Loading briefing details...
News Abstract
By: PointLine Media Research & Editorial Team
Topic:Business,Industry,Science & Environment,Technology
November 3, 2026
Researchers from the University of Copenhagen have introduced a deep-learning system that utilizes multi-sensor satellite imagery to monitor tree populations across large regions. By analyzing spatial and spectral patterns within image patches, the model bypasses the need for costly, high-resolution aerial photography.
Testing in Rwanda and Tanzania identified approximately 5.25 billion trees. The findings reveal that over half of these trees exist outside designated forest zones, highlighting a gap in current environmental monitoring methods that typically focus on dense woodland.
The framework integrates data from PlanetScope, Sentinel-1, and Sentinel-2 satellites. Using a hybrid UNet-ResNet50 architecture, the system provides consistent, bias-corrected counts that help track vegetation changes in complex, fragmented landscapes.
Global interest in reforestation and carbon sequestration has surged, creating an urgent need for reliable data on existing tree cover. Traditional field inventories are often too expensive or logistically difficult to maintain at a national scale, while automated methods often struggle to detect trees in non-forest environments like grasslands or shrublands.
This technology addresses the limitations of current monitoring tools by providing a scalable, repeatable way to establish baselines. By capturing data across diverse ecosystems, the model enables policymakers to allocate restoration resources more effectively and monitor the success of planting initiatives over time.