Wind turbines vibrate under the excitation of environmental loads and rotor movement in normal operating conditions, and vibrations of the tower affect the service life of such system. In order to analyze the vibration and dynamic characteristics of the wind turbine during startup and shutdown process, vibration of a 1.5 MW turbine tower was monitored for a long term. First-order natural frequencies and damping ratios of the wind turbine were identified utilizing a data-driven stochastic subspace identification method. Monitoring data revealed that the wind turbine usually operated near the rotation speed of grid connection. In addition, obvious resonance at the turbine startup was noted as it surpassed the rotation speed of grid connection, but was not apparent at the turbine shutdown. Modal identification results revealed that the first-order natural frequencies varied slightly under different operating conditions, while the damping ratios were found to be smaller when the wind turbine started up, and its rotation speed surpassed that of grid connection, which were found to be greater at turbine shutdown. In addition, the characteristics of different vibration phenomena in the turbine during startup and shutdown processes were explained with the Sommerfeld effect. Our findings contribute to the field of vibration analysis and allow for more efficient optimization of dynamic parameters in wind turbines of similar design.