<p>In VLEO, the continuum assumption breaks down requiring the use of the DSMC method to simulate the gas flow around satellites. However, DSMC simulations relies on simplified GSI models, such as the Maxwell model or the Cercignani–Lampis model, which are based on constant accommodation coefficients. Instead, these coefficients are variable, influenced by multiple factors, which makes their accurate determination challenging. Furthermore, these models are using simplifying assumptions, such as superposition of specular and diffuse reflections, or independent scattering of the normal and tangential components of the velocity. Implementing a high precision GSI model enables to optimize the aerodynamics of VLEO satellites and to design efficient intakes for atmospheric breathing propulsion systems vastly enhance mission planning and fuel requirement calculations, ultimately extending operational lifetimes and reducing costs. We present an approach that integrates MD simulation data into DSMC through a scattering kernel modeled as a Gaussian mixture conditional probability density function. Preliminary tests on synthetic data derived from the Cercignani–Lampis model demonstrate the model’s capability to accurately predict reflected velocities. Additional improvements are needed to enhance the model’s performance in interpolation and extrapolation of unknown incoming velocities in the future.</p>