水体辐射传输模型是水体光谱特性分析的理论基础, 水体固有光学量由水体组成决定, 与水体表面光场无关。 基于统计的半经验算法虽然能获取指定区域水质参数反演结果, 但是缺乏物理意义; 基于生物光学模型的分析算法, 针对城市河网内陆Ⅱ类水体光学特性复杂、 空间分布异质性强、 水体细小、 流动性大等特点, 利用高光谱数据, 研究基于固有光学量的城市河网水质参数反演模型, 对内陆城市浑浊水体光谱特性研究具有重大意义。 提出了适用于内陆城市河网水体的改进QAA算法, 以获取水体固有光学量, 改进包括后向散射估计模型调整和参考波段优化两个方面; 计算参考波段总吸收系数、 颗粒物后向散射系数等固有光学量, 得到浮游植物吸收系数和剔除纯水吸收系数; 对浮游植物吸收系数最优波段比值与叶绿素a浓度进行线性回归分析, 构建叶绿素a水质浓度反演模型, 对剔除纯水吸收系数最优波段比值与悬浮物浓度线性回归分析, 构建悬浮物水质浓度反演模型。 针对内陆河网Ⅱ类水体, 以典型的河网城市嘉兴市为研究区域, 获取了研究区域航空高光谱数据, 以及水质采样化验数据和水面以上光谱数据等地面准同步测量数据; 利用QAA算法和IIMIW算法对实测水面以上光谱进行固有光学量反演, 对比分析两个算法并结合城市河网水体特点, 提出改进QAA算法; 利用改进的QAA算法实现了研究区域水体的固有光学量反演, 基于反演的水体固有光学量建立了叶绿素a浓度和悬浮物浓度两项水色参数定量反演模型, 反演模型决定系数R2分别为0.64和0.71; 并用航空高光谱数据同步区域的4个地面样点实测数据, 对反演结果进行验证分析。 通过水质参数浓度反演值与实测值的对比, 叶绿素a和悬浮物水质浓度反演的平均相对误差分别为9.2%和9.4%, 反演得到的叶绿素a和悬浮物浓度分布图, 也与城市河网的特点和实际情况相符, 为城市河网水质监测提供方法和模型参考。

The water radiative transfer mechanism is the theoretical basis of water spectral characteristic analysis. The inherent optical parameters of water due to the composition of the water is independent of the surface light field of water. The semi-empirical algorithm based on statistics can be used to retrieve water quality parameters in some specified areas. However, that is lacking in physical meaning. It was of great significance for the study of the spectral characteristics of turbid water in inland cities to study the model to retrieve the water quality parameters of urban river network based on inherent optical parameters using hyperspectral data. The study was based on the analysis algorithm of bio-optical model. Taking the consideration of the characteristics of inland type II water of urban river network such as complexity of optical characteristics, strong heterogeneity of spatial distribution, small water and large fluidity. The paper came up with an improved QAA algorithm suitable for water of inland urban river network to obtain the inherent optical parameters of water. The improvements included the following two aspects: adjustment of the backscattering estimation model and optimization of the reference band. By calculating inherent optical parameters, such as the coefficient of total absorption of the reference band, the coefficient of backscattering particles, etc., coefficient of phytoplankton absorption was obtained and coefficient of pure water absorption was eliminated in this paper. A linear regression analysis was carried out on the optimal ratio of coefficient of phytoplankton absorption and concentration of chlorophyll-a to build a model to retrieve water quality of chlorophyll-a concentration. A linear regression analysis was also carried out to eliminate the optimal band ratio of coefficient of pure water absorption and concentration of suspended solids. And a model to retrieve suspended solids concentration for water quality was built. Aiming at type II water bodies of inland river network and taking the typical river network of Jiaxing City as the research area, regional aeronautical hyperspectral data, ground quasi-synchronous measurement data of water sampling data and spectral data above the water surface were collected; the QAA algorithm and IIMIW algorithm were used to retrieve the inherent optical parameters of the measured water spectral data. With the two algorithms compared and the characteristics of urban river network was taken into consideration, the improved QAA algorithm was put forward. The retrieval of the inherent optical parameters of the water in the study area was achieved by using improved QAA algorithm. Based on the inherent optical parameters of water obtained by the inversion, quantitative inversion model of chlorophyll-a concentration and suspended solids concentration were established. The determining coefficients R2 of the inversion model were 0.64 and 0.71, respectively. The retrieved results were validated and analyzed using the actual measured sample data of four ground samples acquired at the same time in the area by the aircrafts which obtained aviation hyperspectral data. Comparing the retrieved values of water quality parameters of concentration with the measured values, the average relative errors of retrieved values for chlorophyll-a and suspended solids were 9.2% and 9.4% respectively. The table of distribution of chlorophyll-a and suspended solids obtained from the retrieval was also consistent with the characteristics and actual conditions of the urban river network. That provided methods and model references for urban river network water quality monitoring.