Improving Heterogeneous Forest Height Maps by Integrating GEDI-Based Forest Height Information in a Multi-Sensor Mapping Process

Forests are one of the key elements in ecological transition policies Europe. Sustainable forest management is needed order to optimise wood harvesting, while preserving carbon storage, biodiversity and other functions. Forest managers public bodies need improved cost-effective monitoring tools. Research studies have been carried out assess use optical radar images for producing height or biomass maps. The main limitations quantity, quality representativeness reference data model training. Global Ecosystem Dynamics Investigation (GEDI) mission (full waveform LiDAR on board International Space Station) has provided an unprecedented number canopy samples from 2019. These could be used improve datasets. This paper aims present validate a method estimating dominant open access satellite (Sentinel-1, Sentinel-2 ALOS-2 PALSAR-2), then GEDI replace field measurements calibration. Our approach combines image features measurements, metrics, Support Vector Machine regression algorithm, with feature selection process. tested mixed uneven-aged broadleaved coniferous forests France. Using training, cross-validation shows 7.3 11.6% relative Root Mean Square Error (RMSE) depending class. When using metrics instead errors increase 12.8–16.7% RMSE. level error remains satisfactory; allow production maps large areas better sample representativeness. Future work will focus confirming these results new study sites, improving filtering processing data, at regional national scale. resulting help resource inventories, as well scientists integrate cartographic into climate models.