基于多时相影像分类的近23 a洪湖水生植被覆盖变化研究  

Trajectory of Aquatic Vegetation Cover in Honghu Lake in Recent 23 Years Based on Multi-temporal Image Classification

作　　者：刘李琼 熊晶 张媛 陆羽仪 蔡晓斌 LIU Li-qiong;XIONG Jing;ZHANG Yuan;LU Yu-yi;CAI Xiao-bin(Key Laboratory for Environmental and Disaster Monitoring and Evaluation in Hubei,Innovation Academy forPrecision Measurement Science and Technology,Wuhan 430077,China;University of Chinese Academy of Science,Beijing 100049,China;Ecological Environment Monitoring Center Station of Hubei Province,Wuhan 430071,China)

机构地区：[1]中国科学院精密测量科学与技术创新研究院环境与灾害监测评估湖北省重点实验室,湖北武汉430077 [2]中国科学院大学,北京100049 [3]湖北省生态环境监测中心站,湖北武汉430071

出　　处：《长江流域资源与环境》2025年第1期126-139,共14页Resources and Environment in the Yangtze Basin

基　　金：国家自然科学基金项目(42171381,U22A20567)。

摘　　要：水生植被作为初级生产者,可以为湖泊提供食物来源和栖息地等重要的生态服务功能。动态监测水生植被对于湖泊保护至关重要。研究基于典型水生植被NDVI物候特征,构建了CART决策树的水生植被分类模型。利用多时相MOD09Q1数据进行分类提取了2000~2022年洪湖水生植被的时空分布信息。结果显示多时相分类方法可有效区分不同类型的水生植被,总体分类精度达85.6%。总体上洪湖水生植被呈现出明显的空间化结构特征,挺水植被菰主要在其西南片区集中分布,莲则在湖岸周边呈零星带状分布。沉水植被则主要分布于挺水植被区向湖心扩散区域及湖泊沿岸湖湾区,整体表现为由外向内的圈层特征。从水生植被的时序变化来看,主要经历了3个阶段。2000~2004年受大面积围网养殖以及高强度人类活动的影响,湖泊的水生植被总体偏少,平均面积在150.56 km2左右。2005~2016年受围网拆除以及湿地生态工程实施,内源污染得到控制,洪湖水生植被逐渐得到恢复,平均面积增至181.85 km2左右,除发生特大旱涝灾害的特殊年份,水生植被面积总体趋于稳定。2017后,受2016年洪水扰动及清漂、圩堤拆除等大范围工程施工引起的底泥扰动、植被破坏,加剧了洪湖水质恶化过程,水生植被面积锐减,平均面积降至100.35 km2,2022年后水生植被面积达到历史最低,近两年水生植被占比维持在10%以下。近23 a水生植被时空变化的监测结果,可为洪湖后续水生植被修复区规划、修复物种的选择提供基础支撑。Aquatic vegetation,as the primary producer in lakes,provides multiple ecological services such as food sources and habitats.Dynamic monitoring of aquatic vegetation is essential for lake protection.The CART(Classification and Regression Tree)approach was utilized to extract the aquatic vegetation based on the phenological characteristics of different types of vegetation of the multi-temporal MOD09QI NDVI(Normalized Differential Vegetation Index)series.The approach was applied to derive the annual aquatic vegetation maps of the Honghu Lake from 2000 to 2022.The results showed that this approach was effective to distinguish the different types of aquatic vegetation with an overall classification accuracy of 85.6%.The aquatic vegetation generally exhibits a distinct spatial structural characteristic.The emergent vegetation,Zizania latifolia,was concentrated in the southwest lake,while Nelumbo nucifera was scattered in areas around the lakeshore.Submerged vegetation was mainly found from the emergent area towards the lake and the shoreline.In terms of the change trajectory,three main phases were identified.From 2000 to 2004,the aquatic vegetation was relatively scarce,with an average area of 150.56km2.For the second phase of 2005 to 2016,with the implementation of wetland ecological projects to control surrounding pollutions,the aquatic vegetation gradually recovered to an area of 181.85km2.The overall aquatic vegetation area was stable in the normal year without extreme drought and flood in the second phase.After 2017,the average area of aquatic vegetation decreased sharply to 100.35km2.This change might be attributed to the continuously deterioration of water quality,the increase of sediment resuspension that was caused by the dike and debris removal projects,and the flood event in summer 2016.After 2022,the aquatic vegetation area reached the historical lowest,which was below 10%in the last two years.This result provided key information on the potential sites and species that could be used in ecological restoratio

关 键 词：物候 MODIS 多时相决策树分类 水生植被覆盖变化 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]