| Autor: |
McArdle, Sean, Russell, Ryan P. |
| Předmět: |
|
| Zdroj: |
Journal of the Astronautical Sciences; Oct2024, Vol. 71 Issue 5, p1-31, 31p |
| Abstrakt: |
This work is the third in a series on point mascon lunar gravity models. Point mascon models are computationally-efficient replacements for the standard spherical harmonics gravity models used in astrodynamics applications. Weighted cubed-sphere mascon gravity models are introduced as runtime-efficient alternatives to the spherical harmonics representation that do not impose the extreme memory costs imposed by other interpolated gravity modeling schemes. Localized models for the lunar gravity field are generated using sets of point-mass potentials and referenced to a cubed-sphere grid. Adjacent localized point mascon gravity models are combined using Junkins weighting functions to form a smooth global model. Three demonstration models are generated that reproduce the GRGM1200A lunar gravity model with equivalent fidelity to degree 70, 300, and 550 spherical harmonics truncation levels. The weighted cubed-sphere models are benchmarked for runtime and memory cost against the standard spherical harmonics models and two recently-introduced types of globally-defined mascon models. The benchmarking results show that the weighted cubed-sphere mascon models enable significant runtime improvements with reasonable memory costs, especially when using OpenMP parallelization. Comparing acceleration runtime with spherical harmonics, the degree 300 equivalent weighted cubed-sphere model shows up to a 30-fold speedup with an 89 megabyte memory footprint and the degree 550 equivalent model shows up to a 90-fold speedup with a 170 megabyte memory footprint. Order of magnitude speedups are demonstrated without parallelization. The runtime models and driver code are available online. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
Full text from SpringerLink