Detection and analysis of forest fire damaged areas using Sentinel-2 imagery in Gyeongsangbuk-do Province, Korea
Mohamed, S.Y.1, Yoon, H.S.2, Lee, S.Y.3, Choung Y.J.4, and Jo M.H.*5

1Researcher, Geo C&I Co., Ltd., South Korea
2Researcher, Geo C&I Co., Ltd., South Korea
3Emerits professor, Geo C&I Co., Ltd., South Korea
4Researcher, Geo C&I Co., Ltd., South Korea
5Director, Geo C&I Co., Ltd., South Korea

Abstract

Forest fires pose substantial threats to both ecological systems and economic stability, particularly in regions with vulnerable natural environments. The ability to accurately and rapidly assess the extent of fire damage is essential for implementing timely and effective recovery strategies. This study aims to detect and map forest fire damage in Gyeongsangbuk-do Province, Korea, where significant fire activity occurred between March 22 and March 29, 2025. To achieve this, high-resolution Sentinel-2 satellite imagery was utilized, capitalizing on its spectral capabilities and frequent revisit times.
The methodology combined supervised classification techniques with the difference Normalized Burn Ratio (dNBR), a widely accepted index for identifying burn severity and extent. Supervised classification was applied to distinguish burned and unburned areas based on spectral signatures. The dNBR was then calculated by subtracting the post-fire NBR values from the pre-fire NBR values.
The analysis revealed that the estimated burned area was 690.19 km2 based on supervised classification and 774.31 km2 using dNBR. Compared to the actual damaged area reported by the Korea Forest Service (902.89 km), the dNBR-based estimate more closely approximates the true extent of the fire-affected region. This suggests that dNBR offers higher accuracy in capturing the full spatial extent of burn severity. While supervised classification is useful for delineating obvious fire damage.
Overall, the study demonstrates that the use of freely available satellite data, combined with well-established image processing techniques, offers a cost-effective and efficient solution for forest fire damage assessment. This methodology can be applied in other forest fire-prone regions to support environmental monitoring, disaster response, and long-term ecological recovery planning.

Keywords: Difference Normalized Burn Ratio (dNBR), Forest fire damage assessment, Gyeongsangbuk-do Province, Sentinel-2 imagery, Supervised classification

Topic: Topic B: Applications of Remote Sensing