Solar irradiance forecasting based on long-wave atmospheric radiation

This work contains information concerning long-wave atmospheric radiation. Artificial neural networks were developed to forecast total mean hourly irradiance based on long-wave atmospheric radiation as cloudiness indicator. It was proved that using this variable in models for forecasting irradiance is wellgrounded. The proof was based on the neural networks sensitivity analysis. It was proved that neural network model is capable to utilize information carried by long wave atmospheric radiation only when the air temperature is provided as additional explanatory variable.