为了实时、准确地判别植烟土壤的干旱程度,利用可见—近红外光谱技术,以云南省寻甸县的红壤、水稻土和紫色土为研究对象,针对遮雨棚内种植烤烟品种K326的根际土壤干旱度开展光谱判别方法研究.通过在烤烟旺长期至现蕾期,对每种土壤类型设正常水分、中度干旱、重度干旱3个不同浇水量处理,采集不同土壤类型、不同土壤干旱程度的反射光谱,分别建立植烟土壤干旱度光谱距离判别分析模型、Bayes判别分析模型、Fisher判别分析模型和支持向量机(SVM)判别分析模型.结果表明,采用支持向量机(SVM)建立的植烟土壤干旱度光谱判别分析模型的效果要优于其他3种方法,其中红壤干旱度建模集判别正确率为94.34%,检验集判别正确率为88.46%;水稻土干旱度建模集判别正确率为90.00%,检验集判别正确率为91.67%;紫色土干旱度建模集判别正确率为100%,检验集判别正确率为92.00%.综上所述,利用可见—近红外光谱技术可以对植烟土壤干旱度进行有效判别.研究将为制订烟叶生产旺长期至现蕾期的灌溉指标提供科学依据,对烟叶生产具有非常重要的利用价值.
In order to get real-time and accurate determination of the drought degree of tobacco-growing soil,we studied red soil,paddy soil and purple soil in Xundian County,Yunnan province with the NIR technology and the spectrum determination of drought de gree of K326-growing soil inside greenhouse.Based on the period of vigorous growth and budding stage of flue-cured tobacco,three different soil types were treated with normal water,moderate drought and severe drought.The reflectance spectrum of soil with different soil types and different degrees of drought were collected and the drought degree of tobacco-growing soil analysis model of spectral distance discriminant,Bayes discriminant analysis model,Fisher discriminant analysis model and support vector machine (SVM)discriminant analysis model were established respec-tively.The results showed that the flue-cured tobacco soil drought degree spectrum discriminant analysis model established with the support vector machine (SVM)was better than the other three methods among which the correct rate of red soil drought degree modeling identification was 94.34%;the correct rate of test set identification was 88.46%;the correct rate of paddy soil drought degree modeling set identification was 90.00%,and the correct rate of the test set identification was 91.67%;the correct rate of purple soil drought degree modeling set identifi-cation was 100%,and the correct rate of the test set identification was 92.00%.Therefore,it is feasible to determine the drought degree of to-bacco-growing soil by visible/near infrared spectroscopy technology.This study will provide a scientific basis for the irrigation index from the period of vigorous growth to bud stage and the important practical value for tobacco production.