欢迎访问《农学学报》,

农学学报 ›› 2016, Vol. 6 ›› Issue (4): 7-15.doi: 10.11923/j.issn.2095-4050.cjas15090013

所属专题: 小麦

• 农艺科学 生理生化 • 上一篇    下一篇

利用高光谱技术估测小麦叶片氮量和土壤供氮水平

谢福来,张士昌,郑军,田玮玮,芦冬涛,杨三维   

  1. 山西省农业科学院小麦研究所,山西省农科院小麦研究所,山西省农业科学院小麦研究所,山西省农业科学院小麦研究所,山西省农业科学院小麦研究所,山西省农业科学院小麦研究所
  • 收稿日期:2015-09-30 修回日期:2015-12-02 接受日期:2015-12-23 出版日期:2016-04-20 发布日期:2016-04-20
  • 通讯作者: 杨三维 E-mail:wonderfully_@163.com
  • 基金资助:
    山西省农业攻关项目“小麦种质中优异基因资源的开发与材料创制”(20150311001-5)。

Estimation of Nitrogen Content in Wheat Leaf and Nitrogen Supply Capacity of Soil by Hyperspectral Reflectance

谢福来,张士昌,,, and   

  • Received:2015-09-30 Revised:2015-12-02 Accepted:2015-12-23 Online:2016-04-20 Published:2016-04-20

摘要: 有效的监测作物氮素营养水平及土壤供氮能力可以为合理施用氮肥提供重要依据。本文以2 年3 点不同氮素水平下不同小麦品种的田间试验数据为基础,运用植被指数和偏最小二乘回归法,比较和分析小麦冠层光谱与叶片氮含量及土壤氮含量的关系。结果表明:小麦冠层光谱与叶片氮含量的相关性分析在可见光波段存在显著负相关,在近红外波段呈显著正相关,而与土壤氮含量的相关性呈相反趋势。基于光谱参数ND705 和GNDVI所建叶片氮含量估算模型的决定系数分别达到0.827 和0.826。基于光谱参数VOG2 所建土壤氮含量估算模型的决定系数达到0.646;与植被指数所建模型相比,综合350~1350 nm光谱波段反射率分别与小麦叶片氮含量、土壤氮含量建立偏最小二乘回归模型的预测精 度均有所提高,决定系数分别达到0.842 和0.654。本研究结果可为小麦氮素营养及土壤供氮水平的诊断监测与合理施肥管理提供了理论依据和技术支持。

关键词: 甘蓝型油菜, 甘蓝型油菜, DH系群体, 脂肪酸组分, 相关性, 主成分分析

Abstract: Effectively monitoring crop nitrogen nutritional level and soil nitrogen supply capability can provide an important basis for the rational application of nitrogen during fertilization. In this study, the method of vegetation index and partial least square regression (PLSR) were used to compare and analyze the relationship between the hyperspectral reflectance of wheat canopy and the concentration of nitrogen in leaf and soil based on data from three growing regions across two growing seasons. The correlation between canopy spectral reflectance and leaf nitrogen content was significantly negative in the visible bands. Moreover, a significantly positive correlation existed in near-infrared bands, while the correlation of canopy spectral reflectance and soil nitrogen content showed the right opposite results. The determination coefficients from the regression model of leaf nitrogen content based on spectral parameters of ND705 and GNDVI were 0.827 and 0.826. According to the parameter ND705 and GNDVI, the determination coefficient obtained by the regression model was 0.646. Compared with the vegetation index, the prediction accuracy of PLSR models for wheat leaf and soil nitrogen content on the spectral reflectance among 350-2500 nm was enhanced with the determination coefficients as 0.842 and 0.654. These results could provide certain theoretical basis and technical support for further study on diagnosing nitrogen nutrition of wheat and soil, and monitoring rational nitrogen fertilization.