Updates to 4.3um CO2 NLTE modeling for nadir hyperspectral infrared sounders

Models for the exchange of radiant energy inside Earth’s atmosphere require knowledge of the excitation temperatures of the emitting energy levels of the atmospheric components. In the lower atmosphere where collisions are frequent, we can safely assume all vibrational levels for all molecules are in Local Thermodynamic Equilibrium (LTE). This is not necessarily true in the upper atmosphere where density is lower and therefore fewer collisions. Lack of knowledge of, or incorrect assumptions about, the upper atmosphere vibrational temperatures for channels whose weighting functions peak high in the atmosphere will impact the accuracy of the modelled top-of-atmosphere radiances. An example is the CO2 4.3 µm (2250-2400 cm-1) bands, which upper states preferentially absorb solar radiation and are therefore in Non-Local Thermodynamic Equilibrium (NLTE). For the current generation of hyperspectral sounders, the effect on observed brightness temperature is on average 8-10 K and can significantly impact the kinetic temperature sounding. In this work we present a comparison between different models in computing NLTE effects. Starting from a very recent set of vibrational temperatures computed for CO2, which includes refinements to the effects of O3 and O collisions on CO2 vibrational temperatures, we have generated synthetic radiances with two different line-by-line models: kCARTA and LBLRTM. With these two models, multiple simulations have been performed across different atmospheric states to capture latitudinal variations and solar angle dependencies. For each atmospheric state, we have computed synthetic convolved monochromatic radiances with and without NLTE effects. The differences between the two results have been parametrized in terms of multiple variables and have been implemented, in different ways, in two fast radiative transfer models: σ- IASI/F2N, implemented by University of Basilicata and under developments in the Italian Space Agency FIT-FORUM project, and SARTA, developed at University of Maryland, Baltimore County. In this work, we will show how this fast parameterization is capable of modeling the 4.3 µm spectral radiances compared to accurate line-by-line models.