The vibration behavior of cable-stayed bridges is time-dependent due to environmental and operational variability, which is one of the reasons why vibration-based condition assessment is challenging in engineering applications. In this study, the dynamic properties of a long-span cable-stayed bridge is analyzed using vibration data collected during a typhoon. The data is obtained through a structural health monitoring system, enabling a comprehensive assessment of the response to varying environmental conditions. The Fast Bayesian Fast Fourier Transform (FFT) method is utilized to identify the frequencies and damping ratios of the first ten modes, which include nine vertical modes and the first lateral mode. Additionally, the variability of dynamic properties under different environmental conditions is analyzed. In the Bayesian approach, dynamic property parameters are considered random variables, and modal identification is treated as an inference problem. Consequently, both the most probable values (MPVs) and the posterior uncertainties (c.o.v.s) of the modal parameters can be theoretically estimated, allowing for a thorough evaluation of modal parameter variations. It is indicated that the Fast Bayesian FFT method yields more accurate identification of modal frequencies compared to damping ratios relatively. Under the typhoon, the modal frequencies are more sensitive and decrease while the modal damping ratios are not changed significantly with the wind speed and vibration amplitude.