Machine learning and deep learning for ground data deformation analytics: a comprehensive critical review

Ground deformation, a pervasive geophysical phenomenon, threatens global infrastructure resilience, environmental sustainability, and socio-economic stability. The growth of remote sensing technologies, particularly Interferometric Synthetic Aperture Radar (InSAR) and dense Global Navigation Satellite System (GNSS) networks, has produced high-resolution spatiotemporal datasets, indicating a paradigm shift from traditional analytical methods toward sophisticated data-driven computational analytics. While Machine Learning (ML) and Deep Learning (DL) algorithms effectively decipher complex patterns in these data cubes, the field lacks a holistic synthesis that categorizes causative factors, maps their interrelationships, and benchmarks research trajectories. This seminal review systematically examines the confluence of ML/DL and ground deformation analytics. The study introduced a novel, tripartite taxonomy for classifying deformation influencers into three categories. A bibliometric analysis spanning 2010 to 2025 was executed on Scopus and Web of Science, yielding an initial corpus of 1,078 documents. This corpus was subsequently refined to 312 publications. A fault tree analysis (FTA) diagram was used to trace the causal pathways from fundamental factors to the ultimate deformation event. This study employs a Fuzzy Analytical Hierarchy Process (FAHP) to mitigate judgmental ambiguity and to implement a Python stack to quantitatively rank the identified factors based on their prevalence in the literature. The findings indicate that “groundwater level fluctuation” is the most extensively investigated factor, followed by “seismic activity” and “precipitation intensity.” The review provides a strategic roadmap for future research, emphasizing the integration of physics-informed neural networks (PINNs), model interpretability, and open-science principles for planetary-scale geohazard mitigation.