For a long time, finite element analysis was widely used as an analytical tool to predict the modal parameters of large civil structures such as frequencies and mode shapes. However, their predicted responses normally do not correlate well with those observed because of the physical uncertainties inherent in the boundary conditions, and material properties that are engaged during the mass and stiffness estimations in the FE model. A modal parameter- or an FRF-based finite element model updating procedure can effectively address the aforementioned issue by varying the structural parameters based on the optimization theory, resulting in a highly reliable FE model. This paper demonstrates how a numerical or mathematical model can be iteratively updated to closely match with the global responses of the physical model. In this study, a slender arch structure for green houses is considered as a test specimen. Using an impulse hammer, an input-output modal test is performed, and its natural frequencies and mode shapes are extracted using an advanced system identification, called PolyMAX. Additionally, the test specimen is numerically modeled using ANSYS and a common in-house program, which can be directly exported to a FE model updating tool. The modal parameters or FRFs extracted from a series of vibration records are used as the reference data for updating the initial finite element model by varying the stiffnesses of the members and base or connection springs in a 3-dofs. Finally, it is found that the updated finite element model shows sufficient improvement in the correlation between the test and analysis results in terms of the modal parameters or FRFs, while the latter case scatters more.