传统的水域水质监测手段不仅成本高,而且空间信息有限,难以对相关水域进行全面的监测与评价.利用遥感技术进行水域水质监测可克服这些局限.本文以新疆艾比湖流域为研究对象,结合2015年5月实测水质数据和从准同步Landsat OLI数据上提取的水体指数值,利用空间分析和多元统计方法进行分析.结果发现,水体指数EWI、AWEIsh、Vegetation index(VI)、NDWI、NWI、NEW与水质指标之间的相关性显著(0.55≤r≤0.88).因此,选用以上6种水体指数与水质指标进行回归分析并建立数学关系估算模型,发现模型均方根误差均较低.利用同期实测数据对估算模型进行精度验证,发现验证判定系数高,验证点的相对平均误差偏低,均方根误差偏小.与此同时,利用2015年10月的40个采样点对模型进行二次验证,
The traditional method of monitoring water quality is not only labor intensive,but also costly,and it is difficult to conduct a comprehensive,accurate monitoring and evaluation over large areas. Remote sensing provides effective means to monitor water quality. Combined with the measured water quality data( COD,TDS,EC,TP,BOD5,SS,pH,Chroma and NTU) and Landsat OLI derived water index,this work presents spatial analysis and multivariate statistical analysis method taking Ebinur Lake watershed in Xinjiang,China as the study site. The results show that correlation between thewater quality parameters and the spectral indices( EWI,AWEIsh,NDWI,Vegetation index,NEW,and NWI) is significant,and therefore we chose the six spectral indices as water indices and the indicators of water quality for regression analysis. The model accuracy was verified by field measured data,and the accuracy of the model was obtained( 0. 55 ≤ r ≤0. 88) with low RMSE. Data collected on October 2015 were used for second test case using 40 samples,and it is found that the coefficient of determination was 0. 22 R^2 0. 81. In estimating model of TP,BOD5,SS,pH,Chroma and NTU data coefficient of determination( R^2) was consistently above 0.5,with low root mean square error. This paper concludes that the combined use of these indices provide an effective method to monitor water quality,including TP,BOD5,SS,pH,Chroma and NTU,in the Ebinur Lake watershed. The study can not only provide a basis for the identification of lakes in arid areas,but also provide a scientific basis for the application of remote sensing technology in the extraction of surface water quality.