A climatological and numerical study of "Karakkaze," a type of local wind in Japan, was conducted. First, winter days under a winter-type synoptic pressure pattern with daily minimum relative humidity of less than 40% were classified according to strong wind (wind speed � 9 ms −1 , Karakkaze day), medium wind (6 ms −1 � wind speed < 9 ms −1 ), and weak wind (wind speed < 6 ms −1 ). Secondly, the spatial patterns of the surface wind in each category are confirmed by the Japan Meteorological Agency (JMA)-Automated Meteorological Data Acquisition System (AMeDAS) observation data. In addition, we compared the boundary-layer wind of the three categories using wind speed data from the observation tower of the Meteorological Research Institute (MRI) in Tsukuba and from the JMA wind profiler in Kumagaya. Finally, we performed one-dimensional numerical experiments using a column Planetary Boundary Layer (PBL) model to evaluate the impact of solar radiation and upper-level wind on the formation of the Karakkaze.The results are summarized as follows. On the strong-wind days, strong northwesterly winds appear in the area along the Arakawa River and the Tonegawa River from Maebashi. The surface wind speed has a clear diurnal variation with a peak in the early afternoon. Such a diurnal variation is observed up to a 200 m level, but this diurnal pattern nearly reverses itself between 200 m and 400 m levels. On weak-wind days, the diurnal variation pattern is similar to that in other two categories, but the reversed pattern appears at a 100 m level, not at a 200 m level.Stronger surface winds appear under a clearer winter-type pressure pattern. The correlation coefficient is 0.632 between the daily maximum surface wind speed and the daily mean wind speed at a height of 2759 m, whereas the correlation is 0.284 between the surface wind speed and daily sunshine duration, which, in a previous study, was suggested to be the most significant factor.Numerical experiments indicate that a large amount of solar radiation is a necessary condition for a strong daytime wind, but this is not a sufficient explanation for the difference in the surface wind speed between weak and strong-wind days.