首页|融合多源遥感影像的城市扩展识别与空间驱动分析

融合多源遥感影像的城市扩展识别与空间驱动分析

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
鉴于当前较难获取最新高精度长时序遥感城市土地监测图,现有城市扩展空间驱动分析实践识别主要驱动因子时未检验因子集的全局解释性且较少关注空间溢出效应.以东莞为实验对象,集合Landsat影像和MODIS归一化植被指数(normalized differential vegetation index,NDVI)时序产品,构建深度学习分类器,获取高精度土地覆盖分类图,识别城市扩展,使用基于logistic回归的探索性回归识别解释东莞全局城市扩展的最优因子集,进而使用auto logistic回归测度空间溢出效应的影响并进行驱动分析.研究发现:①基于深度学习分类器,融合Landsat光谱、纹理和MODIS NDVI时序变化信息可获取高精度(Kappa>93%)土地覆盖分类图;②基于探索性回归可良好识别解释全局城市扩展的最优因子集,受试者工作特征曲线(receiver operating charac-teristic curve,ROC)>0.85;③2000-2020年,最优解释东莞全局城市扩展的主要空间驱动因子有城市规划方案、距建成区距离、空间加权城市密度、城市扩展的空间溢出效应,主要限制因子有高程、坡度、1 km²水体密度和土地可得性.
Identification and Spatial Determinant Analysis of Urban Expansion Through Integrating Multi-source Remote Sensing Images
Currently,it is difficult to obtain the latest high-ac-curacy remote sensing urban land monitoring maps over long-time span,while the studies on spatial determinant analysis of urban expansion ignore the globally explanatory ability of de-terminant sets and pay little attention to spillover effect when identifying major determinants.Taking Dongguan as the ex-perimental area,this study involves Landsat images and MO-DIS normalized differential vegetation index(NDVI)time-se-quence products,constructs deep-learning classifier to obtain high-accuracy land cover classification maps,and then identi-fies urban expansion.Subsequently,exploratory regression backed by logistic regression is employed to identify the opti-mal determinant sets that best explain the urban expansion in Dongguan.Finally,auto logistic regression is used for mea-suring the impact of spillover effect and spatial determinant analysis.The following conclusions are achieved:First,by employing deep-learning classifier and integrating spectral and context information from Landsat images with NDVI dia-chronically variation information from MODIS products,high-accuracy(Kappa>93%)land cover classification maps can be obtained.Second,the optimal spatial determinant sets for globally explaining urban expansion can be well identified by exploratory regression,with receiver operating characteris-tic curve(ROC)over 0.85.Third,the main spatial drivers that optimally explain global urban expansion in Dongguan from 2000 to 2020 include urban planning schemes,distance to build area,spatially weighted urban percentage and spill-over effect of urban expansion.In contrast,the main con-straining factors include elevation,slope,water percentage within 1 km² and land availability.

convolutional neural network(CNN)long short-term memory(LSTM)networksurban expansionspatial determinantauto logisticspillover effect

杨昌兰、关雪峰、李静波、吴华意

展开 >

武汉大学测绘遥感信息工程国家重点实验室,湖北 武汉,430079

卷积神经网络(convolutional neural network,CNN) 长短期记忆(long short-term memory,LSTM)网络 城市扩展 空间驱动因子 auto logistic 空间溢出效应

国家自然科学基金

41971348

2024

10.14188/j.2095-6045.2022128

测绘地理信息
武汉大学

测绘地理信息

CSTPCD
影响因子：0.563
ISSN：1007-3817
年,卷(期)：2024.49(2)
  • 19