Improving the Heliosat-2 method for surface solar irradiation estimation under cloudy sky areas

The purpose of this study was to improve global horizontal irradiation (GHI) estimates under cloudy sky conditions using the Heliosat-2 method (Rigollier et al., 2004) calibrated with Geostationary Operational Environmental Satellite images (GOES) (Albarelo et al., 2015). The Heliosat-2 method using GOES images as input (GOES_H2) was coupled with a cloudy sky radiative transfer parameterisation (RTP) (RTP: radiative transfer parameterisation) to better account for local cloud properties. The key element of this parameterisation is the cloud absorption coefficient, which was the only element to be computed. The GHI estimates obtained when using GOES_H2 coupled with a RTP (RTP_H2) were compared against GHI measurements from six meteorological stations in French Guiana over four years. We used root mean square error (RMSE), mean bias error (MBE), correlation coefficient, and the Kolmogorov Smirnov test integral to assess GHI estimates accuracy. RTP_H2 obtained estimates were also compared with GOES_H2 estimates. Our results show an improvement in the GHI estimates under cloudy, clear, and all sky conditions. Under cloudy skies, the RMSE and MBE of our method RTP_H2 ranged from 43% to 63% and −2% to −22%, respectively, whereas those of GOES_H2 method ranged from 66% to 87% and −48% to −65%, respectively. Another effect of the new method was the improvement in clear sky GHI estimates: the RMSE and MBE ranged from 16% to 24% and −20% to −8%, respectively, while those of GOES_H2 ranged from 19% to 28% and −23% to −10%, respectively. The improvement in GHI estimates under cloudy sky conditions led to better GHI estimates under all sky conditions: the RMSE and MBE of RTP_H2 ranged from 19% to 26% and −7% to −2%, respectively, while the corresponding values for GOES_H2 ranged from 22% to 32% and −15% to −8%, respectively. The suggested cloudy sky RTP offers the advantages of simple implementation and good computation speed. This method requires only one parameter; users need to configure a cloud absorption coefficient suitable for their local conditions. We conclude that coupling the Heliosat-2 method with a cloudy sky RTP improves GHI estimates in French Guiana and reduces discrepancies between satellite-derived irradiation and ground measurements in areas with high cloudiness.