Uncovering New Ionospheric Dynamics with Neural Network Models: The Two-Phase Stormtime Response of Electron Temperature Overshoot

An intense surge in equatorial electron temperature (Tₑ) at sunrise, known as the morning Tₑ overshoot, has been one of the most widely studied ionospheric features since its discovery in the early Space Age. Despite extensive research, its behavior during geomagnetic storms remains poorly understood. Using global electron temperature observations by the CHAllenging Minisatellite Payload (CHAMP) mission in 2002-2010, we develop a global neural network Te model. Due to its excellent performance on independent data, it serves as a digital twin of the global Tₑ distribution. With this model, we uncover a two-phase stormtime response of the morning Tₑ overshoot. During the storm’s main phase, electron temperatures in the overshoot region exhibit a pronounced enhancement, followed by a dramatic depletion exceeding 1000 K and the disappearance of the overshoot during the recovery phase. This evolution corresponds to the initial influence of a westward prompt penetration electric field (PPEF), which reduces electron densities, allowing for more efficient energy exchange between newly ionized sunrise particles and the lower-energy (depleted) ambient plasma. Later in the storm, the eastward disturbance dynamo field flips the ExB drift from downward to upward and lifts more electrons into the F-region. The resulting increase in electron density enhances cooling rates, leading to the overshoot's disappearance in the recovery phase. Our findings shed new light on the dynamics of the morning electron temperature overshoot and highlight the capability of digital twin models of the near-Earth space environment to uncover previously unrecognized physical processes even for the most commonly studied phenomena.