Adduri et al. (2025) Development of hybrid machine learning and deep learning techniques for sea level rise projection in Dubai

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

Identification

• Journal: Geomatics Natural Hazards and Risk
• Year: 2025
• Date: 2025-12-17
• Authors: Siva Durga Adduri, Tarig Ali, Eman Ahmed AlFalasi, Serter Atabay, Md Maruf Mortula, Rabin Chakrabortty
• DOI: 10.1080/19475705.2025.2601266

Research Groups

Not specified in the provided text.

Short Summary

This study developed and evaluated a hybrid deep learning model (Conv1D-LSTM) for near-term sea level rise (SLR) projection along Dubai's coastline, forecasting an increase in sea levels by 2030 and identifying high-risk areas.

Objective

• To develop and evaluate a hybrid deep learning model combining one-dimensional convolutional neural networks (Conv1D) and long short-term memory (LSTM) networks for near-term projection of sea level rise (SLR) at five major locations along Dubai’s coastline, and to extend these forecasts to 2050.

Study Configuration

• Spatial Scale: Five major locations along Dubai’s coastline (Mamzar, Al Jadaf, Dhaw Wharfage, Umm Sugeim, and one other unspecified location).
• Temporal Scale: Historical data from 2004 to 2019; near-term projection up to 2030; extended forecast up to 2050.

Methodology and Data

• Models used: Hybrid Conv1D-LSTM model, K-Nearest Neighbours (KNN) Regressor, Random Forest (RF) Regressor, Dense deep learning model.
• Data sources: Historical sea level data (2004-2019).

Main Results

• Sea levels are projected to increase at all studied sites by 2030.
• The Mamzar area is projected to experience the highest average annual sea level rise of 0.18 mm per year.
• The Conv1D-LSTM model achieved a projection accuracy with a mean squared error (MSE) of 0.21 and a mean absolute percentage error (MAPE) of 14.39%.
• For extended forecasts to 2050, the Dense deep learning model demonstrated the best performance among the tested methods, achieving a mean squared error (MSE) of 0.2143.
• Al Jadaf, Dhaw Wharfage, and Umm Sugeim are identified as high-risk areas for sea level rise.

Contributions

• Development and validation of a novel hybrid deep learning model (Conv1D-LSTM) for near-term sea level rise projection.
• Provision of specific, localized sea level rise forecasts for major locations along Dubai's coastline up to 2030 and 2050.
• Identification of high-risk coastal areas in Dubai, offering critical insights for sustainable planning and mitigation strategies.

Funding

Not specified in the provided text.

Citation

@article{Adduri2025Development,
  author = {Adduri, Siva Durga and Ali, Tarig and AlFalasi, Eman Ahmed and Atabay, Serter and Mortula, Md Maruf and Chakrabortty, Rabin},
  title = {Development of hybrid machine learning and deep learning techniques for sea level rise projection in Dubai},
  journal = {Geomatics Natural Hazards and Risk},
  year = {2025},
  doi = {10.1080/19475705.2025.2601266},
  url = {https://doi.org/10.1080/19475705.2025.2601266}
}