基于高光谱的夏玉米叶片SPAD值估测模型研究  

Study on the estimation model of summer maize leaf SPAD value based on hyperspectrum

作　　者：王俞茜 韩娜娜 周青云 张宝忠[2] 彭致功[2] Wang Yuxi;Han Nana;Zhou Qingyun;Zhang Baozhong;Peng Zhigong(College of Water Conservancy Engineering,Tianjin Agricultural University,Tianjin 300392,China;State Key Laboratory of Simulation and Regulation of Water Cycle in River Basin,China Institute of Water Resources and Hydropower Research,Beijing 100083,China;Sino-US International Joint Research Center for Ecological Agriculture and Water Environment Protection,Tianjin 300392,China)

机构地区：[1]天津农学院水利工程学院,天津300392 [2]中国水利水电科学研究院流域水循环模拟与调控国家重点实验室,北京100083 [3]中美生态农业与水环境保护国际联合研究中心,天津300392

出　　处：《天津农学院学报》2022年第3期1-8,共8页Journal of Tianjin Agricultural University

基　　金：中国水利水电科学研究院基本科研业务费专项(ID0154B082017);国家自然科学基金资助项目(51606170Z)。

摘　　要：快速、准确估测夏玉米叶片SPAD值,为大区域监控夏玉米长势提供数据支撑,通过PROSPECT模型模拟拔节期、抽雄期、灌浆期和乳熟期等不同生育阶段叶片光谱反射率,探讨基于高光谱技术估算SPAD值的建模方法。结果表明:(1)针对每个生育期,提出利用最优高光谱指数估测SPAD值的模型(简称最优模型),对应的最优高光谱指数分别为:REIP、mND705、mND705、G-M,模型的决定系数(R2)分别为:0.6259、0.6912、0.5919和0.5799,均方根误差(RMSE)分别为:14.4、12.3、6.3和9.4,相对误差(RE)分别为:29.1%、22.1%、9.3%和16.9%。(2)针对整个生育期提出利用通用高光谱指数mND705估测SPAD值的模型(简称通用模型),在这四个生育期模型的R2分别为:0.5973、0.6912、0.5919和0.4024,RMSE分别为:12.5、12.0、6.4和7.3,RE分别为:24.2%、21.5%、10.4%和11.8%。对比分析得到,通用模型R2高,RE低,因此,本文提出的通用模型具有较好的估测能力,对夏玉米叶片SPAD值的反演具有较高精度。Quick and accurate estimation of SPAD values of summer maize leaves can provide data support for monitoring the growth of summer maize in a large area.The spectral reflectance of summer maize leaves at different growth stages,such as jointing stage,tasseling stage,filling stage and milky stage,was simulated by PROSPECT model,the modeling method of estimating SPAD value based on hyperspectral technique.The results showed that:(1)the optimal hyperspectral index estimation model for SPAD value was proposed for each growth period,and the corresponding optimal hyperspectral indexes were REIP,mND705,mND705,and G-M.The decision of the model coefficients(R2)are:0.6259,0.6912,0.5919 and 0.5799,the root mean square error(RMSE)are:14.4,12.3,6.3 and 9.4,and the relative error(RE)are:29.1%,22.1%,9.3%and 16.9%.(2)A general hyperspectral index mND705 model was proposed to estimate SPAD value for the whole growth period.The R2 of the model were 0.5973,0.6912,0.5919 and 0.4024,the RMSE were 12.5,12.0,6.4 and 7.3,and the RE were 24.2%,21.5%,10.4%and 11.8%,respectively.Through comparative analysis,it is concluded that the general model R2 is high and RE is low.Therefore,the general model proposed in this paper has better estimation ability and high precision for the inversion of SPAD value of summer maize leaves.

关 键 词：SPAD PROSPECT模型 高光谱指数 通用模型 

分 类 号：S513[农业科学—作物学]