This article describes the process of developing a digital twin aimed at being used for the predictive maintenance of scour on the Espartxo Bridge in San Sebastian, which is continuously monitored in real time. The main objectives of the research are to describe a methodology for creating the digital twin and to establish an effective strategy for the early identification of scour in bridge foundations, which pose a significant risk of structural damage. The developed methodology is based on hydraulic, geotechnical, and structural analyses, combining experimental tests on the bridge with the development of numerical models. As a result of the research, a biunivocal correlation is established between the variation of a specific frequency and the depth of scour, and the analysis of horizontal vibration modes is determined to be an effective strategy for detecting scour.