Due to limitations in maneuverability, Unmanned Surface Vehicle (USV) often struggles to achieve automatic unberthing through a single cohesive operation in constricted harbors. To address this problem, this paper proposed a novel Two-Layer Trajectory Planning Method (TL-TPM) and two Model Predictive Control (MPC) methods. Throughout the trajectory planning process, the TL-TPM considers the USV dynamic constraints along with the cumulative risk in viable areas. It incorporates the residual velocity and yaw angle deviation from the backward phase to derive the USV’s retreating trajectory and optimizes sailing time in the forward phase for a smooth forward trajectory, forming a complete automatic unberthing trajectory. The Efficiency-Priority MPC (EP-MPC) and Accuracy-Priority MPC (AP-MPC) are designed for trajectory tracking, considering tracking positional deviations, yaw angle deviations and surge force variations. The proposed EP-MPC demonstrates high computational efficiency and robustness, while AP-MPC shows superior control accuracy under various environmental disturbances. Extensive simulations validate the effectiveness of these methods in achieving automatic unberthing under wind and current impacts.