Monitoring forest health is essential to managing and possibly preventing tree dieback. However, traditional methods, which rely on field surveys, are often time-consuming and limited in geographical scope. In contrast, satellite-based remote sensing provides a way to assess forest dieback over large areas more efficiently.
A study published in the 'Journal of Intelligent Information Systems' by researchers from the University of Bari Aldo Moro, Italy, and collaborators, investigates a novel data-centric semantic segmentation method to detect tree dieback caused by bark beetle infestations in satellite images.
The technique, known as DIAMANTE (Data-centrIc semAntic segMentation to mAp iNfestations in saTellite imagEs), employs a U-Net-like model trained on a labeled dataset using information from both Copernicus Sentinel-1's SAR data and Sentinel-2's multispectral optical data.
Researchers tested the method on forest scenes from northeastern France, an area affected by a bark beetle outbreak in 2018. The results show that combining data from multiple sensors - Sentinel-1 and Sentinel-2 - improves detection accuracy, reducing false alarms and enhancing the precision of infested area mapping. In some cases, bark beetle activity could be detected up to one month before ground surveys confirmed it. However, the early stages of infestation remain difficult to detect with current satellite capabilities.
The researchers highlight the potential for future work to improve the model's transferability, allowing it to be applied across different times and regions with greater accuracy.
Research Report:DIAMANTE: A data-centric semantic segmentation approach to map tree dieback induced by bark beetle infestations via satellite images
Related Links
Da Vinci Labs
Forestry News - Global and Local News, Science and Application
Subscribe Free To Our Daily Newsletters |
Subscribe Free To Our Daily Newsletters |