GHI and DNI Forecasting Using GFS, Neural Networks and Solar Radiation Estimated from Satellite Data Base
Abstract
This paper presents a methodology to forecast hourly global horizontal solar irradiance (GHI) and direct normal irradiance (DNI) using global forecast system (GFS) model from NOAA and neural networks models. Forecasting horizon of the predictions is 72 hours. Predictions are done operationally with a model which runs 24 hours a day. The predictions are generated 4 times a day at 05:10 UTC, 10:10 UTC, 17:10 UTC, 23:10 UTC, after availability of the predictions from GFS model which begin its run at 00:00UTC, 06:00UTC, 12:00UTC and 18:00UTC.