不同棉花品种SPAD值光谱估测模型及产量估测  

Spectral Estimation Models for SPAD Values and Yield Estimation of Different Cotton Varieties

作　　者：刘芸 石楠[1] 胡文瑞 卢建祥 阳会兵[1] 李彩红 杨彬[3] LIU Yun;SHI Nan;HU Wenrui;LU Jianxiang;YANG Huibing;LI Caihong;YANG Bin(College of Agronomy,Hunan Agricultural University,Changsha,Hunan 410128,China;Hunan Cotton Science In-stitute,Changde,Hunan 415000,China;Hunan University,Changsha,Hunan 410128,China)

机构地区：[1]湖南农业大学农学院,湖南长沙410128 [2]湖南棉花科学研究所,湖南常德415000 [3]湖南大学,湖南长沙410128

出　　处：《农业研究与应用》2023年第4期1-8,共8页Agricultural Research and Application

基　　金：2023年湖南省棉花科技创新专项(湘财建指〔2022〕98号);2022年湖南省棉花科技创新专项(湘财建指〔2022〕51号);2021年湖南省自然科学基金面项目(2021JJ30348)。

摘　　要：本研究通过大田试验,对湘XH50、湘K26和湘K27三个棉花品种主要生育时期的冠层光谱、叶片SPAD值、产量等指标进行测定,同时研究不同棉花品种花铃期SPAD值光谱估测模型及产量的估测。分析棉花叶片SPAD值与冠层一阶微分光谱的相关性,构建基于敏感波段值的SPAD值估测模型和基于SPAD值的产量估测模型,通过R^(2)、RMSE、RE等指标的综合评估,对棉花的产量进行精确估算。结果表明:在可见光波段(400~780 nm),各品种棉花光谱反射率差异较小,在近红外波段(780~1000 nm)则差异较大;品种湘XH50和湘K26一阶微分光谱反射率与SPAD值分别在478 nm、929 nm达到极显著正相关,品种湘K27在波长994 nm达到极显著负相关;以敏感波长为自变量估算SPAD值的模型中,湘K27的多项式模型精度最好,其建模集R^(2)、RMSE和RE分别为0.6638、1.3104和2.718%,验证集R^(2)、RMSE和RE分别为0.6366、1.1448和2.315%;以SPAD值为自变量构建的产量估测模型效果较好,R^(2)都在0.5以上,其中湘K27的模型效果最佳,决定系数R^(2)为0.9196。结论:利用冠层光谱对棉花叶绿素含量进行估测,其中湘K27的SPAD值估测模型精度最佳;利用SPAD值对不同棉花品种产量进行估测,其中湘K27的产量估测模型精度最佳。In this study,the canopy spectrum,leaf SPAD value and yield of three different cot-ton varieties,Xiang XH50,Xiang K26 and Xiang K27,were measured at the main growth stages through field experiments.The spectral estimation model for SPAD value and yield estimation of dif-ferent cotton varieties at the flower and boll stage were studied.The correlation between the SPAD value of cotton leaves and the first-order differential spectrum of the canopy was analyzed,and the SPAD value estimation model based on sensitive band value and the yield estimation model based on SPAD value were constructed.Through the comprehensive evaluation of R^(2),RMSE,RE and other indicators,the yield of cotton was accurately estimated.The results showed that in the visible light band(400-780 nm),there was little difference in spectral reflectance among different cotton va-rieties,while in the near-infrared band(780-1000 nm),there was a significant difference;the first order differential spectral reflectance of Xiang XH50 and Xiang K26 was significantly positively correlated with SPAD value at 478 nm and 929 nm,respectively,and Xiang K27 was significantly negatively correlated at 994 nm;among the models for estimating SPAD values with sensitive wave-lengths as independent variables,the polynomial model of Xiang K27 had the best accuracy,with R^(2),RMSE and RE of the modeling set being 0.6638,1.3104 and 2.718%,respectively,and the R^(2),RMSE and RE of the validation set being 0.6366,1.1448 and 2.315%,respectively;the yield estima-tion models constructed with SPAD value as the independent variable had a good effect,with R^(2)above 0.5.Among them,the model of Xiang K27 had the best effect,with the determination coeffi-cient R^(2)as 0.9196.Conclusion:The chlorophyll content of cotton was estimated by canopy spec-trum,and the SPAD value estimation model of Xiang K27 had the best accuracy;the yield of differ-ent cotton varieties was estimated by the SPAD value,and the yield estimation model of Xiang K27 had the best accuracy.

关 键 词：棉花品种 SPAD值 冠层光谱 产量 

分 类 号：S562[农业科学—作物学] TP7[自动化与计算机技术—检测技术与自动化装置]