AI-Augmented Satellite Altimetry and in-Situ Civil Engineering Data for Coastal Deformation Analysis: A Field Validation Study

Abstract Coastal deformation, driven by sea-level rise, subsidence, and anthropogenic activities, represents a critical challenge for sustainable infrastructure and climate resilience. While satellite altimetry has become an essential tool for monitoring global and regional sea-level variations, its integration with ground-based civil engineering measurements remains underexplored. This study presents an AI-augmented framework for combining multi-mission satellite altimetry data with in-situ civil engineering observations to enhance the precision of coastal deformation analysis. Field validation was conducted using tide gauge records, GNSS measurements, and structural monitoring sensors from selected coastal sites in Europe and Asia, including the North Sea, Mediterranean coastlines, and the Bay of Bengal. Data preprocessing involved bias correction, temporal alignment, and outlier detection. Machine learning models, including convolutional neural networks (CNNs) and random forest regressors, were trained to detect nonlinear correlations between satellite altimetry-derived sea surface height anomalies and local deformation parameters, such as land subsidence rates and seawall displacement. Results demonstrated that the AI-augmented integration improved spatial resolution by nearly 30% and reduced noise levels by up to 25% compared to conventional statistical fusion techniques. Furthermore, validation against independent field datasets confirmed enhanced accuracy in detecting small-scale deformations, particularly in deltaic and harbor regions. The findings highlight the practical benefits of combining geodetic satellite technologies with civil engineering field data through AI, providing a robust decision-support tool for coastal planners and engineers. This research not only addresses gaps in the current literature regarding the joint use of satellite and ground observations but also underscores the potential of AI-driven data fusion to improve coastal risk management strategies. The proposed methodology is scalable, adaptable to diverse coastal environments, and capable of supporting climate adaptation policies in vulnerable regions worldwide.