【摘要】 城市扩张的研究对促进土地资源配置的高效率和合理化有着重要的意义。以南昌市为研究区，基于Google Earth Engine(GEE)平台，利用NPP/VIIRS夜光数据和阈值分割法提取城市建成区后，采用随机森林（random forest,RF）算法解译城市土地信息，分析了建成区范围内2000—2020年20年的土地利用变化。结果表明，基于GEE和随机森林算法得到的分类结果精度较高，平均精度和Kappa系数分别为96.81%和94.56%，且相较于传统遥感分析技术模式效率更高；南昌市20年的城市扩张速率经历了从低速到缓慢的扩张过程。土地利用类型的转移以植被转为建筑用最多，裸地次之，水体最少。社会经济的发展和人口的迁移和增长在一定程度上影响了城市扩长的速率和方向，是城市化的重要影响因素。更多还原

【Abstract】 The study of urban expansion is an important method to promote high efficiency and rationalization of land resource allocation. Taking Nanchang as the research area and using the remote sensing data provided by Google Earth Engine(GEE), we extracted the urban area by NPP/VIIRS night light data and threshold segmentation, the land use change of different five periods from 2000 to 2020 in the urban area can be mapped and analyzed by using the random forest(RF) algorithm,then we evaluated the accuracy of land use classification. The experimental results show that the land use classification based on GEE and random forest algorithm have high accuracy, the average overall accuracy and kappa coefficient are 96. 81% and 94. 56% respectively, the method is more efficient than the traditional remote sensing analysis technology model. In the past two decades, the study area has experienced the urban expansion from low rate to a relatively slower rate respectively.The areas transfer from vegetation to construction land are the largest, followed by bare land and the least water. Development of social economy as well as the migration and growth of population determine the rate and direction of urban expansion to a certain extent, which is an significant influencing factor of urbanization. Big data remote sensing analysis based on cloud platform is the development trend of remote sensing technology in the future.更多还原