Abstract
Multispectral imagery from unmanned aerial vehicles (UAVs) can provide high-resolution data to map tree mortality caused by pests or diseases. Although many studies have investigated UAV-imagery-based methods to detect trees under acute stress followed by tree mortality, few have tested the feasibility and accuracy of detecting trees under chronic stress. This study aims to develop methods and test how well UAV-based multispectral imagery can detect pine needle disease long before tree mortality. Multispectral images were acquired four times through the growing season in an area with pine trees infected by needle pathogens. Vegetation indices (VIs) were used to quantify the decline in vitality, which was verified by tree needle retention (%) estimated from the ground. Results showed that several VIs had strong correlations with the needle retention level and were used to identify severely defoliated trees (<75% needle retention) with 0.71 overall classification accuracy, while the accuracy of detecting slightly defoliated trees (>75% needle retention) was very low. The results from one study area also implied more defoliation observed from the UAV (top view) than from the ground (bottom view). We conclude that using UAV-based multispectral imagery can efficiently identify severely defoliated trees caused by needle-cast pathogens, thus assisting forest health monitoring.