AUTHOR=Zitouni M. Sami , Elneel Leena , Assad Albakri Naseeb , Alkhatib Mohammed Q. , Al-Ahmad Hussain 

TITLE=Forecasting sea level rise using enhanced deep learning models

JOURNAL=Frontiers in Environmental Science

VOLUME=Volume 13 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/environmental-science/articles/10.3389/fenvs.2025.1552834

DOI=10.3389/fenvs.2025.1552834

ISSN=2296-665X

ABSTRACT=Climate change has significantly impacted vulnerable communities globally, with rising temperatures caused by greenhouse gas emissions accelerating global Sea Level Rise (SLR), threatening coastal infrastructure and ecosystems. This study evaluates statistical and deep learning models, including the Autoregressive Integrated Moving Average (ARIMA) and Long Short-Term Memory (LSTM) networks, for predicting SLR and visualizing potentially inundated areas in the United Arab Emirates (UAE) via an interactive web interface. Historical mean sea level (MSL) data from the National Oceanic and Atmospheric Administration (NOAA), spanning 1992 to 2024, were used for training and model evaluation. An interactive web platform was developed to visualize forecasted inundation areas and support decision-making. The enhanced LSTM model, integrated with a Squeeze-and-Excitation (SE) block, achieved the highest accuracy, with a Root Mean Square Error (RMSE) of 2.27, representing an improvement of 8.81% over the standalone LSTM (RMSE 2.47) and 13.66% over ARIMA (RMSE 2.58). The model forecasts sea level changes up to 2100, highlighting critical risks for low-lying coastal regions such as Umm Al Quwain, Abu Dhabi, and Dubai. The findings underscore the value of advanced AI-driven forecasting in enhancing climate resilience, assisting policymakers and urban planners in risk assessment, optimizing emergency response strategies, and implementing coastal adaptation measures. Future work should integrate additional environmental factors influencing MSL.