Recently the increasing damage caused by strong winds and synoptic scale weather as well as damage caused by localized weather phenomena have been reported. Thunderstorm gust of localized weather phenomena, according to the characteristics of local and short-term, is not easy to observe and respond to. This study aims to quantity the frequency and the spatial distribution of thunderstorm wind gust in order to mitigate the wind risk. This study utilized in-situ observation of wind record provided by KMA from January of 2002 to June of 2009. Once daily maximum gusts over 20m/s are obtained from the archives of KMA, any events occurred during the period of typhoon affecting Korea are first excluded. And the wind speed then are corrected for topographical effects by the method specified in Korea Building Code (2009) in order to exclude any high wind caused by topographical effects. The resulting data set consist of only gust wind data in the homogenous condition of flat region presumably only from meteorological phenomena. Next, using meteorological observation of thunderbolts, the present study classifies the wind gust events into thunderstorm ones and non-thunderstorm ones. This study also found that the spatial distribution of cumulative cloud quotient is closely correlated to that of occurrences of thunderstorm wind gust, which suggests the possible utilization of the cloud quotient as weighting factors in assessing wind gust risk. This study also analyze prognostic phenomena using the time series of 1 minute average wind speed of gust based on moving average with duration of 1~12 hour. The moving averaged wind speed of thunderstorm gust in general shows monotonic decrease before sudden change into increase at certain time in advance to the gust. The instances of change can be well represented GEV (Generalized Extreme Value) distribution. These result prepare the probabilistic approach to estimate the prognostic phenomena well before the actual event of gust.