Remote Sensing for Assessing Vegetation Fuels in Wildfire Risk and Propagation Modelling

Wildfires are becoming increasingly frequent and severe due to climate change and land use changes, posing critical threats to ecosystems, infrastructure, and public safety. Accurate and timely assessment of vegetation characteristics— particularly fuel mass, type, and moisture content—is essential for effective wildfire prediction, simulation, and management. This paper presents an integrated approach combining satellite remote sensing and UAV-based data collection to improve the estimation of key vegetation parameters relevant to fire behaviour modelling. UAVs equipped with RGB and LiDAR sensors are used to derive high-resolution canopy volume, biomass, and tree height, while multispectral satellite imagery from Sentinel-2 supports large-scale vegetation type classification and moisture monitoring using spectral indices such as NDVI and NDWI. Machine learning techniques are applied to both UAV and satellite datasets for vegetation classification. The study also explores the integration of these datasets into fire propagation models, highlighting their potential to enhance fire spread predictions and support decision-making in wildfire prevention and response. The methodology is evaluated for its scalability, accuracy, and feasibility in real-world fire-prone landscapes.