摘要
辽河流域水质监测点位有限,不能全面反映辽河流域水质状况,借助遥感技术反演辽河流域水质,为水环境精细化管理提供科学依据。利用高分一号WFV(GF1-WFV)影像,采用最大坡度下降算法选取纯水体像元,采用参数、非参数两类回归模型反演干流及部分支流总氮(Total Nitrogen,TN)浓度。结果表明:①GF1-WFV影像在双分辨率、成像幅宽和波段设置上满足反演需求,8-9月反演中,近红外波段反射率起关键作用;②非参数回归模型反演结果较好,R^(2)高于0.575,RMSE在0.54~1.899之间,MAE在7.27%~26.99%之间,极端梯度提升树模型反演结果优于逐步回归和随机森林模型;③2018年辽河流域干流及部分主要支流TN浓度4-6月明显高于8-9月,清河和浑河沈阳城区段水质不及太子河本溪段下游。
The monitoring points of water quality in Liaohe River Basin are limited,which cannot fully reflect the water quality status.The water quality is retrieved by remote sensing technology,which provides a scientific basis for the fine management of water environment.Based on GF1-WFV images,the maximum gradient algorithm is used to select pure water pixels,and parametric and non-parametric regression models are applied to retrieve TN concentration in the mainstream and some tributaries.The results show that:①GF1-WFV images meet the inversion requirements in terms of dual resolution,imaging width and band setting,and the reflectivity in near infrared bands plays a key role in inversion from August to September.②The inversion results of extreme gradient boosting model are better than stepwise regression and random forest model,R^(2) is higher than 0.575,RMSE is between 0.54~1.899,MAE is between 7.27%~26.99%.③In 2018,the TN concentration in the main stream and some main tributaries of Liaohe watershed is significantly higher from April to June than that in August to September,and the water quality in Qinghe River and Shenyang urban section of Hunhe River is inferior to that in Benxi section of Taizi River.
作者
王文辉
李艳
雷坤
张萌
魏明
WANG Wen-hui;LI Yan;LEI Kun;ZHANG Meng;WEI Ming(Chinese Research Academy of Environmental Sciences,Beijing 100012,China;Shenyang Ligong University,Shenyang 110159,China;Institute of Atmospheric Science,Fudan University,Shanghai 200438,China;Northwest Institute of Eco-environment and Resources,CAS,Lanzhou 730000,China)
出处
《中国农村水利水电》
北大核心
2022年第7期32-40,共9页
China Rural Water and Hydropower
基金
国家重点研发计划项目(2019YFC1407700)
水体污染控制与治理科技重大专项(2018ZX07601-002)。
关键词
辽河流域
遥感
回归
高分一号WFV
总氮
Liaohe Watershed
remote sensing
regression
GF1-WFV
total nitrogen
