基于光谱特征的沉水植物种类识别研究

Study on species identification of submerged aquatic vegetation based on spectral characteristics

沉水植物种类的光谱特征识别可为大尺度遥感监测沉水植被提供种类识别的有效参数.本研究利用高光谱地物光谱仪获取上海市郊最大的天然淡水湖泊淀山湖6种典型沉水植物狐尾藻、竹叶眼子菜、金鱼藻、大茨藻、黑藻和苦草群落冠层的反射光谱,分析不同种类沉水植物的原始光谱和一阶导数光谱特征,通过主成分分析(Principal Components Analysis,PCA)筛选对沉水植物光谱种间差异敏感的植被指数和光谱指数,探寻能够有效识别不同种类水生植物的光谱识别方法.研究结果表明:①不同种类沉水植物的光谱曲线形状在可见光和近红外波段近似,但光谱反射率值差异较大;②光谱指数NAV、REP和NGP是识别6种沉水植物最敏感的特征指数,这些指数综合体现和放大了不同种类沉水植物在形态结构、叶绿素含量、光合和保护色素等方面的特征及所栖居的水体环境特征的种间差异.研究结果可为高光谱遥感影像准确解译与提取沉水植物群落组成、分布和生物多样性的动态变化信息提供理论依据和技术支持.

The among-species spectra characteristics of submerged aquatic vegetation （SAV） could provide effective parameters for species identification to timely monitor the distribution and growth status of SAV using remote sensing technology, in this study, the spectral reflectance the typical SAV plants were measured using a FieldSpecTM Pro JR Spectroradiometer in Dianshan I,ake, the largest natural freshwater lake in Shanghai suburbs, including 6 SAV species Najasmarina, Hydrilla verticillata , Myriophyllum spicatum , Ceratophyllum dernersum , Vallisne- ria natans and Potamogeton malaianus. The spectral characteristics of reflectance curves and first derivative curves for different species were analyzed, while vegetation indexes and spectral indexes were screened for identifying species of SAV using Principal Components Analysis （PCA）. The results show that the reflectancecurves were similar in shape but much variable in magnitude at the visible and near infrared wavebands for different SAV species. The spectral in- dex NAV, REP and NGP were the most sensitive characteristic index for identifying 6 SAV spe-- ties. The index NAV, REP and NGP could well reflect and amplify the differences in morpho- logical structure of plant, physiological and biochemical composition among different SAV spe- cies and the among-species discrimination of their habitat synthetically. The results from this study could be helpful to accurately interpret the community composition, distribution and biodi- versity dynamics of SAV on a large scale from hyperspectral remote sensing image.