Validation of decomposition models for solar irradiance at high latitudes: A preliminary study

Abstract

The accurate estimation of buildings' solar potential contributes to boost the exploitation of solar energy at high latitudes. The decomposition of global irradiation into the direct and diffuse fractions is a fundamental step of the solar irradiance model chain. Diffuse and direct irradiation are, in fact, rarely measured. Previous works recommended Yang4 as the decomposition model with the best overall performance. However, in geographically limited applications, quasi-universal decomposition models such as Yang4 and Engerer4 can be outperformed by local models (i.e., models parametrized with climate-specific data) such as Skartveit3 and Starke3. This makes necessary to perform local validation studies to verify the findings from worldwide validation studies. In this study, the four decomposition models are implemented in Python and experimentally validated against one-minute solar irradiance data (i.e., direct and diffuse irradiance) of Trondheim (Lat. 63°26' N, Norway). Two months representative of clear sky (August) and overcast (October) conditions are considered. The study confirms that the Yang4 model performed the best for high-latitude application: the nMBE ranged from -0.54% (August) to 0.65% (October), the nRMSE from 17.18% (August) to 22.29% (October), and the R2 from 0.96 (August) to 0.97 (October). However, Skartveit3 combines a level of performance close to Yang4 with the lower number of input parameters.