固体表面三维荧光是一种表征固态有机质组分与结构的先进技术, 不需要提取固体样品中的溶解性有机质, 直接对固体样品进行三维荧光光谱检测, 具有可操作性、 实用性、 信息量大的特点。 利用固体表面三维荧光光谱, 结合平行因子(PARAFAC)技术、 聚类分析(HCA)与分类回归树(CART)模型等分析手段, 提取乌梁素海表层底泥中的荧光组分, 识别影响有机质特征的关键因子, 揭示有机组分的空间分异规律。 根据污染程度的不同, 采集了10个表层底泥样品(1^#—10^#), 检测未处理样品和加热处理样品的固体表面三维荧光光谱, 二者之差得到湖泊底泥有机质的固体表面三维荧光光谱。 采用PARAFAC技术, 提取了4个主要的荧光组分(C1—C4)。 C1为类色氨酸物质, 主要来自于内源。 C2为类富里酸物质、 C3为可见区类胡敏酸物质、 C4为紫外区类胡敏酸物质, C2—C4主要来自于陆源。 四个荧光组分的总含量在北部最高, 南部次之, 中部最少。 C2—C4的总含量高于C1, 表明湖区底泥有机质主要来源为陆源。 C1在湖区南部的含量高于北部与中部, 表明C1可能与大量水生植物的生长代谢有关。 C2含量的空间分布为北部>南部>中部。 C3在北部地区的含量远高于南部和中部, 是北部地区底泥的代表性物质。 C4与C2的空间分布规律大致相同。 基于荧光组分HCA, 得出C2与C4可能具有相同的来源; C3与C1是区别底泥有机质特征的关键因子。 基于采样点HCA, 可将湖区底泥分为3个不同区域, 分别为北部重度污染区、 南部中度污染区与中部轻度污染区。 利用CART模型, 进一步验证了C3与C1是识别底泥有机质特征的关键因子, 使湖区底泥的分类结果更为精确, 同时为后续乌梁素海底泥污染特征与污染来源的探究提供了技术支撑。

Solid surface excitation-emission matrix (EEM) fluorescence spectroscopy, a leading technique, is used to characterize structural composition of solid organic matter. The EEM spectra were directly measured using solid samples, instead of extraction of dissolved organic matter. Hence the technique has feasibility, practicality and large amount of information. In this study, the solid surface EEM spectroscopy coupled with parallel factor analysis (PARAFAC), hierarchical cluster analysis (HCA), and classification and regression tree (CART) was applied to extract fluorescent components of the surface sediments in Wuliangsuhai Lake, to track potential factors of the organic matter, and to reveal spatial variations of the components. The ten samples were collected along a pollution gradient. The solid surface EEM fluorescence spectroscopy were measured using the untreated sample and the thermally treated samples, and the spectroscopy of the organic matter were obtained by the difference between the former and the latter. Four fluorescence components (C1 to C4) were extracted by the PARAFAC. The C1 was associated with the tryptophan-like material, which could be derived from in-situ source. The C2 was relative to the fulvic-like material, and the C3 and C4 were concerned with the Vis humic-like and UV humic-like materials respectively, which could be derived from ex-situ source. The total abundances of the four components were the highest in the northern region, followed by the southern and the central regions. The abundances of the C2 to C4 were higher than those of the C1, suggesting that the organic matter was mainly derived from the ex-situ source. The C1 in the southern region was relatively higher than that in northern and central, indicating that the C1 could be relative to the metabolism of many more aquatic plants. The decreasing order of the C2 was northern>southern>central. The C3 was higher in the northern region than those in the southern and central regions, so the C3 was the representative of the sediment in the north. The trend of the C4 was similar to the C2. Based on the fluorescence component HCA, the C2 and C4 might have the terrestrial source, while the C1 and C3 were the potential factors of the organic matter in the sediments. Based on the sample HCA, the sampling sites were grouped into three clusters, i. e. the high-pollution (the northern region), the medium-pollution (the southern region) and the low-pollution (the central region). With the CART model, the C3 and C1 were verified as the potential indicators, which provided clearer classification of the sediments, and offered technical support forsubsequent exploration of pollution characteristics and sources.