Mixed Fidelity Shape Models for Efficient Small Body Gravity Modeling

Abstract

The polyhedral gravity model is the de facto standard for computing the gravity field near the surface of small bodies due to their irregular shapes. However, the computational load to compute the gravitational acceleration can increase significantly for high-resolution shape models. In this work, an algorithm for creating mixed-resolution shapes is developed with the goal of getting similar accuracy in the gravitational acceleration while reducing the computational cost. This method is tested for full integration into low-altitude trajectory simulations. The results show that blended models can provide comparable accuracy while improving the computation performance. Additionally, this approach does not require any fitting or estimation process as it only involves creating shape model connections, which is done in a deterministic manner. Test cases with asteroids Bennu and Itokawa demonstrate that the errors in the gravitational accelerations computed with the hybrid models are comparable to expected uncertainties from unknown heterogeneous mass distributions. For precomputed hybrid shape model strategies, we see savings in computational time on the order of 50% compared to using the high-resolution shape model, while getting much lower propagation errors than using only a low-resolution model alone.