Within the SingleTree project, KOKO Forest has been developing remote sensing methods to monitor forests and detect individual dead trees across different Living Lab areas. The company’s work focuses on building and testing deep learning-based detection models in environments that vary considerably in forest structure, terrain, and environmental conditions. These diverse settings provide an excellent opportunity to evaluate how well dead tree detection methods perform across different forest types, while also exploring how the resulting data can support more informed forest management and decision-making in the future.
High-resolution satellite imagery has been one of the project’s primary data sources. With spatial resolutions below one metre, satellite data enables large forest areas to be monitored consistently, efficiently, and over time. Within the Living Lab framework, this allows forest health and landscape change to be analysed across multiple regions using a shared and scalable approach. The main advantage of satellite imagery lies in its ability to deliver continuous, landscape-level insights across extensive areas.
At the same time, the effectiveness of satellite-based monitoring depends on several environmental and contextual factors. Variations between Living Lab areas can influence model performance and the level of detail that can be reliably captured. Forest density, tree species composition, seasonal conditions, and lighting all affect detection accuracy and interpretation.
Image: Swiss mountain forest imagery. Left: False-colour near-infrared (SkySat) Swiss forest imagery. Upper-Right: True-colour orthophoto; purple outlines show study boundaries. Lower-Right: Detected dead trees (green segments).
Rather than being limitations, these differences highlight the importance of understanding where satellite-based methods are most effective and how they can complement other data sources. In some situations, finer structural information may require additional datasets, while satellite imagery continues to provide a strong foundation for consistent, large-scale forest monitoring and change detection.
From KOKO Forest’s perspective, “satellite imagery offers a unique combination of large-scale comparability and tree-level insight, while also enabling more robust modelling across diverse forest environments”.
