes are Iocalised Therefore the characteristic time for the interactions is of a parlicular significance: it represents a time during whicb the principal energy transfer arises. Beyond the cbaracteristic time the net energy transfer among the interacting waves becomes rather weak, but the local change in the wave energy densities can be Conlsiderable Only a part of the initial energy of the primary wave packet is transferred to the secondary waves during the parametric excilation. The amounts of energy, which each of the two secondary waves extract from the primary wave, are ditthrent, exhibiting a parameter preference in the energy transfer. 1he parametric excitation process can be completed in The propagation lime, For the resonant interaction with two gravity wave packets initially baying large amplitudes, the evolution tale is faster than that in the parametric excitation The primary wave packet can lose most of its energy and finally be reduced to a small fluetuation. The viscous dissipation not only decreases the wave energies but also strongly affects the local energy transfer among the interacting gravity wave packets.