干旱区生态环境知识资源中心

With the progress and development of technology, domestic and foreign scholars have already made a lot of researches on the evaluation and estimation of water quality indexes by different technical means. However, the method of estimating water quality by using two-dimensional fluorescence peak value is still rare. The paper took the surface water of Ebinur Lake basin as the research object, using multivariate statistical analysis and stepwise regression analysis method, and explored the interrelationship of two-dimensional fluorescence peak value and water quality indexes by establishing an estimation model. The results showed that: (1)The water quality was different for the different river of Ebinur Lake basin, the total nitrogen(TN) content of Bortala River was the highest, and the total phosphorus(TP) content of reservoir and canal were the highest. (2)The same water sample has three fluorescence peaks (Peak1, Peak2, Peak3), and the fluorescence peak intensity gradually weakened. The intensity of two-dimensional fluorescence peak value in different rivers was different, among them, the variation range of fluorescence intensity in Bortala River, reservoir and canal were the highest. (3)The correlation analysis between two-dimensional fluorescence peak value and water quality indexes showed that Peak1 and five days biochemical oxygen demand(BOD5), Peak2 and BOD5, Peak2 and chemical oxygen demand(COD), Peak2 and dissolved oxygen(DO) both had good correlation, and the correlation coefficient was 0. 479, 0. 371, 0. 655 and 0. 618 respectively. The estimation model was established by the single fluorescence peak value and water quality indexes, which could not meet the monitoring requirements, because the poor accuracy of the four models were verified. Therefore, based on the single fluorescence peak value model, we established a integrated model of the three peaks and water quality indexes of each water sample for obtaining multiple regression equation. We found that the integrated model of the three groups of fluorescence peak and DO had the best effect through the model validation, the Correlation Coefficient was 0. 756, and the Root Mean Square Error was 1. 001. Therefore, the synthetic estimation model of two-dimensional fluorescence peak value and water quality indexes is feasible, which has a certain reference for the monitoring of water quality in arid area, and provides scientific basis for the development, utilization and protection of water resources in the Ebinur Lake basin.