摘要
叶绿素含量是影响作物生长及产量的主要因素。该研究以2017年6月小型试验田获取的抽穗期春小麦叶绿素含量及其对应的光谱反射率为数据源,对红边(627~780 nm)、黄边(566~589 nm)、蓝边(436~495 nm)、绿边(495~566 nm)、吸收谷和反射峰的最大反射率及反射率总和等16个高光谱特征参数与叶绿素含量之间的相关性进行了分析,并结合偏最小二乘回归法(partial least-squares regression,PLSR)对叶绿素含量进行高光谱建模及验证。结果表明:1)对特定的16个光谱特征参数而言,光谱特征参数绿边最大反射率与春小麦叶绿素质量分数之间的决定系数最低(R^2<0.5);决定系数较高(R^2≥0.5)的光谱特征参数包括蓝边最大反射率、蓝边反射率总和、黄边最大反射率、黄边反射率总和、红边最大反射率、红边反射率总和、绿边反射率总和、820~940 nm反射率总和及最大反射率、500~670 nm归一化吸收深度和560~760 nm归一化吸收深度,其中820~940 nm反射率总和决定系数达到最高(R^2为0.8);2)利用16个特征参量进行PLSR建模后,发现波段范围在820~940 nm的最大反射率及反射率总和所建立的PLSR估算模型为最优模型,其精度参数R^2p=0.8、RMSEp=2.0 mg/g、RPD=3.2。因此,该模型具有极好的预测能力。该研究为相关研究及当地精准农业提供科学支持和应用参考。
Chlorophyll content is one of the major factors that affect crop growth and crop output, and an important parameter to monitor the stresses and health status of vegetation. Currently the spectral feature parameter is one of the methods that have been widely applied to estimate the chlorophyll content of wheat. In order to provide scientific basis for wheat growth monitoring and agronomic decision-making, the spring wheat canopy chlorophyll content was estimated by using hyper-spectral technology （spectral feature parameters） in this paper. The correlation between hyper-spectral characteristic parameters and chlorophyll content of spring wheat （heading date） was analyzed, and the models for estimating chlorophyll content were established based on spectral feature parameters using partial least squares regression （PLSR） method. Data of chlorophyll content and spectral reflectance of spring wheat were obtained from the experimental plots at Ziniquanzi Town, Fukang City, Xinjiang Uighur Autonomous Region, China in June, 2017. The canopy spectral reflectance and chlorophyll content of spring wheat were measured in the experimental plots. After removing the marginal bands （350-400 and 2 401-2 500 nm） and being smoothed by Savitzky-Golay filter, 16 types of hyper-spectral characteristic parameters （such as red edge, blue edge, green edge, total reflectivity, absorption depth, and normalized absorption depth） were derived from the raw hyper-spectral reflectance data. Thereafter, PLSR was employed to build the hyper-spectral estimation models of chlorophyll content. Next, root mean square error （RMSEC and RMSEP） and determination coefficient （R2C and R2P） for calibration set and prediction set and relative prediction deviation （RPD） were used for accuracy assessment. The results showed that： 1） Among the selected spectral feature parameters, correlation coefficient between the maximum reflectivity of green edge and chlorophyll content of spring wheat is lower than 0.5. Spectral characteristic parameters that have the higher correlation coefficient （R2≥0.5） include maximum reflectivity of blue edge, total reflectivity of blue edge （436-495 nm）, maximum reflectivity of yellow edge, total reflectivity of yellow edge （566-589 nm）, maximum reflectivity of red edge, total reflectivity of red edge （627-780 nm）, total reflectivity of green edge （495-566 nm）, maximum reflectivity and total reflectivity within 820-940 nm, normalized absorption depth in 560-670 nm and 560-760 nm. The spectral feature parameters which have the highest correlation coefficient with the chlorophyll content are maximum reflectivity and total reflectivity within 820-940 nm, which reach 0.6 and 0.8, respectively. 2） On the 16 characteristic parameters of PLSR regression, the characteristic parameters （the maximum and sum of reflectance in 820-940 nm） have made a great contribution to the PLSR model, reduce the influence of other parameters on the accuracy of the model, have better performance in predicting chlorophyll content in the study area （R2p=0.8, RMSEp=2.3, RPD=3.0）, and provide scientific support and reference for other related local research and precision agriculture. To achieve more universal and stable inversion model, the next step is to enlarge the sampling area and the number of samples as much as possible to improve and perfect the spring wheat hyper-spectral database.
出处
《农业工程学报》
EI
CAS
CSCD
北大核心
2017年第22期208-216,共9页
Transactions of the Chinese Society of Agricultural Engineering
基金
国家自然科学基金(U170320066
41671348)
