复杂内陆湖泊水体遥感反射率质量控制与光学分类
Quality Control and Optical Classification of Remote Sensing Reflectance of Complex Inland Lake Waters

本文以典型的草型湖泊乌梁素海为研究区，基于实测归一化遥感反射率数据集(N = 595组)构建了具有代表性的主要水类型遥感反射率的质量控制与光学分类方法。质量控制系统可以很好地识别出有问题的数据，提高数据的鲁棒性，对归一化遥感反射率光谱数据集进行k-means非监督分层聚类分析，定义了9种不同的光谱类型，发现各类间平均反射率光谱存在显著差异，大致分为四个湖泊特征：纯水体类型(1、2类，主要以水体透明度情况而定)、水下淹没类型(3~7类，与水生植被高度参数有关)、水面以上水草类型(8类)和挺水类型(9类)，最后分析了遥感反射率光学类型及特征。
This article takes the typical grassy lake Wulansuhai as the research area and constructs a representative quality control and optical classification method for remote sensing reflectance of major water types based on the measured normalized remote sensing reflectance dataset (N = 595 sets). The quality control system can effectively identify problematic data, improve the robustness of the data, and perform k-means unsupervised hierarchical clustering analysis on the normalized remote sensing reflectance spectral dataset. Nine different spectral categories were defined, and significant differences in average reflectance spectra were found among them, which can be roughly divided into four lake characteristics: pure water types (1 and 2 categories, mainly determined by the transparency of the water body), submerged underwater types (3~7 categories, related to the height parameters of aquatic vegetation), aquatic grass type above water surface (8 categories), and emergent types (9 categories). Finally, the optical types and characteristics of remote sensing reflectance were analyzed.