中文摘要：

利用地物高光谱遥感技术，在室内进行小球藻、聚球藻及其混合藻高光谱测量．得到这3组藻的反射光谱特征，同时进行叶绿素a（Chl．a）浓度测量．利用Matlab软件中的神经网络工具箱对得到的高光谱数据进行了曲线拟合．并用拟合结果和光谱测量实际结果分别建立了两种单一藻类的定量模型．单一小球藻Chl．a最优的定量模型为用反射率实际值建立的小球藻单波段反射率模型Chl．a=1×10^7（R687）。一37016R687＋53．64．单一聚球藻Chl．a最优的定量模型为利用反射率实际值建立的聚球藻两波段模型Chl．a=853．15×[R^-1（669）一R^-1（730）]×R（730）＋505．78．在对两种单一藻类定量模型研究的基础上分别用单波段反射率分离模型、三波段分离模型和两波段分离模型对由小球藻和聚球藻组成的混合藻进行了Chl．a浓度分离．其中单波段反射率分离模型和两波段分离模型得到了较好的分离结果，单波段反射率分离模型结果要优于两波段分离模型结果．利用神经网络模型拟合值构建的模型要优于直接用反射率测量值构建的模型，而三波段分离模型的分离结果不理想，不适用于本研究．

英文摘要：

Based on the hyperspectral remote sensing technique, the hyperspectral characteristics of Chlorella vulgaris, Synechococcus sp. and their mixed algae were obtained in the laboratmy. The concentrations of Chl. a for the three groups of algae were measured. Neural network technique was used to fit the curves of algae. The fitted spectral values and measured spectral values were used to establish the quantitative models of Chl. a. The optimal hyperspectral model of Chlorella vulgaris was Chl. a = 1 × 10^7 （R687） 2 -37016 R687 ＋ 53.64, which was created using measured values, and the optimal hyperspectral model of Synechococcus sp. was Chl. a = 853.15 × [ R ^- 1 （669） - R ^- 1 （730） ] × R （730） ＋ 505.78, which was built using measured values too. Based on the models of two individual algae, single band, two bands spectral ratio and three bands spectral ratio separation models were created to separate the mixed algae. The results showed that single band models were better than two bands spectral ratio models and the models using fitted spectral values were better than models using measured values, and the results of three bands spectral models are not good and thus not appropriate for this research.