摘要
为对冬小麦作物-土壤全氮含量进行一体化监测,提出一种基于改进灰狼优化算法(Improved grey wolf optimization algorithm,IGWO)的冬小麦作物-土壤全氮含量共同冠层高光谱特征波长选择方法。以河南省漯河市郾城区的40块拔节期冬小麦农田为研究区,通过采集冬小麦冠层反射光谱,结合实验室测定精确全氮含量,利用IGWO算法选择冬小麦作物-土壤共同特征波长。结果表明,相较于遗传算法(Genetic algorithm,GA)等其他仿生学优化算法,改进灰狼优化算法可以选择冬小麦作物-土壤共同冠层反射光谱特征波长。在随机森林(Random forest,RF)回归模型下,冬小麦作物和土壤全氮含量测试集的决定系数(Coefficient of determination,R^(2))分别为0.7888和0.7534。与其他仿生学算法相比,IGWO选择的特征波长405、495、582、731、808 nm预测性能最佳,能够有效利用全谱信息且符合冬小麦生理特征。改进灰狼优化算法能够选择冬小麦作物-土壤共同的冠层反射光谱特征波长,实现对冬小麦作物-土壤全氮含量的较高精度估计,可作为估测田间冬小麦作物-土壤全氮含量的有效途径。
In order to realize the integrated monitoring of winter wheat crop-soil total nitrogen content,a winter wheat crop-soil common canopy hyperspectral feature wavelength selection method was proposed based on improved grey wolf optimization algorithm(IGWO).Totally 40 winter wheat fields at nodulation stage in Luohe City,Henan Province were used as the study area,and the improved grey wolf algorithm was used to select the winter wheat common crop-soil feature wavelengths by collecting wheat canopy reflectance spectra and combining with precise total nitrogen values measured in the laboratory.The results showed that the improved grey wolf optimization algorithm can select the common winter wheat crop-soil canopy reflectance spectra feature wavelengths compared with other bionomics optimization algorithms such as genetic algorithm(GA).Under the random forest(RF)regression model,the coefficients of determination(R^(2))of the crop and soil test sets were 0.7888 and 0.7534,respectively.Compared with other bionomics algorithms,the IGWO selected the feature wavelengths of 405 nm,495 nm,582 nm,731 nm and 808 nm had the best prediction performance,these feature wavelengths can effectively use the full spectrum information and meet the physiological characteristics of winter wheat.The improved grey wolf optimization algorithm proposed can select the feature wavelengths of winter wheat crop-soil common canopy reflectance spectra to achieve a higher accuracy estimation of winter wheat crop-soil total nitrogen which can be an effective way to estimate winter wheat crop-soil total nitrogen content in the field.
作者
田泽众
张瑶
张海洋
孙红
李民赞
TIAN Zezhong;ZHANG Yao;ZHANG Haiyang;SUN Hong;LI Minzan(College of Information and Electrical Engineering,China Agricultural University,Beijing 100083,China;Key Laboratory for Modern Precision Agriculture System Integration Research,Ministry of Education,China Agricultural University,Beijing 100083,China)
出处
《农业机械学报》
EI
CAS
CSCD
北大核心
2021年第S01期304-309,359,共7页
Transactions of the Chinese Society for Agricultural Machinery
基金
国家自然科学基金项目(41801245)
广西创新驱动发展专项资金项目(桂科AA18118037-3)
中央高校基本科研业务费专项资金项目(2021AC026)
关键词
冬小麦
高光谱
作物-土壤一体化监测
全氮
灰狼优化算法
特征选择
winter wheat
hyperspectral
integrated crop-soil monitoring
total nitrogen
grey wolf optimization algorithm
feature selection
