The invention relates to a method for predicting hourly scattering ratios on the basis of astronomical and meteorological environmental factors. The method includes steps of (1), acquiring radiation data, astronomical data and meteorological environmental data; (2), dividing weather types including fine weather types, fine-to-cloudy weather types, fine-to-overcast weather types, cloudy-to-overcastweather types and rain, snow and haze weather types; (3), selecting preset models according to the weather types to predict the hourly scattering ratios, to be more specific, predicting the hourly scattering ratios by the aid of PCA-LMBP (principal component analysis-Levenberg Marquardt back propagation) neural network models when the weather types are the fine weather types, the fine-to-cloudy weather types and the fine-to-overcast weather types, predicting the hourly scattering ratios by the aid of LMBP neutral network models when the weather types are the cloudy-to-overcast weather types,and predicting the hourly scattering ratios by the aid of linear regression models when the weather types are the rain, snow and haze weather types. The PCA-LMBP neutral network models, the LMBP neutral network models and the linear regression models are prediction models screened on the basis of astronomical factors, meteorological factors and the weather types. Compared with the prior art, the method has the advantage of accurate and reliable prediction results.