Sampling design optimization of ground radiometric stations

Abstract

This chapter presents a methodology to identify optimal site locations to establish a surface radiometric monitoring network once the raw solar estimations are produced from satellite images or numerical models. The site selection is done considering the long-term solar resource, its spatial distribution, variability and technical and logistics aspects. The methodology presented here is an adaptive sampling strategy under an assumed population model derived from satellite images or numerical models. The objective is to install the radiometric stations in optimal locations to correct the systematic biases of the modelled solar radiation, improving the estimates and minimizing the number of stations needed. To achieve that, we need to identify the area with a similar dynamic in terms of solar radiation. Inside the areas identified, the most optimal locations will be used to place the radiometric stations. The methodology is divided into three phases. The first phase divides the geographical extension under study to identify the areas with a similar dynamic in terms of monthly solar radiation. The selection of the number of cluster/areas is done with an information criteria technique. Once the optimal number of clusters and the extension of each area is defined, the second phase is the production of a long list of candidate sampling sites with GIS techniques based on constraints to identify the best locations in each area to place the radiometric stations. The third phase is based on ranking the long list of candidate locations with site visits and a checklist criterion based on BSRN recommendations to produce a final shortlist of optimal sites to measure solar radiation in each area. Finally, two tiers of radiometric stations are proposed to place in each area depending on the ranking level of each location.