Empirical Assessment of Storm‐Time Thermospheric Density Inversion Methods From LEO POD Data

Abstract

Thermospheric mass density is one of the largest sources of operational uncertainty for spacecraft in low Earth orbit, particularly during geomagnetic storms. The growing population of Global Navigation Satellite System‐equipped satellites presents a data set of opportunity: precise orbit determination (POD) data streams can be used to estimate thermospheric density along their paths. We present a multi‐timescale benchmark of the two main cooperative density inversion methods, the Energy Dissipation Rate method and POD‐accelerometry, against accelerometer‐derived effective densities from the CHAMP and GRACE‐FO‐A satellites over 49 geomagnetic storms. At orbit‐effective cadence, both methods achieve and a log‐normal scatter , and are statistically indistinguishable in typical‐orbit scatter. However, the Energy Dissipation Rate method tracks storm‐time variations more faithfully and produces fewer large‐error orbits ( 0.992 vs. 0.985; RMS error 12.0% vs. 13.5%), at roughly lower computational cost. Increasing the fit‐span to three orbits reduces to 13%–14% for both methods against matched‐interval accelerometer effective densities, but against unaveraged (10 s) accelerometer data drops sharply beyond about half an orbit. Optimizing the arc length below one full orbit offers an attractive compromise, achieving –0.94 and – against unaveraged accelerometer data. Retrieval accuracy depends strongly on drag‐acceleration magnitude, with RMS errors below 10% where drag exceeds but growing above 30% below . These results provide empirical observation‐error statistics for storm‐time POD‐derived density retrievals, directly relevant to next‐generation assimilative models.