On 4 February 2022, 38 Starlink satellites were destroyed by the geomagnetic storm, which brought significant financial, aerospace and public influences. In this letter, we reveal the space weather process during 3–4 February 2022 geomagnetic disturbances, from the Sun all the way to the satellite orbiting atmosphere. Initiated by an M1.0 class flare and the following coronal mass ejection (CME), a moderate geomagnetic storm was stimulated on 3rd February by the CME arrival at Earth. Subsequently, another moderate storm was triggered on 4th February by the passage of another CME. Model simulations driven by solar wind show that the first geomagnetic storm induced around 20% atmospheric density perturbations at 210 km altitude on 3rd February. The unexpected subsequent storm on 4th February led to a density enhancement of around 20%–30% at around 210 km. The resulting atmospheric drag can be even larger, since the regional density enhancement was over 60% and the satellite orbits were continuously decaying. This event brings forth the urgent requirements of better understanding and accurate prediction of the space weather as well as collaborations between industry and space weather community.

This study investigates midlatitude ionospheric variations during the super geomagnetic storm on 10–11 May 2024, utilizing multi‐instrument data from ground‐based sources (Global Navigation Satellite Systems receivers and a Fabry–Perot Interferometer) and space‐based measurements (Swarm and DMSP). We observed several distinct density gradient structures in the midlatitude ionosphere, with the main findings summarized as follows: (a) Significant zonal plasma density enhancements developed continuously in local dusk across the American‐Pacific‐Asian longitude sectors around geomagnetic latitude. These midlatitude peaks exhibited a wide longitudinal extension exceeding 150 and a prolonged duration of 12–15 hr during the late main phase and early recovery phase of the storm. (b) Strong storm‐enhanced density (SED) was observed in both hemispheres yet with different longitudinal and universal time preferences. In the Northern Hemisphere, significant SED occurred over the American longitude sector during 20:30–22:30 UT on May 10. In the Southern Hemisphere, pronounced SED was observed not only in the American longitudes during 20:30–22:30 UT on May 10 but also in the Australian longitude sector during 02:00–04:00 UT on May 11.

Stream interaction regions (SIRs) are important sources of geomagnetic storms. In this work, we first extend the end time of the widely used SIR catalog developed by Jian et al. (2006, https://doi.org/10.1007/s11207-006-0132-3), which covered the period from 1995 to 2009, to the end of 2016. Based on this extended SIR catalog, the geoeffectiveness of SIRs is discussed in detail. It was found that 52% of the SIRs caused geomagnetic storms with Dst_min ≤−30 nT, but only 3% of them caused intense geomagnetic storms with Dst_min ≤−100 nT. Furthermore, we found that 10 of the intense geomagnetic storms caused by SIRs were associated with complex structures due to interactions between SIRs and interplanetary coronal mass ejections (ICMEs). In such a structure, an ICME is embedded in the SIR and located between the slow and fast solar wind streams. In addition, we found that the geoeffectiveness of SIRs interacting with ICMEs is enhanced. The possibility of SIR‐ICME interaction structures causing geomagnetic storms is markedly higher than that of isolated SIRs or isolated ICMEs. In particular, the geoeffectiveness of SIR‐ICME interaction structures is similar to that of the Shock‐ICME interaction structures, which have been demonstrated to be the main causes of geomagnetic storms.

Solar flare formation mechanisms and their corresponding predictions have commonly been difficult topics in solar physics for decades. The traditional forecasting method manually constructs a statistical relationship between the measured values of solar active regions and solar flares that cannot fully utilize the information related to solar flares contained in observational data. In this article, we first used neural-network methods driven by the measured magnetogram and magnetic characteristic parameters of the sunspot group to learn the prediction model and predict solar flares. The prediction fusion model is based on a deep neural network, convolutional neural network, and bidirectional long short-term memory neural network and can predict whether a sunspot group will have a flare event above class M or class C in the next 24 or 48 hr. The real skill statistics (TSS) and F1 scores were used to evaluate the performances of our fusion model. The test results clearly show that this fusion model can make full use of the information related to solar flares and combine the advantages of each independent model to capture the evolution characteristics of solar flares, which is a much better performance than traditional statistical prediction models or any single machine-learning method. We also proposed two frameworks, namely F1_FFM and TSS_FFM, which optimize the F1 score and TSS score, respectively. The cross validation results show that they have their respective advantages in the F1 score and TSS score.

This study investigates the responses of the equatorial and low-latitude ionosphere in the American–Atlantic longitude sector during the super geomagnetic storm that occurred on 10–11 May 2024. The investigation utilizes multi-instrument datasets, including ground-based observations (GNSS TEC, ionosonde, and Fabry–Perot interferometer) as well as space-borne satellite measurements (GOLD, Swarm, DMSP, and TIMED). Our findings reveal significant day-to-day variations in the storm-time equatorial ionization anomaly (EIA), summarized as follows: (1) During the main phase of the storm, the low- and mid-latitude ionosphere experienced a positive storm, with TEC drastically enhanced by 50–100% within a few hours. The EIA crests exhibited a substantial poleward expansion, reaching as high as ±35° MLAT. This expansion was caused by the enhanced fountain effect driven by penetration electric fields, along with increased ambipolar diffusion due to transient meridional wind surges. (2) During the recovery phase of the storm, the global ionosphere was characterized by a substantial negative storm with a 50–80% depletion in TEC. The EIA crests were notably suppressed and merged into a single equatorial band, which can be attributed to the composition change effect and the influence of disturbance dynamo electric fields. These results illustrate the complex processes of magnetosphere–ionosphere–thermosphere coupling during a superstorm, highlighting the significant impacts of space weather on the global ionosphere.

An empirical model that predicts the AU index, a measure of the Earth's east electrojet, derived from magnetometers in the Northern Hemisphere, is introduced together with an improved AL model which, combined with the AU model, produces an AE model. All models are based on solar wind and interplanetary magnetic field parameters and the solar F_10.7 index for the years 1995 to 2001. The linear correlation coefficient (LC) between the 10 min averaged AU index and the model is 0.846 for the years 1995–2001. The LC for the updated AL model is 0.846, and using AE=AU−AL, the LC for the AE model is 0.888. The better LC of the AE model over AU and AL models is because AU and AL are better correlated than their errors. The models show that (1) solar ultraviolet intensity plays a significant role in auroral activity by changing the ionospheric conductivity and scale height. Increasing solar ultraviolet intensity increases the eastward electrojet as measured by AU but decreases the westward electrojet as measured by AL; (2) solar wind dynamic pressure also affects the auroral electrojet indices, although they are much more strongly dependent on the solar wind velocity and the interplanetary magnetic field; (3) AU and AL behave differently during geomagnetic storm main phases: AU, unlike AL, can drop to a small value during storms; (4) the longer averaged auroral electrojets indices can be predicted well but shorter timescale variations are less predictable.