Silvia Ma Lu , Sebastian Zainali, Pietro Elia Campana, Bengt Stridh, Anders Avelin, Stefano Amaducci, Michele Colauzzi

Decomposition models of solar irradiance estimate the magnitude of diffuse horizontal irradiance from global horizontal irradiance. These two radiation components are well-known to be essential for the prediction of solar photovoltaic systems performance. In open-field agrivoltaic systems, that is the dual use of land for both agricultural activities and solar power conversion, cultivated crops receive an unequal amount of direct, diffuse and reflected photosynthetically active radiation (PAR) depending on the area they are growing due to the non-homogenously shadings caused by the solar panels installed (above the crops or vertically mounted). It is known that PAR is more efficient for canopy photosynthesis under conditions of diffuse PAR than direct PAR per unit of total PAR. For this reason, it is fundamental to estimate the diffuse PAR component in agrivoltaic systems studies to properly predict the crop yield. Since PAR is the part of electromagnetic radiation in the waveband from 400 to 700 nm that can be used for photosynthesis by the crops, several stand-alone decomposition models of solar irradiance are selected in this study to partition PAR into direct and diffuse. These models are applied and validated in three locations in Sweden: Lanna, Hyltemossa and Norunda, using the coefficients stated on the original publications of the models and locally fitted coefficients. Results showed weaker performances in all stand-alone models for non-locally fitted coefficients (nRMSE ranging from 29% to 95%). However, performances improve with re-parameterization, reaching highest nRMSE of 37.94% in Lanna. YANG2 decomposition model is the best-performing one, reaching lowest nRMSE of 24.31% in Norunda applying re-estimated coefficients. Country level sets of coefficients for the best-performing models, YANG2 and STARKE, are given after parameterization using joined data of the three locations in Sweden. These Sweden-fitted models are tested and showing nRMSE of 25.56% (YANG2) and 28.36% (STARKE). These results can be used to perform estimations of PAR diffuse component in Sweden where measurements are not available, and the overall methodology can be similarly applied to other countries.