Abstract: The automatic observation data in one minute interval are used to analyze 52 fog events that affected the takeoff and landing of aircraftsat Beijing Capital International Airport (BCIA) from 2000 to 2011. It was found that the visibility evolution of dense fog in the processesof its formation, development and dissipation is abrupt and oscillatory. These oscillations of dense fog can be classified to the leading oscillation,the middle oscillation and the posterior position oscillation. There were 86.5%, 62.5% and 79.2% cases with the leading oscillation, themiddle oscillation and the posterior position oscillation. The temperature oscillation in cold season fog is more distinct than that of warm seasonfog. In the 34 cold season fog events affecting BCIA, there are 61.8% cases including the temperature oscillation. In radiation fog events,the temperature oscillation is slightly ahead of the visibility oscillation. There is a temperature oscillation in a few hours just before the arrivalof each advection fog event in cold season, while the temperature oscillation is not distinct at the time of the advection fog occurrence. Basedon the statistical results, a quantitative index of the oscillation in dense fog is defined as DFOI (Dense Fog Oscillation Index). It was found that±0.2 can be used as the criterion of nowcasting prediction of dense fog in BCIA, with which an alarm of dense fog can be issued 1-2 hoursearlier for the events during 2000 to 2011. According to the result of actual tests in January 2012, it was found that using this index alonecould not distinguish fog from snow or haze, other indices must be introduced in the process of early fog warning.