Abstract: The error of meteorological forecast for complex terrain is mainly related to the model accurateness for describing terrain features and the physical dynamic processes. Integrating the mesoscale numerical weather forecasting system WRF and the CalMet model on micro-scale wind velocity simulation, it may account the dynamic processes on large scale, mesoscale and micro-scale, also may describe the local terrain effect upto the maximum extent. The results of the integrating model comparing to the mesoscale prediction show that the integrating forecasting method has a better capability of predicting the gale in complex terrain, and gives important reference and a significant guidance for forecasting, early warning and analying the windstorm in complex terrain.