Mapping interannual changes of evergreen forests in Beijing, China at 30-m spatial resolution during 1990–2023

The lack of long-term, high-resolution annual maps of evergreen forests has constrained understanding of their spatiotemporal dynamics and hindered effective management and conservation. To address this gap, we generated a 34-year (1990–2023) annual evergreen forest dataset for Beijing at 30 m resolution and analyzed its long-term dynamics. We proposed a novel time-series evergreen forest mapping framework integrating automated training sample extraction, machine learning classification, and temporal refinement. Accuracy assessments for 2010, 2016, and 2022 showed consistently high performance, with overall accuracy ranging from 92.32% to 95.00%. Temporal reliability was further evaluated using an earliest change year map, with overall accuracy of 87.24% within ± 2 years. The results revealed a substantial increase in evergreen forest area from 28.09 km² in 1990 to 1,073.79 km² in 2023, characterized by slow growth before 2005 and rapid expansion thereafter. Major gains were occurred in Miyun District (+208.14 km²), followed by Huairou, Yanqing, and Pinggu.