This paper addresses the flocking control problem for a group of unmanned surface vehicles (USVs) to follow a leader USV guided by a parameterized path in an environment with obstacles. The locations and shapes of the obstacles are unknown, and only the closest collision points and collision vectors can be locally measured. A safety-critical flocking control method is proposed for achieving collision avoidance and establishing a flocking behavior for USVs. Specifically, control inputs of virtual reference points are designed based on an artificial potential function. For the leader USV, a nominal path following guidance law is designed to follow a predefined parameterized path. For the follower USVs, a flocking guidance law is designed to follow the virtual reference points. An exponential control barrier function is designed based on the closest collision points. A constrained quadratic programming problem incorporating exponential control barrier function is established to calculate optimal yaw rate guidance signals. The effectiveness of the proposed control method for safety-critical flocking control of USVs is validated through simulation results.