Robust Variational Inversion: with simulated ATOVS radiances

by Zhiquan Liu and Chengli Qi National Satellite Meteorological Center, China Meteorological Administration Beijing, China Abstract Least-square principal is applicable for dealing with data with Gaussian distribution errors, but it is not satisfying with Non-Gaussian distribution errors which are often the feature of the actual observation data. This study examines a "Robust Variational Inversion (RVI)" method based upon the Huber-estimators. On one side, the method is of variational due to the use of the adjoint technique. On the other hand, the method is robust by taking into account the Non-Gaussian feature of data. The method is implemented in the framework of the 1DVAR retrieval with simulated ATOVS data. The preliminary results show the potential of the method.