论文部分内容阅读
叶绿素浓度是水体富营养化状态的重要指标,也是水色遥感反演的水质参数之一。水体中叶绿素浓度的遥感反演主要是建立实测光谱和实测水质参数二者之间的关系模型,利用遥感影像进行叶绿素浓度的信息提取。传统的叶绿素浓度遥感反演受区域性和季节性的影响,反演精度不高,而且反演模型不具普适性,需对叶绿素光谱特征进行分析,建立高精度的反演模型。本文采用Hydrolight数据模拟了不同叶绿素浓度(1~200μg·L-1)的水体在可见光近红外的反射波谱曲线,通过分析叶绿素的光谱特征选取了特征波段或波段组合,并建立了叶绿素浓度反演模型。研究表明,除反射峰波长模型外,反射峰面积模型、三波段模型、红光线高度模型等均能较好地反演叶绿素浓度。在不同叶绿素反演模型中,除红光线模型外,最优的是反射峰面积模型,其决定系数为0.9689,反演误差为25.25μg·L-1;其次是三波段模型,其决定系数为0.9637,反演误差为10.66μg·L-1。究其原因,三波段模型考虑了水体中非色素悬浮物、黄色物质及水体后向散射对叶绿素浓度反演造成的影响;反射峰面积模型除此之外还综合考虑了叶绿素散射效率的影响。
Chlorophyll concentration is an important indicator of eutrophication of water bodies and is also one of the water quality parameters of aquatic remote sensing inversion. Remote sensing inversion of chlorophyll concentration in water mainly establishes the relationship model between measured spectra and measured water quality parameters, and extracts the chlorophyll concentration information by using remote sensing images. The conventional retrieval of chlorophyll concentration by remote sensing is affected by regional and seasonal effects, the inversion accuracy is not high, and the inversion model is not universal. The chlorophyll spectral characteristics need to be analyzed to establish a high-precision inversion model. In this paper, Hydrolight data were used to simulate the reflectance spectra of water with different concentrations of chlorophyll (1 ~ 200μg · L-1) in the near-infrared region of visible light. The characteristic bands or band combinations were selected by analyzing the spectral characteristics of chlorophyll and the chlorophyll concentration was established model. The results show that except for the reflection peak wavelength model, the peak area model, the three-band model and the red light height model can all accurately reflect the chlorophyll concentration. In the different chlorophyll inversion models, except for the red light model, the optimal reflection peak area model has a coefficient of determination of 0.9689 and an inversion error of 25.25 μg · L -1, followed by a three-band model with a determination coefficient of 0.9637, the inversion error is 10.66μg · L-1. The reason for this is that the three-band model considers the effects of non-pigment suspended solids, yellow matter and water backscattering on the concentration of chlorophyll in the water. Reflectance peak area model also considers the effect of chlorophyll scattering efficiency.