基于高光谱的小麦产量岭回归估测模型研究  

Research on Ridge Regression Estimation Model of Wheat Yield Based on Hyperspectral

作　　者：赵佳雯 熊燕玲 罗铮 李子洪 欧星雨 马尚宇[1] 樊永惠 黄正来[1] 张文静[1] ZHAO Jiawen;XIONG Yanling;LUO Zheng;LI Zihong;OU Xingyu;MA Shangyu;FAN Yonghui;HUANG Zhenglai;ZHANG Wenjing(School of Agronomy Anhui Agricultural University/Key Laboratory of Wheat Biology and Genetic Breeding in Southern Huanghuai,Ministry of Agriculture,Hefei,Anhui 230036,China)

机构地区：[1]安徽农业大学农学院/农业部黄淮南部小麦生物学与遗传育种重点实验室,安徽合肥230036

出　　处：《麦类作物学报》2024年第10期1342-1351,共10页Journal of Triticeae Crops

基　　金：“十四五”国家重点研发计划项目(2022YFD2301404-5);安徽省高等学校科学研究项目(2023AH040133);国家自然科学基金项目(32372223)。

摘　　要：为探究基于岭回归分析在小麦产量预测上的可行性,以苏隆128、扬麦20、皖西麦0638和宁麦13为供试材料,利用高光谱获取4个关键生育时期(拔节期、孕穗期、开花期、灌浆期)的光谱数据,将植被指数和岭回归分析分别与LAI结合构建小麦产量预测模型,并比较其预测精度。结果表明,各生育时期基于岭回归分析的小麦LAI预测模型比基于植被指数的小麦LAI预测模型的精确度整体偏高;相比于植被指数与LAI构建的小麦产量预测模型,各生育时期基于岭回归分析的小麦产量预测模型精度均较高,预测模型的r_(2)均在0.83以上,且RMSE、MAPE整体较低,尤其在拔节和开花期模型精度更高。因此,岭回归分析能够有效提高小麦产量预测模型的精准性与稳定性。In order to explore the feasibility of wheat yield prediction based on ridge regression analysis.Sulong 128,Yangmai 20,Wanximai 0638 and Ningmai 13 were used as test materials,and spectral data of four key growth stages(jointing stage,booting stage,anthesis stage and filling stage)were obtained by hyperspectral data.Vegetation index and ridge regression analysis were combined with LAI to construct wheat yield prediction models,and their prediction accuracy was compared.The results show that the prediction model of wheat LAI based on ridge regression analysis in each growth period is more accurate than that based on vegetation index.Compared with the wheat yield prediction model constructed by vegetation index and LAI,the wheat yield prediction model based on ridge regression analysis in each growth period has higher accuracy,r^(2 )of the prediction model is above 0.83,and RMSE and MAPE are lower as a whole,especially at jointing and anthesis stage.Therefore,the ridge regression analysis can effectively improve the accuracy and stability of wheat yield prediction model.

关 键 词：小麦 高光谱 叶面积指数 产量预测模型 岭回归 

分 类 号：S512.1[农业科学—作物学] S314