Reproduction of substorm auroral currents in the ionosphere using a machine-learning-based emulator and SuperDARN data assimilation

Abstract

Auroral electrojet dynamics during isolated substorms are reproduced using a machine-learning-based emulator with SuperDARN data assimilation. Representing such short-timescale variability is important for improving space-weather prediction relevant to technological systems. Reproducing this variability remains challenging for the SMRAI emulator (Surrogate Model for the REPPU Auroral Ionosphere; Kataoka et al., 2024, https://doi.org/10.1029/2023SW003720), where REPPU is the REProduce Plasma Universe global magnetohydrodynamic simulation. SMRAI is trained on ionospheric outputs from a REPPU MHD simulation and primarily produces smooth and gradual variations in auroral currents during substorms. Therefore, we evaluate the performance of SMRAI with SuperDARN line-of-sight velocity data assimilation (Nakano et al.,2025, https://doi.org/10.1029/2025SW004488), referred to as the SMRAI–SD framework. Two isolated substorm events on 30 and 31 October 2023 are analyzed to assess the performance of the SMRAI–SD. The SMRAI–SD reproduces auroral current variability in closer agreement with observations than SMRAI, as indicated by improved correspondence with the AL and AU indices. Although reproducing AE indices is not the primary objective, they are used to evaluate the performance of the SMRAI–SD approach and to identify its strengths and limitations through comparison with observations and examination of two-dimensional maps of ionospheric electric potential. SMRAI–SD reproduces abrupt changes in the ionospheric potential that are more pronounced than in the SMRAI. This study therefore evaluates whether such abrupt electrodynamic changes are physically consistent and under what conditions they are reliable. The results demonstrate that SuperDARN data assimilation enhances the ability of the SMRAI model to reproduce ionospheric auroral electrodynamics during substorm and supports its application to space-weather forecasting studies.