根据黄东海及珠江口附近海域17个航次的实测水下光谱剖面数据，建立了基于水体遥感反射比光谱数据与490 nm波段水体漫衰减系数Kd的遥感反演算法。经实测数据检验，经验算法反演结果的平均相对误差为28.27%，均方根误差为0.35，相关系数为0.91；半分析算法反演结果的平均相对误差为31.53%，均方根误差为0.48，相关系数为0.83。通过采用实测数据对新算法和现有算法的比较，发现新建立的算法能取得更好的反演效果，新建立的模型中经验算法能取得比半分析算法更高的精度。考虑到半分析算法建立过程独立于实测的水体漫衰减数据，因此可以认为半分析算法比经验算法有更大的适用性。

Optical data collected in the 17 cruise surveys in the Yellow Sea, the East China Sea and the Pearl River Estuary are used to establish new retrieval algorithms to evaluate diffuse attenuation coefficient Kd at 490 nm by using the relationship between remote sensing reflectance and Kd. Compared with in situ observation, the retrieval from the empirical algorithm shows that the correlation coefficient is 0.91, the root mean square error is 0.35 and the average relative error is 28.27%. And comparison between retrieval from the semi-analytical method and in situ observation shows that the correlation coefficient, root-mean-square error, average relative error of the results obtained are 0.83, 0.48, 31.53% respectively. Furthermore, both the empirical algorithm and the semi-analytical method can attain better performance than the existing algorithms by comparing retrieval with in situ observation. Although the empirical algorithm shows more precise than the semi-analytical method, the semi-analytical method is more independent to the in situ observation than the empirical algorithm during the process of algorithm establishment. Therefore, the semi-analytical method is more suitable in retrieving Kd in the Yellow Sea, the East China Sea and the Pearl River Estuary.