Exploring global dark sky quality and potential dark sky park site selection: Integrating multi-source spatial data with random forest model

With the continuous advancement of global urbanization and industrialization, the pollution issue (e.g., light pollution) has become increasingly severe, posing a serious threat to the global dark sky quality. Evaluating the global dark sky quality and conducting site selection for potential dark sky parks not only contribute to the protection of natural ecosystems and the safeguarding of ecological security, but also provide scientific reference for sustainable ecotourism. Thus, our study integrated remote sensing nighttime light, cloud cover, and other data to develop a global dark sky quality identification based on the random forest model, effectively achieving interannual and seasonal dynamic identification of global dark sky quality. Then, the study defined global construction potential constraints and conducted site selection research for potential conventional dark sky parks and potential seasonal dark sky parks. The results show that in 2023, areas with high dark sky quality accounted for only 20.6 % globally, mainly distributed in high-altitude or arid regions worldwide, including the Qinghai-Tibet Plateau, inland Australia, and other areas. From 2012 to 2023, global dark sky quality exhibited an interannual fluctuation pattern of “decline–rise–decline” and further showed significant seasonal differences. In addition, the identified potential conventional dark sky parks are mainly distributed in southern Africa, the southwestern United States, and other regions, while potential seasonal dark sky parks are mainly located in the Korean Peninsula, southern Qinghai-Tibet Plateau, and other areas. Our study can provide a scientific reference for global light pollution control, ecotourism development, and policymaking on nighttime environmental protection, supporting strategic progress assessment of the United Nations Sustainable Development Goals (e.g., SDG 11: Sustainable Cities and SDG 15: Life on Land). • Random forest model outperforms in global dark sky quality assessment; • High dark sky quality areas concentrate in remote and high-altitude regions; • Global dark sky quality shows “decline–rise–decline” inter-annual trends; • Potential conventional dark sky parks and potential seasonal dark sky parks were identified.