四川农业大学学报 ›› 2021, Vol. 39 ›› Issue (2): 149-156.doi: 10.16036/j.issn.1000-2650.2021.02.003

• • 上一篇    下一篇

基于小波变换的带状套作玉米叶面积指数光谱估测

张佳伟, 王仲林, 谭先明, 王贝贝, 杨文钰, 杨峰*   

  1. 四川农业大学农学院/农业部西南作物生理生态与耕作重点实验室/四川省作物带状复合种植工程技术研究中心,成都 611130
  • 收稿日期:2020-10-20 出版日期:2021-04-28 发布日期:2021-05-07
  • 通讯作者: * 杨峰,教授,主要从事作物生理生态研究,E-mail:f.yang@sicau.edu.cn。
  • 作者简介:张佳伟,硕士研究生。

Strip Intercropping Corn Based on Wavelet Transform Leaf Area Index Spectrum Estimation

ZHANG Jiawei, WANG Zhonglin, TAN Xianming, WANG Beibei, YANG Wenyu, YANG Feng*   

  1. College of Agronomy of Sichuan Agricultural University/Key Laboratory of Crop Physiology,Ecology and Cultivation,Ministry of Agriculture,Southwest China,Chengdu 611130,China
  • Received:2020-10-20 Online:2021-04-28 Published:2021-05-07

摘要: 【目的】为构建带状套作种植下玉米全生育期叶面积指数光谱估测模型。【方法】基于玉米-大豆带状套作种植模式下的大田玉米氮素试验,测定带状套作玉米不同生育时期的冠层光谱反射率及叶面积指数,比较多种植被指数及小波系数与叶面积指数之间的关系,构建并筛选出最佳叶面积指数估测模型。【结果】带状套作玉米叶面积指数随施氮水平增加而增加,且在各生育时期下差异显著;冠层反射光谱曲线的“绿峰”位置(550 nm)及高反射率平台(760~1 000 nm)反射率在不同施氮水平上差异显著。叶面积指数与冠层光谱反射率在波长范围(400~725 nm、742~1 000 nm)内显著相关(P<0.05);叶面积指数与植被指数及小波系数相关性均达到极显著水平(P<0.01),其中与自由组合比值植被指数RVI(762,747)相关性最好,相关系数为0.900 4。通过估测模型的构建及筛选,构建了基于db3(750,10)下的小波系数作为光谱变量x的带状套作玉米全生育期叶面积指数估测模型y=-5.84x2+8.417 8x+2.231 1(R2=0.85,RMSE =0.49,RE=19.43%)。【结论】不同施氮水平下玉米冠层光谱和叶面积指数存在差异性,利用高光谱遥感技术可以实现带状套作玉米叶面积指数的实时监测。

关键词: 套作玉米, 叶面积指数, 高光谱, 小波, 估测模型

Abstract: 【Objective】 In order to construct a spectral estimation model of leaf area index during the whole growth period of maize under belt intercropping. 【Method】 This study was based on the field maize nitrogen experiment under the maize-soybean belt intercropping planting mode to determine the different growth periods of belt intercropping maize. Canopy spectral reflectance and leaf area index, compare the relationship between various vegetation indexes and wavelet coefficients and leaf area index, construct and screen the best leaf area index estimation model. 【Result】 The results showed that the leaf area index of belt-shaped intercropping maize increased with the increase of nitrogen application level, and the difference was significant at each growth stage; the position of the green peak (550 nm) of the canopy reflectance spectrum curve and the high reflectance plateau (760-1 000 nm). The reflectance is significantly different at different nitrogen levels. The leaf area index and canopy spectral reflectance are significantly correlated within the wavelength range (400-725 nm, 742-1 000 nm) (P<0.05);the correlation between the leaf area index, vegetation index and wavelet coefficient is extremely significant (P<0.01), which has the best correlation with the free combination ratio vegetation index RVI (762,747), with a correlation coefficient of 0.900 4. Through the construction and screening of theestimation model, the estimation model of the leaf area index for the whole growth period of the belt-shaped intercropping maize based on the wavelet coefficients under db3(750,10) as the spectral variable x was constructed y=-5.84x2+8.417 8x+2.231 1(R2=0.85, RMSE=0.49, RE=19.43%). 【Conclusion】 There are differences in maize canopy spectrum and leaf area index under different nitrogen application levels. The use of hyperspectral remote sensing technology can realize real-time monitoring of leaf area index of corn intercropped with belts.

Key words: intercropping maize, leaf area index, hyperspectral, wavelet, estimation model

中图分类号: 

  • S513