Accuracy evaluation and adaptability analysis of NRLMSIS 2.1 model during different geomagnetic disturbances at different orbital heights

Thermospheric density is a critical parameter for understanding the space environment. Despite extensive research and the development of various atmospheric models, their accuracy remains limited, particularly during geomagnetic storms when model deviations can exceed 100%. This study systematically evaluates the performance of the NRLMSIS 2.1 model during five geomagnetic storm events of varying intensity (weak, moderate, strong, severe, and extreme) from 2018 to 2024. The evaluation utilizes thermospheric density data derived from GRACE-FO (at approximately 500 km altitude) and Swarm-C (at approximately 450 km altitude) satellite accelerometers, focusing on the model’s response characteristics at different altitudes and phases of geomagnetic storms. The results reveal significant systematic biases in the NRLMSIS 2.1 model during storm periods. Specifically, the model generally underestimates density peaks during the main phase and overestimates density during the initial phase, and tends to underestimate during the recovery phase due to thermal inertia. Furthermore, the model inadequately captures the high-latitude density enhancements and rapid diffusion processes of atmospheric density at high latitudes. We also find that the discrepancies between the modeled and observed densities become substantially larger during intense and extreme geomagnetic storm events, indicating that the NRLMSIS 2.1 model has difficulty capturing rapid and highly localized storm-time thermospheric responses. Moreover, altitude-dependent analysis reveals that the larger absolute errors at 450 km result from a more severe systematic background overestimation combined with atmospheric density scaling effects, rather than inherently poorer model applicability. In conclusion, the NRLMSIS 2.1 model exhibits considerable limitations in accurately characterizing the magnitude, spatial structure, and temporal delay of thermospheric responses during geomagnetic storms.