Most standalone atmospheric models do not perform well in simulations of coastal boundary layer jets (BLJs), important weather processes that can trigger heavy rain in coastal areas by supplying both moisture and dynamic lifting. We compared 33-year simulations with a coupled atmosphere–ocean model and its standalone atmospheric component, the REgional atmosphere MOdel (REMO), forced by the prescribed sea surface temperature (SST). We validated our results using the Tropical Rainfall Measuring Mission SST and the ERA5 hourly reanalysis data set. We found that the coupled model gave a more realistic SST standard deviation than the REMO on BLJ days and corrected the overestimated air temperature over land during the day. The coupled atmosphere–ocean model showed a lower land–sea thermal contrast in the boundary layer. This increased the effects of inertial oscillation, which caused the ageostrophic flows to veer southwest, which is the direction of the maximum wind speed on BLJ days. This reproduced a more reasonable land–sea thermal contrast in the boundary layer as a result of strong air–sea mixing in coastal weather processes, which led to a more robust inertial oscillation and a larger SST standard deviation over the central South China Sea. These findings deepen our understanding of the influence of a fully mixed air–sea boundary on coastal weather processes. These results show that operational numerical weather prediction models can be improved by applying atmosphere–ocean coupling to advance their ability to forecast the weather (e.g., BLJ events) in coastal areas.