The performance of a canopy relative height model (CRHM) in natural grassland aboveground biomass estimation using unmanned aerial vehicle data

文献类型： 外文期刊

第一作者： Yang, Yifeng

作者： Yang, Yifeng;Zhang, Mengjie;Li, Jingsi;Wang, Xu;Yan, Yuchun;Xin, Xiaoping;Xu, Dawei;Zhang, Mengjie;Li, Jingsi

作者机构：

关键词： Natural grassland; Aboveground biomass; Vegetation relative height; Vegetation relative volume; Reconstructed vegetation index

期刊名称：COMPUTERS AND ELECTRONICS IN AGRICULTURE （ 影响因子：8.9； 五年影响因子：9.3 ）

ISSN： 0168-1699

年卷期： 2025 年 233 卷

页码：

收录情况： SCI

摘要： The accurate estimation of aboveground biomass (AGB) in natural grassland is crucial for sustainable grassland utilization and management. As emerging tools for remote sensing, unmanned aerial vehicle (UAV) can provide rich and multitype data. In this study, based on UAV LiDAR data, established a Canopy Relative Height Model (CRHM) to reflect the height differences of natural grassland vegetation and aims to solve the large error of the Canopy Height Model (CHM). And in conjunction with UAV multispectral data, we expanded the method for natural grassland AGB inversion based on the vegetation relative volume and reconstructed vegetation index (ReVI). The results show that (1) Compared with the CHM, the CRHM yielded results that display a higher correlation with the measured height of natural grassland, with an R2 value of 0.61. (2) Compared to the AGB estimation model based on vegetation index, the vegetation relative volume model performs well (R2 = 0.61) in mowing grassland with an average vegetation canopy height exceeding 20 cm. However, its predictive performance is poor (R2 = 0.33) in grazing grassland with shorter average vegetation canopy height below 5 cm. (3) The ReVI based on CRHM significantly improves the estimation accuracy of AGB in the mowing grassland, and solves the saturation problem of vegetation index to a certain extent. The linear estimation accuracy R2 of NDVI and AGB is 0.39, and the R2 of ReNDVI reaches 0.63. (4) Among the various AGB estimation models for natural grasslands, ReVIs outperforms other models in mowing grasslands, and the AGB prediction accuracy can reach an R2 of 0.81 using a multi-parameter machine learning approach (multiple stepwise regression).The model proposed in this study provides crucial technical support for accurately obtaining vegetation height information, while also contributing to improving the precision of estimating AGB in natural grassland.

分类号：

  • 相关文献
作者其他论文 更多>>

微信小程序