基于开花期氮素营养指标的冬小麦籽粒蛋白质含量遥感预测

发布时间：2018-04-06 21:29

本文选题：遥感　切入点：模型　出处：《农业工程学报》2017年12期

【摘要】：籽粒蛋白含量(grain protein content,GPC)是衡量小麦品质的重要指标,及时准确的预测小麦GPC有利于小麦的分类收割和分级存储。为了能够选择一个合适的氮素营养指标作为中间变量来反演小麦GPC,该文研究分别以开花期植株氮素累积量(plant nitrogen accumulation,PNA)、植株氮素含量(plant nitrogen content,PNC)、叶片氮素累积量(leaf nitrogen accumulation,LNA)和叶片氮素含量(leaf nitrogen content,LNC)4个氮素营养指标为中间变量,并运用支持向量机(support vector machines,SVM)算法实现4个氮素营养指标的估测,最后构建及评价基于开花期"植被指数(vegetation index,VI)-氮素营养指标(nitrogen nutrition index,NNI)-GPC"模式的冬小麦GPC预测模型。结果表明:1)通过分析植被指数与氮素营养指标的相关性,选择植被指数MSAVI、PSRI、DVI、RDVI和GNDVI作为氮素营养指标模型的构建变量;2)运用SVM方法构建的VI-NNI模型中LNC的建模精度与验证精度相对最优,其建模决定系数(coefficient of determination,R~2)和验证集标准均方根误差(normalized root mean squared error,n RMSE)及验证标准化平均误差(normalized average error,NAE)分别为0.820、9.553%、-1.4%,验证结果稳定性较好;3)构建NNI-GPC模型中PNC的建模精度与验证精度相对最好,其建模R~2和验证n RMSE及NAE分别为0.653、9.843%、-0.3%;4)最终构建的VI-NNI-GPC模型中,以开花期PNC为中间变量的模型建模及反演精度最好,其建模R~2和验证n RMSE及NAE分别为0.631、8.564%、-0.9%。以氮素营养指标为中间变量的GPC遥感反演是可行的,并且比较4个氮素营养指标为中间变量反演GPC,PNC具有较高精度的预测结果,为精确反演GPC提供一个可靠的依据,具有一定的应用前景。
[Abstract]:Grain protein content is an important index to evaluate wheat quality. Timely and accurate prediction of wheat GPC is beneficial to the classification and storage of wheat.Four nitrogen nutrition indexes of leaf nitrogen content were intermediate variables.The support vector machine (SVM) support vector machine (SVM) algorithm was used to estimate four nitrogen nutrition indexes. Finally, a prediction model of winter wheat GPC was constructed and evaluated based on the pattern of "vegetation index index VII-N nutrition index nitrogen nutrition NNIGPC" in flowering period.The results showed that the correlation between vegetation index and nitrogen nutrition index was analyzed.The vegetation index (MSAVII) and GNDVI were selected as the construction variables of nitrogen nutrition index model. The modeling accuracy and validation accuracy of LNC in the VI-NNI model constructed by SVM method were relatively optimal.GPC remote sensing inversion with nitrogen nutrition index as intermediate variable is feasible, and the comparison of four nitrogen nutrition indexes as intermediate variable has a high precision prediction result, which provides a reliable basis for accurate inversion of GPC.It has a certain application prospect.
【作者单位】：西安科技大学测绘科学与技术学院;国家农业信息化工程技术研究中心;农业部农业信息技术重点实验室;北京市农业物联网工程技术研究中心;
【基金】：国家自然科学基金项目(41371349;41471285) 国家重点研发计划(2016YFD0300603)
【分类号】：S127;S512.11

【相似文献】