FORMSpoT provides annual forest canopy height maps at
1.5 m spatial resolution over metropolitan France for the period
2014–2024, derived from SPOT-6/7 satellite imagery.
FORMSpoT-Δ complements these maps with annual forest disturbance polygons, enabling tree-level monitoring of forest dynamics across the country.
The canopy height maps were generated using a hierarchical vision transformer model (PVTv2) trained on co-registered SPOT-6/7 imagery and Airborne Laser Scanning (ALS) reference data from the French LiDAR HD program. The resulting annual height time series was further processed to ensure temporal consistency before deriving disturbance polygons based on year-to-year height losses.
Input Data
- SPOT-6/7 satellite imagery was obtained from the annual, cloud-free mosaics distributed by the DINAMIS platform and used as predictors to produce the height maps. SPOT-6/7 data is available here : https://openspot-dinamis.data-terra.org/.
- Airborne Laser Scanning (ALS) data from the IGN LiDAR HD program were used exclusively for model training and independent validation. LiDAR HD data is available here : https://diffusion-lidarhd.ign.fr/.
Computing Resources
Computational resources were provided by IDRIS under the HPC allocations 2024-AD010114718 and 2025-AD010114718R1 granted by GENCI.
Products
The dataset is organized into two complementary products:
- FORMSpoT: Annual canopy height maps at 1.5 m resolution over France (2014–2024), provided as Cloud-Optimized GeoTIFFs (COGs). Heights are encoded in decimeters (0.1 m).
- FORMSpoT-Δ: Annual forest disturbance polygons (2014–2024) derived from the FORMSpoT height time series.
Reference
For a detailed description of the methodology, validation, and applications, please refer to the associated publication
Schwartz & al., 2025[1].
