Monitoring land cover change in urban and pen-urban areas using dense time stacks of Landsat satellite data and a data mining approach
Monitoring land cover change in urban and pen-urban areas using dense time stacks of Landsat satellite data and a data mining approach
复制标题
DOI：
10.1016/j.rse.2012.06.006
复制
发表时间：
2012-09-01
影响因子：
13.5
通讯作者：
Schneider, Annemarie
中科院分区：
文献类型：
--
作者：
Schneider, Annemarie
鉴于全球许多地区城市扩张的速度和规模，城市环境在日常生活质量问题、生态过程、气候、物质流动和土地改造中发挥着越来越重要的作用。遥感已成为监测城市扩张速度和模式的有力工具，但许多早期挑战 - 例如区分新城市土地与裸地 - 仍未解决。为了处理定居点内的高时间和空间变化以及复杂的多特征类别，本文提出了一种新方法，该方法使用多日期复合变化检测技术来利用陆地卫星图像密集时间堆栈中的多季节信息。该方法的核心前提是，城市地区内/附近的土地在变化发生之前和之后都具有不同的时间轨迹，并且这些导致了几个光谱区域的特征时间特征。该方法依赖于监督分类，该分类利用从 Google 地球图像解释的稳定/变化区域的训练数据，以及提供所有可用陆地卫星数据作为输入的“强力”方法，包括由于扫描线校正器 (SLC) 问题而存在数据间隙的场景。测试了三种分类算法（最大似然、增强决策树和支持向量机）在三个研究区域（规模、生态气候条件和发展速度/模式不同）五个时期（1988-1995、1996-2000、2001-2003、2004-2006、2007-2009）监测扩张的能力。决策树和支持向量机的性能均优于最大似然分类器（总体准确度为 90-93%，而前者为 65%），但决策树在处理缺失数据方面更胜一筹。向 Landsat 数据堆栈添加转换后的特征（例如波段度量）可将准确度提高 1-4%，而减少特征数量（旨在模拟噪声或缺失数据）的实验会导致准确度下降 1-9%。事实证明，该方法对于监测城市核心区以外的笔式城市化特别有效，覆盖了 98% 以上的村庄住区。 (c) 2012 Elsevier Inc. 保留所有权利。
Given the pace and scale of urban expansion in many parts of the globe, urban environments are playing an increasingly important role in daily quality-of-life issues, ecological processes, climate, material flows, and land transformations. Remote sensing has emerged as a powerful tool to monitor rates and patterns of urban expansion, but many early challenges - such as distinguishing new urban land from bare ground - remain unsolved. To deal with the high temporal and spatial variability as well as complex, multi-signature classes within settlements, this paper presents a new approach that exploits multi-seasonal information in dense time stacks of Landsat imagery using a multi-date composite change detection technique. The central premise of the approach is that lands within/near urban areas have distinct temporal trajectories both before and after change occurs, and that these lead to characteristic temporal signatures in several spectral regions. The method relies on a supervised classification that exploits training data of stable/changed areas interpreted from Google Earth images, and a 'brute force' approach of providing all available Landsat data as input, including scenes with data gaps due to the Scan Line Corrector (SLC) problem. Three classification algorithms (maximum likelihood, boosted decision trees, and support vector machines) were tested for their ability to monitor expansion across five time periods (1988-1995,1996-2000,2001-2003,2004-2006,2007-2009) in three study areas that differ in size, eco-climatic conditions, and rates/patterns of development. Both the decision trees and support vector machines outperformed the maximum likelihood classifier (overall accuracy of 90-93%, compared to 65%), but the decision trees were superior at handling missing data. Adding transformed features such as band metrics to the Landsat data stack increased accuracy 1-4%, while experiments with a reduced number of features (designed to mimic noisy or missing data) led to a drop in accuracy of 1-9%. The methodology also proved particularly effective for monitoring pen-urbanization outside the urban core, capturing >98% of village settlements. (c) 2012 Elsevier Inc. All rights reserved.