Abstract: Correlation analysis between wheat powdery mildew occurrence data of 7 stations in Hebei province and meteorological data in corresponding period from 1983 to 2010 was made. It was confirmed that the average relative humidity during late March to middle April was the key factor for the epidemic disease. There was no significant difference between slight and serious areas in Hebei province by using of the T test for the difference of disease occurrence.It is not necessary to establish different prediction models for different regions. Discriminant classifying models for wheat powdery mildew were established by using of Fisher criterion. The model accuracy reached 82.5% based on evaluation for the historical data, and the forecasting result in the year of 2010 displayed that the accuracy was 100%. In general, the forecasting accuracy of powdery mildew was raised by establishing the models, and these models provide a solid foundation for wheat powdery mildew forecasting and early control on wheat.

Key words: Wheat powdery mildew, T test, Fisher criterion, Mid term prediction