High-resolution remote sensing of heathland vegetation using multi-sensor and multi-season data fusion to support habitat conservation monitoring
Quantifying the distribution and abundance of plants is of fundamental importance to plant ecology and conservation biology. Recently, UAV (Unmanned Aerial Vehicles) technologies have evolved rapidly, allowing to cover of large areas with centimetr-level image resolution. UAVs are thus a new asset in the toolbox of researchers and managers to monitor the spatial distribution of plants and, thanks to time series, their evolution over time. I will present here a research project we are working on in the "Vallée du Canut' Natural 2000 site, aiming to test the potential for UAV data to assess the vegetation conservation status of heathland habitats. In detail, we fed a machine learning algorithm for supervised vegetation classification with multispectral and photogrammetric data collected in 2021. The random forest model prediction was used to assess the vegetation conservation status at 20 meters resolution. Mapping products might support management decisions.