ИСТИНА

The paper discusses the problem of determining the characteristics of clouds (lower boundary, optical thickness, etc.) and aerosol (optical thickness, the vertical distribution, etc.) according to the characteristics of measured diffuse solar radiation (O4 optical thickness, color index, absolute intensity, etc.). Besides that approaches to estimation of the air mass factors used for the transfer of the slant column depth to the vertical column of gases are discussed. The neural network is used as a method for solving the emerging nonlinear estimation problems, the accuracy of the estimation is determined on the training sample, and the control sample is used to characterize the agreement of the assessment results (that is, how much you can trust the parameter estimates and its errors).