Abstract: In order to assess wind energy resource during a long time (20-30 years), a new weather classification method was built. The classification method was based on surface and sounding observational data with the three classification factors as wind speed, wind direction and daily maximum mixing layer height, and 5% days of each type were stochastically selected as typical days for simulation. The following conclusions were drawn by comparison analysis of the wind energy parameters of typical days with those averaged over the 30 years in whole country: (1) The absolute error of the average wind speed was less than 0.1 m/s between selected typical days and 30 years-mean for all the classification stations, with the absolute error all less than 6.5%. The average value of the wind direction frequency absolute error at all the 16 wind directions was 0.28%-0.48%, while that of the wind speed frequency absolute error from 0 to 14 m/s wind speed was 0.09%-0.54%. (2) It was indicated by comparison of the wind direction frequency and the wind speed frequency for the other sounding weather stations in the classification station area: the wind direction frequency error was 0.27%-0.63% in the coastal area, and the wind speed frequency error was 0.14%-0.49%; while in the inland mountain area that it were less than 0.57% of the the wind direction frequency and 0.22%-0.60% of the wind speed frequency. The result shows that one sounding station was able to be selected as the classification station that could represent the wind character in the whole simulation area. And, (3) it was shown through comparison of the classification results from the sounding weather stations in the overlap area that the wind parameter error of weighted average results over the typical days of all the classification stations was much less that of the average results of the respective simulative area.