辽西半干旱区玉米产量预报方法对比研究

Comparative Study of Maize Yield Forecasting Method for Semi-arid Region in the West of Liaoning Province

基于辽西半干旱区阜蒙县1986-2016年气象要素与玉米产量、发育期数据,使用产量丰歉相似年法、关键气象因子法、作物模型法开展了玉米产量预报研究,分析比较了3种产量预报方法的趋势预报正确率和定量预报准确率,结果表明,产量丰歉相似年法虽然可以实现玉米产量动态预报,但大概率法和加权分析法对玉米产量趋势的预报正确率均较低(33.3%~66.7%),其中加权平均法准确率高于大概率法的,但二者总体预报准确率都不高,且表现不稳定,不能满足业务服务需求。关键气象因子法提高了2014和2015年(生长季降水偏少年)的产量预报准确率,2年的产量预报趋势正确率为100%,产量预报准确率平均为89.5%。作物模型法随着发布预报日期的推迟,产量预报值更接近实际产量,降水正常年份预报效果好于降水偏少年份的预报效果。综上,将关键气象因子法与作物模型法相结合是准确预报半干旱区玉米产量的有效办法。两种方法相结合,对2011-2016年的产量预报准确率平均值达91.6%,能够满足业务服务需求。

Using meteorological data,maize yield and phenology data from 1986 to 2016 in Fumeng county located in the west of Liaoning Province,as well as the similar year method of yield’s bumper or poor harvest(SY),critical meteorological factors method(CF)and crop model method(CM),we analyze the maize yield tendency accuracy and quantitative forecast accuracy comparatively.The results show that the SY method can predict maize yield dynamically,but the accuracy of predicting maize yield trend by maximum probability method and weighted analysis method is lower in the range of 33.3%-66.7%.The forecast accuracy of the weighted analysis method is higher than that of the maximum probability method,but the overall accuracy of the two methods is not high and the performance is unstable,so they cannot meet the needs of forecasting operation.In contrast,the CF method increases the forecast accuracy of maize yield in 2014 and 2015,during which precipitation was less than in normal years.The tendency accuracy of maize yield forecast for 2014 and 2015 reaches 100% and the average forecast accuracy is 89.5%.In terms of the CM method,with the delay of prediction date,the forecast maize yield is closer to the actual yield,and the forecast accuracy of normal precipitation years is higher than that of drought years.In conclusion,combining CF and CM methods to predict maize yield in the west of Liaoning Province is an effective solution for the maize yield forecasting in semi-arid regions.With the combined method of the two,the average forecast accuracy of maize yield for 2011-2016 reached 91.6%,which can satisfy the needs of forecasting operation.