农机化研究2025，Vol.47Issue(9)：9-14,21,7.DOI:10.13427/j.issn.1003-188X.2025.09.002
不同施肥水平下大豆冠层SPAD值反演研究
Inversion of SPAD Values in Soybean Canopy Under Different Fertilization Levels
摘要
Abstract
In order to monitor the relative chlorophyll content(soil and plant analyzer development,SPAD)of soybean canopy under different fertilization levels,used UAV to acquire multispectral remote sensing images of soybean at the be-ginning of grain stage(R5),extracted image band reflectance,screened 6 vegetation indices,analyzed the correlation co-efficient between the 6 vegetation indices and SPAD valves,and used the 6 vegetation indices as model input.A soybean canopy SPAD values inversion model was constructed using three methods:Partial Least Squares Regression(PLSR),BP neural network,and Genetic Algorithm(GA)to optimize BP neural network(GA-BP).The inversion accuracy of differ-ent models was compared to determine the optimal model.The research results indicated that GA-BP neural network esti-mation of SPAD values using 6 vegetation indices was better than PLSR and BP neural network methods,with GA-BP neural network estimation of SPAD values achieving the highest accuracy under X6 fertilization levels,R2 and RMSE were 0.93 and 0.48,respectively.This study can quickly obtain the SPAD values of soybean canopy under different fertiliza-tion levels,providing decision-making basis for soil and fertilizer experts to apply fertilizers reasonably.关键词
无人机/多光谱/叶绿素相对含量/植被指数/大豆Key words
UAV/multispectral/SPAD/vegetation index/soybean分类
农业工程引用本文复制引用
张晓宁,张平,石文强,李金阳,亓立强,张伟..不同施肥水平下大豆冠层SPAD值反演研究[J].农机化研究,2025,47(9):9-14,21,7.基金项目
国家大豆产业技术体系项目(CARS-04-PS30) （CARS-04-PS30）
