高光谱技术已广泛运用于水质检测领域。 探讨不同指标浓度下水质光谱变化规律及其光谱特征， 能够为水质指标遥感光谱精准识别与定量提取提供理论基础。 选取琅琊山景区不同水体景观共47个典型站位进行水质指标与光谱同步测量， 提取每个检测点的7个水质指标及350~950 nm波段， 探讨不同浓度水质指标光谱特征变化规律， 分析水质指标与光谱反射率、 反射率一阶微分、 任意两波段反射率比值及差值之间的关系。 结果表明： 各水质指标光谱曲线变化趋势一致， 但各有差异， 区分度最大的波段在可见光范围； 不同盐度、 溶解性总固体、 电导率含量的水质光谱曲线变化较为接近， 含量最高的样本光谱反射率最高， 且变化最显著； 浊度含量较高的水质样本光谱反射率变化较显著， 700~950 nm波段不同浊度含量的水质样本光谱反射率区分不明显； 溶解氧浓度为4~49 mg·L-1的水质光谱反射率在350~900 nm波段内明显低于其余样本； 在350~380 nm波段范围， 光谱反射率不随叶绿素含量变化而变化， 叶绿素含量接近0的样本在400~950 nm波段低于其余样本； 不同蓝绿藻藻蓝蛋白含量的样本光谱曲线相比其余水质指标在350~730 nm波段变化较大， 交叉点较多。 此外， 水质指标与原始光谱反射率相关性较低， 光谱一阶微分、 差值指数、 比值指数与各水质指标相关性整体有所提升。 该研究可为水质高光谱遥感检测提供一定的理论基础。

It is possible to provide theoretical basis for accurate identification and quantitative extraction of water quality indicators remote sensing spectroscopy by studying the variation of water quality spectrum and its spectral characteristics under different target concentrations with Hyperspectral technology, which has been widely used in water quality testing. A total of 47 typical stations in Langya Mountain Scenic Area were selected for water quality testing and spectral synchronization in this paper. Then, seven water quality indexes and 350~950 nm bands of each test point were extracted to explore the variation of spectral characteristics of different concentration water quality indexes, and to analyze the relationship between water quality index and spectral reflectance, first derivative reflectance, any two-band reflectance ratio and difference. The results showed that the spectral curves of the water quality indexes were consistent, but their changing rules were different.What’s more, bands with the highest degree of discrimination were in the visible range. The spectral curves of water quality with different salinity, total dissolved solids and conductivity content were close to each other, and reflectivity of water samples that change the most significantly was the highest. The spectral reflectance of the water samples with higher turbidity content was more obvious, but there was no difference of spectral reflectance of different turbidity content samples in the range of 700~950 nm. The spectral reflectance of water with a dissolved oxygen concentration of 4~49 mg·L-1 was significantly lower than that of the remaining samples in the range of 350~900 nm. In the range of 350~380 nm, the spectral reflectance did not change with the chlorophyll content, and the samples with chlorophyll content close to zero were significantly lower than those of the remaining samples in the 400~950 nm bands. The spectral curves of different BGA-PC concentrations water samples were more complex than other water quality indexes in the range of 350~730 nm. In addition, the correlation between the water quality index and its original spectral reflectivity was low, but it could be improved by the combination of band reflectivity, such as the first derivative reflectance, any two-band reflectance ratio and difference. This study aims at providing a theoretical support for water quality monitoring of hyperspectral remote sensing.