Abstract: In nuclear safety analysis and environmental impact assessment, hourly weather data must be used. However, the calculation method and the statistic method of hourly wind direction from automatic weather stations have not been defined as normal rules in the corresponding regulations similar to “ground observational criterion” and “nuclear safety guide” in China, so the statistical method of the hourly wind direction was discussed in this paper. Four methods were often used to the calculation of average wind direction, including the methods of an arithmetic mean, a sliding average, a vector average and a highest frequency. The first two methods were modified since they produce the error when wind direction across 0° was calculated. The results indicate that the modified method for wind direction across 0° is simple and accurate. The modified arithmetic mean method is more reliable when the data 10-min ahead normal hours are used as the hourly data of wind direction. However, it is easy to produce error using this method when wind speed is zero. Thus, a vector average method is recommended when the frequency of light wind and calm is high. The hourly wind direction statistic methods from “RG1.23 guide” presented by U.S and from “nuclear safety guide” and “ground observational criterion” in China differ each other, so the results from the above methods are compared with the observational data by using the same calculation algorithm. Using the wind direction with a 10-min average and a 15-min average before normal hours or other period of time as the hourly wind direction, the consistency of correlation matrix is 97.87%; using average of each 15-min value in one hour or average of each 10-min value in one hour as the hourly wind direction, their consistency is 99.96%. The consistency between the two classification methods and the method of average of a 10-min and a 15-min in one hour is only 86.00 %. Using a 60-min average as the hourly wind direction, the consistency is the worst among all methods.

Key words: Average wind direction, Wind direction amendment, Hour data, Arithmetic mean, Slipping average, Vector average, Highest frequency