Abstract: In order to realize the frost fine monitoring and forecast by using the observed data of regional meteorological stations and high resolution numerical prediction product, based on meteorological observation data of daily minimum temperature, precipitation, relative humidity, wind, sunshine, total cloud cover, weather phenomenon of 97 ground meteorological stations in Hunan Province from the time the stations were established to April 2012, on the analysis of frost formation based on the meteorological conditions, the frost monitoring model was established. And the model was compared with frost observation results. It showed that in the days the minimum temperature less than or equal to 3℃ condition, the accuracy of frost simulation was improved by considering different meteorological elements critical threshold value; the error between the established frost monitoring model and frost observations was less than that with general monitoring indicators; the spatial distribution pattern and decadal changing trendency of the first frost date and ending date, frost-free period, frost days simulated were mainly consistent with observations, and the model can be used in frost monitoring and forecasting.