Three-dimensional reconstruction of ionospheric disturbance structures during the pre-storm phase of the 10–11 May 2024 geomagnetic storm using a compressed sensing-based tomographic technique

• High-resolution 3D electron density perturbations were reconstructed over the continental United States during the 10–11 May 2024 geomagnetic storm. • Pronounced MSTIDs were observed in the F-region with distinct horizontal and vertical propagation features. • Compressed sensing tomography outperformed ART and SIRT in reconstructing complex ionospheric structures. • The proposed framework enables continental-scale monitoring of ionospheric disturbances using dense GNSS networks. • The results provide insights into pre-storm ionosphere-thermosphere coupling processes. The pre-storm phase of intense geomagnetic storms is associated with ionospheric perturbations, yet its three-dimensional structure remains poorly understood. In this study, a compressed sensing–based three-dimensional ionospheric disturbance tomography (CST) technique is applied to investigate ionospheric perturbations preceding the extreme geomagnetic storm of 10–11 May 2024. Using detrended slant total electron content (dSTEC) derived from approximately 1,700 GNSS stations over the continental United States, high-resolution three-dimensional electron density perturbations (dNe) are reconstructed with a horizontal resolution of 1° × 1° and a vertical resolution of 50 km. The results reveal pronounced traveling ionospheric disturbances (TIDs) observed more than 10 h before the storm main phase, characterized by dTEC amplitudes up to ±0.6 TECU, horizontal phase velocities of ∼400–700 m/s, and dominant periods of 30–60 min. Three-dimensional reconstructions indicate that these disturbances are primarily confined to the F region between 200 and 300 km, with peak dNe amplitudes of approximately ±1.0 × 10 1 ⁰ el/m 3 and clear upward and horizontal propagation signatures. Quantitative validation shows that CST achieves lower RMSE and higher correlation compared with traditional algebraic reconstruction technique (ART) and simultaneous iterative reconstruction technique (SIRT) methods in this study, achieving an RMSE of 0.139 TECU and a correlation coefficient of 0.963.