Assessing Burn Severity Indexes for Fire Mapping using Multispectral Satellite Imagery

Project Assignment of the Earth Observation course at Politecnico di Milano.

The main Google Earth Engine source code is avaiable at this link.

The complete paper can be found in the report file.

Accurate burn severity assessment is crucial for understanding fire behavior, ecology, and post-fire management. While NASA Landsat satellites are commonly used, the newer European Space Agency Sentinel-2 satellites provide higher temporal and spatial resolution globally and free data access. Burn severity estimates was compared from Landsat and Sentinel in four different regions in western North America and South Europe affected by wildfires. Sentinel often outperforming Landsat, utilizing Sentinel’s 20 m resolution reduced high-severity patch interior mapping compared to Landsat (30 m), indicating improved fine-scale fire effect detection. However, caution is advised when comparing severity patterns at different resolutions. Integrating Sentinel imagery enhances burn severity mapping by increasing availability and enabling finer-scale fire effect analysis. In the end, the accuracy assessment of the estimated burned area from Sentinel 2 was performed using the newly integrated QGIS plugin BAD (Burn Area Detector) as the ground truth.