A comprehensive long-term (1981–2020) radiation dataset for China: Ensuring consistency, comparability, and incorporating uncertainty

Abstract Comprehensive, reliable, and long-term radiation data is imperative for ecological model development and improvement. However, the existing radiation datasets either lack comprehensiveness or have relatively short temporal coverage, and few provide information regarding their uncertainties. In the present study, we generated a comprehensive radiation dataset for China from 1981 to 2020 by combining in-situ observations from the China Meteorological Administration (CMA) and the Chinese Ecosystem Research Network (CERN) with an advanced Gaussian Process Regression (GPR) approach, covering global radiation, diffuse radiation, photosynthetically active radiation (PAR), and diffuse PAR, along with corresponding uncertainty estimates, and independently validated the robustness of the dataset. The results indicated that the generated radiation dataset effectively captured the inter-annual and seasonal dynamics of the observations, with R 2 values of 0.69 and 0.89, respectively; and the spatial uncertainties of the dataset were relatively low, around 1%. The dataset exhibited evident spatial heterogeneity, higher in the southwest and lower in the northeast, and the annual mean values of global radiation, diffuse radiation, PAR, and diffuse PAR were 5355.63, 2411.11, 2075.78, and 1018.99 MJ/m 2 ·year, respectively. From 1981 to 2020, global radiation and PAR exhibited decreasing trends primarily driven by water vapor, with decline rates of −3.62 and −1.37 MJ/m 2 ·year, respectively, whereas changes in their diffuse components were modest, at −0.39 and −0.21 MJ/m 2 ·year, mainly attributable to the combined influence of aerosol optical depth (AOD) and cloud cover. Our study provides a comprehensive and consistent spatiotemporal dataset, which is of great value for investigating the impact of radiation on terrestrial ecosystems and for conducting ecosystem modeling and uncertainty analyses.