Ionosphere-Thermosphere Responses to 2023 March Geomagnetic Storm Using Observations and TIEGCM Driven by Data Assimilated Aurora and Electric Fields

The geomagnetic storm that began on March 23, 2023, and peaked with a Dst index of -169 nT in the early hours of March 24 was marked as the strong storm event (Kp=8) of solar cycle 25. During this event, the GOLD measurement showed a substantial increase in neutral temperature by over 230K along with a marked decrease in the O/N2 ratio by more than 0.5 at high latitudes compared to preceding days. Similarly, the Incoherent Scatter Radars (ISRs) at Poker Flat and Tromso, Norway, show a reduction in electron density and substantial fluctuation in both electron and ion temperature during storm time.
In this study, we focus on high latitude forcings: aurora precipitation and electric field. We utilized the aurora flux data from the ground (THEMIS/ASIs), satellite (DMSP/SSUSIs) and empirical models and the line-of-sight ion drift velocity from SuperDARN radars and electric field from ISRs (PFISR and EISCAT) and SuperDARN Spherical Harmonic Fitting (SHF) potential model. The diverse data sets were integrated using the lattice kriging data assimilation method to create auroras and electric field maps with mesoscale structures, offering a more realistic representation of magnetospheric forcing. We then employed these maps to drive the TIEGCM model to simulate ionospheric and thermospheric responses to the storm. The model's predictions were evaluated against GOLD measurements of O/N2 and neutral temperature, as well as GNSS-derived Total Electron Content (TEC). Compared to the default run, the assimilated run of the TIEGCM captured the O/N2 reduction (~0.4) and neutral temperature enhancement (>200K). Additionally, we examined the electron density, and ion/electron temperatures from ISRs to assess localized storm-driven impacts. We found that the TIEGCM (both default run and assimilated run) overestimates the electron density at Poker Flat but the enhancement of electron density due to storm effect was captured only in the assimilated run. The simulated results captured the ion temperature enhancement (~3000K at PokerFlat and ~2000k at Tromso) accompanied by strong electric fields at both locations. The assimilated run also better represented electron temperature fluctuations during the storm, while the default run only produced large-scale structures. Our work demonstrates that data-assimilated model drivers help to produce realistic storm-time I-T responses of dynamic variability across different scales.