摘要
遥感技术已被广泛应用于大面积地表水观测,但由于水面光谱特征在空间和时间上的高度变化,使得遥感在大面积、多时相地表水监测方面存在局限性,因此提出一种融合光谱和时相特征的地表水遥感信息提取模型,以解决中分辨率遥感影像条件下植被、云、地形阴影等地表覆盖背景中开阔水域的提取问题。基于Google Earth Engine(GEE)平台构建地表水体分类模型系统,处理辅助数据集以及Landsat卫星影像的多光谱/时相属性,从长期连续的Landsat影像多维数据中提取地表水分布,探索干旱地区地表水的时空变化动态。利用三个独立的数据集,验证地表水提取的总体精度和kappa系数分别为0.91和0.81。该模型通用、高效且可靠,应用于中国内陆干旱区的实验表明,该方法对多种环境条件都有效,可为较复杂地表环境下水体提取和时空演化分析提供技术参考。
Remote sensing has been applied to large-area surface water observations to obtain higherresolution,long-term continuous surface water observation records.However,due to the high spatial and temporal variability of water surface spectral characteristics,remote sensing still has limitations in large-area,multi-temporal surface water monitoring.In this study,a surface water remote sensing information extraction model that integrates spectral and temporal characteristics is proposed to solve the problem of extracting open water in surface coverage backgrounds such as vegetation,clouds,and terrain shadows under mediumresolution remote sensing images.Based on the proposed model and combined with the Google Earth Engine(GEE) platform,a surface water classification system was constructed to process the auxiliary data sets and multispectral/temporal attributes of the Landsat satellite images,and then the equations of the cluster shell of ground objects in the multidimensional spectral feature space were incorporated into the classification process,surface water distribution is extracted from long-term continuous Landsat image multi-dimensional data.Based on this,the spatiotemporal changes of surface water in arid areas over decades is explored.Using three independent data sets,the overall accuracy and kappa coefficient of surface water extraction were verified to be 0.91 and 0.81,respectively.The proposed method is universal and it can be applied to other sensors with similar spectral bands with good reliability and efficiency.Experiments applied in arid inland areas of China shows that this method is effective for various environmental conditions.The research results record the dynamic changes of surface water with good spatial details and accuracy level,which can provide technical reference for water extraction and spatiotemporal evolution analysis in complex surface environments.
作者
王西营
张帅娟
WANG Xiying;ZHANG Shuaijuan(Hunan Provincial Institute of Land and Resources Planning,Changsha 410119,China;Henan Zhong'an Planning and Design Co.,Ltd.Hunan Branch,Changsha 410118,China)
出处
《内蒙古师范大学学报（自然科学版）》
CAS
2024年第5期488-496,共9页
Journal of Inner Mongolia Normal University(Natural Science Edition)
基金
宁夏回族自治区自然科学基金资助项目“宁夏平原湿地水资源的季节性变化及其对年内季相干旱效应的响应机理”(2023AAC03041)
宁夏回族自治区青年科技人才托举工程资助项目“第七批自治区青年科技人才托举工程”(2023AA071)。
关键词
多维光谱特征空间
光谱聚簇外壳方程
地表水提取模型
地表水体时空变化
multidimensional spectral feature space
shell equations of spectral clusters
surface water extraction model
spatiotemporal variation of surface water
