Spectral radiative-transfer modeling with minimized computation time by use of a neural-network technique

A new approach based on a neural-network technique for reduction in the com putation time of radiative-transfer models is presented. This approach give s high spectral resolution without significant loss of accuracy. A rigorous radiative-transfer model is used to calculate radiation values at a few se lected wavelengths, and a neural-network algorithm replenishes them to a co mplete spectrum with radiation Values at a high spectral resolution. This m ethod is used for the UV and Visible spectral ranges. The results document the ability of a neural network to lean this specific task. More than 20,00 0 UV-index values for all kinds of atmosphere are calculated by both the ri gorous radiative-transfer model alone and the model in combination with the neural-network algorithm. The agreement between both approaches is general ly of the order of +/-1%; the computation time is reduced by a factor of mo re than 20. The new algorithm can be used for all binds of high-quality rad iative-transfer model to speed up computation time. (C) 2001 Optical Societ y of America OCIS codes: 030.5620, 200.4260, 040.7190, 010.1320.