基于无人机多光谱数据的水质参数反演与评价——以海口市永庄水库为例  

Inversion and evaluation of water quality parameters based on UAV multispectral data:a case study of Yongzhuang Reservoir in Haikou

作　　者：张乐 雷金睿 陈毅青 陈宗铸 周鹏[2] 何荣晓[2] 吴庭天 陈小花 李苑菱 潘小艳 ZHANG Le;LEI Jin-rui;CHEN Yi-qing;CHEN Zong-zhu;ZHOU Peng;HE Rong-xiao;WU Ting-tian;CHEN Xiao-hua;LI Yuan-ling;PAN Xiao-yan(Hainan Academy of Forestry(Hainan Academy of Mangrove),Haikou 571100,China;College of Forestry,Hainan University,Haikou 570228,China;Haikou Wetland Protection Engineering Technology Research and Development Center,Haikou 571100,China)

机构地区：[1]海南省林业科学研究院(海南省红树林研究院),海南海口571100 [2]海南大学林学院,海南海口570228 [3]海口市湿地保护工程技术研究开发中心,海南海口571100

出　　处：《中国环境科学》2023年第S1期258-267,共10页China Environmental Science

基　　金：国家自然科学基金资助项目(32260106);海南省财政科技计划项目(jscx202024)

摘　　要：以海口市水源保护地永庄水库为研究对象,利用无人机搭载多光谱传感器获取水体光谱反射数据,与对应的水质实测数据进行拟合建模,分别构建水库旱季和雨季的总磷(TP)、总氮(TN)、浊度(TUB)、叶绿素a(Chl-a)4种水质参数反演最优模型,并开展水质状况空间评价.结果表明:永庄水库旱、雨两季水体TP、TN、Chl-a反演模型均以多项式模型精度最高,决定系数R2均在0.5以上;而TUB反演模型在旱季以多项式函数模型精度最高,雨季以线性回归模型精度最高.永庄水库旱季和雨季4种水质参数在空间分布上有较大差异,总体上旱季水库西侧水质状况较好,东侧水质状况较差;雨季水库东侧水质状况较好,往西呈现出水质状况变差的趋势.建议在水库东侧城乡结合部加强截污管道建设,减少输入性污染源直排入库;水库西侧设置生态缓冲区域截流农业面源污染,降低水体富营养化风险.研究结果可为城市水源保护地的有效监测与保护提供科学依据.The subject of this paper was Yongzhuang Reservoir,a water source reserve in Haikou.Unmanned aerial vehicles(UAVs)equipped with multi-spectral sensors were utilized to obtain spectral reflectance datasets of water bodies,and fitting modeling was performed with corresponding measured data.Based on this,this paper established optimal inversion models for four water quality parameters,namely total phosphorus(TP),total nitrogen(TN),turbidity(TUB),and Chlorophyll a(Chl-a),for the reservoir's dry and rainy seasons,and conducted spatial water quality evaluation.The results indicated that:the polynomial models all presented the highest accuracy for the TP,TN,and Chl-a inversion models of the water bodies in the Yongzhuang Reservoir's dry and rainy seasons,with an R2 greater than 0.5.The accuracy of the TUB inversion model with the polynomial function model was the highest during the dry season,while the accuracy of the linear regression model was the highest during the rainy season.There were notable variations in the spatial distribution of the four water quality parameters in Yongzhuang Reservoir's dry and rainy seasons.In general,the water quality on the west side of the reservoir was good during the dry season,while that on the east side was poor.Additionally,the water quality on the east side of the reservoir was good during the rainy season,while the water quality gradually deteriorated towards the west.It was recommended that sewage intercepting pipelines be strengthened at the urban-rural junction on the reservoir's east side to reduce the direct discharge of imported pollution sources.On the west side of the reservoir,an ecological buffer zone was recommended to be established to intercept agricultural non-point source pollution and reduce the danger of water eutrophication.The findings of this paper could serve as a scientific foundation for the effective monitoring and protection of urban water source reserves.

关 键 词：无人机 多光谱 水质参数 反演 综合评价 永庄水库 

分 类 号：X524[环境科学与工程—环境工程]