基于Landsat-8的城市湖泊水体总悬浮物吸收系数的遥感反演——以杭州西湖为例

近年来城市湖泊水质受到广泛关注.总悬浮物(total suspended matter,TSM)是水质和水环境评价的重要参数之一,其直接决定着水下光场分布,进而影响水体的初级生产力.本研究基于在杭州西湖的各湖区收集的遥感反射率(R_(rs),sr~(-1))数据和总悬浮物的吸收系数(a_p(440),m~(-1))进行高光谱建模,比较不同经验函数形式和不同模型输入的效果,得到高光谱下的最佳函数是指数函数,最佳模型输入是B1/B2或(B1-B2)/(B1+B2).其中,当模型输入为B1/B2和(B1-B2)/(B1+B2)时,模型R~20.6,波段B1和B2范围分别是580～690 nm和515～535 nm,及700～720 nm和515～615 nm.参照最佳波段范围和Landsat-8的波段设置,选定Landsat-8红色和绿色波段反演模型.并根据卫星数据验证结果,选定最优模型a_p(440)=51.17e~(-7.75x),x=(OLI3-OLI4)/(OLI3+OLI4),在3幅Landsat-8图像上应用模型,观察西湖TSM的变化.研究结果发现人类活动可能对西湖的TSM有很大的影响,并且我们的研究和结果有望为未来城市湖泊水质管理提供方法、数据和指导.

Remote sensing inversion of total suspended matter absorption coefficient in coastal urban lakes using Landsat-8: Case study of west lake in Hangzhou

In recent years, more and more attentions have been paid on water quality of urban lakes. The total suspended matter (TSM) is one of the important parameters for evaluating water quality and environment. TSM directly determines underwater light field distributions, and then affects primary productivity of water. In this study, we collected and measured hyperspectral remote sensing reflectance (R_rs, sr^-1) and TSM absorption coefficients (a_p(440), m^-1) of the West Lake in Hangzhou. We then simulated the relations between R_rs and a_p(440), using combinations of different bands and band ratios as well as different empirical functions. As the results of the hyperspectral data modeling, the best function is the exponential model and the best band combination is B1/B2 or (B1-B2)/(B1+B2). The accuracy of exponential model is acceptable (R² > 0.6), the wavelength ranges of B1 and B2 are respectively 580~690 nm and 515~535 nm, and 700~720 nm and 515~615 nm for (B1-B2)/(B1+B2). Therefore, the red and green bands of Landsat-8 were used to develop models based on the hyperspectral best band ranges and Landsat-8 band settings. According to the validation results of using satellite data, the best Landsat-8-based model is a_p(440)=51.17e^-7.75x, x=(OLI3-OLI4)/(OLI3+OLI4). Then the best TSM model was applied to three Landsat-8 images of West Lake, and some TSM seasonal distribution patterns and dynamics were observed from the images. The results show that human activities may strongly influence on the TSM of West Lake, and our study and results are expected to provide method, data and guidance for future water quality management of urban lakes.