生物入侵严重威胁生物多样性与生态安全，成为全球环境变化研究中的热点问题。互花米草是我国海岸带最典型的入侵植物，在滨海地区分布范围较广，且有向高纬度扩散的趋势。在复杂的滨海环境背景下，如何充分挖掘遥感信息，精准获取互花米草入侵状况是目前亟待解决的技术问题之一。本项目以环渤海海岸带为研究区，基于湿地植物群落野外调查和野外光谱测量，解析互花米草高光谱遥感机理；基于国产GF-5高光谱影像数据，集成数据降维与U-net深度学习模型方法，进行环渤海海岸带互花米草遥感识别；分析互花米草空间格局对环境因素的响应规律，利用MaxEnt生态位模型，预测渤海湾海岸带互花米草潜在分布区，明确研究区互花米草入侵的高风险区域。研究成果将为入侵植物高光谱遥识别提供思路和方法借鉴，为海岸带入侵植物管控、滨海湿地生态系统管理等提供数基础和科学依据，也将推动国产高光谱卫星数据的行业化应用。

Biological invasion has been a research hotspot of global environmental change and poses significant threats to biodiversity and ecological security. As the most typical invasive plant in the coastal salt marsh of China, Spartina alterniflora presents a great invasion risk in high latitude coastal areas of China. Under the background of the complex coastal environment, how to fully mine the remote sensing image information and accurately obtain the information of the distribution of S. alterniflora is one of the technique problems to be solved urgently. This project selects the coastal zone of Bohai Bay as the research area. GF-5 hyperspectral images and deep learning method combined with dimension reduction algorithm are used to get the distribution of S. alterniflora in tidal flats.Based on the distribution of S. alterniflora and meteorological data set,.MaxEnt modeling is established to predict the potential distribution of S. alterniflora and analyze the response of the spatial pattern of S. alterniflora to environmental factors. The project will provide ideas and methods for hyperspectral remote sensing recognition of invasive plants. Implementation of this project will benefit and provide scientific and data basis to the control of bio-invasion and conservation of coastal wetlands, at the same time promote the application of domestic hyperspectral satellite data.