The growing number of space objects in low-Earth orbit necessitates accurate orbit predictions to decrease the likelihood of operational disruptions. The challenges in accurately capturing how gas particles interact with the objects’ surfaces result in uncertainties in their aerodynamic coefficients, directly affecting the accuracy of orbital perturbation models. Currently, gas–solid boundary interactions are accounted for by empirical models like those proposed by Sentman and Cercignani-Lampis-Lord. These models have one or two adjustable parameters, typically tuned based on orbital tracking and acceleration data. However, these models are inadequate in accurately representing crucial processes at the gas–solid interface such as multiple reflections, shadowing, and backscattering resulting from the roughness of real surfaces. We propose a new, physics-based gas-surface interaction model that leverages electromagnetic wave theory to incorporate macroscopic effects on the gas particle scattering distribution resulting from surface roughness. Besides better describing the physics of gas-surface interaction, this model’s parameters can be determined by combining ground measurements to characterise the surface roughness and molecular dynamics simulations to specify the atomic-scale interaction. The model is verified for the entire parameter range using a test-particle Monte Carlo approach on a simulated rough surface. In addition, we successfully replicate several experimental results available in literature on the scattering of Argon and Helium on smooth and rough Kapton and Aluminium surfaces. We conclude by demonstrating the model’s effect on the aerodynamic coefficients for simple shapes and comparing these results with those produced with the Sentman and Cercignani-Lampis-Lord models, thereby demonstrating that previously observed inconsistencies between these models and tracking data of spherical satellites can be explained by surface roughness.