悬浮颗粒物粒径是重要的海洋光学参数，在以悬浮颗粒物为主的黄渤海海域，对水体生物、化学过程等起着重要的作用，该参数的卫星遥感反演也对海洋光学与水色遥感研究具有重要意义。利用2014年5月和11月的黄渤海47组实测数据，建立了基于静止海洋水色卫星(GOCI)波段设置的遥感反射率(Rrs)与悬浮颗粒物中值粒径(D50)反演模型，555 nm 波段幂函数的反演效果最佳，决定系数R2为0.72, 绝对误差SMAPE 为6.35%，经实测数据检验，均方根误差SRMSE约为0.17，相对误差变化范围为-5%~5%。对反演模型引入5%误差进行敏感性检验，绝对误差、均方根误差分别控制在2%以及0.002以内，具有较好的稳定性。将此模型运用于2013年6月GOCI卫星数据，反演出悬浮颗粒物中值粒径D50的时空分布图，呈现从近岸向远海粒径逐渐变大的趋势。

Suspended particle size is an important marine optical parameter. In Yellow Sea and Bohai Sea, suspended particle size plays an important role in marine biogeochemical processes and ocean color remote sensing, since a large quantity of sediments dominate water composition in this study area. Using 47 in situ samples collected in May and November 2014, based on Geostationary Ocean Color Imager (GOCI) band, a novel inversion model is developed to estimate the median diameter (D50) of suspended particle sizes, by means of remote sensing reflectance (Rrs). The obtained results indicate that a power model by Rrs(555 nm) performs best, coefficient of determination R2 is 0.72, mean absolute percentage error SMAPE is 6.35%. As for model validation, the root mean square error (SRMSE) by using the independent dataset shows a value of about 0.17, and the relative errors ranged from -5% to 5%. The robustness of the developed models is investigated by introducing 5% random error into original data, and the corresponding results show that SMAPE is below 2%, and SRMSE is less than 0.002. Additionally, the model is successfully applied to derive distribution map of D50 from GOCI data (Jun, 2013), and the results show that the diameters of suspended particles increas from near-shore region to off-shore waters.