Journal article
A general and robust active fire detection method using Sentinel-2 and Landsat shortwave infrared measurements
IEEE transactions on geoscience and remote sensing, Vol.64, 4100314
2026
DOI: 10.1109/TGRS.2025.3647717
Abstract
Accurate detection of active fires (AFs) is crucial for improving estimates of biomass burning emissions and environmental management. Satellite remote sensing at shortwave infrared (SWIR) bands with 10–30 m spatial resolution has great potential in identifying the numerous small AFs missed by fire-monitoring systems using coarse spatial resolution data, but is challenging due to heterogeneous backgrounds and multiple interferences. By combining the enhanced threshold method and abundant prior information, we developed a generalized and robust fire detection algorithm using the daytime Sentinel-2 SWIR observations over complex surfaces in China. To discriminate with high reflecting surface targets such as rooftops and solar panels, more thresholds with visible and SWIR bands, along with land cover datasets, are used. Moreover, industrial thermal anomalies are excluded by time series satellite images. Validations of Sentinel-2 AF detection through visual inspection show very low commission rates (<1%–3%) over selected regions of the world. Since burned areas in fire boundaries with higher temperature than the background were taken as AFs, the average omission rates increase to ~3%–4%. A maximum omission rate of ~7%–8% occurs in African grassland, which has a short burning time. In spite of consistent spatial patterns, 1 km moderate resolution imaging spectroradiometer (MODIS) and 375 m visible infrared imaging radiometer Suite (VIIRS) detections performed in the medium infrared (MIR) bands missed more than 90% of Sentinel-2 AFs. With the flexible framework, our method also has robust performance in similar Landsat 8/9 SWIR observations. By combing Sentinel-2 2B/2C and Landsat 8/9 observations, our reliable AF detection at fine scales (20–30 m) can provide a significant supplement for the biomass burning emission estimation with a temporal resolution of 2–3 days.
Details
- Title: Subtitle
- A general and robust active fire detection method using Sentinel-2 and Landsat shortwave infrared measurements
- Creators
- Lu Gui - China University of GeosciencesRong Li - Hubei UniversityMinghui Tao - China University of GeosciencesZhe Jiang - Institute of Atmospheric PhysicsLunche Wang - China University of GeosciencesJun Wang - University of IowaJiaxin Huang - China University of GeosciencesYi Wang - China University of Geosciences (Beijing)Liangfu Chen - State Key Laboratory of Remote Sensing ScienceSanto V. Salinas - National University of Singapore
- Resource Type
- Journal article
- Publication Details
- IEEE transactions on geoscience and remote sensing, Vol.64, 4100314
- DOI
- 10.1109/TGRS.2025.3647717
- ISSN
- 0196-2892
- eISSN
- 1558-0644
- Publisher
- IEEE
- Grant note
- National Natural Science Foundation of China: 42171354, 42271382 CAS Project for Young Scientists in Basic Research: YSBR-037
This work was supported in part the National Natural Science Foundation of China under Grant 42171354 and Grant 42271382 and in part by CAS Project for Young Scientists in Basic Research under Grant YSBR-037.
- Language
- English
- Electronic publication date
- 12/23/2025
- Date published
- 2026
- Academic Unit
- Electrical and Computer Engineering; Civil and Environmental Engineering; Physics and Astronomy; Chemical and Biochemical Engineering
- Record Identifier
- 9985112980502771
Metrics
21 Record Views