实测数据支持下的地表水浊度与总氮遥感反演——以长江流域中段为例

Remote Sensing Inversion of Turbidity and Total Nitrogen of Surface Water Supported by Measured Data—A Case of the Middle Section of the Yangtze River Basin

针对以实地采集数据进行地表水水质反演存在的空间范围小、耗时长、难度大、所构建模型在地域及季节方面很难具有良好普适性等问题,以长江流域中段为研究区,利用Landsat 8 OLI遥感影像数据及国家地表水水质自动监测站实测数据,建立浊度、总氮与反射率的反演模型,采用决定系数(R^(2))、平均绝对值误差(MAE)、均方根误差(RMSE)及平均绝对百分比误差(MAPE)进行精度检验,并制作浊度、总氮反演分布图,分析浊度和总氮空间分布特征。结果表明:浊度反演模型的反演精度较高,R^(2)为0.945,MAE为4.945 NTU,RMSE为7.117 NTU,MAPE为29.4%;总氮反演模型的反演精度一般,R^(2)为0.502,MAE为0.427 mg/L,RMSE为0.541 mg/L,MAPE为31.0%。这说明使用国家地表水水质自动监测站监测数据进行浊度反演具有一定的可行性。

In view of the problems of small spatial range,long time and high difficulty in retrieving surface water quality from field data,it is difficult to establish a model with good universality in terms of geography and seasons.Taking the middle section of the Yangtze River Basin as the research area,the inversion models of turbidity,total nitrogen and reflectance were established by using the remote sensing image data of Landsat 8 OLI and the measured data of the national surface water quality automatic monitoring station.The coefficient of determination(R^(2)),mean absolute error(MAE),root mean square error(RMSE)and mean absolute percentage error(MAPE)were used to test the accuracy.Turbidity and total nitrogen inversion distribution maps were made to analyze the spatial distribution characteristics of turbidity and total nitrogen.The results showed that the accuracy of turbidity inversion was relatively high,with R^(2)of 0.945,MAE of 4.945 NTU,RMSE of 7.117 NTU and MAPE of 29.4%;the inversion accuracy R^(2)of total nitrogen was 0.502,MAE was 0.427 mg/L,RMSE was 0.541 mg/L,and MAPE was 31.0%,indicating that it is feasible to use the data of the national automatic monitoring station for turbidity inversion.