有色溶解有机物( Chromophoric Dissolved Organic Matter，简称CDOM)，也称黄色物质（Yellow Substance），是一类广泛分布于自然水体中的溶解有机物,是一种重要的水质参数，与悬浮物、浮游植物同属于水色遥感的主要研究对象。遥感技术的应用促进了大尺度时间和空间范围的水环境变化研究的开展。相对于悬浮物和叶绿素a这两种水质参数而言，CDOM的遥感反演研究相对较少，并且研究区多是选择光学性质较简单的海洋水体。光学性质较复杂的近海和内陆水体CDOM的遥感反演研究相对较少。本文选取官厅水库和太湖两个典型内陆水体为研究区，以CDOM的吸收系数为浓度指标，开展CDOM遥感反演研究。 本文的主要创新点如下： （1）结合QAA_V4、QAA_V5两个QAA改进版本和QAA-CDOM方法，创新性的提出了QAA(V4)-CDOM，QAA(V5)-CDOM两个方法，用于CDOM遥感反演。 （2）依据光谱特征将太湖和官厅水库水面光谱数据分为两类，分别用QAA(V4)-CDOM和QAA(V5)-CDOM方法反演。在一定程度上解决了反演方法的季节性和区域性的限制。 本文的主要内容及结论如下： （1）本文用QAA-CDOM、QAA-E、董强提出的方法（本文中称其为QAA-FOC）以及本文提出的QAA(V4)-CDOM，QAA(V5)-CDOM方法，共五种半解析方法用于CDOM反演。确定了官厅水库用QAA(V4)-CDOM方法反演效果最好。对于太湖，QAA(V5)-CDOM方法反演结果精度最好，但仍有待提高。最终对太湖220个采样点光谱数据分为了V4、V5两类，分别用QAA(V4)-CDOM和QAA(V5)-CDOM方法进行反演。分类反演后，较分类前反演精度得到了提升。 （2）按照同样的方法，对官厅水库的60个采样点进行分类。综合官厅水库和太湖两个研究区，分析两类数据任两个波段比值与CDOM的相关性及高光谱数据的波段设置后，分别选定用Rrs(665)/Rrs(443）和Rrs(620)/Rrs(443）两个波段比值，建立了基于水面光谱数据的经验反演模型，反演结果的平均相对误差分别为12.1%和18.8%，均方根误差RMSE分别为0.157 m-1和0.192 m-1。 （3）分别按季节、研究区和V4、V5类，分别逐波段分析了遥感反射率和CDOM的相关性，并建立了相应的单波段CDOM反演经验模型。结果显示，单个研究区春季和冬季的模型表现最好。夏季和秋季由于受浮游植物的干扰，反演效果不理想。综合了不同研究区和不同季节的数据的反演模型效果不理想。由于水体区域性、季节性的差异影响，用单一波段建立模型反演CDOM是很困难的。 （4）将分类反演的半解析方法和分类建立的基于波段比值的经验反演模型，应用于官厅水库CHRIS和太湖MERIS高光谱数据的CDOM反演。官厅水库没有同步实测数据进行反演精度评定，仅对比了半解析方法和经验方法反演的结果，两种方法的反演结果分布规律统一。太湖有75个准同步点，用于精度检验，检验结果显示，反演结果精度尚可。太湖五次实验覆盖了春夏秋冬四个季节，其中春、冬两个季节反演结果较好，夏、秋两个季节反演结果精度仍有待提升。

Chromophoric Dissolved Organic Matter(CDOM),also called yellow substance,is a kind of dissolved organic matter widely distributed in natural waters.CDOM together with suspended solids and phytoplankton are all the main objects of ocean color remote sensing study. The application of remote sensing technology promoted the research of water optical properties changes in large scale of time and space range.Compared with suspended solids and phytoplankton,the studies of CDOM inversion is relatively small.And the optical properties of sea waters are more simpler,and often chosen as the research areas. Conversely,the studies in Coastal and inland waters are relatively small amount.This article select Guanting Reservoir and taihu lake,two typical inland waters as the research areas,take the absorption coefficient of CDOM as concentration index. In this paper, the main innovation points are as follows: (1) Combining QAA_V4, QAA_V5 ，two improved version of QAA and QAA-CDOM, this paper proposed two innovative methods, QAA(V4)-CDOM and QAA(V5)-CDOM, used in remote sensing inversion of CDOM. (2) This paper divided water surface spectral data of taihu lake and guanting reservoir into two kinds,based on the spectral characteristics.Then,for each kind,QAA(V4)-CDOM and QAA(V5)-CDOM were taken for CDOM inversion respectively.To a certain extent, it solved the of seasonal and regional restrictions of the inversion method. In this paper, the main contents and conclusions are as follows: (1) In this paper,QAA-E,Dong Qiang’s method(called QAA-FOC in this paper), QAA-CDOM,QAA(V4)-CDOM and QAA(V5)-CDOM,totally five methods used for CDOM inversion.The results showed that,QAA(V4)-CDOM is the best method for Guanting reservoir,and QAA(V5)-CDOM is the best method for Taihu lake.But the precision of inversion results of taihu lake still need to increase. So this paper divided water surface spectral data of taihu lake into two kinds, and take QAA(V4)-CDOM and QAA(V5)-CDOM for CDOM inversion respectively.The inversion accuracy was improved after classification. (2) The same with Taihu lake,60 points of Guanting reservoir was also divided into two kinds.Integrated Guanting reservoir and taihu lake, after analysising the correlation of each two band ratio with CDOM of each kinds respectively, and also taking hyperspectral data band set into account, two band ratios Rrs(665)/Rrs(443）和Rrs(620)/Rrs(443）were selected,and two empirical models were established. The average relative error (MRE) of the inversion results for each kind are 12.1% and 18.8% respectively and the root mean square error(RMSE) are 0.157 m-1and 0.192 m-1. (3) This paper analyzed the correlation between CDOM and remote sensing reflectance at each band respectively, according to seasons, study areas and V4 ,V5 classes, and establish experience CDOM inversion model based on single band.The results showed that the model of single study area of spring and winter are the best.Due to the interference of phytoplankton, the model for summer and autumn is not ideal. The modelsrelated to different research areas and different seasons data are also not ideal.Above all, it is difficult to invert CDOM based on a single band. (4) Classification semi-analytical inversion method and band ratio experience inversion model, are all used in hyperspectral data CDOM inversion.There are two type hyperspectral data:CHRIS and MERIS. There is no quasi-synchronous measured data to evaluate the accuracy of the inversion results for Guanting reservoir,so this paper just compares the semi-analytical method and the experience method inversion results.The result show that the inversion results of the two methods’result have unified distribution. Taihu lake has 75 quasi-synchronous measured points used for precision inspection.Five experiments cover four seasons.Spring and winter have better inversion results.Because of the influence of chlorophyll, summer and autumn’s inversion results accuracy remain to be promoted