水域多源遥感提取成果的空间一致性分析
Spatial consistency analysis of surface water extracted from multi-source remote sensing data in lake area
许珊 1邹滨 2翟亮 3桑会勇 3周希雅1
作者信息
- 1. 长沙理工大学水利与环境工程学院,长沙 410114
- 2. 中南大学地球科学与信息物理学院,长沙 410083
- 3. 中国测绘科学研究院,北京 100036
- 折叠
摘要
针对同一区域多源遥感数据水域提取成果可能存在的空间差异,该文提出了一种空间一致性度量指标.以东洞庭湖为例,基于Landsat、哨兵卫星影像与随机森林模型开展水域遥感提取成果精度对比与空间一致性度量,并借助相关性与重要性分析手段探索地理特征因子对指标的指征规律.结果表明,不同遥感数据源水域提取成果精度不同,空间一致性存在差异,且差异随时间变化.地表反射率、地表温度和植被覆盖度能在一定程度上表征多源遥感水域提取成果的空间一致性(相关系数:-0.772～0.428;重要性:0.101～0.697)及差异,但存在时间效应.因此,在水域时空变化监测过程中基于多源遥感数据进行融合或填补时,应当顾及此种空间一致性差异及其时间情景特征,考虑应用特征地理因子提升效果.
Abstract
To address the potential spatial differences in the extraction results of surface water from multi-source remote sensing data in the same lake area,a spatial consistency measurement index was proposed in this paper.Taking the East Dongting Lake area as an example,surface water was extracted from images of Landsat and Sentinel satellites with random forest models.The extraction accuracy was compared and spatial consistency between extraction results was measured.The relationship between spatial consistency and geographical factors was explored using correlation and importance analysis methods.The results indicated that the land surface reflectance,land surface temperature,and vegetation coverage could indicate the spatial consistency and difference in the extraction results of surface water from multi-source remote sensing data to some extent(correlation coefficient:-0.772～0.428,importance:0.106～0.697),but there were differences between periods.Therefore,in the process of monitoring spatial-temporal changes in surface water of lake area based on multi-source remote sensing data fusion or filling,this spatial consistency difference and its temporal scenario characteristics should be taken into account and the related geographical factors may improve the monitoring reliability.
关键词
水域/空间差异/随机森林/关联特征/地理因子Key words
lake water area/spatial difference/random forest/relationship features/geographic factors引用本文复制引用
基金项目
地理信息工程国家重点实验室、自然资源部测绘科学与地球空间信息技术重点实验室联合资助项目(2022-02-10)
出版年
2024
NETL
NSTL
万方数据