Abstract: With the rapid development of satellite observation technology, scientists have made relatively accurate estimates of the energy budget at the top of the Earth's atmosphere. However, the level of estimation error in the Earth's surface energy imbalance is still more than one order of magnitude greater than the changes in radiative forcing caused by the increase in greenhouse gases. Among various components of energy imbalance estimation on the Earth's surface, the downward shortwave solar radiation contributes one of the most important sources of error. Therefore, accurate observation of surface solar radiation (SSR) has become a key variable in reducing this error. At the same time, SSR is not only an important driving factor for simulating land surface processes, but also an important indicator for the utilization of renewable solar photovoltaic energy. Its long-term changes have always been a key scientific issue for scientists in many fields. This paper first compares existing SSR data products, then discusses the bias and uncertainty issues in the long-term changes of SSR, and finally clarifies the importance of developing high-quality SSR observation baseline data products. On this basis, the paper introduces a series of research in recent years, which are based on the most complete SSR station observation data to date and systematically considered its inhomogeneity and sampling problems, performed systematic homogenization processing and AI reconstruction of site sequences, and estimated the long-term changes and uncertainty levels of global and regional SSRs. These studies provide new evidences for global and regional climate change observations, detection attribution, and future projection. Finally, this paper provides an outlook on existing problems and future development trends in current SSR data and long-term change research.