A neural network approach for a fast and accurate computation of a longwave radiative budget

The authors have investigated the possibility of elaborating a new generati on of radiative transfer models for climate studies based on the neural net work technique. The authors show that their neural network-based model, Neu roFlux, can be used successfully for accurately deriving the longwave radia tive budget from the top of the atmosphere to the surface. The reliable sam pling of the earth's atmospheric situations in the new version of the TIGR (Thermodynamic Initial Guess Retrieval) dataset, developed at the Laboratoi re de Meteorologie Dynamique, allows for an efficient learning of the neura l networks. Two radiative transfer models are applied to the computation of the radiative part of the dataset: a line-by-line model and band model. Th ese results have been used to infer the parameters of two neural network-ba sed radiative transfer codes. Both of them achieve an accuracy comparable t o, if not better than, the current general circulation model radiative tran sfer codes, and they are much faster. The dramatic saving of computing time based on the neural network technique (22 times faster compared with the b and model), 10(6) times faster compared with the line-by-line model, allows for an improved estimation of the longwave radiative properties of the atm osphere in general circulation model simulations.