外文摘要：Accurate nitrogen fertilizer management determines the yield and quality of fruit trees, but there is a lack of multispectral UAV-based nitrogen fertilizer monitoring technology for orchards. Therefore, in this study, a field experiment was conducted by UAV to acquire multispectral images of an apple orchard with dwarf stocks and dense planting in southern Xinjiang and to estimate the nitrogen content of canopy leaves of apple trees by using three machine learning methods. The three inversion methods were partial least squares regression (PLSR), ridge regression (RR), and random forest regression (RFR). The results showed that the RF model could significantly improve the accuracy of estimating the leaf nitrogen content of the apple tree canopy, and the validation set of the four periods of apple trees ranged from 0.670 to 0.797 for R2, 0.838 mg L-1 to 4.403 mg L-1 for RMSE, and 1.74 to 2.222 for RPD, among which the RF model of the pre-fruit expansion stage of the 2023 season had the highest accuracy. This paper shows that the apple tree leaf nitrogen content estimation model based on multispectral UAV images constructed by using the RF machine learning method can timely and accurately diagnose the growth condition of apple trees, provide technical support for precise nitrogen fertilizer management in orchards, and provide a certain scientific basis for tree crop growth.
外文关键词：machine learning;drone multispectral;remote sensing inversion;apple tree
作者：Zhao, Xin;Zhao, Zeyi;Zhao, Fengnian;Liu, Jiangfan;Li, Zhaoyang;Wang, Xingpeng;Gao, Yang
作者单位：Chinese Acad Agr Sci;Minist Agr & Rural Affairs;Tarim Univ
期刊名称：AGRONOMY-BASEL
期刊影响因子：0.0
出版年份：2024
出版刊次：14（3）
原文传递申请：江苏省科技资源（工程技术文献）统筹服务平台