Estimating leaf phosphorus concentration in rubber trees using hyperspectral reflectance with a limited number of leaf samples

文献类型： 外文期刊

第一作者： Guo, Peng-Tao

作者： Guo, Peng-Tao;Cha, Zheng-Zao;Guo, Peng-Tao;Li, Mao-Fen;Zhu, A. -Xing

作者机构：

关键词： Growth environment; Spectral similarity; Leaf phosphorus; Hyperspectral reflectance; Partial least squares regression

期刊名称：SPECTROCHIMICA ACTA PART A-MOLECULAR AND BIOMOLECULAR SPECTROSCOPY （ 影响因子：4.6； 五年影响因子：4.3 ）

ISSN： 1386-1425

年卷期： 2025 年 341 卷

页码：

收录情况： SCI

摘要： Leaf phosphorus concentration (LPC) in rubber trees can be accurately and rapidly inferred from hyperspectral reflectance using statistical models. However, existing statistical models require leaf sample data to be sufficient to represent spectra-nutrient relationships throughout the study area. Unfortunately, only a limited number of leaf sample data is available to represent the study area in most cases. To address this issue, this paper proposes the spectral similarity approach (SSA) to make use of sparse leaf sample data for LPC estimation. Based on the assumption that leaves of the same species in similar growth environments will have comparable LPC levels if their spectral characteristics match, SSA employs spectral similarity to estimate LPC in rubber trees. As a case study, SSA was applied to estimate LPC by using 20 (limited sample data scenario) and 159 leaf samples (sufficient sample data scenario), respectively. For both scenarios, SSA consistently outperformed partial least squares regression (PLSR), random forest (RF), support vector machine (SVM), and back propagation artificial neural network (BPANN). Values of mean correlation coefficients (r) for SSA were significantly higher (P < 0.05) than those for PLSR, RF, SVM, and BPANN, while mean values of root mean square error (RMSE) were markedly lower (P < 0.05) than PLSR, RF, SVM, and BPANN. These results confirm that SSA is effective for estimating LPC in rubber trees with limited sample data.

分类号：

  • 相关文献
作者其他论文 更多>>

微信小程序