As a prominent maritime country, China's sustainable harnessing of marine resources is profoundly impacted by the fluctuating conditions of marine water quality. Traditionally, remote sensing has concentrated on monitoring optical water quality indicators, often neglecting the study of non-optically active ones. By correlating optically active and non-optically active parameters, we employ satellite data and machine learning to develop predictive models, achieving continuous spatiotemporal monitoring for four key marine water quality parameters in the Bohai and Yellow Seas. Results indicate that the R2 for DO, COD, IN, and AP are 0.71, 0.53, 0.72, and 0.58, respectively. The RMSE are 0.61 mg/L, 0.31 mg/L, 0.07 mg/L, and 0.002 mg/L, respectively. The model fitting results and stability are satisfactory, demonstrating the significance of this work for the protection of the marine environment in the Bohai and Yellow Seas.