This study discusses the application of data-driven approaches based on Dynamic Mode Decomposition (DMD) for real-time digital twinning of complex dynamical systems, with specific applications in three domains: naval ship motions, floating offshore wind turbines (FOWTs) performance, and urban dynamics. In the maritime domain, the Bayesian Hankel-DMD (BH-DMD) is utilized to forecast ship motions under wave-induced conditions [1]. This variant of DMD integrates probabilistic forecasting, crucial for digital twins where reliability is paramount. It demonstrates robust accuracy, capturing the course-keeping performance of naval vessels in irregular waves by dynamically updating models with minimal data requirements. Compared to deterministic DMD, the Bayesian approach enhances both accuracy and robustness by quantifying prediction uncertainty in real-time, meeting essential criteria for operational decision support and long-term system resilience. The second application addresses the digital twinning of FOWTs, which operate in highly stochastic marine environments [2]. Due to wind and wave interactions, FOWTs experience complex dynamic loads, which pose challenges to energy efficiency and structural stability and integrity. Here, BH-DMD is applied to forecast the dynamics of a floating turbine prototype, specifically focusing on energy output and load distribution. BH-DMD incorporates variability in algorithm hyperparameters, allowing it to capture the nonlinear nature of marine conditions more robustly than standard DMD approaches. Results show a notable improvement in predictive accuracy, up to 10% in normalized root mean square error, supporting enhanced energy yield and reduced maintenance costs for FOWTs through informed predictive control and load mitigation. Finally, in the context of urban systems, DMD’s application extends to analysing and forecasting urban occupancy and mobility patterns, using anonymized data from mobile network cells. The urban digital twin decomposes dynamic occupancy data, providing insights into population distributions, movement trends, and temporal patterns critical for city management. The decomposition reveals periodic cycles and helps predict occupancy shifts, thus aiding in infrastructure planning, resource allocation, and real-time emergency management. DMD’s computational efficiency and flexibility make it well-suited to handle the high-dimensional data typical of urban dynamics, facilitating rapid responses to [...]