巴西大豆产量预报方法研究

Prediction Model of Soybean Yield for Brazil

利用1961～2007年巴西大豆平均单产资料、西太平洋月平均海温、北半球500hPa平均高度场环流资料以及巴西大豆种植区域代表气象站的逐日气温资料,建立了基于海温、环流资料和基于地面气象要素的巴西大豆产量预报模型。利用3种模型分别对1996～2005年巴西大豆平均单产进行预报检验;综合3种模型的模拟效果,利用加权方法建立了巴西大豆产量预报集合模型。1996～2005年预报检验和2006～2007年预报试验的准确率基本都在90%以上,能够满足业务服务的需要。

Based on the data of per unit soybean yield from 1961 to 2007,the main factors of monthly average temperature of the West Pacific Ocean surface and the 500 hpa height circulation data of the Northern Hemisphere and the daily average temperature of the representative weather stations in the soybean-planted area of Brazil,three predicting models that predict soybean yield per unit before harvest for Brazil are built.Then the integrative predicting model is built based on the result of each model by a weighted method.The results of historical predicting examination from 1996 to 2005 and test from 2006 to 2007 indicate that the accuracy of the integrative model is over 90% and could meet the needs of operational service.