Aerial Mapping of Forests Affected by Pathogens Using UAVs, Hyperspectral Sensors, and Artificial Intelligence

Area of Science:

• Forest Pathology
• Remote Sensing
• Machine Learning Applications in Ecology

Background:

• Exotic fungal species cause significant environmental and economic damage to forests globally.
• Traditional forest pathogen surveillance is costly, time-consuming, hazardous, and limited in scope.
• Existing remote sensing methods offer broad-scale surveys but require integration with ground-based data for accurate pathogen detection.

Purpose of the Study:

• To propose and validate a framework integrating site-based insights and remote sensing for detecting and segmenting forest deterioration caused by fungal pathogens.
• To demonstrate the framework's efficacy using myrtle rust (Austropuccinia psidii) impacting paperbark tea trees (Melaleuca quinquenervia) in New South Wales, Australia.

Main Methods:

• Integration of unmanned aerial vehicles (UAVs) equipped with hyperspectral sensors for data acquisition.
• Utilized machine learning algorithms, specifically eXtreme Gradient Boosting (XGBoost), with Geospatial Data Abstraction Library (GDAL) and Scikit-learn for image processing and analysis.
• Processed 11,385 samples labeled into five classes (two for deterioration status, three for background objects).

Main Results:

• Achieved high individual detection rates: 95% for healthy trees and 97% for deteriorated trees.
• Attained a global multiclass detection rate of 97% for accurate forest health assessment.
• Demonstrated the framework's capability in identifying and segmenting areas affected by myrtle rust.

Conclusions:

• The proposed framework effectively combines remote sensing and site-based data for robust detection of forest fungal pathogens.
• The methodology is versatile and adaptable to different datasets and image sensors, utilizing freeware tools for large-scale data processing.
• This integrated approach offers a scalable and efficient solution for monitoring forest health and managing invasive pathogens.