中国稻米 ›› 2021, Vol. 27 ›› Issue (3): 21-29.DOI: 10.3969/j.issn.1006-8082.2021.03.005

• 专论与研究 • 上一篇    下一篇

基于冠层高光谱遥感的杂交水稻植被指数氮素营养诊断模型

王晓珂1(), 刘婷婷2, 许桂玲1, 冯跃华1,3,*(), 彭金凤1, 李杰1, 罗强鑫1, 韩志丽1, 卢苇1, PHONENASAY Somsana1   

  1. 1贵州大学农学院,贵阳 550025
    2黔西南州农业农村局,贵州 兴义 562400
    3 贵州大学/山地植物资源保护与种质创新教育部重点实验室,贵阳550025
  • 收稿日期:2021-02-22 出版日期:2021-05-20 发布日期:2021-05-20
  • 通讯作者: 冯跃华
  • 作者简介:

    第一作者:411282a22he.cdb@sina.cn

  • 基金资助:
    国家自然科学基金(31360311;31160263);公益性行业(农业)科研专项子项目(201503118-03);贵州省农业科技攻关项目(黔科合支撑[2019]2303号;黔科合支撑[2016]2563号;黔科合NY[2013]3005号;黔科合NY[2011]3085号);贵州省特色粮油作物栽培与生理生态研究科技创新人才团队(黔科合平台人才[2019]5613号);贵州省高层次创新型人才项目(黔科合平台人才[2018]5632);贵州省普通高等学校粮油作物遗传改良与生理生态特色重点实验室项目(黔教合KY字[2015]333);贵州省生物学一流学科建设项目(GNYL[2017]009)

Nitrogen Diagnosis Model of Vegetation Indices Based on Canopy Hyperspectral Remote Sensing for Hybrid Rice

Xiaoke WANG1(), Tingting LIU2, Guiling XU1, Yuehua FENG1,3,*(), Jinfeng PENG1, Jie LI1, Qiangxin LUO1, Zhili HAN1, Wei LU1, Somsana PHONENASAY1   

  1. 1College of Agronomy, Guizhou University, Guiyang 550025, China
    2 Qianxinanzhou Agriculture and Rural Affarirs Bureau, Xingyi, Guizhou 562400, China
    3Laboratory of Plant Resource Conservation and Germplasm Innovation in Mountainous Region of (Ministry of Education), Guiyang 550025, China
  • Received:2021-02-22 Online:2021-05-20 Published:2021-05-20
  • Contact: Yuehua FENG
  • About author:

    1st author: 411282a22he.cdb@sina.cn

摘要:

以杂交水稻为研究对象,进行两因素裂区试验,主区为品种,副区为施氮水平,分析了4个植被指数(VIs)分别与叶片氮素含量(LNC)、叶片氮素积累量(LNA)和地上部氮素积累量(APNA)之间的相关性,并建立了以VIs为自变量的氮素营养诊断模型。结果表明,4个VIs和LNC、LNA之间均存在决定系数大于0.7的波段区域且波段区域一致,4个VIs和APNA之间的决定系数均较低,仅在0.2左右;比值植被指数(RVI)和LNC之间的决定系数最大值为0.886,对应的波段组合为 694 nm和763 nm;垂直植被指数(PVI)和LNC之间的决定系数最大值为0.869,对应的波段组合为 864 nm和483 nm;差值植被指数(DVI)和LNC之间的决定系数最大值为0.883,对应的波段组合为1 292 nm和1 258 nm;归一化植被指数(NDVI)和LNC之间的决定系数最大值为0.881,对应的波段组合为1 296 nm和1 220 nm。最佳的氮素营养诊断模型为叶片氮素含量诊断模型,其模型表达式为LNC=1E+03NDVI2- 132.55NDVI+3.72,建模集R2、RMSE和RE分别为0.879、0.357%和16.267%,测试集R2、RMSE和RE分别为0.895、0.331%和15.136%。

关键词: 杂交水稻, 氮素营养, 高光谱遥感, 植被指数, 诊断模型

Abstract:

A split plot experiment of hybrid rice with two factors was carried out. There were two cultivars in the main plot and five nitrogen application levels in the sub plot. The correlation between four vegetation indices (VIs) and leaf nitrogen content(LNC), leaf nitrogen accumulation(LNA) and aerial part nitrogen accumulation (APNA) were analyzed respectively, and nitrogen diagnosis models with VIs as the independent variable were established. The results showed that there were band regions with coefficient of determination (r2) greater than 0.7 between the four VIs and LNC and LNA, respectively, and the band regions are consistent. The r2 between the four VIs and APNA were low, only about 0.2; the maximum r2 between ratio vegetation index (RVI) and LNC was 0.886, and the corresponding band combination is 694 nm and 763 nm. The maximum r2 between perpendicular vegetation index(PVI) and LNC is 0.869 and the corresponding band combination was 864 nm and 483 nm, the maximum r2 between difference vegetation index (DVI) and LNC was 0.883 and the corresponding band combination was 1 292 nm and 1 258 nm, the maximum r2 of normalized vegetation index (NDVI) and LNC was 0.881 and the corresponding band combination was 1 296 nm and 1 220 nm. The optimal nitrogen diagnosis model was LNC diagnosis model, and the model expression was LNC = 1E+03NDVI2-132.55NDVI+3.72; the correlation index(R2), RMSE and RE of training set were 0.879, 0.357% and 16.267%, respectively; the R2, RMSE and RE of test set were 0.895, 0.331% and 15.136%, respectively.

Key words: hybrid rice, nitrogen nutrition, hyperspectral remote sensing, vegetation indices, diagnosis model

中图分类号: