A topographic geostatistical approach for mapping monthly mean values of daily global solar radiation: A case study in southern Spain

Local topography influences total incoming solar radiation at ground surface in mountainous areas, and so it becomes a key factor for the spatial distribution of plants. However, radiometric stations are often clustered only around farmland or populated areas, usually throughout valleys and flat regions. In this work, we use residual kriging methods to account for cloud- and terrain-related effects, especially when availability of measurements in mountains is scarce. Terrain-related effects have been considered through the terrain elevation and a topographic clear-sky solar radiation model that, additionally, also allow us to consider local clouds effects. Mesoscale-level phenomena were considered through the distance to the coast and the geographical longitude, that partially explain the atmospheric circulation in the studied region. The study has been conducted in the region of Andalusia, in southern Spain, using a target grid support of 1 km of grid-spacing and based on a 10-year length experimental dataset of 63 stations. Two different residual kriging approaches were evaluated and compared against ordinary kriging estimates. Overall, all kriging methods showed good skills in predicting the spatial regionalization of the monthly averages of daily solar radiation. The use of the distance to the coast and the geographical longitude enhanced the performance of residual kriging methods. Elevation proved to be important during summer months, while clear-sky solar radiation estimates were helpful especially during winter months. Overall, the RMSE value for ordinary kriging at the validation sites was about 3%. The residual kriging methods were able to outperform ordinary kriging around a 5% in winter and up to a 18% in summer, in relative terms.