Based on the reflectance data of the fifteenth and sixteenth channels in near-infrared bands and spatial geometrical angles of the moderate resolution imaging spectro-radiometer, a physical iterative algorithm was proposed, using the Junge power-law size distribution to approximate the actual atmospheric aerosol model. The algorithm was used for simultaneous determination of the aerosol optical thickness and the exponent of the Jung power law. The results show that the aerosol optical thickness over most of the studied ocean is in the range of 0.02~0.17, the exponent of the Jung power law is in the range of 2.8~3.8, and both the aerosol optical thickness and exponent of the Jung power law have the trend of descending from coast to open sea. Therefore, the reasonable spatial distributions of the exponent of the Junge power law and aerosol optical thickness were obtained. Comparing the retrieval results with the moderate resolution imaging spectro-radiometer product and aerosol robotic network measurements, the retrieval algorithm whose results are more approach aerosd robotic network data than moderate resolution imaging spectro-radiometer product, is more accurate than moderate resolution imaging spectro-radiometer algorithm. Therefore, the retrieval method which is applied for the studied region is feasible and reliable.