基于小波理论的干旱区内陆湖泊叶绿素a的TM影像遥感反演

发布时间：2018-10-12 07:44

【摘要】：叶绿素a(Chl-a)是衡量湖泊富营养化的重要指标,利用遥感技术动态监测面积较大的湖区水体中Chl-a浓度对了解湖区水质具有重要意义。以内蒙古乌梁素海为例,提出利用TM影像中的水体实测光谱进行小波去噪和光谱信号重构,并结合水质采样实测数据进行神经网络拟合,建立光谱反射率比值与Chl-a浓度的反演模型的方法。结果显示:小波理论和神经网络相结合的模型可以适用于估算乌梁素海Chl-a浓度,去噪后Chl-a浓度与光谱信号的相关系数(-0.575)较去噪前(-0.417)明显增强,去噪后的采样点光谱信号与Chl-a浓度之间表现出比原始信号更强的负相关性,证明了去噪后的观测值可进一步减弱随机误差的干扰和去除噪声,使观测数据更加逼近Chl-a浓度的真实情况,图像去噪重构结果显示重构后的光谱范围较之前有所缩窄,部分信号点得到了增强,但基本剖面结构并没有产生较大变化,反演模型的平均相对误差为0.142,与其他研究相比差别不大。反演得出的乌梁素海Chl-a浓度分布反映了污染源的分布,同时说明了乌梁素海Chl-a浓度在时空分布上呈现一定的差异,表现为丰水期呈现浅水区Chl-a浓度值高于湖心区,来水区高于其他湖区的分布趋势,枯水期乌梁素海中部呈现由西向东Chl-a浓度逐步降低的分布规律,西部呈均一化分布。反演模型基本可以满足实际预测的需要。但模型在具体应用中在影像数据采集、数据量及算法方面还有很大的改进空间,该方法的提出为干旱区大型内陆水体富营养化的实时定量遥感监测提供了新的解决方案。
[Abstract]:Chlorophyll a (Chl-a) is an important index to measure eutrophication of lakes. It is very important to use remote sensing technology to dynamically monitor the concentration of Chl-a in lake waters with large area. Taking Wuliangsuhai in Inner Mongolia as an example, the wavelet denoising and spectral signal reconstruction based on the measured spectrum of water body in TM image is proposed, and the neural network fitting is carried out by combining the measured data of water quality sampling. The inversion model of spectral reflectance ratio and Chl-a concentration is established. The results show that the model combined with wavelet theory and neural network can be used to estimate the concentration of Chl-a in Wuliangsuhai. The correlation coefficient between Chl-a concentration and spectral signal after denoising (- 0.575) is significantly higher than that before denoising (- 0.417). The negative correlation between the spectral signal and the Chl-a concentration of the sample point after denoising is stronger than that of the original signal, which proves that the observed value after denoising can further reduce the interference of random error and remove noise. The results of image denoising and reconstruction show that the spectral range of the reconstructed image is narrower than before, and some of the signal points are enhanced, but the structure of the basic profile has not changed greatly. The average relative error of the inversion model is 0.142, which is not different from other studies. The inverse Chl-a concentration distribution in the Wuliangsu Sea reflects the distribution of the pollution sources. It also shows that the Chl-a concentration in the Wuliangsu Sea is different in time and space distribution, which shows that the concentration of Chl-a in shallow water is higher than that in the center of the lake in the high water period. The distribution trend of the coming water area is higher than that of other lakes. In the dry season, the concentration of Chl-a in the middle of Wuliangsu Sea decreases gradually from west to east, and the distribution of Chl-a in the west is uniform. The inversion model can basically meet the needs of actual prediction. However, there is still much room for improvement in image data acquisition, data volume and algorithm in the practical application of the model. This method provides a new solution for real-time quantitative remote sensing monitoring of eutrophication of large inland water bodies in arid areas.
【作者单位】：巴彦淖尔市环境科学研究所;山西大学环境与资源学院;山西大学黄土高原研究所;
【基金】：国家水污染防治专项(乌梁素海综合整治项目)
【分类号】：X87

【参考文献】