基于气象因子影响的云南烤烟主要化学成分预测

Forecasting for Main Chemical Compositions of Flue-cured Tobacco in Yunnan Based on Effects of Meteorological Factors

以2004年云南省83个植烟县烟叶主要化学成分(总糖、还原糖、总氮、烟碱和蛋白质)与大田期(5～8月)月值气象数据,分析了烟叶各化学成分与气象因子的关系,并计算出各化学成分的气象因子影响综合指数,在此基础上建立了烟叶各化学成分的预测模型。根据2005年气象数据,预测了各植烟县烟叶主要化学成分含量,并与实测值进行了比较。结果表明,总糖、还原糖、总氮、烟碱和蛋白质预测的根均方差百分率(RRMSE)分别为9.30%、9.87%、10.55%、60.40%和12.35%。总糖、还原糖、总氮和蛋白质预测值相对误差的空间分布表明,全省大部分烟区该4种化学成分的相对误差小于10%。结果说明,模型对总糖、还原糖、总氮和蛋白质的预测效果较好,而对烟碱的预测效果差。但是当烟叶化学成分含量明显偏高或偏低时,模型的预测误差较大。

Relationships between chemical compositions of flue-cured tobacco leaf （total sugar, reducing sugar, total nitrogen, nicotine and protein） and mcteorelogical factors were analyzed, with the data of tobacco chemical compositions and monthly meteorological data in field period （from May to August） of tobacco in 83 tobacco-planting counties of ~unnan Province in 2004. And the meteorological factors impact comprehensive index of chemical compositions were calculated. On this foundation, forecasting model of tobacco chemical compositions was established. Main chemical compositions of tobacco leaf in the counties in 2005 were simulated with the corresponding meteorelogical factors, and the forecasting values were compared with observed values. The results showed that percent relative root mean square error （RRMES） of forecasted total sugar, reducing sugar, total nitrogen, nicotine and protein values were 9.30 %, 9.87 %, 10, 55 %, 60.40 % and 12.35 %, respectively. Spatial distribution of the relative errors （RE） of total sugar, reducing sugar, total nitrogen and protein showed that REs of the four chemical compositions in most of tobacco-planting areas of Yunnan Province were lower than 10 %. The results suggested that predictive effects of total sugar, reducing sugar, total nilrogen and protein were preferable while the predictive effect of nicotine was poor. But when the content of tobacco leaf chemical compositiota was far too high or low, the forecast error of the model was larger.