Abstract: In order to improve the forecasting capability and accuracy of regional heavy dust pollution weather, the HYSPLIT backward trajectory model was used to predict a dust weather process in the Shihezi region. Based on the multi-source data, the cause analysis and tracing of the dust weather process were carried out. During the sand-dust weather, the changes in AQI, PM_2.5, and PM₁₀ concentrations were consistent, with coarse particles being dominant. Due to the lack of precipitation in autumn and winter in northern Xinjiang, there was no stable snow cover in the upper reaches of the weather area. The surface soil formed sand dust under wind operation, and it was transported along the northern slope of Ten-zan, forming severe sand and dust weather in Shihezi and other places.