A Survey of Flood and Landslide Prediction Using Machine Learning

Floods and landslides rank among the most devastating natural disasters, leading to widespread damage to lives, properties, and essential infrastructure worldwide. Accurate and early prediction of these disasters is critical to mitigating their impacts. However, traditional prediction methods often fall short due to the complexity and variability of environmental factors. Emerging innovations in machine learning (ML) have opened new avenues for effectively tackling these challenges, leveraging advanced algorithms to address environmental complexities. This paper provides a comprehensive review of state-of-the-art ML techniques for flood and landslide prediction, focusing on methodologies, datasets, strengths, and limitations. Methods: The methodology is divided into two main sections: article selection and detailed review. A systematic search was conducted in databases such as IEEE Xplore, Scopus, and Google Scholar, covering publications from 2011 to 2023. Selected articles were analyzed based on ML techniques, datasets used, performance metrics, and key contributions. Results: ML models have demonstrated high accuracy in flood and landslide prediction. Techniques like convolutional neural networks (CNNs), and ensemble methods have been widely applied. Transfer learning and hybrid models reduce training time and enhance model robustness. However, challenges like data scarcity, lack of real-time data, and computational requirements remain. Keywords: Flood prediction, Landslide prediction, Machine learning, Convolutional Neural Networks (CNNs)