Verifying the spatial consistency of the CAMS Radiation Service and HelioClim-3 satellite-derived databases of solar radiation using a dense network of measuring stations: the case of The Netherlands
The present work deals with two well-known databases of hourly mean of solar irradiance that are derived from satellite imagery: the CAMS Radiation Service version 3.2, abbreviated as CAMS-Rad and part of the Copernicus Atmosphere Monitoring Service version 3.2 (CAMS), and the HelioClim-3 version 5, abbreviated as HC3v5. It adds up to the continuous documentation of these two databases that demonstrates that both databases capture the temporal and spatial variability of the solar radiation and are reliable sources of data. The spatial consistency of the uncertainties of these databases is verified against measurements performed within a dense network of ground stations in the Netherlands from the Royal Meteorological Institute KNMI for the period 2014–2017. For the CAMS-Rad database, the correlation coefficients between ground measurements and estimates are around 0.94–0.97 for irradiance E and 0.85–0.89 for clearness index KT. The bias ranges between −12 and 27 W m−2 (−4 % and 10 %). The standard deviation for E is almost constant at all stations and around 67 W m−2 (24 %) except at Valkenburg (57 W m−2, 20 %) and De Bilt (73 W m−2, 27 %). For this database, the statistical indicators are constant at inland stations showing the spatial consistency of the performances of CAMS-Rad dataset. At seashore stations, statistical indicators are more variable. The sea proximity influences CAMS-Rad performances and a weak spatial consistency is observed near seashore. For the HC3v5 database, the correlation coefficients are around 0.96–0.98 for E and 0.86–0.91 for KT. The bias ranges between −3 and 15 W m−2 (−1 % and 5 %). The standard deviation ranges between 48 (17 %) and 60 W m−2 (22 %). It is constant at inland stations and around 54 W m−2 (20 %) with peaks at Ell (59 W m−2, 21 %) and De Bilt (60 W m−2, 22 %). For this database, statistical indicators are constant at all stations as a whole, showing a strong spatial consistency of uncertainties. An exception is the standard deviation which tends to decrease at the seashore stations and is close to 50 W m−2 (17 %); this could be related to the general trend of decreasing standard deviation with increasing KT already reported in the scientific literature. As a whole, it is found that both databases are reliable sources on solar radiation in the Netherlands.