An AI-Driven Framework for Coastal City Monitoring via Deep Learning and Earth Observation

Xuan Zheng^*

• College of Science, Chongqing University of Technology, Chongqing, China

Coastal urban environments are among the most dynamic and vulnerable ecosystems on Earth, where the intricate interaction between anthropogenic activities and natural variability poses substantial challenges to reliable environmental monitoring. Traditional remote sensing approaches often struggle with limited responsiveness, sparse sensor deployment, and poor adaptability to rapid environmental changes. To address these challenges, we propose an AI-driven framework that integrates deep learning and earth observation data for intelligent coastal city monitoring. Our approach models environmental systems as dynamic spatio-temporal graphs, enabling adaptive learning of evolving patterns influenced by factors such as wind, currents, and pollution. We further introduce a multi-resolution prioritization module that allocates sensing resources efficiently based on predicted uncertainty and environmental variability. This unified system supports high-fidelity monitoring under bandwidth and energy constraints. Extensive experiments on benchmark coastal datasets demonstrate that our framework significantly outperforms existing baselines in prediction accuracy and resource efficiency. The results highlight the potential of combining adaptive graph-based learning with data-driven sensing strategies to advance intelligent environmental monitoring in complex coastal ecosystems. Our model outperforms state-of-the-art baselines by up to 3.2% in F1 Score and 4.0% in AUC on coastal environmental monitoring tasks.