Methodology for predicting solar radiation based on typical days: comparative analysis of average and median profiles
Abstract
Methodology for predicting solar radiation based on typical days: comparative analysis of average and median profiles
Incoming article date: 29.05.2025The article discusses a methodology for predicting global horizontal radiation for the day ahead, based on the concept of characteristic days – typical daily profiles of global horizontal insolation. The method is aimed at use in the absence of operational meteorological information and is intended as an alternative (backup) method of short-term forecasting in autonomous energy systems. As a test case, data from the Murino station (Khanty-Mansiysk Autonomous Okrug), obtained from the PVGIS service based on ERA5 reanalysis, was used. The average and median characteristic days for each month are constructed and the accuracy of forecasts using the RMSE, MAE, MdAE and R2 metrics is compared. It is shown that median profiles provide more stable results in conditions of increased weather variability. The method can be used for approximate planning of photovoltaic installations in remote and northern regions with limited access to meteorological data.
Keywords: solar radiation, global horizontal insolation, ERA5, statistical forecasting methods, forecasting in energy systems, forecasting horizon, forecasting for the day ahead, methods for constructing characteristic days, average and median profiles