Ozone limb retrieval for the. SCIAMACHY instrument

Kurzfassung

Due to the strict requirements regarding the computation time, the off-line processor of the SCIAMACHY (Scanning Imaging Absorption Spectrometer for Atmospheric CHartographY) instrument is forced to operate with several approximations. These approximations are incorporated in both the forward and the inversion models. Here we would like to investigate the impact of these approximations. The main simplification which is done in the forward model concerns the treatment of the multiple scattering. Only the single scattering term accounts on the actual atmospheric state, and it is apparent that this approximation is valid if the a priori state is sufficiently close to the true atmospheric state. Note that not only the forward model but also the Jacobian are affected by the multiple scattering approximation. The regularization method which is used in the inversion process is the Tikhonov regularization with an a priori regularization parameter. This means that the regularization parameter, which should balance the residual and the constraint, is chosen in advance and is not correlated with the true measurement. The a priori selection of the regularization parameter is performed for synthetic data, and therefore the method can be problematic especially when the measurement is affected by large systematic errors. The scientific processor developed at the German Aerospace Center is the counterpart of the offline processor, which is not, however, limited by any time constraints. This brings the opportunity to employ more time-consuming approaches and study their impact. The processor uses the Picard iteration method to simulate the radiance field in a full spherical atmosphere and includes polarization as well as Ring effects. A large class of regularization methods as for instance, the Tikhonov regularization, the iteratively regularized Gauss-Newton method, the regularizing Levenberg-Marquardt method, the asymptotical regularization approach and the regularized total least-squares method can be used for a specific application. Here a comparison between limb retrieval results obtained by the operational and the scientific processor for the SCIAMACHY instrument is presented.