Estimated spatiotemporal variability of total, direct and diffuse solar radiation across China during 1958–2016

Abstract

The long‐term variability of total, direct and diffuse solar radiation across China during 1958–2016 is investigated based on a ground‐measured daily radiation dataset. Missing data are estimated using a 3‐day average moving window and a backpropagation artificial neural network (BP network). The BP network achieves better estimates of direct (R2 = 0.32–0.96) than diffuse radiation (R2 = 0.00–0.81). A dimming period during 1958–1990 and a “From Dimming to Brightening” transition between 1990 and 1993 have been detected across China. The declining ratio of direct to diffuse radiation suggests a degrading air quality caused by increasing aerosols in eastern China. To study the aerosol effect on radiation, two empirical models are developed from 2000–2016 using the ground‐measured total radiation, sunshine duration and satellite‐retrieved total aerosol concentration. Both models perform well in the estimate of direct (R2 = 0.71–0.89) and diffuse radiation (R2 = 0.63–0.95). The increasing total radiation in eastern China since 2000 is mainly contributed by diffuse radiation. Besides, small anthropogenic aerosols can increase diffuse fraction, the proportion of diffuse in total radiation, whereas large natural aerosols may reduce it. The BP network and empirical models exhibit a better agreement in the estimate of direct than diffuse radiation in eastern China, which highlights the impact of aerosols on diffuse radiation in the recent decade.