E.H. Shiguemori, J.D.S. da Silva, H.F. de Campos Velho, J. C. Carvalho (2005): Atmospheric temperature retrieval by a Radial Basis Function neural network, Journal of Intelligent and Fuzzy Systems (submitted).

Abstract: Vertical temperature profiles are obtained from measured satellite radiance data by using a Radial Basis Function neural network (RBF-NN). The RBF-NN is trained with data provided by the direct model characterized by the Radiative Transfer Equation (RTE). The RBF results are compared to the ones computed using regularized inverse solutions. In addition to synthetic data (corrupted by noise), real radiation data from the HIRS/2 (High Resolution Infrared Radiation Sounder) is used as input for the RBF to generate temperature profiles that are compared to radiosonde measured temperature profiles. Analysis of the RBF results reveals the generated profiles closely approximate the results of Carvalho et al. and Ramos et al., showing the adequacy of neural network based models for solving the inverse problem of temperature retrieval from satellite data. RBF neural network based systems are useful because of the intrinsic features of parallelism and implementation simplicity, as well as the possibility of hardware device that may imply in onboard fast processing systems.