Optimal combination of gridded and ground-observed solar radiation data for regional solar resource assessment

Proper quantification of the available solar resource is essential for the diverse development phases of any solar power plant. At local scale, the solar resource is best assessed from ground radiometric stations. At regional-to-continental scales, however, satellite-based techniques are currently the most suitable approach. Solar radiation evaluation using numerical weather prediction (NWP) models presents some advantages over satellite-based techniques. Nonetheless, gridded solar radiation estimates using either satellite-based or NWP-based techniques still produce biased estimates which are often much higher for the latter. Therefore, a correction needs to be applied before these gridded values are usable for solar applications, especially for NWP-based estimates. This contribution introduces an original method based on the optimal interpolation technique to adjust gridded solar radiation estimates consistently with concomitant radiometric ground observations. The method's performance is demonstrated using NWP-based gridded estimates with 10-km spacing of global and direct monthly irradiation data during the 10-year period from 2003 to 2012 over continental Spain and the Balearic Islands. It is shown that the proposed methodology produces adjusted gridded values that are (or nearly always are, in the case of direct irradiation) within the expected measurement uncertainty of the ground observations, provided a sufficiently large number of observations is available for correction. For the studied region and these NWP-based gridded datasets, our findings suggest that a homogeneous mean distance between ground observations of 100-150 km can result in unbiased gridded estimates.