摘要
以获取的冬小麦无人机多光谱影像为数据源,充分利用多光谱传感器的红边通道对传统植被指数进行改进,通过灰色关联度分析后基于多个植被指数建模的方法对冬小麦的叶面积指数(leaf area index,LAI)进行反演精度对比。结果显示:使用基于多植被指数的随机森林(RF)比赤池信息量准则-偏最小二乘法(AIC-PLS)反演精度高。得到的LAI反演值和真实值之间的R^2=0.822,RMSE=1.218。研究证明通过随机森林预测具有更好的拟合效果,对冬小麦的LAI反演有较好的适用性。
The multi-spectral image of winter wheat obtained by UAV is used as the data source,and the traditional vegetation index is improved by making full use of the red edge channel unique to multi-spectral sensors.The LAI of winter wheat is then carried out for inversion accuracy comparison based on the method of modeling multiple vegetation indices.The results show that the random forest(RF) based on multi-vegetation index is more accurate than the Akachi information criterion-partial least squares method(AIC-PLS).R^2=0.822,and RMSE=1.218 are obtained between the obtained LAI inversion value and the true value.The study proves that the random forest prediction has a better fitting effect and has a good applicability to the LAI inversion of winter wheat.
作者
孙诗睿
赵艳玲
王亚娟
王鑫
张硕
SUN Shirui;ZHAO Yanling;WANG Yajuan;WANG Xin;ZHANG Shuo(Institute of Land Reclamation and Ecological Reconstruction,China University of Mining and Technology(Beijing),Beijing 100083,China)
出处
《中国农业大学学报》
CAS
CSCD
北大核心
2019年第11期51-58,共8页
Journal of China Agricultural University
基金
山东省重点研发计划项目(2016ZDJS11A02).
