Climate models contain atmospheric and oceanic components that are coupled together to simulate the thermodynamic and dynamic processes during air-sea interactions. Community Earth System Model (CESM version 1.2.1) is a state-of-the-art coupled model that is widely used and participates in Coupled Model Intercomparison Projects. Community Atmospheric Model (CAM), the atmospheric component of CESM, is based on the finite-volume dynamic core, which utilizes staggered Arakawa-D grids. However, the dynamics-physics (D-P) coupling in CAM causes the prognostic winds of the dynamic core be interpolated onto non-staggered locations, which affects the wind structure for computing the air-sea interaction and dynamical coupling. In this study we propose a new scheme that eliminates the extra interpolation during D-P coupling for the atmosphere-ocean interaction. We show that it improves the simulated climatology in key regions including eastern boundary upwelling regions and Southern Oceans. Compared with the default scheme, the new approach simulates strong surface wind near coast in eastern boundary upwelling regions. As a result, existing problems of the model, such as warm SST biases in these regions, are reduced. Meanwhile, for Southern Ocean, the prevailing westerlies are enhanced in new scheme, resulting in meridional sea ice transport. As a result, the overestimation of sea ice extent and negative bias in SST is reduced. This new scheme is generally applicable to coupled models with staggered dynamics-physics, such as spectral-element method based CAM.Plain Language Summary Coupled models simulate air-sea interaction by coupling components models together. However, in state-of-the-art models such as CESM, the atmospheric model contains dynamic core and physic parameterization that are not differentiated during air-sea interaction. We propose a new dynamical coupling scheme specific for finite-volume version of the atmospheric component in CESM. Model biases in key regions are reduced with the new scheme, including upwelling regions and sea ice modeling in the Southern Oceans. The scheme can be applied to other models to better exploit the dynamic core's capability by reducing the numerical diffusion during the dynamics-physics coupling for air-sea interactions.