Abstract: To get high quality of ice grape, the adequate temperature during grape harvest season should be kept between -13～-8℃. So the first day of daily minimum temperature ＜-8℃ was taken as beginning harvest date for ice grape. For forecasting the beginning harvest date for ice grape, the subordinative fuzzy function sets and multiple regression model of daily minimum temperature decreasing to below -8℃ were established by using correlation analysis, subordinative fuzzy function and regression analysis methods, based on the daily minimum temperature and the related annual climate data from 1953 to 2010 in Huanren county, Liaoning province. The subordinative fuzzy function set was 88% through the examination of historical information. The historical correct rate was 86% through multiple regression prediction model. The results indicated that combined two methods could get good accuracy for forecasting the beginning harvesting date for ice grape, which could provide meteorological basic for ice grape harvesting and ice wine producing.

Key words: Ice grape, Harvest season, Forecasting, Subordinative fuzzy function, Multiple regression