基于多源遥感数据的鬼城现象分析方法

Analysis method of ghost city phenomenon based on multi-source remote sensing data

遥感技术对于城市化进程监测,把握人与自然和谐共生等方面起着至关重要的作用。为进一步推动遥感技术在城市专题研究中的深入应用,本文提出一种基于多源遥感数据的鬼城现象分析方法,将美国军事气象卫星(DMSP)夜间灯光数据和Landsat遥感影像数据进行特征融合与优势互补,在此基础上建立支持向量机(SVM)监督分类中的样本精炼与迭代分类机制,并将获得的城市建成区精细化结果,引入至现有鬼城指数(GCI),以分析“鬼城现象”发生区域的时空分布特征。为了测试该方法,本文选取广西壮族自治区各下辖市作为试验区域,分析2008—2015年“鬼城现象”,从可视化效果和精度两个方面验证本文方法的有效性。实验结果表明,所提出的方法相比传统SVM监督分类方法可以获得更优的鬼城现场时空分布特征。

Remote sensing technology plays a crucial role in monitoring the urbanization process and understanding the harmonious coexistence of humans and nature.To further promote the in-depth application of remote sensing technology in urban thematic research,this paper proposed an analysis method of ghost city phenomenon based on multi-source remote sensing data.This method integrated features and complemented advantages between night-time light data from the defense meteorological satellite program(DMSP)and data from Landsat remote sensing images.On this basis,the paper established a sample refinement and iterative classification mechanism within the supervised classification framework of support vector machine(SVM).The refined urban built-up area results obtained were then incorporated into the existing ghost city index(GCI),so as to analyze the spatio-temporal distribution characteristics of ghost city phenomena within the study area.To test this method,multiple cities under the jurisdiction of the Guangxi Zhuang Autonomous Region were selected as the study areas to analyze the ghost city phenomenon from 2008 to 2015.The effectiveness of the proposed method was validated in terms of visualization effects and accuracy.Experimental results indicate that the proposed approach can obtain more precise spatio-temporal distribution characteristics of ghost city phenomena compared to traditional SVM-based supervised classification methods.