A mesoscale coupled atmosphere–ocean model has been developed based on the GRAPES(Global and Regional Assimilation and Prediction System) regional typhoon model(GRAPES_TYM) and ECOM-si(estuary, coast and ocean model(semi-implicit)). Coupling between the typhoon and ocean models was conducted by exchanging wind stress, heat, moisture fluxes, and sea surface temperatures(SSTs) using the coupler OASIS3.0. Numerical prediction experiments were run with and without coupling for the case of Typhoon Muifa in the western North Pacific. To investigate the impact of using more accurate SST information on the simulation of the track and the intensity of Typhoon Muifa, experiments were also conducted using increased SST resolution in the initial condition field of the control test. The results indicate that increasing SST resolution in the initial condition field somewhat improved the intensity forecast, and use of the coupled model improved the intensity forecast significantly, with mean absolute errors in maximum wind speed within 48 and 72 h reduced by 32% and 20%, respectively. Use of the coupled model also resulted in less pronounced over-prediction of the intensity of Typhoon Muifa by the GRAPES_TYM. Moreover, the effects of using the coupled model on the intensity varied throughout the different stages of the development of Muifa owing to changes in the oceanic mixed layer depth. The coupled model had pronounced effects during the later stage of Muifa but had no obvious effects during the earlier stage. The SSTs predicted by the coupled model decreased by about 5–6°C at most after the typhoon passed, in agreement with satellite data. Furthermore, based on analysis on the sea surface heat flux, wet static energy of the boundary layer, atmospheric temperature, and precipitation forecasted by the coupled model and the control test, the simulation results of this coupled atmosphere–ocean model can be considered to reasonably reflect the primary mechanisms underlying the interactions between tropical cyclones and oceans. A mesoscale coupled atmosphere-ocean model has been developed based on the GRAPES (Global and Regional Assimilation and Prediction System) regional typhoon model (GRAPES_TYM) and ECOM-si (estuary, coast and ocean model (semi-implicit)). Coupling between the typhoon and ocean models was conducted by exchanging wind stress, heat, moisture fluxes, and sea surface temperatures (SSTs) using the coupler OASIS 3.0. Numerical prediction experiments were run with and without coupling for the case of Typhoon Muifa in the western North Pacific To investigate the impact of using more accurate SST information on the simulation of the track and the intensity of Typhoon Muifa, experiments were also conducted using increased SST resolution in the initial condition field of the control test. The results indicating that increasing SST resolution in the initial condition field somewhat improved the intensity forecast, and use of the coupled model improved the intensity forecast significantly, with mean absolu Use of the coupled model also resulted in less pronounced over-prediction of the intensity of Typhoon Muifa by the GRAPES_TYM. Moreover, the effects of using the coupled model on the intensity varied throughout the different stages of the development of Muifa due to changes in the oceanic mixed layer depth. The coupled model had pronounced effects during the later stage of Muifa but had no obvious effects during the earlier stage. The SSTs predicted by the coupled model decreased by about 5-6 ° C at most after the typhoon passed, in agreement with satellite data. Further, based on analysis on the sea surface heat flux, wet static energy of the boundary layer, atmospheric temperature, and precipitation forecasted by the coupled model and the control test, the simulation results of this coupled atmosphere-ocean model can be considered reasonably reflect the primary mechanisms underlying t he intereactions between tropical cyclones and oceans