Abstract We examine GNSS-derived total electron content (TEC) variations preceding four large megathrust earthquakes: the 2011 Tohoku–Oki (Mw 9.0), 2014 Iquique (Mw 8.2), 2015 Illapel (Mw 8.3), and 2009 Dusky Sound (Mw 7.8) events. Using stacking, backward-looking moving-average filtering, and Akaike Information Criterion (AIC)-based change-point detection, we evaluate the spatial and temporal characteristics of previously reported TEC anomalies within several hours before the mainshocks. For the Tohoku–Oki and Dusky Sound earthquakes, signals previously interpreted as preseismic anomalies exhibit systematic north-to-south migration at speeds of about 400 m/s, consistent with large-scale traveling ionospheric disturbances. Similar propagation patterns occur frequently on non-earthquake days, indicating that these signals reflect common background ionospheric variability. Comparable behavior is observed for the Iquique and Illapel earthquakes, although sparse GNSS station coverage limits spatial resolution. To assess the robustness of fitting-based precursor identification, we conduct synthetic experiments in which coseismic TEC disturbances of varying amplitudes are superimposed on background time series. Polynomial fitting generates apparent precursor-like anomalies whose amplitudes and inferred lead times scale with the imposed coseismic disturbance, reproducing previously reported magnitude-dependent trends through methodological effects alone. Spatially dense station analysis further reveals substantial station-to-station variability in apparent precursor metrics, with distributions that overlap those obtained on non-earthquake days. Overall, our results indicate that many reported TEC precursors are more plausibly explained by naturally occurring ionospheric disturbances and methodological artifacts rather than by earthquake preparation processes. We find no compelling evidence for robust preseismic TEC anomalies and emphasize the need for spatially resolved analyses and careful treatment of coseismic contamination when evaluating potential ionospheric precursors. Graphical abstract
Koyama et al. (Sat,) studied this question.