| 摘要: |
| 【目的】利用高光谱数据对抽穗期冬小麦冠层叶绿素含量进行估测，旨在为叶绿素含量快速准确估测提供参考。【方法】利用ASD便携式野外光谱仪和SPAD-502叶绿素仪实测了冬小麦抽穗期冠层光谱反射率及叶绿素含量，并对原始光谱反射率及其一阶导数光谱与叶绿素相对含量进行了相关分析，建立了基于敏感波段、红边位置、原始光谱峰度和偏度、一阶导数光谱峰度和偏度的叶绿素估算模型，并进行检验，从中筛选出精度最高的模型。【结果】冬小麦冠层光谱曲线特征与叶绿素含量之间有着密切联系。基于原始光谱一阶导数偏度和峰度的冬小麦（抽穗期）叶绿素含量估算模型拟合精度优于其他4种估算模型，决定系数R2分别为0.847和0.572，均方根误差RMSE分别为0.397和0.697，相对误差RE分别为61.0%和119.0%，拟合精度优于其他4种估算模型。【结论】原始光谱一阶导数的偏度和峰度作为自变量能很好地估测抽穗期小麦冠层叶绿素含量。 |
| 关键词: 高光谱估测 叶绿素含量 冬小麦 抽穗期 估算模型 |
| DOI： |
| 分类号: |
| 基金项目:国家科技支撑计划项目(2012BAH29B04-00)；国家“863计划”项目(2013AA102401-2)；河南省科技攻关计划项目(132102110210) |
|
| Hyper-spectral estimation of chlorophyll content in canopy of winter wheat |
|
WANG Xiao-xing,CHANG Qing-rui,LIU Meng-yun,et al
|
| Abstract: |
| 【Objective】This study estimated chlorophyll content in winter wheat canopy at heading stage using hyper-spectral data to provide reference for quick and accurate estimation of chlorophyll content.【Method】The spectral reflectance of canopy and chlorophyll content of selected winter wheat at heading stage were measured by ASD handheld spectrometer and SPAD-502 in field,respectively.The correction between hyper-spectral reflectance,its first derivative and chlorophyll content was analyzed.Then models to estimate chlorophyll content based on sensitive wave,red edge position,kurtosis and skewness of original hyper-spectral curve and its first derivative were established and tested.【Result】The spectral characteristics of winter wheat canopy were closely related to chlorophyll content.The models based on the kurtosis and skewness of the first derivative of spectral reflectance were superior to other models with determination coefficients (R2) of 0.847 and 0.572,root mean square errors (RMSE) of 0.397 and 0.697 and relative errors (RE) of 61.0% and 119.0%,respectively.【Conclusion】Using skewness and kurtosis of first derivative of spectral data as independent variables could estimate chlorophyll content in winter wheat canopy at heading stage. |
| Key words: hyper-spectral estimation chlorophyll content winter wheat heading stage estimation model |
