The accuracy of the atmospheric modeling results depends upon the accuracy both of physical processes parameterization and the physical parameter calculation. The solar radiation energetics of the atmospheric models is the main source of the atmosphere motions. The dependence of the solar radiation fluxes and flux divergence on the atmospheric, clouds and underlying surface properties is investigated. The development solar radiation calculation algorithm is based upon the two-stream delta-Eddington method with the description of the stretch form of cloud phase function by delta function. The requested by radiation algorithm parameters, that are the humidity distribution, cloud amount, cloud water content, cloud phase and mean radii of cloud particles are calculated by humidity transformation model. This model was developed earlier by Dmitrieva-Arrago (Dmitrieva-Arrago et al., 1985) and contains the microphysical algorithm modified additionally for radiation problem. The main goal of the paper is to estimate the variation of the atmospheric radiation characteristics because of the errors and uncertainty of some atmospheric and cloud parameters by numerical experiment. The cloud amount, cloud water content, specific humidity, mean radius of cloud particles, the asymmetry factor, surface albedo were varied in the limits of their possible errors of calculations or experimental uncertainty.