The freeze-thaw and de-icing salt resistance of cementitious materials is heavily influenced by the arrangement of aggregates and the resulting Interfacial Transition Zone (ITZ), particularly in the near-surface boundary layer. This study presents a combined experimental and numerical approach to investigate the relationship between aggregate size, spatial distribution, and local surface scaling intensity. Experimental investigations were conducted on mortar specimens with constant cement paste volume but varying quartzitic aggregate fractions (ranging from 0.67-2 mm up to 8-11 mm). Following the standard laboratory testing protocol for freeze-thaw and de-icing salt expositions (CDF), surface scaling was quantified using the RAPTOR method. In contrast to traditional weighing of surface scaling, RAPTOR utilizes high-resolution 3D laser scanning point clouds to generate detailed damage topographies and depth distributions. The results confirm that surface scaling depth increases proportionally with increasing aggregate size. To interpret these findings mechanically, the internal geometry of the investigated samples was recreated using the Distribution-Optimized Mesostructure Estimation (DOME) method. DOME is a computationally efficient algorithm that, in contrast to many random packing models, explicitly simulates the wall effect, which results in the accumulation of finer particles in the near-surface formwork boundaries. The study correlates the simulated aggregate distributions in these near-surface zones with the empirical surface scaling depth distributions obtained from RAPTOR. By comparing the specific aggregate packing at the surface with the localized damage profiles, the effective aggregate sizes most relevant for the surface scaling process are characterized. The results demonstrate that the specific mesostructural configuration of the edge zone, as predicted by DOME, is a critical factor in the development of freeze-thaw and de-icing salt induced damage, offering a robust tool for evaluating material performance.