Abstract: Suspended sediment particle size distribution characteristics not only represent the state of being of the suspended particle，but can also be used to imply the procedure and strength of the hydro-dynamics and re-suspension.Mie theory is used to build the suspended sediment average grain size retrieval model. Results show that suspended sediment backscattering coefficient and the cube of average grain size has ideal linear relationship，with the determination coefficients of the fitting equations all above 0. 93 at four bands(412 nm，443 nm，555 nm，667 nm).The maximal and minimal fitting errors are 16. 632 2%(412 nm)and 20. 314 3%(667 nm)，respectively.The suspended sediment backscattering coefficient in the study area was estimated using the quasi-analytical algorithm(QAA) from a MODIS remote sensing reflectance image. Surface water average grain size was estimated by combining the retrieval values of the suspended sediment concentration and the suspended sediment backscattering coefficient.The result shows that values estimated from the inversion model are in good agreement with experimental data from near shore areas with high suspended sediment content. This study can provide significant supporting data for deep research on land-ocean interaction marine ecological system evolution，and marine participates in the global carbon cycle.