The study focused on the annual dust storm time series of past 36 years in Siziwang County, China. Based on the MHF wavelet method, the temporal-frequency multi-time scale variations and jumping features of dust storm days was analyzed. Then, reveal the periods and turning points of dust storm series in different time-scale and forecast dust storm activities in the following years. The results show that the variations of dust storm activities in Siziwang County, China. It also found that approximate 17-year period-oscillation of dust storm activities variations is the strongest. In final the wavelet coefficients patterns exhibit a probable increasing trend in dust storm in the following years. It shows dust storm activities will increase in short periods.