| 英文摘要： |
Information regarding the dynamic change of urban green space (UGS) at fine spatiotemporal resolutions is important for developing sustainable urban environments and understanding urban ecosystems in the future. As the capital of China, due to population explosion, economic growth, and redevelopment of inner-city areas, Beijing underwent extremely rapid urbanization in the past few years, which is intensifying the contradiction between human activities and UGS. Timely monitoring of Beijing's dynamic change of UGS is very important for the planning of the capital's living environment. Hence, a novel framework was proposed to map the dynamic change of UGS from 2000 to 2018 by integrating social sensing data and time series remote sensing data. First, we generated a dynamic mapping of the vegetation and a mapping of the annual maximum value of the Nominalized Difference Vegetation Index (NDVImax), using time series Landsat images on a Google Earth Engine platform. Then, a temporal segmentation approach was implemented to identify the turning years and using a decision tree approach determine the historical time of social sensing data. Finally, the UGS mapping and its dynamics were generated according to the identified time parameter of social sensing data, as well as remote sensing data. Our results showed that the temporal segmentation approach is reliable for detecting urban changes and identifying the historical time of social sensing data, with an overall accuracy of 90%. The overall classification accuracies of UGS are 0.87 (both Level 1 and Level 2). Moreover, the UGS area continuously increased from 2000 to 2018, of which the dominant types are residential green space and theme parks. The proposed framework provides an effective tool for better understanding the dynamic change of UGS, and the results from this work present vital insights for planners and government departments to develop targeted measures and protect UGS effectively. |
在线留言
联系我们
所长信箱