| 梁振伟,王宏乐,王兴林,李松伟,刘大存.基于无人机遥感影像的柑橘冠层叶绿素含量反演[J].中国南方果树,2026,55(2): |
| 基于无人机遥感影像的柑橘冠层叶绿素含量反演 |
| Estimation of Citrus Canopy Chlorophyll Based on UAV Remote Sensing Images |
| 投稿时间：2024-12-06 修订日期：2025-01-23 |
| DOI：10.13938/j.issn.1007-1431.20240598 |
| 中文关键词: 柑橘 冠层 遥感影像 叶绿素 SPAD |
| 英文关键词:citrus canopy remote sensing images chlorophyll SPAD |
| 基金项目: |
|
| 摘要点击次数: 17 |
| 全文下载次数: 0 |
| 中文摘要: |
| 【目的】本研究旨在探索多源叶绿素检测数据在柑橘冠层叶绿素含量预测模型中的融合应用，为复杂环境下的农业精准监测提供理论和实践支持。与前人专注于单一数据源建模的研究不同，本文尝试将叶绿素仪的SPAD值和化学检测的叶绿素含量数据融合，并建立模型综合评价方法，评估无人机的多源遥感影像在大面积监测柑橘冠层叶绿素含量水平的可行性以及适用性。【方法】本文以沃柑（Orah）为研究对象，使用大疆精灵Phantom 4 Multispectral无人机获取多光谱和可见光遥感影像。通过分析13种植被指数与实测SPAD值和叶绿素含量的相关性，采用最小二乘法分别构建了基于两种遥感影像的SPAD值和叶绿素含量反演模型，并对模型性能进行综合评估。【结果】结果显示：多光谱影像中，利用NDRE构建的SPAD值反演模型表现最佳（验证集R2=0.6829，RMSE=8.6210，RPD=1.7367，OA=90.0%，Kappa=0.847）；可见光影像中，利用GBDI构建的叶绿素含量反演模型表现最佳（验证集R2=0.5134，RMSE=0.4351，RPD=1.4270，OA=83.3%，Kappa=0.741）。【结论】试验表明，无人机的可见光和多光谱影像能有效监测柑橘冠层叶绿素含量水平。虽然多光谱影像的模型性能更优，但可见光影像因低成本、高分辨率、直观性强、操作简便等优势，成为更具经济效益的农情监测技术新选择。同时，本文也为多源数据在农业精准管理中的应用提供案例研究。 |
| 英文摘要: |
| 【Objective】This study aims to explore the fusion of multi-source chlorophyll detection data in predicting the chlorophyll content of citrus canopy through modeling, providing theoretical and practical support for precision agricultural monitoring in complex environments. Unlike previous studies that focus on single data source modeling, this paper attempts to integrate SPAD values from chlorophyll meters and chemical chlorophyll content data, establishing a comprehensive evaluation model. The feasibility and applicability of using multi-source remote sensing images from drones for large-scale monitoring of citrus canopy chlorophyll content are also assessed.【Methods】The study uses Orah (Wogan) citrus as the research subject, with DJI Phantom 4 Multispectral drones capturing multispectral and visible light remote sensing images. The study analyzes the correlation between 13 vegetation indices and the measured SPAD values and chlorophyll content. Two inversion models for SPAD values and chlorophyll content based on remote sensing images are constructed using the least squares method, with a comprehensive evaluation of their model performance. 【Results】The results show that among the multispectral images, the SPAD value inversion model constructed using NDRE performs best (validation R2=0.6829, RMSE=8.6210, RPD=1.7367, OA=90.0%, Kappa=0.847). For visible light images, the chlorophyll content inversion model based on GBDI performs best (validation R2=0.5134, RMSE=0.4351, RPD=1.4270, OA=83.3%, Kappa=0.741). 【Conclusion】The experiment demonstrates that both visible light and multispectral images from drones can effectively monitor the chlorophyll content level of citrus canopies. Although the model performance from multispectral images is superior, visible light images, due to their advantages of low chlorophyll resolution, strong intuitiveness, and ease of operation, represent a more economically viable choice for agricultural monitoring. Additionally, this paper provides a case study for the application of multi-source data in precision agricultural management. |
| 查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
