‘清香’核桃叶片叶绿素含量高光谱估测模型研究  

Hyperspectral Estimation Models for Chlorophyll Content of Juglans regia'Qingxiang'Leaves

作　　者：陈学兄[1] 杨天剑 张小军[2] 刘群龙[2] CHEN Xue-xiong;YANG Tian-jian;ZHANG Xiao-jun;LIU Qun-long(College of Resources and Environment,Shanxi Agricultural University,Taigu 030801,Shanxi,China;College of Horticulture,Shanxi Agricultural University,Taigu 030801,Shanxi,China)

机构地区：[1]山西农业大学资源环境学院,山西太谷030801 [2]山西农业大学园艺学院,山西太谷030801

出　　处：《西北林学院学报》2023年第5期125-133,共9页Journal of Northwest Forestry University

基　　金：中央引导地方科技发展专项资金项目(YDZX20181400004769);山西省高等学校科技创新项目(2020L0136);晋中市科技重点研发计划(农业)(Y212013);教育部产学合作协同育人项目(202102245010)。

摘　　要：为实现快速、无损、精确地估测核桃叶片叶绿素含量,以山西威特核桃示范园的‘清香’核桃为试验对象,基于田间人工定量施肥试验,利用皮尔逊(Pearson)相关分析法筛选与果实不同生育期叶片绿色度(SPAD值)呈极显著相关的光谱特征参量,并以筛选出的光谱特征参量为自变量,采用回归分析构建‘清香’核桃果实不同生育期叶片SPAD值的估测模型。结果表明,核桃果实不同生育期叶片原始光谱曲线的总体变化趋势基本一致;不同生育期叶片SPAD值与原始光谱和一阶微分光谱的相关性差异显著,其中果实速生期叶片SPAD值与原始光谱的所有波段呈负相关关系,在波段705 nm处相关系数高达0.4565;红边蓝边面积的归一化值[(S_(Dr)-S_(Db))/(S_(Dr)+S_(Db))]可较好地估测速生期和脂化期叶片的SPAD值,在果实速生期(S_(Dr)-S_(Db))/(S_(Dr)+S_(Db))与叶片SPAD值的关系以指数模型为最佳,而在果实脂化期对叶片SPAD值估测以二次多项式模型最优,2个时期检验模型的均方根误差分别为0.4768、0.5323,相对误差分别为0.8862和0.7766,均较小;MERIS陆地叶绿素指数(M_(TCI))可较好地估测果实硬核期和成熟期叶片的SPAD值,其中果实硬核期以线性模型的检验精度更高,均方根误差和相对误差分别为0.5040和0.8442,而果实成熟期以M_(TCI)的二次多项式模型估测叶片SPAD值最优,检验模型决定系数(R^(2))为0.8211,均方根误差和相对误差分别为0.9576、1.4819,说明以上模型能较好地估测核桃叶片叶绿素含量,且不同生育期的适用模型有所不同。To realize a rapid,non-destructive and accurate estimation of chlorophyll content in walnut leaves,a cultivar Juglans regia'Qingxiang'occurring in Weite Walnut Demonstration Orchard,Shanxi was selected as the object in this study.Based on the field test with artificial quantitative fertilization,Pearson correlation analysis was conducted to screen out the spectral characteristic parameters that were extremely and significantly correlated with the leaf green degree manifested by the values of SPAD(soil and plant analyzer development)during the fruit growth periods.The estimation model for SPAD value of the leaves in different growth periods of the fruit was established through regression analysis with the screened spectral characteristic parameters as independent variables.The results indicated that the overall change trend of original spectrum curve of the leaves in different growth periods remained basically the same.There were significant differences in the correlations of the SPAD value of the leaves in different growth periods with the original spectrum and first-order differential spectrum.Specifically,the SPAD value of the leaves in the rapid growth period of the fruit exhibited a negative correlation with all wavebands of the original spectrum,and the correlation coefficient was as high as 0.4565 at the 705 nm waveband.The normalized values of red and blue edge areas[(S_(Dr)-S_(Db))/(S_(Dr)+S_(Db))]could preferably estimate the SPAD value of the leaves in the rapid growth and fat-change periods of fruit.Besides,the exponential model was optimum for the relationship between(S_(Dr)-S_(Db))/(S_(Dr)+S_(Db))and the SPAD value of the leaves in the rapid growth period of the fruit,while the quadratic polynomial model was optimum for estimating the SPAD value of the leaves in the fat-change period of the fruit.The root mean square errors of the test models for the two periods were 0.4768 and 0.5323,and the relative errors were 0.8862 and 0.7766,respectively,both of which were at low levels.MERIS terrestrial chloro

关 键 词：核桃叶片 叶绿素含量 光谱参数 高光谱特征 估测模型 

分 类 号：S718.43[农业科学—林学]