To improve the numerical simulation of the severe snow storms occurred in the south of China and the middle/lower reaches of Changjiang River during January of 2008, the observations from the automatic weather stations (AWS) over the Qinghai-Xizang Plateau (QXP) and its surrounding areas were assimi-lated into the Weather Research and Forecasts (WRF) model using multi-cycle 3-dimensional varia-tional data assimilation (3DVAR). Due to the large-scale special topography of the QXP and its sur-rounding areas which may reach up to the mid-troposphere, the AWS located at different height on the deep slope of the plateau are different to those located on plains and take a role analogous in some extent to that of radio soundings in obtaining the vertical "profile" information of the atmosphere, and have the advantages in the aspects of sampling frequency, location/height fixing, and synchronization. The information captured by these AWS may carry the early-warning "strong signals" in the upstream sensitive area for the downstream weather systems to the east of the plateau and thus the assimilation of these AWS data is expected to lead to significant improvements on the simulation of the severe weather system occurred in its downstream areas through adjusting the 3-dimensional structures of the atmospheric thermal-dynamics for the initial conditions of the model. This study indicates that the assimilated information of moisture, temperature and pressure carried in the observations of AWS over the Qinghai-Xizang Plateau and its surrounding areas is very important and useful in the forecasting of precipitation in its downstream areas.
