Journal article · 2025
Remote Sensing-Based Deep Learning Approach for Identifying Burned Forest Areas
Turkish titleYanmış Orman Alanlarının Belirlenmesi için Uzaktan Algılama Tabanlı Derin Öğrenme Yaklaşımı
Çukurova Üniversitesi Mühendislik Fakültesi Dergisi, 40(1), 33–48
- Year
- 2025
- Publication date
- Journal or conference
- Çukurova University Journal of the Faculty of Engineering
- Volume
- 40
- Issue
- 1
- Pages
- 33–48
Abstract
The study maps the burned areas and fire severity associated with forest fires in Samandağ, Hatay, between 5 and 10 September 2020. Sentinel-2 satellite imagery was used to construct an image dataset and train a deep learning model. The resulting model was compared with NBR, dNBR, BAIS2 and dBAIS2 burned-area indices and General Directorate of Forestry fire records. The reported proportional accuracy for identifying burned forest areas in the Samandağ study region is 98.36%.
Keywords
Core method
Deep learning classification from Sentinel-2 imagery, evaluated alongside NBR, dNBR, BAIS2 and dBAIS2 burned-area indices and official fire records.
Data used
Sentinel-2 imagery covering the Samandağ, Hatay study area before and after the 5-10 September 2020 forest fires.
Verified key results
- Reported proportional accuracy in the Samandağ study area: 98.36%
APA citation
Paşaoğlu, R., Arık, A. E., & Emrahaoğlu, N. (2025). Remote sensing-based deep learning approach for identifying burned forest areas. Çukurova Üniversitesi Mühendislik Fakültesi Dergisi, 40(1), 33-48. https://doi.org/10.21605/cukurovaumfd.1665481
BibTeX
@article{PasogluArikEmrahaoglu2025BurnedForest,
author = {Paşaoğlu, Reha and Arık, Ahmet Ertuğrul and Emrahaoğlu, Nuri},
title = {Remote Sensing-Based Deep Learning Approach for Identifying Burned Forest Areas},
journal = {Çukurova Üniversitesi Mühendislik Fakültesi Dergisi},
year = {2025},
volume = {40},
number = {1},
pages = {33--48},
doi = {10.21605/cukurovaumfd.1665481}
}