利用三维荧光技术进行水质监测对干旱区绿洲河流水质的有效管理具有重要的意义。以三维荧光技术为手段，以艾比湖流域地表水为研究对象，结合平行因子(PARAFAC)法和自组织特征映射神经网络(SOM)方法，探讨了艾比湖流域地表水溶解性有机质的三维荧光特征及其与地表水水质指标之间的关系。通过PARAFAC法，有效提取了艾比湖流域地表水样中的4种荧光组分，C1荧光峰对应物质为紫外区类富里酸，C2荧光峰对应物质为类富里酸，C3包括2个峰C3(T1)和C3(T2)，其中C3(T1) 荧光峰对应物质为类蛋白，C3(T2)荧光峰对应物质为类腐殖酸，C4荧光峰对应物质为类腐殖质。经SOM训练，在不同聚类层中探讨水质参数分布情况，水质状况由差到好的顺序依次为博河上游、精河绿洲、乌苏周边农田、艾比湖周边。在艾比湖流域丰水期，酸碱度(pH)、电导率(EC)、溶解氧(DO)、化学需氧量(COD)和五日生化需氧量(BOD5)与水样的三维荧光峰具有较为显著的相关性，而总磷(TP)、总氮(TN)及氨氮(NH+3-N)与各荧光峰相关性较弱。分别建立pH、EC、DO、COD及BOD5与各荧光组分间的多元线性回归方程，求得相关系数R分别为0.579、0.632、0.502、0.762和0.785，可以在一定程度上利用各荧光组分模拟水质参数的变化情况。在利用PARAFAC探讨地表水荧光特征的基础上，SOM网络作为一种有效的水体荧光光谱分析工具，可为干旱区水质监测和河流水质污染治理提供科学依据。

Using three-dimensional fluorescence technique for water quality monitoring has great significance to manage the river water quality in oases of arid areas. The major objectives of this study are to apply parallel factor (PARAFAC) method and self-organizing map (SOM) to assessing fluorescence properties, and to characterize and quantify the relationship between fluorescence and water quality indexes in the Ebinur Lake watershed. Fluorescence components are identified by the PARAFAC in the Ebinur Lake watershed. C1 corresponds to ultraviolet humic-like fulvic acid and C2 corresponds to humic-like fulvic acid. C3 includes two peaks, C3(T1) and C3(T2), C3(T1) corresponds to protein-like substance and C3(T2) corresponds to humic-like acid. C4 corresponds to humic-like substance. SOM technique is employed for the exploratory analysis of fluorescence components of water samples in the Ebinur Lake watershed. Discussing the distribution of water quality parameters in different clustering layers shows that the water quality tends to be better as a sequence of Bortala River upstream, JingHe Oasis, farmland surrounding Wusu, and Ebinur Lake nearby. There exist certain relationships between water quality indicators and fluorescence peaks. In wet season, the pH, electricity conductivity (EC), dissolved oxygen (DO), chemical oxygen demand (COD) and biological oxygen demand in five days (BOD5) have significant correlations with fluorescence peaks, while the total phosphor (TP), total nitrogen (TN) and NH+3-N have weak correlations with fluorescence peaks. Multiple linear regression equations are established between the pH, EC, DO, COD, BOD5 and fluorescent components, and correlation coefficients R are 0.579, 0.632, 0.502, 0.762 and 0.785, respectively. In general, based on the investigation of the fluorescence characteristics using PARAFAC, the SOM network can be used as an effective tool for analyzing water fluorescence spectra, and it can help to provide a scientific basis for water quality monitoring and river pollution controlling in arid regions.