基于高光谱的核桃-大豆复合系统大豆叶绿素估算研究  

Estimation of soybean chlorophyll in walnut-soybean hybrid system based on hyperspectral

作　　者：施光耀 王若伦 桑玉强[2] 张劲松[1] 孟平[1] SHI Guangyao;WANG Ruolun;SANG Yuqiang;ZHANG Jinsong;MENG Ping(Research Institute of Forestry, Chinese Academy of Forestry/ Key Laboratory of Tree Breeding and Cultivation, State Forestry Administration, Beijing 100091, China;College of Forestry of Henan Agricultural University, Zhengzhou 450002, China)

机构地区：[1]中国林业科学研究院林业研究所/国家林业局林木培育重点实验室,北京100091 [2]河南农业大学林学院,河南郑州450002

出　　处：《河南农业大学学报》2020年第5期762-769,共8页Journal of Henan Agricultural University

基　　金：中央级公益性科研院所基本科研业务费专项(CAFZC2017M005);国家科技支撑计划(2015BAD07B050602)。

摘　　要：针对核桃—大豆农林复合系统,利用高分辨率光谱仪对林下大豆进行了不同物候期冠层光谱的探测,根据冠层叶绿素含量提取敏感波段以及红边位置,基于已往研究计算得到了特定反射率,一阶微分,红边参数,叶片叶绿素指数,比值植被指数,归一化植被指数等6类高光谱监测模型,通过精度检验,筛选出监测农林复合系统下大豆叶绿素含量的最优估算模型。结果表明,6类光谱参数均可用于叶片叶绿素含量的监测且精度均达到极显著水平,其中光谱参数R696′,R737′和叶片叶绿素指数与叶片色素含量之间的相关系数显著高于其他高光谱参数,因此选用以上3种参数分别进行模型拟合以期提高叶绿素含量估测的精度。通过对3种光谱参数建立回归模型,发现叶绿素A、叶绿素B和总叶绿素含量估算以光谱参数R737′表现最优,大豆叶片叶绿素含量的估算模型分别为:y=717.89x+0.634、y=0.517 ln(x)+3.735、y=877.18 x+0.849,其决定系数(R2)分别为0.742,0.575,0.717(P<0.01),说明以光谱参数R737′构建的光谱模型估算农林复合系统中大豆叶片叶绿素A、叶绿素B、总叶绿素含量具有更高的精度。通过对构建模型预测值与实测值的分析比较,发现叶绿素A、叶绿素B、总叶绿素含量的估测结果与实测值均不存在显著差异,决定系数分别为0.856,0.838,0.853,模型的精度高且均方根误差较小。High-resolution spectrometer was used to detect the canopy spectra of soybean in different phenological periods in walnut soybean agroforestry system.According to the content of chlorophyll in the canopy,the sensitive band and red edge position were extracted.Six kinds of hyperspectral data were calculated,such as sensitive bands,first-order differential,red edge parameter,leaf chlorophyll index,ratio vegetation index and normalized vegetation index.Through the accuracy test,the optimal estimation model for monitoring soybean chlorophyll content in agroforestry system was selected.The results showed that:six kinds of spectral parameters could all be used to monitor the chlorophyll content of leaves,and their estimation accuracy reached a very significant level.The correlation coefficients between the spectral parameters R696′,R737′and the leaf chlorophyll index with leaf pigment content were significantly higher than other hyperspectral parameters.Therefore,the three spectral parameters were selected for model fitting to improve the estimation of chlorophyll content.The regression model of three groups of spectral parameters showed that the spectral parameter R737′was the best in estimating chlorophyll A,chlorophyll B and the total chlorophyll content.The estimation models of soybean leaf chlorophyll content were y=717.89x+0.634,y=0.517ln(x)+3.735,and y=877.18 x+0.849,respectively.The coefficients of determination(R2)were 0.742,0.575 and 0.717(P<0.01),respectively,which indicated that the spectral model based on the spectral parameter R737′had higher accuracy in estimating the contents of chlorophyll A,chlorophyll B and the total chlorophyll content in soybean leaves in agroforestry system.Through the analysis and comparison of the predicted value and the measured value of the constructed model,it was found that there was no significant difference between the estimated and measured values of the three chlorophyll contents,and the determination coefficients were 0.856,0.838 and 0.853,respectively.The accur

关 键 词：叶绿素含量 大豆 农林复合系统 高光谱 估算模型 

分 类 号：S529[农业科学—作物学]