Abstract: Downbursts are extremely hazardous to engineering structures. In order to study the wind field characteristics of downbursts, the time-varying mean wind and pulsating wind characteristics of downbursts are statistically analyzed based on 29 sets of downburst events recorded at the 356 m meteorological gradient tower in Shenzhen. The results show that the duration of downburst wind speed is the shortest in the ramp-up stage and the longest in the peak stage. The time-varying mean wind patterns of the downbursts can be classified into four categories. The nose shape of the wind profiles appears mainly at velocity ramp-up stage and peak stage. With the increase of height, the mean turbulence intensity of the downburst samples decreases slightly, but the decreasing amplitude is obviously smaller than that of the synoptic winds, and the standard deviation increases first and then decreases. Both the mean and standard deviation of the turbulence integral scale increase with height, and the mean value is also significantly lower than that of the synoptic winds. The differences in power spectra density between the downburst records are greatest in the ramp-up stage and least in the peak stage, and the wind spectra in the ramp-up stage is in good agreement with the von Karman and Solari spectra.