关键词: 模式预报, 褐飞虱, 回归分析

Abstract: Used the historical data over 15 years from five cities/counties, a group of forecasting models was established by means of stepwise regression. The models could forecast both emergent quantity and degree of the main endangering generation of brown planthopper (BPH) on late-season rice. Through eight-year forecasting and verifying, 73 models were chosen from 469 ones. These models, a complete group of forecasting models, could be used as long, middle and short term forecasting. The accurate rate of forecast was 93.98% for the whole, and 87.5%, 95.83 % and 100% for either. This forecasting method suited to more than 400 thousand ha paddy-field covered 23 cities/counties in Zhejiang. The middle and long term model could forecast 25 days and 55 days earlier than conventional forecasting method, respectively.

Key words: model forecast, Nilaparvata lugens, regression analysis