Sentinel-3 temporal data improves burn scar detection after wildfires
Temporal Modelling for Burn Scars on Sentinel-3
Computer Vision and Pattern Recognition
Summary
Quickly finding burn scars after wildfires helps understand the damage. The authors used satellite images from Sentinel-3, which takes pictures daily in many colors. They found that looking at images from before and after fires together helps catch burn scars better than looking at each picture alone. Also, using just five colors gives results as good as using all twenty-one, making the process simpler.
What this means in practice
- •For emergency response teams: Identify and map wildfire burn scars rapidly for post-fire damage assessment using paired pre- and post-fire Sentinel-3 imagery.
- •For forest management services: Use a simplified five-band Sentinel-3 input to monitor burn scars over time with improved temporal analysis for better resource allocation.
Authors
Luca Barco, Edoardo Arnaudo, Andrea Bragagnolo, Claudio Rossi, Paolo Garza
Abstract
Rapid and accurate burn scar delineation from satellite imagery is essential for post-fire damage assessment. Sentinel-3 OLCI, with daily revisit and 21 spectral bands, suits rapid mapping, yet most pipelines treat acquisitions independently, leaving the pre/post-fire change signal unexploited. We present a dataset of 246 wildfire activations (2016-2025) from the Copernicus Emergency Management Service, with Sentinel-3 OLCI temporally paired acquisitions. We benchmark spatial and temporal (ConvLSTM-augmented) variants of three backbones (U-Net, SegFormer, ConvNeXt-UPerNet) under two input modes and spectral configurations. Temporal modeling improves segmentation only when pre-fire frames are included, and a 5-band subset matches the full 21-band OLCI configuration.