Ruidas et al. (2026) Enhancing Flood Prediction Accuracy Through LSTM-CNN Fusion Model with Satellite Imagery and Weather Data

Identification

• Journal: Lecture notes in networks and systems
• Year: 2026
• Date: 2026-01-01
• Authors: Amit Ruidas, Harshit Kumar Sahu, Asim Kumar Mahadani, Pabitra Pal
• DOI: 10.1007/978-981-96-9239-2_26

Research Groups

• Department of Computer Science and Engineering, IIT Bhilai, Durg, Chhattisgarh, India
• Department of Computer Science and Engineering, Bankura Unnayani Institute of Engineering, Pohabagan, Bankura, WB, India
• Department of Computer Applications, Maulana Abul Kalam Azad University of Technology, Kolkata, WB, India

Short Summary

This paper proposes a CNN-LSTM fusion model to predict flood levels in already flooded areas, integrating satellite imagery, weather data, and elevation data. The model achieved a 90% accuracy on testing data, demonstrating its potential for flood management.

Objective

• To enhance flood prediction accuracy by developing a machine learning-based approach that predicts flood levels in already flooded areas.

Study Configuration

• Spatial Scale: Already flooded areas in India.
• Temporal Scale: Not explicitly stated, but implies short-to-medium term flood level prediction.

Methodology and Data

• Models used: Convolutional Neural Network (CNN) combined with Long Short-Term Memory (LSTM) networks (CNN-LSTM fusion model).
• Data sources: Image data from the ISRO Bhuvan website (satellite imagery), weather data based on geographic coordinates, and elevation data.

Main Results

• The proposed CNN-LSTM fusion model achieved an accuracy of 98% on the training data.
• The model demonstrated a testing accuracy of 90% for flood level prediction.

Contributions

• Development of a novel CNN-LSTM fusion model specifically designed for flood level prediction.
• Effective integration of diverse data sources, including satellite imagery, weather data, and elevation data, to improve prediction accuracy.
• Demonstration of high predictive performance (90% testing accuracy) for flood levels in real-world scenarios.
• Provides a valuable tool with potential to significantly aid in flood management and mitigation efforts.

Funding

• Not explicitly mentioned in the provided paper text.

Citation

@article{Ruidas2026Enhancing,
  author = {Ruidas, Amit and Sahu, Harshit Kumar and Mahadani, Asim Kumar and Pal, Pabitra},
  title = {Enhancing Flood Prediction Accuracy Through LSTM-CNN Fusion Model with Satellite Imagery and Weather Data},
  journal = {Lecture notes in networks and systems},
  year = {2026},
  doi = {10.1007/978-981-96-9239-2_26},
  url = {https://doi.org/10.1007/978-981-96-9239-2_26}
}