中国科学院机构知识库网格
Chinese Academy of Sciences Institutional Repositories Grid
首页 机构 成果 学者
  
  1. CAS IR Grid
  2. 海洋研究所
  3. 中国科学院海洋研究所
  4. 海洋环流与波动重点实验室
Label-free mangrove mapping from temporally consistent PlanetScope imagery with interpretable deep unfolding network

文献类型：期刊论文

作者Nie, Xiangyu1,2,4; Xue, Zhaohui3; Li, Xiaofeng5
刊名ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING
出版日期2026-05-01
卷号235页码:19-37
关键词Mangrove Label-free PlanetScope Interpretability Deep unfolding
ISSN号0924-2716
DOI10.1016/j.isprsjprs.2026.02.035
通讯作者Xue, Zhaohui(zhaohui.xue@nuaa.edu.cn)
英文摘要Mangroves play a vital role in maintaining coastal ecosystem stability and supporting environmental sustain-ability, but they have experienced widespread degradation due to human activities and climate change. As such, accurate, efficient, and reliable mapping results are essential for mangrove monitoring and effective management. However, with the increasing demand for detailed monitoring, the limited spatial and temporal resolution of widely utilized Landsat and Sentinel series imagery is proving insufficient. Meanwhile, existing mapping methods based on spectral indices and traditional machine learning struggle with low accuracy and limited generalization. Although deep learning-based methods perform better, they are constrained by the high demand for labeled samples and the black-box nature. To address these issues, this study conducts a systematic exploration of imagery selection, label-free sample acquisition, and model interpretability. First, low-tide and temporally consistent PlanetScope imagery is selected as the data source, providing a solid foundation for fine-scale mangrove mapping. Second, by leveraging spectral index features and existing mangrove datasets, a label-free sample annotation strategy is proposed to efficiently acquire high-quality training samples. Finally, inspired by the representation model, a novel deep unfolding network with an interpretable decision-making mechanism is developed for mangrove mapping, which not only enhances the accuracy of the results but also strengthens its credibility in policy-making and management. To validate our research, a national-scale mangrove mapping experiment is conducted in China using temporally consistent PlanetScope imagery acquired in 2024. The mapping results reveal that China's mangrove area is 30,455 ha, with a total of 14,830 patches. Notably, about 45% of the patches are smaller than 0.1 ha, while those exceeding 10 ha account for only 3.5%. Evaluation across nine representative test areas shows that the overall accuracy ranges from 93.51% to 97.63%. Furthermore, without the need for manual editing, the results of this study exhibit significant advantages over existing mangrove datasets in terms of spatial detail and coverage completeness.
WOS关键词CLIMATE-CHANGE ; FORESTS
资助项目National Natural Science Foundation of China[42271324] ; Key Laboratory of Land Satellite Remote Sensing Application, Ministry of Natural Resources of the People's Republic of China[KLSMNR-G202505]
WOS研究方向Physical Geography ; Geology ; Remote Sensing ; Imaging Science & Photographic Technology
语种英语
WOS记录号WOS:001710732800001
出版者ELSEVIER
源URL[http://ir.qdio.ac.cn/handle/337002/204985]  
专题海洋研究所_海洋环流与波动重点实验室
通讯作者Xue, Zhaohui
作者单位1.Hohai Univ, Jiangsu Prov Engn Res Ctr Watershed Geospatial Int, Nanjing 211100, Peoples R China
2.Hohai Univ, Key Lab Soil & Water Proc Watershed, Nanjing 211000, Peoples R China
3.Nanjing Univ Aeronaut & Astronaut, Coll Astronaut, Nanjing 210016, Peoples R China
4.Hohai Univ, Sch Earth Sci & Engn, Nanjing 211100, Peoples R China
5.Chinese Acad Sci, Key Lab Ocean Circulat & Waves, Inst Oceanol, Qingdao 266071, Peoples R China
推荐引用方式
GB/T 7714		 Nie, Xiangyu,Xue, Zhaohui,Li, Xiaofeng. Label-free mangrove mapping from temporally consistent PlanetScope imagery with interpretable deep unfolding network[J]. ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING,2026,235:19-37.
APA Nie, Xiangyu,Xue, Zhaohui,&Li, Xiaofeng.(2026).Label-free mangrove mapping from temporally consistent PlanetScope imagery with interpretable deep unfolding network.ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING,235,19-37.
MLA Nie, Xiangyu,et al."Label-free mangrove mapping from temporally consistent PlanetScope imagery with interpretable deep unfolding network".ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING 235(2026):19-37.

入库方式： OAI收割

来源：海洋研究所

浏览0
下载0
收藏0
其他版本

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。