A fusion method for creating sub-hourly DNI-based TMY from long-term satellite-based and short-term ground-based irradiation data

In order to correctly perform Concentrated Solar Power (CSP) plant electric energy output estimation, a standard approach is to consider Typical Meteorological Year (TMY) as a reference input data in CSP simulation software such as System Advisor Model or Greenius. These TMYs may be built from satellite derived irradiation databases. In order to correctly estimate the CSP electricity production, it is highly recommended to use sub-hourly DNI information. Due to limitation in spatial and temporal resolution of geostationary satellite images, satellite-based irradiation data lack good representativity in term of any sub-hourly temporal variability. To overcome this limitation, we propose an innovative fusion method to combine a one year short time series of ground-based sub-hourly irradiation data and the long-term satellite-based one to create calibrated, sub-hourly and long-term based TMY irradiation data. This method has been successfully applied in the planning of the CSP plant project in Morocco: one year and a half of high quality 10-minute irradiation data from pyranometric ground stations belonging to the Moroccan Agency for Solar Energy (MASEN) has been used with long-term hourly satellite-based irradiation data to create calibrated 10-minute DNI based TMY. The ground-based irradiation data have passed the standard quality check procedure recommended by the Baseline Solar Radiation Network for the World Climate Research Program. The satellite hourly irradiation data has been calibrated on a one year learning period of ground station data and this calibration has been verified on a subsequent half year ground station data. This calibration has been applied to the hourly long-term satellite irradiation time series from which a TMY was computed. The final and innovative step consists in introducing the site specific sub-hourly variability into the whole set of hourly daily profiles of irradiation of the TMY time series, as needed to obtain a better estimation of the CSP producible. The method uses the whole 10-minute measured irradiation data as a store of available days, normalized in terms of time between sunrise and sunset.