基于多源数据多特征组合的白洋淀生态修复工程前后湿地变化  

Wetland changes before and after Baiyangdian ecological restoration project based on multi-source data and multi-feature combination

作　　者：郑晓玲 于一雷 吕翠翠[3] 李文彦[3] 姜纪沂 王仕琴[4] Zheng Xiaoling;Yu Yilei;Lv Cuicui;Li Wenyan;Jiang Jiyi;Wang Shiqin(Academy of Ecological Environment,Institute of Disaster Prevention,Sanhe 065201,Hebei,China;School of Eco-Environment,Hebei University,Baoding 071002,Hebei,China;Xiongan Institute of Innovation,Xiongan 071899,Hebei,China;Center of Agricultural Rescources Research,Institute of Genetics and Developmental Biology,Chinese Academy of Sciences,Shijiazhuang 050022,Hebei,China)

机构地区：[1]防灾科技学院生态环境学院,河北三河065201 [2]河北大学生态环境系,河北保定071002 [3]中国科学院雄安创新研究院(筹),河北雄安071899 [4]中国科学院遗传与发育生物学研究所农业资源中心,河北石家庄050022

出　　处：《北京林业大学学报》2025年第2期152-162,共11页Journal of Beijing Forestry University

基　　金：国家重点研发计划(青年科学家项目)(42071053);地下水污染控制与修复教育部工程研究中心开放基金(GW203312)。

摘　　要：【目的】白洋淀湿地是华北平原最大的淡水湿地生态系统,对改善和调节雄安新区乃至整个京津冀的气候和生态环境具有重要作用。探究白洋淀湿地生态修复前后下垫面的变化特征,旨在为后续的生态修复工程提供理论参考。【方法】本研究通过选取生态修复前后(2017年和2022年)的Sentinel-1 SAR影像和Sentinel-2多光谱影像,提取光谱指数特征、SAR指数特征和纹理特征,结合随机森林方法,对白洋淀湿地进行分类;从下垫面类型、面积和景观格局指数变化,分析生态修复前后白洋淀湿地的变化情况。【结果】(1)同时使用光谱指数特征、SAR指数特征和纹理特征3种特征进行分类的总体精度最高,光谱指数特征、SAR指数特征和纹理特征的重要性贡献率分别为40.713%、27.269%和32.018%。(2)2017—2022年,耕地面积明显减少,湿地面积显著增加,湿地面积的增长主要来自耕地的转换。(3)生态修复后,2022年林地和水体的破碎化程度均有减小的趋势,水体、挺水植物和林地的聚集度和连通性都有不同程度的提升,其中最明显的为林地。【结论】生态修复之前白洋淀湿地景观趋于破碎化,生态修复后白洋淀湿地的水体、挺水植物以及林地各自的连通性和聚集度明显提升。淀区内的退耕还淀、生态清淤和生态补水等工程增加了白洋淀湿地面积,提升了白洋淀水体的连通性。[Objective] Baiyangdian wetland is the largest wetland ecosystem in the North China Plain,and plays an important role in improving and regulating the climate and ecological environment of Xiong'an New Area and even the whole Beijing Tianjin Hebei Region.Exploring the changing characteristics of Baiyangdian wetland underlying surface before and after ecological restoration aims to provide theoretical reference for the subsequent ecological restoration project.[Method] Sentinel-1 SAR images and Sentinel-2multi-spectral images before and after ecological restoration(2017 and 2022) were selected to extract spectral index features,SAR index features and texture features.Combined with random forest method,the Baiyangdian wetland was classified,and the changes of Baiyangdian wetland before and after ecological restoration were analyzed from the changes of underlying surface type,area and landscape pattern index.[Result](1) The overall accuracy of classification using spectral index features,SAR index features and texture features was the highest.The importance contribution rates of spectral index features,SAR index features,and texture features were 40.713%,27.269%,and 32.018%,respectively.(2) From 2017 to 2022,the area of cultivated land decreased significantly,while the area of wetland increased significantly.The increase of wetland area mainly comes from the conversion of cultivated land.(3) After ecological restoration,the fragmentation degree of forest land and water body in 2022 had a decreasing trend,and the aggregation degree and connectivity of water,emergent plants and forest had been improved to varying degrees,among which the most obvious was forest.[Conclusion] The landscape of Baiyangdian wetland tends to be fragmented before the ecological restoration,while after ecological restoration,the connectivity and aggregation of water bodies,emergent plants,and forest land have been improved,which shows that the projects of returning farmland to lake,ecological dredging and water replenishment have increased the

关 键 词：白洋淀湿地 生态修复 湿地分类 特征组合 景观指数 随机森林 

分 类 号：P237[天文地球—摄影测量与遥感]