<p>Identifying the dynamic characteristics of bridges under environmental influences is critical for effective structural health assessment. While previous studies have investigated various approaches to relate the dynamic characteristics with structural damages, limited studies have considered associated uncertainties and the impact of environmental variations on physical parameters. This research addressed these gaps by developing a combined data-driven uncertainty estimation and finite element (FE) model-based surrogate modeling approach, taking a single-span prestressed concrete girder bridge as an example, to examine the effects of environmental variations and potential damage scenarios on its dynamic characteristics. Vibration data, along with temperature and relative humidity measurements collected over one year, were used to identify the dynamic characteristics of the structure under real-world conditions. A Gaussian Process Regression (GPR) model was developed to predict natural frequencies with associated uncertainties using environmental parameters. In addition, a GPR-based surrogate was developed using an FE model to predict natural frequencies using concrete and bearing stiffnesses influenced by environmental factors and different damage scenarios as inputs. A damage index based on the natural frequencies was formulated, and its effectiveness and limitations in detecting damages were discussed using the surrogate model with various simulated damage cases under environmental variations. The damage index showed its capability, within the range of damages investigated, in detecting 15% reduction in the concrete stiffness at the mid-span of a single girder, 10% reduction in the bearing stiffness, and 5% reduction of all concrete and bearings in the bridge.</p>