This thesis presents ROOTS, a vibration-based structural health monitoring framework for urban trees of the genus Platanus, the most prevalent and, in aggregate, highest-consequence species in the urban forest of Torino. To overcome the absence of continuous, high-frequency field data, a physics-grounded digital twin was developed to synthesise a population of 1,000 virtual trees across a five-year window, embedding tree biomechanics, canopy reconfiguration, temperature-dependent stiffness, and non-linear root-soil decay. An automated Operational Modal Analysis (OMA) pipeline, built on PyOMA2 and combining Stochastic Subspace Identification with Enhanced Frequency Domain Decomposition, extracts natural frequency, damping ratio, and residual offset from raw, output-only acceleration streams; its robustness was cross-validated on the independent Manitou reference dataset.