The experiment was set up for examining the physiological and biological indexes quickly and exactly, for obtaining information of tobacco-field fertilizing and tobacco growing. The ASD Field spec FR 2500 was used to ...The experiment was set up for examining the physiological and biological indexes quickly and exactly, for obtaining information of tobacco-field fertilizing and tobacco growing. The ASD Field spec FR 2500 was used to measure spectra reflectance of flue-cured tobacco and the relationship between hyperspectral parameters and biochemical contents （total nitrogen, chlorophyll, carotenoid）, and physiological indexes （fresh weight, dry weight, moisture content） of flue-cured tobacco leaves was studied by correlation and stepwise regression statistic methods at different nitrogen and potassium levels. The results indicated that the spectra curves of different treatments had obvious rules and great diversities. There were high correlations between different types of spectra parameters and ten physiological and biochemical indexes of flue-cured tobacco leaves. Hyperspectral characteristic variables of ten physiological and biochemical indexes were found through stepwise regression, and SDr/SDb was the characteristic variable closest to seven biochemical contents. Simultaneously, the R2 and regression coefficient of equations reached 0.05 significant level and the equations had good estimating effects through the examination of other samples. Accordingly, this study suggested that the ten physiological and biochemical indexes could be estimated quickly by the estimating models, at the same time nitrogen-potassium fertilization and growth condition of flue-cured tobacco could be inspected.展开更多
【目的】基于江汉平原麦后移栽棉花适宜氮素诊断指标建立追肥模型,为棉花氮素精准管理提供依据。【方法】于2019年采用田间小区试验,设置6个施氮量处理(0、90、180、270、360、450 kg N/hm^(2)),分析蕾期、花铃期倒4叶叶绿素含量(SPAD值...【目的】基于江汉平原麦后移栽棉花适宜氮素诊断指标建立追肥模型,为棉花氮素精准管理提供依据。【方法】于2019年采用田间小区试验,设置6个施氮量处理(0、90、180、270、360、450 kg N/hm^(2)),分析蕾期、花铃期倒4叶叶绿素含量(SPAD值)、叶柄硝酸盐含量(NIT)及24个高光谱参数与前期施氮量的关系,研究麦后移栽棉适宜氮素营养诊断关键指标,结合氮肥效应函数建立氮素追肥模型。【结果】随施氮量的增加,棉花倒4叶SPAD值、叶柄NIT含量显著增大,棉花冠层光谱绿光波段形成的反射峰变缓,但当氮肥增加到360 kg/hm^(2)时不再显著增加(蕾期NIT除外)。蕾期、花铃期倒4叶SPAD值、叶柄NIT含量、冠层高光谱参数(RSI和mND705)与前期施氮量均呈极显著一元二次方程关系(R2>0.8,RMSE<1),可作为江汉平原麦后移栽棉氮素营养诊断指标,以产量潜力的95%为临界值,蕾期麦后移栽棉SPAD值、叶柄NIT含量、冠层RSI和mND705临界值分别为34.802、2.307、1.526和0.549,花铃期相应为34.841、4.174、1.589和0.619。【结论】4个指标为江汉平原麦后移栽棉氮素营养诊断指标。麦后移栽棉最适宜施氮量为310.64 kg/hm^(2),产量潜力为4662.53 kg/hm^(2)。展开更多
为了进一步提高玉米叶绿素含量的高光谱估算精度,该文测定了西北地区玉米乳熟期叶片的光谱反射率及其对应的叶绿素相对含量(soil and plant analyzer development,SPAD)值,分析了一阶微分光谱、高光谱特征参数与SPAD的相关关系,构建了...为了进一步提高玉米叶绿素含量的高光谱估算精度,该文测定了西北地区玉米乳熟期叶片的光谱反射率及其对应的叶绿素相对含量(soil and plant analyzer development,SPAD)值,分析了一阶微分光谱、高光谱特征参数与SPAD的相关关系,构建了基于一阶微分光谱、高光谱特征参数和BP神经网络的SPAD估算模型,并对模型进行验证;再结合主成分回归(principal component regression,PCR)、偏最小二乘回归(partial least squares regression,PLSR)以及传统回归模型与BP神经网络模型进行比较。结果表明:SPAD值与一阶微分光谱在763nm处具有最大相关系数(R=0.901);以763 nm处的一阶微分值、蓝边内最大一阶微分为自变量建立的传统回归模型可用于玉米叶片SPAD估算;将构建传统回归模型时筛选到的光谱参数作为输入,实测SPAD值作为输出,构建BP神经网络模型,其建模与验模R2分别为0.887和0.896,RMSE为2.782,RE为4.59%,与其他回归模型相比,BP神经网络模型预测精度最高,研究表明BP神经网络对叶绿素具有较好的预测能力,是估算玉米叶片SPAD值的一种实时高效的方法。展开更多
文摘The experiment was set up for examining the physiological and biological indexes quickly and exactly, for obtaining information of tobacco-field fertilizing and tobacco growing. The ASD Field spec FR 2500 was used to measure spectra reflectance of flue-cured tobacco and the relationship between hyperspectral parameters and biochemical contents （total nitrogen, chlorophyll, carotenoid）, and physiological indexes （fresh weight, dry weight, moisture content） of flue-cured tobacco leaves was studied by correlation and stepwise regression statistic methods at different nitrogen and potassium levels. The results indicated that the spectra curves of different treatments had obvious rules and great diversities. There were high correlations between different types of spectra parameters and ten physiological and biochemical indexes of flue-cured tobacco leaves. Hyperspectral characteristic variables of ten physiological and biochemical indexes were found through stepwise regression, and SDr/SDb was the characteristic variable closest to seven biochemical contents. Simultaneously, the R2 and regression coefficient of equations reached 0.05 significant level and the equations had good estimating effects through the examination of other samples. Accordingly, this study suggested that the ten physiological and biochemical indexes could be estimated quickly by the estimating models, at the same time nitrogen-potassium fertilization and growth condition of flue-cured tobacco could be inspected.
