清远市越北腹露蝗虫害发生程度的气象等级预测

Forecast Model for Occurrence Degree of Fruhstoferiola Tonkinensis Using Meteorological Data in Qingyuan

本文以多年来清远地区越北腹露蝗虫害发生的程度和常规气象因子为研究对象，选取秩相关系数较高且稳定性好的初选气象要素作为判别因子，然后运用逐步判别分析和逐步回归方法建立了越北腹露蝗虫害发生程度的气象等级预测模式。然后将两种模式进行组合预测。组合预测模式对2003-2006年的拟舍和2007-2008年的预报均取得了良好的效果。

In this paper, based on long-term field investigation data of Fruhstoferiola tonkinensis occurrence and the related meteorological data in Qingyuan, the critical factors controlling Fruhstoferiola tonkinensis occurrence were selected by rank correlation method. Forecasting model was established to predict the possibility of the development of locusts by regression method and Bayes＇ stepwise discrimination during the growth period. The paper applies regression method and Bayes＇ stepwise discrimination method to set up Combination forecasting model. The application of the combination forecast model was proved a viable and effective in long-term prediction of locust from 2003 to 2008.