Algorithms for separating diffuse and beam irradiance from data over the East Asia-Pacific region: A multi-temporal-scale evaluation based on minute-level ground observations

Diffuse irradiance and direct normal irradiance are two components that are critical to both energy and ecological systems; however, these are rarely available in most ground observations, which only provide the global horizontal irradiance (GHI), leaving the separation of them from the GHI an inevitable process. Although many separation models have been developed to provide solutions, a large proportion of these has been constructed based on site-specific datasets at different time resolutions, limiting their possibilities to be adopted to other sites without previously being extensively evaluated. In this study, we evaluated a total of 80 separation algorithms with specific attention to their time resolutions, based on the finest minute-scale observations from 13 irradiation sites across six Ko center dot ppen-Geiger-Photovoltaic climate classification zones in East Asia-Pacific. We found that there was a dramatic difference in the performance of hourly-scale models and that the Starke3 model gave the best performance of the minute-scale models. Further residual analyses revealed that the clearness index, solar zenith angle, and atmospheric parameters played important roles. Since data from the geostationary Himawari-8 satellite are available to provide minute-scale atmospheric parameters, there is the possibility of reducing the residuals by incorporating the information with high temporal resolutions into separation models. The results of this study will provide a more comprehensive understanding of the utilization and improvement of separation models.