Evaluating Near-Real Time Thermospheric Density Retrieval Methods from Precise Low Earth Orbit Spacecraft Ephemerides During Geomagnetic Storms

TL;DR

This paper compares two near-real-time thermospheric density retrieval methods during geomagnetic storms, demonstrating that the POD accelerometry method outperforms the EDR method and some models, offering a promising tool for satellite operations.

Contribution

It evaluates and demonstrates the superior performance of the POD accelerometry method over EDR and certain models for near-real-time thermospheric density estimation during storms.

Findings

POD accelerometry surpasses EDR and DTM2000 in accuracy.

POD accelerometry is comparable to JB2008 and NRLMSISE-00.

The method shows potential for operational satellite and thermosphere modeling.

Abstract

Characterizing the density of the thermosphere during geomagnetic storms is critical for both thermosphere modelling efforts and satellite operations. Accurate near-real time density estimates can feed into data assimilation schemes and provide operators with an early warning system for storm-triggered drag increases. This study evaluates two methods for generating near-real time thermospheric density estimates: the Energy Dissipation Rate (EDR) method and the Precise Orbit Determination (POD)-accelerometry method. Using accelerometer-derived densities from the Gravity Recovery And Climate Experiment Follow-On (GRACE-FO) and Challenging Minisatellite Payload (CHAMP) spacecraft as truth over 45 geomagnetic storms, the POD accelerometry method was found to surpass EDR density retrieval as well as one commonly used atmospheric density model (DTM2000) in terms of mean absolute percentage error (by 113.30\% and 130.64\%, respectively). The POD accelerometry method is comparable, albeit slightly worse, than two other models: JB2008 (-8.74\%) and NRLMSISE-00 (-22.74\%). These results highlight the potential for near-real-time density inversion to rival models driven by post-processed indices, which outperform these same models in an operational setting, where they rely on forecasted or nowcasted indices. By applying the POD accelerometry method along the orbits of three LEO satellite orbits during 80 geomagnetic storms (2001--2024), this study illustrates the potential of POD accelerometry as a near-real-time resource for the thermosphere and satellite operations community. The accompanying codebase facilitates broader adoption of these techniques, advancing both storm-time modelling and operational response capabilities.