Bhosale et al. (2025) A Hybrid Framework for Forest Fire Detection and Severity Prediction using Sequential Deep Learning on Multitemporal Satellite Imagery

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Springer Link (Chiba Institute of Technology)
• Year: 2025
• Date: 2025-11-20
• Authors: Rohini Bhosale, Poonam Railkar
• DOI: 10.1051/epjconf/202534101057/pdf

Research Groups

[Information not provided in the paper text.]

Short Summary

This study aims to detect and predict forest fires using a deep learning-based hybrid approach applied to multi-temporal satellite images. The proposed model, combining change detection, LSTM, and attention mechanisms, demonstrates high accuracy in identifying fire-prone zones and providing early warnings, particularly during the pre-monsoon period and in protected areas.

Objective

• To detect and predict forest fires based on multi-temporal images captured by satellites.

Study Configuration

• Spatial Scale: Regional, focusing on surveyed areas with high concentrations of protected zones.
• Temporal Scale: Multi-temporal, analyzing seasonal patterns (pre-monsoon, especially March) for early detection and prediction.

Methodology and Data

• Models used: Deep learning-based hybrid approach combining change detection, Long Short-Term Memory (LSTM) networks, and an attention mechanism.
• Data sources: Multi-temporal satellite images, pre-processed for analysis.

Main Results

• The majority of forest fires were observed to occur during the pre-monsoon period, specifically in March.
• High-risk areas for fires were identified in regions with the largest concentration of protected zones.
• The proposed deep learning hybrid model achieved high accuracy in detecting and predicting forest fires, outperforming traditional methods.
• The model aids in early warning systems and supports decision-making for fire management authorities.

Contributions

• Proposes a novel deep learning-based hybrid approach for forest fire detection and prediction, integrating change detection, LSTM, and attention mechanisms.
• Demonstrates a robust system that leverages remote sensing and deep learning to achieve high accuracy, surpassing traditional methods.
• Provides a framework for identifying and prioritizing fire-prone zones, essential for mitigating the impact of forest fires.

Funding

[Information not provided in the paper text.]

Citation

@article{Bhosale2025Hybrid,
  author = {Bhosale, Rohini and Railkar, Poonam},
  title = {A Hybrid Framework for Forest Fire Detection and Severity Prediction using Sequential Deep Learning on Multitemporal Satellite Imagery},
  journal = {Springer Link (Chiba Institute of Technology)},
  year = {2025},
  doi = {10.1051/epjconf/202534101057/pdf},
  url = {https://doi.org/10.1051/epjconf/202534101057/pdf}
}