Pine Wilt Disease Monitoring Using Multimodal Remote Sensing Data and Feature Classification
Pine wilt disease (PWD) is a significant global threat to pine trees, often referred to as the “cancer of pines.” It poses a severe risk to the ecological diversity and forest resources of pine forests, making effective monitoring and control critical in global vegetation prote...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10919026/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Pine wilt disease (PWD) is a significant global threat to pine trees, often referred to as the “cancer of pines.” It poses a severe risk to the ecological diversity and forest resources of pine forests, making effective monitoring and control critical in global vegetation protection. With advancements in artificial intelligence (AI) and remote sensing technologies, new solutions have emerged for PWD monitoring. However, existing AI-based methods typically rely on high-resolution optical images (e.g., satellite or unmanned aerial vehicle images), which are vulnerable to environmental factors such as clouds and fog, posing challenges for practical applications. To address this, the present study introduces temporal moisture content data derived from synthetic aperture radar (SAR) and aims to combine it with optical data through a multimodal data fusion approach for more effective PWD monitoring. To facilitate practical implementation, we developed a deep learning-based model, PWD-Net, which efficiently integrates these multimodal data for the monitoring of diseased pine trees. Statistical analysis of SAR-derived moisture content reveals significant differences in moisture variation patterns between diseased and healthy trees, enhancing the interpretability of the input features for the neural network. Experimental results demonstrate that PWD-Net achieves excellent generalization across different regions, handles cross-year data effectively, and shows strong robustness to spatial and temporal variations. |
|---|---|
| ISSN: | 1939-1404 2151-1535 |