Algorithms for nonlinear retrieval problems in atmospheric remote sensing using regularization methods

In this paper, ute present a retrieval algorithm fur nonlinear retrieval problems based on regularization Cheery. The proposed method is based on the Gauss-Newton method fur nonlinear least square problems. In the proposed algorithm, Tikhonov and truncated singular value decomposition techniques are used to regularize the solution of the linearization problem used to compute the Gauss-Newton search direction. The dependency of the performance and behavior of the proposed algorithms on the initial guess, stopping criterion, and regularization parameter is studied by means of simulations. Results are presented for atmospheric temperature retrievals based on radiometry from the HIRS/2 and MSU instruments in NOAA TOVS.