Estimating surface solar irradiance from geostationary Himawari-8 over Australia: A physics-based method with calibration. (15th May 2021)

• Record Type: Journal Article
• Authors: Qin, Yi; Huang, Jing; McVicar, Tim R.; West, Sam; Khan, Maryam; Steven, Andrew D.L.
• Is Part Of: Solar energy. Volume 220(2021)
• Page Start: 119
• Page End: 129
• Publication Date: 2021-05-15
• Subjects: Solar energy -- Solar irradiance -- Solar radiation -- Solar insolation -- Solar flux -- Solar exposure -- Remote sensing -- Himawari -- AHI; Solar energy -- Periodicals; Solar engines -- Periodicals; 621.47
• DOI: 10.1016/j.solener.2021.03.029
• Languages: English
• ISSNs: 0038-092X

Abstract

Highlights: Physics-based method estimating all components of surface solar irradiance (SSI). Calibration removing requirement of any ancillary data during operation. Time series based radiative transfer model to simultaneously retrieve properties of clouds, aerosols and surface. Comprehensive validation showing consistently low bias and relatively small RMSE for all components. For examples, <4 W/m 2 bias for all components and time scales, and 58 W/m 2 GHI hourly RMSE. Abstract: Surface Solar Irradiance (SSI) is required for solar energy planning and adoption, and is a fundamental parameter in modelling weather, climate, ecosystem and agricultural activities. Herein a time series based radiative transfer model was developed to simultaneously retrieve properties of clouds, aerosols and surface albedo, which were in turn used to calculate the components of SSI: i.e., global, direct and diffuse irradiance. The calculated results were calibrated across the Australian continent against in-situ measurements to account for minor factors (e.g., water vapor) not considered by the physics-based method. Detailed validation of the SSI components was performed against three years of in-situ measurements at 11 sites across Australia, at a range of time scales (i.e., instantaneous, hourly, daily and monthly) and under both all-sky and cloudy-sky conditions. The main advantage of the present method is the reliable separation of the direct and diffuse components with consistently low biases (~4 W/m 2 ) at all four time scales, while still maintaining relatively low RMSE (root-mean-square-error) and MAE (mean-absolute-error). Once calibration has been performed the model does not require any ancillary data when implemented operationally: that is the model only requires geostationary satellite data. This model may be implemented across the globe using widely available next generation geostationary satellite data with a handful of ground-data for calibration.