Electromagnetic anomalies before earthquakes exhibit significant short-term characteristics. In this study, we first employed the non-negative matrix factorization (NMF) method to extract anomalies in the electric field, magnetic field, and electron density within the ionosphere. Then, to obtain earthquake-related anomalies, we propose a random matrix theory (RMT) anomaly distinction approach, which involves comparing the eigenvalue distribution of the ionospheric anomaly’s correlation matrix with that of a random matrix to remove earthquake-unrelated anomalies. Temporally, we observed that the pre-earthquake anomalies displayed sigmoidal growth between −65 and −25 days, followed by power-law growth from −20 days until the seismic event. Spatially, the anomalies exhibited a concentration pattern, migrating from the periphery of the study region toward the epicenter. Finally, the ionospheric anomalies were compared with multi-layer anomalies before the Madoi earthquake. We found three distinct phases, in the early phase, anomaly propagation through the lithosphere, atmosphere, and ionosphere followed an asynchronous chain process. In contrast, in the middle phase, anomalies appeared synchronously across the multi-layer. In the impending earthquake phase, the anomalies were directly coupled from the lithosphere to the ionosphere, which we hypothesize to be a new mechanism related to the dramatic decrease in Earth's resistivity.
Yang et al. (Tue,) studied this question.