Examination of Correction Method of Long-term Solar Radiation Forecasts of Numerical Weather Prediction

To optimize operation plans for storage batteries with solar power systems, it is necessary to predict solar radiation for up to one week with high accuracy. However, it is known that the further a prediction system forecasts into the future, the lower the prediction accuracy. In this paper, we report on the results of testing a method for correcting the solar radiation prediction value statistically from a numerical weather prediction model by using the tendency of past error occurrence for predicting seven days ahead.