In order to quickly assess bridge scour during operation period, a dynamic-based identification method was proposed based on the measured vibration modes and model updating technique, and was applied to the pylon scour detection of the Hangzhou Bay Bridge. Two field measurements were conducted in 2013 and 2016 respectively to record the acceleration data of the superstructure of the Hangzhou Bay Bridge under ambient vibration. A modal analysis was carried out to obtain the natural frequencies and modes of vibration for each measurement. The superstructure was numerically simulated by a fish-bone finite element model and the pile-soil effect of the substructure was simulated by soil springs. The stiffness of equivalent springs around piles in the simulation model was firstly updated based on the measured natural frequencies of the scour-insensitive modes. Until the simulated natural frequencies corresponded to the measurement, the stiffness of the springs in the simulation could be seen as the identification of the real pile-soil effect of the bridge. Then, the scour depth in the simulation model was updated based on the variation of the measured natural frequencies of the scour-sensitive modes. Until the simulated and the measured variations of the natural frequencies were the same, the scour depth in the simulation model could be regarded as the identification of the real situation of the foundation during the three years. The identification accuracy was finally verified by the results of underwater terrain map around the foundation of the bridge. Results show that it is feasible to update the foundation scour by tracing the measured modal variation of the superstructure to identify a correct scour depth. This method can resolve the long-term difficulty of the traditional scour detection because it does not need underwater operation.