We describe a new line-by-line (LBL) algorithm for radiative excitation in infrared bands in a non local thermodynamic equilibrium (non-LTE) planetary atmosphere. As a specific application, we present a predictive model for the terrestrial CO2 15-μm emission that incorporates this generic algorithm, and validate the model by comparing its results with emission spectra obtained in a limb-scanning rocket experiment. The radiative-excitation algorithm has certain unique features. Being a completely monochromatic calculation, it includes the detailed layer-by-layer variation of the shapes of the individual lines in its evaluation of atmospheric transmittivity; and, being an iterative algorithm, it avoids the need to construct and invert large matrices, so that a fine layering scheme can be implemented. It also includes a simple correction procedure to minimize the most serious error due to having discrete layers. These features contribute to accurate radiative transfer results and reliable atmospheric cooling rates. For altitudes above 40 km, we present results of model calculations of CO2(&ngr;2) vibrational temperatures, 15-μm limb spectral radiances, and cooling rates, for the main band as well as for weaker hot and isotopic bands. We calculate the excitation and deexcitation rates due to different processes, including the radiative pumping due to individual atmospheric layers.

We compare the predicted limb radiance with earthlimb spectral scans obtained in the SPIRE rocket experiment over Poker Flat, Alaska, and get excellent agreement as a function of both wavelength and tangent height. This constitutes the first validation of a long-wavelength CO2 non-LTE emission model using an actual atmospheric data set, and it verifies the existence of certain aeronomic features tht have only been predicted by models. It also constrains the previously unknown value of the very important rate constant for deactivation of the CO2 bending mode by atomic oxygen to the range of 5--6¿10-12 cm3/(mol s) at mesospheric and lower thermospheric temperatures. We discuss the significance of this large value for the terrestrial and Venusian thermospheres. We also discuss the convergence rate of the iterative scheme, the model's sensitivity to the background atmosphere, the importance of the lower boundary surface contribution, and the effects of the choice of the layer thickness and the neglect of line overlap. ¿ American Geophysical Union 1992