Teja et al. (2026) Flood Prediction and Forecasting for Nellore City in Andhra Pradesh Using Hybrid Deep Learning Framework

Identification

• Journal: Lecture notes in civil engineering
• Year: 2026
• Date: 2026-01-01
• Authors: B. V. S. Ravi Teja, Subrahmanya Kundapura
• DOI: 10.1007/978-981-95-3775-4_9

Research Groups

National Institute of Technology Karnataka, Surathkal, Mangalore, India

Short Summary

This study develops a hybrid deep learning framework for precise flood prediction and forecasting in Nellore city, Andhra Pradesh, a region prone to flash flooding. By integrating hydrometeorological, geological, and remote sensing data, the hybrid LSTM-CNN model enhances forecasting accuracy, providing a valuable tool for proactive disaster risk reduction.

Objective

• To enhance flood prediction and forecasting accuracy for Nellore city in Andhra Pradesh, a region prone to sudden flash flooding, by leveraging hybrid deep learning frameworks.

Study Configuration

• Spatial Scale: Nellore city, Andhra Pradesh, India, located in the Pennar River Basin.
• Temporal Scale: Utilizes historical data for model training to enable forecasting for imminent flood threats.

Methodology and Data

• Models used: Standalone Long Short-Term Memory (LSTM) model, Standalone Convolutional Neural Network (CNN) model, and a Hybrid LSTM–CNN model.
• Data sources: HydroSHEDS, WRIS, and IMD.
  • Data types: Hydrometeorological data (precipitation, temperature, wind speed, wind direction), geological data (elevation, slope, aspect), remote sensing data (land use and land cover), and historical discharge data.
  • Preprocessing: Lagged features, cyclic transformations, and hyperparameter tuning.

Main Results

• The study demonstrates the significant potential of AI and machine learning to enhance flood forecasting accuracy.
• The developed hybrid deep learning framework, particularly the LSTM-CNN model, is identified as a valuable tool for local authorities.
• The model exhibits adaptability to different types of floods and effectively considers rare occurrences.
• It supports proactive responses to imminent flood threats and contributes to disaster risk reduction efforts in the Nellore city region.

Contributions

• Development and application of a novel hybrid deep learning framework (LSTM-CNN) specifically tailored for flood prediction and forecasting in the high-risk urban environment of Nellore city.
• Integration of a comprehensive set of diverse data types, including hydrometeorological, geological, remote sensing, and historical discharge data, to improve model robustness and accuracy.
• Offers a promising and adaptable alternative to traditional hydrological models, particularly beneficial for regions with limited data resources, by effectively processing large datasets and identifying complex, nonlinear hydrological relationships.
• Provides a practical and valuable tool for local authorities to enable proactive disaster management and enhance climate-driven hydrological change resilience.

Funding

• No specific funding projects or programs were mentioned in the provided text.

Citation

@article{Teja2026Flood,
  author = {Teja, B. V. S. Ravi and Kundapura, Subrahmanya},
  title = {Flood Prediction and Forecasting for Nellore City in Andhra Pradesh Using Hybrid Deep Learning Framework},
  journal = {Lecture notes in civil engineering},
  year = {2026},
  doi = {10.1007/978-981-95-3775-4_9},
  url = {https://doi.org/10.1007/978-981-95-3775-4_9}
}