A Survey of Artificial Intelligence-Driven Flood Risk Assessment and Forecasting with Remote Sensing Data

Abstract

Floods are among the most devastating natural disasters, causing significant damage to lives, property, and infrastructure. Traditional flood prediction methods, such as hydrological models and statistical approaches, often struggle with real-time data complexity and environmental variability. Recent advancements in artificial intelligence (AI) and machine learning (ML) have revolutionised flood risk assessment and forecasting by leveraging remote sensing data, IoT networks, and satellite imagery. This survey comprehensively reviews AI-driven flood prediction techniques, highlighting the efficacy of models like Random Forests, Support Vector Machines, and deep learning architectures such as Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs). The study also explores ensemble methods, hybrid models, and metaheuristic optimisations to enhance prediction accuracy. Challenges such as data scarcity, computational constraints, and model interpretability are discussed, along with emerging solutions like quantum machine learning and semi-subjective analytic frameworks. The review underscores the transformative potential of AI in flood management, particularly in regions like India, where climate variability and rapid urbanisation exacerbate flood risks. By synthesising the latest research, this paper provides insights into future directions for improving flood forecasting systems and mitigating disaster impacts.