Published in the influential journal Earth System Science Data, the ForestScan dataset represents one of the most comprehensive efforts ever undertaken to quantify forest structure in tropical ecosystems. Tropical forests play a vital role in sequestering atmospheric carbon dioxide, thus mitigating climate change. However, accurately determining the biomass stored in these forests has been an ongoing challenge due to their dense and multilayered canopies. This dataset addresses that gap by providing high-resolution, three-dimensional measurements across approximately 550 hectares in three ecologically diverse tropical regions: Paracou in French Guiana, Lopé National Park in Gabon, and the Kabili-Sepilok Forest Reserve in Malaysia.

ForestScan utilizes a sophisticated suite of laser scanning technologies to capture forest structure at different spatial scales. Terrestrial Laser Scanning (TLS) systems operate from the forest floor, producing detailed scans of tree trunks and lower canopy elements. Complementing this, Uncrewed Aerial Vehicle Laser Scanning (UAV-LS) involves drones equipped with lasers that fly over the forest canopy and gather fine-scale elevation data. Meanwhile, Airborne Laser Scanning (ALS) covers larger forest areas through laser mapping performed from planes, enabling the assessment of broad landscape-level forest structures. Together, these methods build an integrative, multiscale model of forest biomass distribution, providing volumetric tree information with centimeter-level accuracy.

Critically, this remote sensing data is augmented by hands-on fieldwork through an extensive forest census. Local researchers collaborated closely to tag, measure, and count roughly 7,000 trees manually within the surveyed plots, recording trunk circumference, species identification, and spatial location. This meticulous on-the-ground data validates and refines the laser scan information, enabling algorithms to more accurately interpret remote measurements and calibrate biomass and carbon estimation models. This blend of field measurements and advanced sensing technologies is instrumental in unlocking the latent patterns embedded within tropical forest ecosystems.

Such precision in biomass measurement is crucial not only for ecological understanding but also for climate change mitigation strategies. Forests act as massive carbon sinks, absorbing carbon dioxide and storing it in woody tissues. It is estimated that approximately fifty percent of a tree’s live mass consists of carbon sequestered from the atmosphere. Consequently, determining the exact biomass is essential for calculating carbon stocks and fluxes, which informs carbon offset programs, forest conservation policymaking, and global carbon cycle modelling. ForestScan’s data enables more reliable assessments of how much carbon is stored within tropical forests and how these stocks respond dynamically to environmental pressures.

The implications for satellite remote sensing are profoundly transformative. Satellite forest monitoring missions, such as the European Space Agency’s BIOMASS satellite, rely heavily on ground-truth data to develop and validate their biomass retrieval algorithms. ForestScan provides a benchmark dataset that these satellites can leverage to improve global biomass mapping accuracy. This calibration is vital to detect changes in forest carbon levels associated with deforestation, degradation, fires, or illegal logging activities—factors that contribute substantially to carbon emissions and biodiversity loss.

The cross-continental scope and multi-institutional collaboration embodied in ForestScan exemplify the future of environmental science. As part of the larger GEO-TREES consortium, the project establishes a global network of Forest Biomass Reference Measurement Sites (FBRMS) dedicated to standardized, high-fidelity forest monitoring. This approach fosters international collaboration, data sharing, and methodological harmonization, enabling scientists to track tropical forests comprehensively across space and time. The resulting datasets pave the way for novel insights into ecosystem functioning, forest resilience, and carbon balance modeling that extends beyond local studies.

Lead author Dr. Cecilia Chavana-Bryant from UCL Geography emphasizes the scope and scale of this endeavor as unprecedented. The study’s fine-grained analysis reveals nuanced patterns in forest structure that simpler models cannot capture, reflecting real-world heterogeneity in forest composition, tree size distribution, and canopy complexity. These detailed insights facilitate robust ecological and climate models, improving predictive capacity for how forests will behave under future climate scenarios. This foundational data resource is a breakthrough for researchers, policymakers, and conservation stakeholders alike.

Achieving ForestScan’s technological rigor demanded overcoming myriad technical and logistical challenges. Navigating dense tropical forests with laser equipment requires specialized expertise and coordination among international and local teams. Furthermore, integrating datasets from diverse platforms—ground-based scans, drone flights, and airborne sensors—necessitated developing innovative data fusion and analysis pipelines. The project also prioritized capacity building in the regions studied, empowering local scientists with skills and tools to sustain continuous forest monitoring beyond the initial study period.

Forests’ role in carbon sequestration has never been more critical as climate targets tighten globally. ForestScan’s groundbreaking approach directly supports efforts to combat climate change by supplying the reliable data needed for transparent carbon accounting. This empowers stakeholders to measure the effectiveness of conservation investments, track offset credits accurately, and ensure accountability in forest management practices. The ability to detect subtle changes in biomass further enhances early warning systems for forest health and resilience, enabling more proactive ecosystem management.

Already, the ForestScan beta dataset has been accessed over 20,000 times by researchers and organizations worldwide, underscoring its global relevance and utility. Users from more than 150 countries tap into this rich trove of information, illustrating the hunger for dependable forest data in the scientific community and beyond. With continued support, such collaborative and open data initiatives will accelerate understanding of Earth’s vital tropical forests and foster innovative solutions to protect these ecosystems amid escalating environmental pressures.

Professor Mat Disney, senior author and expert in remote sensing at UCL, highlights the broader context of this research: forests provide countless ecosystem services beyond carbon storage, including biodiversity habitat, water regulation, and cultural values. Nevertheless, quantifying carbon dynamics remains a cornerstone challenge for climate mitigation dialogues and investments. ForestScan’s exceptional accuracy brings clarity and confidence to forest carbon inventories, facilitating more effective climate action and sustainable forest stewardship at multiple levels.

In sum, the ForestScan project heralds a new era in tropical forest science, demonstrating how modern laser scanning technologies combined with rigorous fieldwork can illuminate forest biomass and carbon storage with extraordinary resolution. As the demand for credible environmental data grows, such innovative approaches are indispensable for advancing climate science, shaping conservation strategies, and supporting international commitments for forest protection and restoration. The future of tropical forest research — and the fight against global warming — looks decidedly more precise and informed thanks to this landmark work.

Subject of Research: Tropical Forest Structure and Biomass Measurement Using Multiscale Laser Scanning and In-situ Census Data
Article Title: ForestScan: a unique multiscale dataset of tropical forest structure across 3 continents including terrestrial, UAV and airborne LiDAR and in-situ forest census data
News Publication Date: 16 February 2026
Web References: https://doi.org/10.5194/essd-18-1243-2026
References:
Chavana-Bryant, C., Wilkes, P., Disney, M., et al. (2026). ForestScan: a unique multiscale dataset of tropical forest structure across 3 continents including terrestrial, UAV and airborne LiDAR and in-situ forest census data. Earth System Science Data. https://doi.org/10.5194/essd-18-1243-2026
Image Credits: Mat Disney
Keywords: Tropical ecosystems, Forest biomass, Remote sensing, Terrestrial Laser Scanning, UAV Laser Scanning, Airborne Laser Scanning, Carbon storage, Forest structure, Ecology, Forest monitoring

Traditionally, assessments of tropical forest coverage have relied heavily on conventional satellite imagery. While these tools provide invaluable data, they often fail to capture smaller or more isolated tree clusters, especially in heterogeneous landscapes where crop fields, secondary growth, and degraded forests intermingle. The result has been a substantial underestimation of tree density and, consequently, inaccuracies in carbon cycling models that hinge on robust biomass estimations.

The research team surmounted these limitations by integrating a high-resolution LiDAR (Light Detection And Ranging) dataset with multispectral and hyperspectral satellite imagery across multiple tropical regions spanning South America, Africa, and Southeast Asia. LiDAR technology, which uses laser pulses to generate three-dimensional representations of forest canopy structure, allowed for discriminating individual tree crowns from the surrounding vegetation with unprecedented precision. By processing terabytes of LiDAR and spectral data through advanced machine learning algorithms, the group achieved fine-grained identification of tree presence down to single specimens previously undetectable by standard methods.

Key to their analysis was the development of a novel computational pipeline that fused disparate data sources to correct for atmospheric interference, topographic distortion, and spectral noise. This approach involved training convolutional neural networks on labeled forest plots to recognize characteristic tree spectral signatures and canopy geometries. Through iterative refinement and cross-validation with in-situ field measurements, the model achieved over 90% accuracy in detecting individual trees across different forest types and successional stages.

Their pan-tropical tree map revealed startling patterns. Contrary to earlier reports suggesting significant net losses in tropical forest area over recent decades, the new data emphasize nuanced trends: while some primary forests experience deforestation, regrowth and natural reforestation processes have resulted in net gains in tree numbers and canopy cover in many degraded regions. These overlooked changes highlight the dynamic mosaic nature of tropical landscapes, challenging the binary forest/non-forest classification that dominates ecological monitoring frameworks.

One particularly novel finding was the prominent role of secondary forests and agroforestry systems in bolstering tropical tree cover. Often dismissed as imperfect substitutes for primary forests, these ecosystems harbor unexpectedly high densities of smaller and younger trees, which contribute substantially to regional carbon sequestration and biodiversity support. The demarcation of such vegetation types as integral components of tropical forest cover offers a paradigm shift in conservation prioritization and policy formulation.

Moreover, the enhanced spatial resolution enabled the detection of “hidden” refuge areas – small forest patches or isolated trees embedded within agricultural matrices that serve as critical habitat islands for wildlife and genetic reservoirs. These refugia play a vital role in maintaining ecological connectivity and resilience under rapid land-use changes caused by human expansion.

The implications of this refined mapping extend deeply into global climate change models. Tropical forests act as Earth’s lung, sequestering vast amounts of CO2. Accurate representation of their biomass dynamics is fundamental to predicting future atmospheric carbon trajectories. Current models that underestimate tree density by excluding smaller or regenerating trees risk skewing climate projections and misinforming mitigation strategies. The study suggests that enhanced integration of high-resolution tree cover data into Earth system models can sharpen predictions about carbon fluxes and feedback loops.

The study also underscores the importance of leveraging artificial intelligence (AI) in environmental monitoring. The intricate task of parsing heterogeneous spectral and structural data to pinpoint individual trees requires computational power beyond traditional statistical methods. The successful deployment of convolutional neural networks in this context paves the way for AI-assisted biodiversity assessments and real-time forest monitoring, which are essential under accelerating climate and anthropogenic pressures.

Notably, this research bridges the gap between technological innovation and practical conservation efforts. The refined tree maps offer actionable insights for governments, NGOs, and land managers seeking to optimize reforestation campaigns, enforce protection laws, and design sustainable agriculture practices that incorporate native tree species. By pinpointing areas undergoing subtle but positive forest transformations, policymakers can allocate resources more efficiently to promote ecosystem restoration.

The research was conducted through a collaborative network spanning institutions across continents, combining expertise in remote sensing, ecology, computer science, and climate modeling. Field campaigns were integral to validate remote sensing outputs and involved extensive ground-truthing to sample tree height, diameter, and species composition in representative plots. This multi-disciplinary effort ensures the robustness and applicability of the resulting pan-tropical tree map.

Despite these advances, the authors acknowledge certain limitations. The mapping at this scale inevitably faces challenges related to temporal resolution—capturing rapid changes such as illegal logging or wildfire damage requires frequent revisits that current satellite missions may not provide. Additionally, differentiating native trees from invasive species remains complex, necessitating further spectral and genetic analyses. The team envisions future integration of drone-based surveys and next-generation satellite sensors to overcome these hurdles.

Looking forward, the study opens avenues for continuous monitoring frameworks that combine remote sensing, AI, and citizen science. Such systems could enable near real-time alerts on deforestation hotspots, illegal logging activities, or successful regeneration, transforming tropical forest conservation from reactive to proactive. The technological blueprint laid out by Liu and colleagues serves as a foundational step towards this vision.

In conclusion, this pioneering research redefines our understanding of tropical forest cover dynamics by uncovering previously hidden tree populations and rejuvenating estimates of forest change. By enhancing the granularity and accuracy of ecological data, it equips scientists, policymakers, and conservationists with better tools to combat biodiversity loss and climate change. The tropical biome, long under the spotlight as a vulnerable yet vital global resource, can be managed with renewed precision and optimism thanks to these innovative methods.

The publication sets a new standard for environmental mapping and exemplifies the power of interdisciplinary collaboration in addressing some of the planet’s most pressing challenges. As climate targets become more urgent worldwide, such breakthroughs in mapping and monitoring will be indispensable for verifying commitments and guiding restoration efforts at scale. Tropical forest ecosystems, acting as carbon sinks and biodiversity hotspots, remain essential allies in the global fight against climate change and ecosystem degradation.

With this refined understanding, humanity better grasps the intricate dance of loss and gain shaping tropical forest landscapes. The revelation of overlooked trees is more than a data correction—it is a hopeful testament to nature’s resilience and the potential for informed stewardship. The challenge now lies in scaling these advances into global policies that preserve and amplify the health and diversity of tropical forests for generations to come.

Subject of Research: Mapping pan-tropical tree cover using high-resolution remote sensing and machine learning to reveal overlooked changes in tropical forests.

Article Title: Mapping previously undetected trees reveals overlooked changes in pan-tropical tree cover.

Article References:
Liu, S., Zhang, J., Wang, L. et al. Mapping previously undetected trees reveals overlooked changes in pan-tropical tree cover. Nat Commun 16, 5561 (2025). https://doi.org/10.1038/s41467-025-60662-z

Image Credits: AI Generated