基于主成分分析和随机森林回归的冬小麦冠层叶绿素含量估算  

Estimation of Winter Wheat Canopy Chlorophyll Content Based on Principal Component Analysis and Random Forest Regression

作　　者：王琪 常庆瑞[1] 李铠 陈晓凯 缪慧玲 史博太 曾学亮 李振发 WANG Qi;CHANG Qingrui;LI Kai;CHEN Xiaokai;MIAO Huiling;SHI Botai;ZENG Xueliang;LI Zhenfa(College of Natural Resources and Environment,Northwest A&F University,Yangling,Shaanxi 712100,China;College of Land Resources and Environment,Jiangxi Agricultural University,Nanchang,Jiangxi 330045,China)

机构地区：[1]西北农林科技大学资源环境学院,陕西杨凌712100 [2]江西农业大学国土资源与环境学院,江西南昌330045

出　　处：《麦类作物学报》2024年第4期532-542,共11页Journal of Triticeae Crops

基　　金：国家863计划项目(2013AA102401-2)。

摘　　要：为提高冬小麦冠层光谱对叶绿素含量的估算精度,以陕西省乾县冬小麦为研究对象,利用SVC-1024i光谱仪和SPAD-502型叶绿素仪实测了冬小麦冠层反射率和叶绿素含量,分析了一阶导数光谱、10种特征参数和9种植被指数与叶绿素含量的相关性,并利用主成分分析(PCA)对叶绿素敏感的可见光波段(390~780 nm)一阶导数光谱进行降维,将特征值大于1的主分量结合特征参数和植被指数形成不同的输入变量,用偏最小二乘回归和随机森林回归构建冬小麦冠层叶绿素估算模型,并利用独立样本对模型进行验证。结果表明,小麦冠层叶绿素含量与一阶导数光谱在751 nm处的相关性最高(r=0.71),特征参数中红边蓝边归一化(SDr-SDb)/(SDr+SDb)与叶绿素含量的相关性最高(r=0.66),植被指数(VI)中修正归一化差异指数(mND705)相关性最高(r=0.74)。在输入变量相同的情况下,基于随机森林(RF)回归的预测模型均优于偏最小二乘回归(PLSR)模型,其中PCA-VI-RF模型的各精度指标均达到最优(r^(2)=0.94,RMSE=1.05,RPD=3.70),是冬小麦冠层叶绿素含量估算的最优模型。To further improve the accuracy of estimation of chlorophyll content by canopy spectra,winter wheat canopy reflectance and chlorophyll content were measured empirically using SVC-1024i spectrometer and SPAD-502 chlorophyll meter in Qian County,Shaanxi Province.The correlations between the first-order derivative spectra,10 characteristic parameters and 9 vegetation indices and chlorophyll content were analyzed;the chlorophyll-sensitive first-order derivative spectra in the visible band(390-780 nm)were downscaled using principal component analysis(PCA),and the principal components with eigenvalues greater than 1 were combined with characteristic parameters and vegetation indices to form different input variables using partial least squares regression(PLSR)and random forest(RF)regression to construct a winter wheat canopy chlorophyll content estimation model,and the model was validated using independent samples.The results showed that canopy chlorophyll content had the highest correlation with the first-order derivative spectrum at 751 nm(r=0.71);the highest correlation was achieved between the normalized value of red-edge and blue-edge(SDr-SDb)/(SDr+SDb)and canopy chlorophyll content in the characteristic parameters(r=0.66)and between the modified normalized difference index(mND705)in the vegetation index(r=0.74).The PCA-VI-RF model was the best model for canopy chlorophyll content estimation in winter wheat(r^(2)=0.94,RMSE=1.05,RPD=3.70),as the random forest(RF)regression based model outperformed the PLSR model with the same input independent variables.

关 键 词：冬小麦 冠层叶绿素 主成分分析 偏最小二乘法 随机森林回归 

分 类 号：S512.1[农业科学—作物学] S314