Abstract: In order to determine the diurnal variations of wind parameters in typical seasons in a high-altitude deep gorge, the field measurement was conducted on the bridge site, a long-span bridge across the Yarlung Zangbo River on the Qinghai-Tibet Plateau. Typical periods without strong weather systems were selected in winter and summer. The mean wind parameters were analyzed, such as wind speed, wind direction, wind attack angle, wind speed profile, and so on. The time-varying average wind speed was extracted using the db8 wavelet, and the fluctuating wind parameters were analyzed by the non-stationary wind speed model, such as turbulence intensity, integral scale, turbulent power spectrum, and the like. The results show that the wind speed and direction of the gorge wind field present regular fluctuations with a daily cycle. The wind speed is relatively small in the hours before dawn and morning, and the rising and falling times are inconsistent in different seasons. The wind speed in summer and winter began to increase gradually at 10 and 12 o'clock, respectively, and began to decrease rapidly after 20 and 21 o'clock. The distribution of wind directions is basically stable in summer and suddenly changes at 15 o'clock in winter. The wind attack angle is significantly related to the wind speed and direction, such that it first decreases and then increases in winter, and the opposite in summer. The maximum negative value of the wind attack angle in winter is taken at 15 o'clock. The wind speed profile changes with time, such that it tends to be stable in summer and change significantly in winter, and the exponential law model cannot accurately describe the vertical distribution of wind speed in the gorge wind field. The turbulent intensity changes with time, such that it is basically stable in the summer, and it first decreases and then increases in winter. The minimum value of the turbulent intensity in winter is taken at 12 o'clock. The ratio of the measured crosswind and vertical turbulence intensity to the downwind is higher than the standard recommended value. The turbulence integral scale in winter and summer has the same trend of change with time and is consistent with the trend of wind speed, and the measured integral scale is much smaller than the standard recommended value. The peak frequency of the normalized power spectrum first decreases and then increases with time. The power spectrum can be described in a unified form by considering the offset effect of its distribution in the frequency domain with the change of season and time.