文章摘要
基于高光谱的小麦旗叶净光合速率的遥感反演模型的比较研究
Comparison of Estimation Methods for Net Photosynthetic Rate of Wheat’s Flag Leaves Based on Hyperspectrum
投稿时间:2017-06-26  修订日期:2017-08-10
DOI:10.13254/j.jare.2017.0173
中文关键词: 小麦旗叶,高光谱,净光合速率,遥感反演
英文关键词: flag leaves of wheat, hyperspectrum, net photosynthetic rate, remote sensing inversion
基金项目:山东省重点研发计划“农田肥料高校利用数据智能采集处理关键技术”(2015GNC1101010);国家研发计划项目(SQ2017ZY060105);山东省“双一流”计划项目(SYL2017XTTD02)
作者单位E-mail
吕玮 山东农业大学资源与环境学院, 土肥资源高效利用国家工程实验室, 山东 泰安 271000  
李玉环 山东农业大学资源与环境学院, 土肥资源高效利用国家工程实验室, 山东 泰安 271000 yuhuan@sdau.edu.com 
毛伟兵 山东农业大学水利土木工程学院, 山东 泰安 271000 maoweibing316@126.com 
宫雪 山东农业大学资源与环境学院, 土肥资源高效利用国家工程实验室, 山东 泰安 271000  
陈士更 山东农大肥业科技有限公司, 山东 泰安 271000  
摘要点击次数: 1819
全文下载次数: 1024
中文摘要:
      植物净光合速率是植物生产的基础,是体现植物生长状况的重要生理指标。本文将小麦旗叶高光谱波段反射率进行一阶导数变换后与净光合速率(Pn)进行相关性分析确定敏感波段,分别采用二次多项式逐步回归(QPSR)、偏最小二乘法(PLSR)、BP神经网络法(BPNN)3种方法构建小麦旗叶的净光合速率反演模型,并对3种模型的预测精度进行比较分析。结果表明:(1)将小麦旗叶的原始光谱进行一阶导数变换后与Pn进行相关性分析确定的敏感谱区集中在750~925 nm之间,确定的6个敏感波段分别是:760、761、767、814、815、889 nm;(2)基于QPSR、PLSR、BPNN3种方法以及敏感波段的反射率一阶导数构建的Pn估测模型预测精度都较高,说明用这3种方法以及敏感波段对Pn的估测是可行的,其中模型估算能力顺序为QPSR > BPNN > PLSR,说明小麦旗叶Pn的最佳高光谱分析模型为小麦叶片750~925 nm反射率一阶导数变化后的QPSR模型。
英文摘要:
      Net photosynthetic rate of plants is the basis of plant production, and is an important physiological index to reflect the growth of plants. In this paper, hyperspectral reflectance of flag leaves of wheat was transformed with the first derivative and then correlated with net photosynthetic rate(Pn) to determine the sensitive bands, adopting three methods quadratic polynomial stepwise regression(QPSR), partial least squares regression(PLSR), back propagation neural network(BPNN) respectively to construct the inversion model of Pn for flag leaves of wheat, and to compare and analyze the prediction accuracy of the three models. The result showed that:(1) After the first derivative transformation of the original spectra of wheat leaves, and analysis with Pn in correlation the determined sensitive zone concentrate occured on 750~925 nm, and the six sensitive bands were determined as 760, 761, 767, 814, 815 nm and 889 nm.(2) Based on QPSR, PLSR and BPNN the Pn estimation model constructed was highly forecasting precision. This illustrates the three methods and sensitive band was feasible to estimate Pn. Among them the order of the ability to estimate the model was QPSR > BPNN > PLSR, which indicated the best hyperspectral model for flag leaf Pn of wheat was QPSR model whose first derivative changed after 750~925 nm reflectivity of wheat leaf.
HTML   查看全文   查看/发表评论  下载PDF阅读器
关闭