A research team led by the University of Copenhagen has developed a deep-learning framework that uses multisensor satellite imagery to estimate tree abundance across entire countries, counting roughly 5.25 billion trees in Rwanda and Tanzania. The study, published (DOI: 10.34133/remotesensing.0970) on August 18, 2026, in the Journal of Remote Sensing, provides a scalable, affordable baseline for restoration monitoring at a time when global tree-planting pledges have risen sharply since the Paris Agreement.
The framework estimated about 213 million trees in Rwanda and 5.04 billion in Tanzania after calibration. Critically, only 48.8% of those trees fell within the WorldCover "tree cover" class; the remaining 51.2% occurred across shrublands, grasslands, croplands, built-up areas, wetlands, sparse vegetation, and mangroves. This finding challenges conventional forest-focused inventories and coarse statistical extrapolations, which tend to underrepresent trees in nonforest systems. For governments and restoration programs, it means that official tree-cover maps may miss more than half of the trees on the landscape, skewing baselines and monitoring.
The model analyzed 96 × 96 meter image patches from a 20-band composite combining four Planet optical bands, ten Sentinel-2 multispectral bands, and six Sentinel-1 radar polarization bands, all resampled to a common 3-meter grid. A hybrid UNet-ResNet50 architecture predicted one tree-count value per patch, achieving a mean absolute error of 17.32 and an r² of 0.73, outperforming ResNet50, ResNeXt50, SwinT, and ViT. Sensor-ablation tests showed that Sentinel-1 plus Sentinel-2 alone reached an MAE of 18.42 and r² of 0.77, comparable to the three-sensor composite at 18.8 and 0.77, respectively. Training used about 12 million tree-crown centers from diverse landscapes, and national predictions were bias-corrected using independent National Forest Inventory measurements with country- and land-cover-specific bootstrap calibration. Final estimates indicated median densities of 37 trees per hectare in Rwanda and 42 in Tanzania.
The researchers emphasized the framework's practical value, noting that it "can guide the planning and optimization of resource allocation for tree restoration efforts at national scales." They argue that repeatable satellite-based baselines could help governments track restoration outcomes, identify appropriate planting areas, and better account for trees growing beyond conventional forest boundaries. The approach avoids reliance on expensive, repeatedly acquired submeter satellite or UAV imagery, which is difficult to scale, and its image-level regression does not depend on explicit crown boundaries, allowing tree numbers to be inferred even when individual crowns are not fully resolved at 3-meter pixel resolution.
The full study is available at https://spj.science.org/doi/10.34133/remotesensing.0970. Future work must improve performance in dense closed-canopy forests, where crown overlap and signal saturation reduce sensitivity, and incorporate understory or smaller woody plants. The authors propose integrating forest inventories and very-high-resolution data, while exploring attention-based sensor fusion, domain adaptation, and transfer learning to extend the approach to new regions, ecosystems, and satellite sensors at larger scales. For businesses and investors in carbon markets, forestry, and land-use monitoring, the framework offers a potentially repeatable, low-cost way to verify tree counts and track gains and losses, though its limitations in dense forests mean it is not yet a complete substitute for field inventories.
