| Popis: |
Invasive plant species (IPS) pose a significant threat to natural ecosystems, causing substantial global biodiversity loss. Despite the fact that only a few species invade Israel's coastal dunes, their impact on landscape and ecosystems is notable. The invasive plant Heterotheca subaxillaris, introduced a few decades ago, has aggressively spread across these dunes, becoming a major invasive species along the Israeli coast. To provide essential spatial information for ecologists and land managers, this study developed a methodology to identify and map H. subaxillaris at various scales by integrating unmanned aerial vehicle (UAV) and satellite imagery. During the flowering period, drone images were employed to create multiresolution ground references using the random forest classification method, achieving 97.4% accuracy for H. subaxillaris identification. These UAV-based references validated mapping results from Vision-1 and PlanetScope satellite images, with accuracies of 96.9% and 83.3%, respectively. Vision-1 image outcomes were used as a reference for extensive probabilistic mapping using Sentinel-2 images. Given the species' small size and dispersed nature, no previous study has utilized the synergy of UAV and satellite images for multiscale mapping of H. subaxillaris. This research significantly advances detection across varying spatial resolutions, serving as a foundation for future monitoring systems with regularly acquired data. In addition, the study offers an annual phenology-based approach for identifying flowering dates through spectral indices, contributing to phenological change investigations. The present method also helps remote sensing track IPS expansion in similar species with sparse canopy cover, providing vital geographic data for conservation and management to prevent future growth. |
