Abstract: The thermospheric density plays a crucial role in the space environment and is closely related to the atmospheric drag in low-orbit vehicles. Accurately calculating atmospheric density can reduce errors in precise orbit determination and orbital prediction for spacecraft. The thermospheric density was simulated using empirical models (MSIS2.0, DTM2020, JB2008) and physical model TIEGCM2.0 under historical severe, major and minor storms. The simulated density among each model was evaluated by comparing the in-situ satellite observations from CHAMP and GRACE. The simulation results indicate that: The empirical models show better performance than the physical model TIEGCM2.0. MSIS2.0 and JB2008 overestimate the atmospheric density by 28%~44% and 13%~30% compared with satellite observations, respectively. DTM2020 exhibits the smallest differences ranging from-16% to 11%.During geomagnetic quiet or unsettled periods, DTM2020 shows the best performance with a normalized root mean square error of 0.3~0.6. However, JB2008 exhibits the best simulations in geomagnetic active or stormy periods, the correlation coefficient between simulated and observed values is about 0.9. Compared to the TIEGCM with 5° resolution, using 2.5°spatial resolution in TIEGCM can reduce the relative error from 20% to 17%. These results suggest that different models have varying applicability in different space environments. In space engineering applications, considering a combination of multi-model results may improve the accuracy of density and orbit determination.

Key words:  , thermosphere, atmospheric density, geomagnetic storm, empirical model, MSIS, TIEGCM