Abstract: Sea fog is a kind of disaster weather with strong local characteristics. It has brought more serious losses as people's activities over the sea become more frequent. Therefore, the forecast of sea fog is becoming more and more important. There is no effective method to directly and widely obtain the real data of meteorological fields over the sea at present. However, the microwave scatterometer aboard a satellite is the most popular sensor that provides accurate global sea surface winds which are used widely. At the same time, the progress and popularity of variational assimilation method also make a great deal of unconventional data be used in numerical models, which optimize initial condition of models.To evaluate the impact of assimilating the sea surface wind data aboard QuikSCAT on the sea fog simulation, 3 sensitive experiments are carried out for the sea fog process during 3—5 April 2006. The effect of the different boundary layer parameterization schemes, such as Medium Range Forecast Model (MRF), Yonsei University (YSU) and Mellor-Yamada-Janjic (MYJ) on sea fog simulation are discussed. It is found that YSU boundary layer physical parameterization scheme is more suitable for simulating sea fog, the MRF scheme takes the second place and the MYJ scheme is the worst. Then by the three dimensional variational data assimilation (3DVAR) method based on the Weather Research and Forecast (WRF) model the sea surface wind aboard QuikSCAT is assimilated to investigate an advection fog and a radiation fog occurred over the sea around China, respectively.The results with and without assimilating the QuikSCAT sea surface wind data are used as the initial fields for WRF model to simulate the sea fog, respectively. The simulated results are compared with the satellite nephogram and the ground observation. Preliminary results show that the three dimensions variational assimilation of sea surface wind data by WRF-3DVAR system can make the sea surface wind data affect other element fields at low levels of the model, which have an obviously positive impact on the area forecast of sea fog. In particular, some detailed results are obviously better compared with those of the control experiment without assimilating the sea surface wind data. But for high levels the impact is limited. Since the sea fog data are sparse, results of only two cases are insufficient to generalization, so more cases will be discussed to validate the conclusions in the future.