This paper presents an approach for the identification of material properties and the model calibration of composite sandwich panels. More specifically, this study focuses on the influence of spatial variability of the thickness due to the manufacturing process. The calibration is performed using a classification-based technique referred to as "fidelity maps", which has the ability to treat a large number of correlated responses to match. The material properties of the skin layers of a foam core sandwich panel are to be identified using modal data. In order to check the influence of geometric uncertainty, the thickness is first treated as spatially constant and added to the parameters to estimate. In a second stage a fixed thickness distribution is obtained from measurements. Finally, the thickness is treated as a random field whose features are characterized through proper orthogonal decomposition (POD). The results demonstrate that calibration results might be highly sensitive to uncertainty in the thickness distribution.