Journal Article

The Thermosphere Was Poorly Predictable Not Only During but Also Before and After the Starlink Storm on 3–4 February 2022

Kotov, D. V., Reznychenko, A. I., Domnin, I. F., Richards, P. G., Goncharenko, L. P., Zhivolup, T. G., Bogomaz, O. V., Reznychenko, M. O.

2025

Overview

Observation‐based simulations of the ionosphere were performed with the NRLMSISE‐00 model for six locations around the globe during 1–9 February 2022, which includes the so‐called Starlink Storm. Unlike other studies, we focused on the magnetically quiet days around the storm. Unexpectedly, the observed values of the F2‐layer peak density were ∼50% larger than the simulated values. We show that this implies that the daytime O density in the thermosphere was systematically ∼30% larger than the NRLMSISE‐00 predicts. Further investigation shows that this discrepancy is not an exclusive feature of the period around the Starlink Storm and a similar problem happens for some periods for different years. It is unclear if the reason is an actual increase of the O density or its underestimation by the model. Resolving this problem is critical for providing accurate predictions of the atmosphere to avoid the degradation of normal operation or even loss of space assets. Plain Language Summary In early February 2022, dozens of Starlink satellites were lost after a moderate magnetic storm which was not expected to be able to cause significant perturbations in the Earth's atmosphere but did that. Numerous investigations showed the Starlink loss happened because of a remarkable increase of density of the upper atmosphere (thermosphere) caused by the storm. Teams of researchers were focused on the storm days 3 and 4 February. We got curious about the conditions in the thermosphere around the storm days. We employed observational data collected by ionospheric radars around the globe and conducted simulations with a physical model of the ionosphere‐thermosphere system, which allowed us to obtain the thermosphere composition. Our results indicate that the density of atomic oxygen in the thermosphere was one third larger than predictions of the standard model of the atmosphere during the periods before and after the storm. This means that the thermosphere was already poorly predictable prior to the storm occurrence. It is possible that this affected the thermosphere behavior during the storm. Also, we found that a similar problem with the prediction of the thermosphere by the standard model happens from time to time posing threat to the safety of spacecraft. Key Points For magnetically quiet days, the observed daytime NmF2 values were ∼50% larger than those simulated using the NRLMSISE‐00 model This implies ∼30% larger neutral O density comparing with the NRLMSISE‐00 model prediction or the density underestimation by the model Similar problem with the O density prediction happens for some periods for different years

• Publisher: John Wiley & Sons, Inc,Wiley
• Subject: Atmosphere / Atomic oxygen / Density / Geomagnetic storms / Ionosphere / Ionospheric composition / Ionospheric models / magnetic storm / Magnetic storms / magnetically quiet conditions / neutral composition / observation‐based simulations / Predictions / Satellite constellations / Spacecraft / Starlink / Storms / Thermosphere / Upper atmosphere