基于光学卫星遥感的海岸带水体盐度反演方法研究  

Research on retrieval method of coastal water salinity based on optical satellite remote sensing

作　　者：孙艳艳 曹洪涛 张甲波 刘宪华[3] 么嘉棋 崔铁军[1] SUN Yanyan;CAO Hongtao;ZHANG Jiabo;LIU Xianhua;YAO Jiaqi;CUI Tiejun(Academy of Eco-civilization Development for Jing-Jin-Ji Megalopolis,Tianjin Normal University,Tianjin 300387,China;Marine Ecological Restoration and Smart Ocean Engineering Research Center of Hebei Province,Qinhuangdao 066000,China;School of Environmental Science and Engineering,Tianjin University,Tianjin 300350,China)

机构地区：[1]天津师范大学京津冀生态文明发展研究院,天津300387 [2]河北省海洋岸线生态修复与智慧海洋监测工程研究中心,河北秦皇岛066000 [3]天津大学环境科学与工程学院,天津300350

出　　处：《海洋信息技术与应用》2024年第4期211-223,共13页JOURNAL OF MARINE INFORMATION TECHNOLOGY AND APPLICATION

基　　金：河北省海洋岸线生态修复与智慧海洋监测工程研究中心开放基金课题(53H23039);高分专项政府综合治理应用与规模化产业化示范项目(66-Y50G03-9001-22/23)。

摘　　要：海岸带区域地表水体的盐度信息可以反映海水侵袭、海岸线退化以及土地盐碱化等生态状况,对滨海区域的海洋和陆地生态系统的研究、保护和资源开发有重要意义,但其监测主要依赖于现场测验,技术手段较为单一。卫星遥感方法具有高效、大范围的优势,本文采用高空间分辨率的光学遥感卫星开展海岸带地表水体盐度的反演方法研究,在分析单波段反射率、波段比值及常用的光谱指数与水体盐度相关性基础上,确定了水体盐度的敏感波段,构建了经验模型、半经验模型和随机森林模型,并对盐度反演的精度和结果进行比较与分析。结果表明,随机森林模型反演结果的检验精度较高,决定系数R^(2)为0.81,RMSE为193.01μS/cm,且精度稳定;半经验辐射传输模型检验的决定系数R^(2)为0.53,RMSE为303.82μS/cm;经验统计模型检验的决定系数R^(2)为0.16,RMSE为407.46μS/cm。该研究为掌握海岸带地表水质空间特征、分析水质时间变化、评价水资源可利用性提供了重要的技术途径,对海岸带生态环境的调查监测、修复治理具有重要意义。The salinity information of surface water in the coastal zone can reflect the ecological status of seawater invasion,coastline degradation and land salinization,which is of great significance to the research,protection,and resource development of marine and terrestrial ecosystems in the coastal region.However,its monitoring mainly relies on field tests,and the technical means are relatively simple.Since satellite remote sensing has the advantage of high efficiency and wide range,this paper adopts optical remote sensing satellite with high spatial resolution to carry out research on inversion method of surface water salinity in coastal zones.Based on analysis of single-band reflectance,band ratio,and correlation between commonly used spectral index and water salinity,the sensitive band of water salinity is determined.The empirical model,semi-empirical model,and random forest model are constructed,and the accuracy and results of salinity inversion are compared and analyzed.The results show that the random forest model has high accuracy,with the determination coefficient R^(2)of 0.81 and the RMSE of 193.01μS/cm,and its accuracy is stable.The semi-empirical radiative transfer model has the determination coefficient R^(2)of 0.53 and the RMSE of 303.82μS/cm.The empirical statistical model has the determination coefficient R^(2)of 0.16 and the RMSE of 407.46μS/cm.This study provides an important technical approach to grasp the spatial characteristics of surface water quality in coastal zones,analyze the temporal changes of water quality,and evaluate the availability of water resources.It is of great significance to the investigation,monitoring,restoration,and management of coastal ecological environment.

关 键 词：海岸带遥感 水体盐度 经验统计模型 机器学习 SERT模型 

分 类 号：P714[天文地球—海洋科学] TP79[自动化与计算机技术—检测技术与自动化装置]