Evaluation of spectroscopic databases through radiative transfer simulations compared to observations. Application to the validation of GEISA 2015 with IASI and TCCON

The quality of spectroscopic parameters that serve as input to forward radiative transfer models are essential to fully exploit remote sensing of Earth atmosphere. However, the process of updating spectroscopic databases in order to provide the users with a database that insures an optimal characterization of spectral properties of molecular absorption for radiative transfer modelling is challenging. The evaluation of the databases content and the underlying choices made by the managing team is thus a crucial step. Here, we introduce an original and powerful approach for evaluating spectroscopic parameters: the Spectroscopic Parameters And Radiative Transfer Evaluation (SPARTE) chain. The SPARTE chain relies on the comparison between forward radiative transfer simulations made by the 4A radiative transfer model and observations of spectra made from various observations collocated over several thousands of well-characterized atmospheric situations. Averaging the resulting ‘calculated-observed spectral’ residuals minimizes the random errors coming from both the radiometric noise of the instruments and the imperfect description of the atmospheric state. The SPARTE chain can be used to evaluate any spectroscopic databases, from the visible to the microwave, using any type of remote sensing observations (ground-based, airborne or space-borne). We show that the comparison of the shape of the residuals enables: (i) identifying incorrect line parameters (line position, intensity, width, pressure shift, …), even for molecules for which interferences between the lines have to be taken into account; (ii) proposing revised values, in cooperation with contributing teams; (iii) validating the final updated parameters. In particular, we show that the simultaneous availability of two databases such as GEISA and HITRAN helps identifying remaining issues in each database. The SPARTE chain has been here applied to the validation of the update of GEISA-2015 in 2 spectral regions of particular interest for several currently exploited or planned Earth space missions: the thermal infrared domain and the short-wave infrared domain, for which observations from the space-borne IASI instrument and from the ground-based FTS instruments at the Parkfalls TCCON site are used respectively. Main results include: (i) the validation of the positions and intensities of line parameters, with overall significantly lower residuals for GEISA-2015 than for GEISA-2011; (iii) the validation of the choice made on the parameters (such as pressure shift and air-broadened width) which has not been given by the provider but completed by ourselves. For example, comparisons between residuals obtained with GEISA-2015 and HITRAN-2012 have highlighted a specific issue with some HWHM values in the latter that can be clearly identified on the ‘calculated-observed’ residuals.