Experiments have been conducted with a two-dimensional timedependent, numerical model of the general circulation of the atmosphere up to the mesopause. A scheme for the calculation of cooling rates due to the 15(mu)m band of carbon dioxide is developed. It uses the Curtis matrix approach which incorporates cooling-to-space, transfer of radiation between atmospheric layers and non-equilibrium effects in the upper mesosphere. The sensitivity of the cooling rate calculations to the choice of collisional relaxation time is investigated. An 'almost exact' scheme to calculate heating rates due to the absorption of solar radiation by ozone and molecular oxygen is presented. Use of both new radiation schemes enables the diabatic heating rate to be calculated to the upper boundary of the model. Other heat sources in the region of the mesopause are discussed. Incorporation of the new schemes considerably improves the modelled temperature structure of:the stratosphere and lower mesosphere. The upper mesosphere is not well reproduced with no indication of the observed cold summer mesopause. The heat and momentum budgets of the mesosphere are studied. Eddy momentum fluxes derived from satellite observations of planetary waves are found to be significant for the circulation and transport properties of the stratosphere but incapable of producing the required distribution of angular momentum in the mesosphere. A Rayleigh friction parameterisation is included in the mesosphere to reproduce the observed zonal wind and temperature structure. Momentum deposition by tides and gravity waves is discussed. Curtis matrices are calculated with higher mixing ratios of carbon dioxide and the effects of increased atmospheric CO2 on stratospheric temperatures and ozone is investigated. Temperature decreases of up to about 10K are predicted with increases in ozone concentration in the upper stratosphere. In the lower stratosphere the ozone increases are restricted to high latitudes and a decrease shown in equatorial regions. The latitudinal variations are reflected in the ozone column density. An experiment is conducted in which chlorofluorocarbons are released into the model atmosphere and the effects on stratospheric ozone are in exactly the opposite sense to those predicted for the CO2 case. A run in which CO2 and CFCs are introduced simultaneously shows that the two effects are not linearly additive. A simple photochemical theory is used to investigate the temperature dependence of ozone and to explain the non-linearity of the coupled experiment.

Observations of the temperature structure of the stratosphere made by the selective chopper radiometer on the Nimbus 5 satellite are used to analyse the energetics of the sudden warming of January/February 1973. A method of retrieving vertical profiles of zonal Fourier coefficients of temperature from the Fourier coefficients of the measured radiances is described. Some retrievals are compared with conventional observations and quite good agreement is obtained. Comparison of independent estimates of the mean meridional circulation shows that the retrieved temperature and height fields are not sufficiently accurate to give precise values for the zonal mean vertical and meridional velocities. The implication of this result for the energy budget is demonstrated. The sudden warming is analysed first in terms of wave structure and changes in temperature and zonal wind. No evidence for an upward propagating temperature or geopotential disturbance is found. Wave amplifications are observed to occur simultaneously at all levels or to propagate downwards. Largest temperature changes occur in the upper middle stratosphere and maximum zonal flow accelerations in the upper stratosphere. The energy cycle of the lower middle stratosphere is found to be in agreement with observations of previous warmings except in that little increase in eddy energy occurs during the event. A marked baroclinic energy cycle below 10 mb in high latitudes is shown to enhance vertical energy propagation prior to the warming, leading to increased eddy available potential energy between 10 and 2 mb. The variation with latitude of the energetics during the warming is shown to be significant. In the upper stratosphere barotropic conversion from zonal to eddy kinetic energy dominates in mid-latitudes causing deceleration of the zonal flow there first. In high latitudes the deceleration of the zonal flow occurs through the action of the induced mean meridional circulation, while the major source of eddy kinetic energy here is convergence of the vertical eddy energy flux. Examination of the relationship between the latitudinal distributions of the vertical eddy energy flux and the zonal flow reveals that maximum upward propagation of energy is centred con- sistently to the north of the polar night jet until the onset of the warming when coincidence occurs. Some aspects of the observed warming are compared with numerical simulations of sudden warmings. Although the 1973 event is characterised by a wavenumber one disturbance some striking resemblances are found in two numerically simulated wavenumber two type warmings.

The spectral width of single-frequency radiation scattered in the atmosphere may be used to determine air temperature. In general, the measurement is complicated by pressure dependent changes in the spectral profile of the scattered radiation and by the inherently low received signal levels. A fixed-delay Michelson interferometer minimizes both of these problems by: (1) exhibiting a low sensitivity to pressure induced changes in the scattered spectrum and (2) optimally utilizing the available signal. By frequency shifting the signal in one arm of the interferometer relative to the other it is possible to efficiently modulate the output of the interferometer and make it insensitive to small changes in the center frequency of the scattered spectrum. Laboratory results obtained using a fixed-delay, frequency-shifted Michelson interferometer demonstrate the ability of this instrument to remotely measure air temperatures in the range 290 K to 310 K with an uncertainty of ±2 K with averaging times on the order of seconds at a received signal level of 6 x 10⁻¹⁰ watts.