南极冰盖表面融水遥感识别方法研究综述  

A review of the Antarctic surface meltwater identification via remote sensing technology

作　　者：牛立杭 唐学远 杨树瑚 张云[1] NIU Lihang;TANG Xueyuan;YANG Shuhu;ZHANG Yun(College of Information Technology,Shanghai Ocean University,Shanghai 201306,China;Key Laboratory for Polar Science,Ministry of Natural Resources,Polar Research Institute of China,Shanghai 200136,China;School of Oceanography,Shanghai Jiao Tong University,Shanghai 200030,China)

机构地区：[1]上海海洋大学信息学院,上海201306 [2]自然资源部极地科学重点实验室,中国极地研究中心(中国极地研究所),上海200136 [3]上海交通大学海洋学院,上海200030

出　　处：《极地研究》2025年第1期105-117,共13页Chinese Journal of Polar Research

基　　金：国家自然科学基金(42276257,41941006,41876230)资助。

摘　　要：表面融水深刻影响着南极冰盖物质平衡和全球海平面上升。受地理环境和气候条件影响,南极冰盖表面融水的年际变化差异较大,对环南极表面融水的动态监测方法与技术提出了诸多挑战。本文旨在对南极冰盖表面融水遥感识别方法的相关研究进行综述:(1)简介冰盖表面融水影响南极冰盖物质平衡的主要途径;(2)阐述用于南极表面融水提取任务的遥感卫星产品,包括多光谱图像和合成孔径雷达图像;(3)突出介绍目前实现南极表面融水提取的主要技术,包括阈值法、机器学习和深度学习等;(4)比较上述各种方法的优缺点,重点展望深度学习在南极冰盖表面融水的应用前景。Surface meltwater has profound impact on both the mass balance of Antarctica and the rise in global sea levels.Owing to the influence of geographical and climatological conditions,there is substantial interannual variability in surface meltwater in Antarctica,making dynamic monitoring of surface meltwater in Antarctica a challenging task.This paper reviews the methods used for remote sensing identification of surface meltwater on the Antarctic ice sheet.First,the main ways in which surface meltwater affects the mass balance of Antarctica are introduced.Second,the remote sensing satellite products used for extraction of surface meltwater in Antarctica are described,which include multispectral and synthetic aperture radar images.Third,the three main technical approaches currently used for extraction of surface meltwater in Antarctica are introduced:thresholding methods,machine learning,and deep learning.Finally,the advantages and disadvantages of the above methods are compared,and the prospects for application of deep learning to extraction of surface meltwater in Antarctica are discussed.

关 键 词：表面融水 遥感卫星 深度学习 机器学习 阈值法 

分 类 号：P23[天文地球—摄影测量与遥感]