Understanding how vegetation changes over time is essential for managing forests, assessing carbon stocks, detecting illegal clearing, and protecting biodiversity. At Geoneon, we have developed a method to monitor forest clearing by comparing vegetation height models derived from satellite imagery over time.
In this post, we demonstrate how our technology tracks vegetation loss and volume cleared using Sentinel-2 satellite data and validated with Maxar’s very-high resolution imagery.
Deforestation and land-use change are among the leading contributors to biodiversity loss and carbon emissions. But traditional monitoring tools like NDVI or land cover classification often miss the full picture—especially when the structure and volume of vegetation are key.
That’s where Geoneon’s approach comes in: by estimating vegetation height from satellite imagery, we can measure how much vegetation has been cleared, not just where.
To create our vegetation height datasets, we apply a deep learning model to Sentinel-2 imagery at 10m resolution. These models are trained to recognise patterns in 3-band (Red-Green-Blue) imagery that correspond to known vegetation heights.
To ensure consistency and avoid cloud interference, we generate cloud-free mosaics using three months of imagery. Read more about how our vegetation height model works in this earlier post.
By comparing datasets, we can detect where vegetation height has dropped—and estimate the volume cleared, thanks to the availability of vertical structure.
Figure 1: Large Scale Vegetation Loss
We begin by mapping the vegetation height. Areas where vegetation height has decreased are highlighted in shades of yellow to red, indicating the volume of vegetation removed.
This map allows users—such as forest managers, government agencies, or carbon offset auditors—to quickly pinpoint where clearing has occurred across large areas.
Figure 2: Zooming In on a Clearing Area
Next, we zoom in to a specific area where a significant clearing event was detected. Using Sentinel-2 imagery, we show the difference before and after clearing generated from our height maps, clearly illustrating the change in vegetation structure over time.
To validate and further analyse this clearing event, we used very-high resolution satellite imagery from Maxar Technologies. As an official integrator partner of Maxar, Geoneon has access to Analysis-Ready Data (ARD) that enables enhanced processing and comparison.
Figure 3: Maxar Before and After Imagery (Imagery © 2025 Maxar Technologies. Used with permission.)
In these images, you can see the exact same location before and after clearing, with sub-metre resolution revealing roads, tree shadows, machinery tracks, and cleared patches in fine detail.
While Sentinel-2 allows for scalable, regional analysis, Maxar imagery gives us the precision to quantify change at the individual tree or plot level.
Figure 4: Geoneon Height Model Applied to Maxar Imagery
We ran our vegetation height algorithm on the Maxar data, enabling sub-metre resolution height estimation before and after the clearing. From this, we calculated:
This level of precision is invaluable for regulatory compliance, carbon accounting, and detailed change assessment.
Figure 5: Area and Volume Removed. (Imagery © 2025 Maxar Technologies. Used with permission.)
This consistent approach allows us to build a dynamic view of vegetation change over time, offering a scalable and reliable solution to environmental monitoring.
Interested in accessing this data or applying it to your region or project?
Contact us to discuss applications in forest monitoring, carbon accounting, compliance, and conservation.
Subscribe to our newsletter for updates like this one and new case studies.