Paper Details

Name: Isha Malik Arora, Mandeep Kaur & Kamal Malik
Country: India

Year: 2025
Volume: Volume-12, Issue-2 (July-December)
Page Number: 349-359
DOI: https://doi.org/10.5281/zenodo.17920452

ABSTRACT
Hurricanes are highly destructive tropical systems that pose significant threats to human life,
coastal infrastructure, and global economies, making accurate forecasting an essential
component of disaster preparedness. Because traditional numerical weather prediction models
often struggle with the nonlinear dynamics governing hurricane behaviour, artificial
intelligence (AI) has emerged as a promising approach for improving predictive accuracy.
This systematic review examines advancements in AI-based hurricane prediction by
synthesizing research published across major scientific databases, including Web of Science,
Scopus, and IEEE Xplore, following PRISMA guidelines for study identification, screening,
and selection. The review highlights how machine learning and deep learning models—such
as artificial neural networks, support vector machines, random forests, convolutional neural
networks, LSTMs, and hybrid architectures—have been applied to predict hurricane
formation, track, intensity, and rapid intensification patterns. These models leverage satellite
imagery, atmospheric reanalysis data, oceanographic variables, and historical storm records
to uncover complex spatial and temporal relationships that conventional methods may
overlook. The findings indicate that deep learning approaches, particularly CNN–LSTM
hybrids and transformer-based networks, outperform traditional techniques by capturing both
multi-dimensional spatial features and sequential atmospheric dependencies. Despite these
advancements, several challenges persist, including limited labeled hurricane datasets,
inconsistencies across ocean basins, model interpretability issues, climate change-driven
variability, and difficulty integrating AI systems with operational forecasting frameworks.
Overall, the review demonstrates that AI-driven approaches have substantial potential to
enhance early warning systems, improve risk assessment, and support more informed
decision-making by disaster management authorities. Continued progress will depend on
expanding high-quality datasets, developing physically interpretable models, and integrating
hybrid systems that combine data-driven learning with established meteorological principles.
Keywords :-Hurricane Prediction, Artificial Intelligence, Machine Learning, Deep Learning,
Tropical Cyclone Forecasting, Meteorological Modelling