The geomagnetic field observations presented in this study were obtained from SuperMAG, a global network comprising of around 600 ground-based magnetometers (Gjerloev, 2009). The SuperMAG data is processed using a uniform data processing technique and made available to the public. For this study, we leverage existing tools and methods, e.g. (Ngwira et al., 2023, 2025), to perform a routine quality check of the geomagnetic field measurements to ensure the extreme perturbations (“spikes”) were not due to artifacts in the data. Details of the ground magnetometer stations that have been used in this study are displayed in Table 1.

For this study, the Spherical Elementary Current System (SECS) method was adopted to compute the ionospheric currents. In the SECS technique, a two-dimensional picture of the ionospheric currents can be derived with an array of well-spaced ground magnetometers (Amm, 1997; Amm et al., 2002; Weygand et al., 2012). The method has regularly been applied to calculate the equivalent ionospheric currents (EICs), which are parallel to Earth’s surface, and the spherical elementary current (SEC) amplitudes, which are a proxy for the field-aligned-currents (FACs). In general, the EICs are calculated from a matrix inversion of the ground magnetic disturbances. One of the important features of this technique is that it requires no integration time of the magnetometer data. There are two outputs from the SECS inversion. The first output is the EICs, which are a combination of the real Hall and Pedersen currents. The temporal and spatial resolutions of the EICs are 10 s, and 6.9 ° geographic longitude (GLong) by 2.9 ° geographic latitude (GLat), respectively. The second output is the SEC amplitudes, which are a proxy for the field-aligned-like currents, with a 10 s temporal resolution and 3.5 ° GLong by 1.5 ° GLat spatial resolution. Both set of currents are derived at an altitude of 100 km. For more details on the SECS method, see the following resources: (Amm, 1997; Weygand et al., 2011, 2012, and references therein).

In the present study, an estimate of the auroral oval equatorward boundary location is performed using the SECS technique outlined by Weygand et al. (2023). In that paper the equatorward boundary location was fitted visually by selecting points along a contour of about 0.5 μ A m p / m 2 . However, the equatorward boundary for the Gannon storm was frequently below the preset SECS grid for calculating the currents. Therefore, to get a good fit, a rough boundary was selected not at the typical 0.5 μ A m p / m 2 value used in previous studies. It must be noted that in the present case, getting a fit was difficult because currents were mostly in disarray due to the extreme nature of the storm. But one of the challenges to extending the grid further south is the limited number of stations ( ∼ 3) that might help to obtain a good fit of the boundary location at μ A m p / m 2 . We refer readers to Weygand et al. (2023) for more details on the auroral boundary estimation process.

The Missouri Skies Observatory (40.25 N, 94.32 W, www.missouriskies.org) is an amateur-run facility located within the city limits of rural Albany, Missouri. Its main instrument is a 50 cm reflector telescope. As part of citizen science activities, off-the-shelf surveillance camera equipment mounted on the observatory and other points around the property are used. The all sky cameras (ASCs) are unmodified and have settings tweaked to record, at maximum sensitivity, real-time movies of satellite phenomena, airglow, auroras, and meteors. The ASC field of view is 180°. Sometimes, the Observatory hosts live streams of important auroral events for the public interest, such as the Gannon storm. During the May 2024 great auroral event, the Observatory recorded color ASC, wide-angle color north-looking, and black and white east-looking video with an exposure time of 1-s and recording frame rate of 1/30-sec. Selected frames from the Missouri Skies ASC video showing the auroral sequence on 11 May are presented in Section 4 of this study.

In May 2024, solar cycle 25 (SC-25) hit a new high when the monthly sunspot number reached 172 beating the previous high of 160 recorded in June/July of 2023. In comparison to the previous two cycles, the maximum monthly sunspot number reported for SC-25 (217, August 2024) is about 67% higher than the 146 recorded for SC-24 in February 2014, while SC-23 had a maximum of 244 in July 2000. Interestingly, during the period 8–9 May 2024, the Sun continuously unleashed several CMEs in quick succession including Earth-directed ones. All CMEs were linked to X-class or strong M-class solar flares dominated by two powerful X-class flares with magnitude X2.25 and X1.12 (Tulasi Ram et al., 2024). Shortly there after on 10–11 May, another set of powerful X-class flares with magnitude X3.9 and X 5.8 were also launched. According to the NOAA Space Weather Prediction Center (SWPC), the solar Active Region AR3664 was the candidate for Earth-directed CMEs.

In general, the observed CMEs had average speeds ranging from 456 km/s to a maximum exceeding 2000 km/s. Given that, it is very likely that later CMEs may have caught up with preceding ones, leading to collisions or merging events (CME cannibalism). This type of interaction can amplify CME speed, plasma density, and magnetic field strength, substantially intensifying the shock strength upon reaching Earth’s magnetosphere, which can trigger G5-level GMDs. Unlike other extreme events (e.g., Halloween Storm of 2003), the Gannon Storm’s rapid sequence of CMEs led to stronger magnetospheric shocks through CME–CME interactions, thus creating a powerful, long-duration geomagnetic storm. The solar wind and the interplanetary magnetic field (IMF) conditions during the Gannon storm are presented in Figure 1. The data were obtained from the OMNIWeb service available through the NASA/GSFC Space Physics Data Facility provided by CDAWeb (https://omniweb.gsfc.nasa.gov/). The figure contains the IMF magnitude (a), the IMF-By (blue) and Bz (red) components (b), the solar wind density (c) and speed (d), and the geomagnetic field SYM-H index (e), a high-resolution equivalent of the Dst index.

Following the arrival of the first ICME forward shock (CME-1) around 17:05 UT on 10 May (Figure 1), the IMF magnitude (a), speed (c), the density (d), and the dynamic pressure (e), were all observed to suddenly increase, as traditionally expected for most CMEs. It must be noted that the density and speed were slightly elevated before the first CME arrival starting around 15:00 UT, which is also reflected in the SYM-H index in Figure 1f. It is not clear at present what role these elevated levels of density and speed may have played in the subsequent storm dynamics and strength. Nevertheless, this topic related to pre-conditioning of the magnetosphere will not be addressed in the present paper. A second ICME (CME-2) forward shock was detected shortly after 22:00 UT, as also indicated by the second dashed line in Figure 1. In the early phase and main phase of the storm (first 10 h), the IMF-Bz was predominately southward with a peak value around −53 nT. The high IMF-Bz levels and the sustained southward orientation is one of the reasons for the development of an extreme geomagnetic storm, as is well-known.

On the Earth’s surface, the first CME arrival is reported to have produced an extreme sudden commencement (SSC), as recorded by ground magnetometers (Piersanti et al., 2025). A comparison of the Gannon storm features to other historical extreme events is displayed in Table 2. The SSC perturbation ( Δ SSC = peak SSC value–value before CME arrival) in column two of the table is derived from SYM-H index data. In their investigation of the geomagnetic response to the Gannon storm focusing on the European sector, Piersanti et al. (2025) found that the strong solar wind pressure coupled with a large sustained southward IMF-Bz and a favorable shock inclination angle resulted in an extreme SSC with an amplitude of roughly 180 nT recorded in the near–dusk sector at the Grocka (40.55 ° , MLT = 18:30) ground magnetometer station in Serbia. They attributed this extreme SSC to both ionospheric and magnetosphere drivers. Regarding the timing of the SSC, we note the existence of two different reported arrival times at 16:34 UT and 17:05 UT by Tulasi Ram et al. (2024) and Piersanti et al. (2025), respectively. Based on the currently available level 2 OMNI solar wind and the ground magnetometer data, we adopt 17:05 UT as the arrival time of the CME.

Evidently on the Dst scale, the Gannon storm is one of the most powerful geomagnetic storms of recent times ranking as the third largest storm in modern history (1985–2024) following the March 1989 and November 2003 extreme events, as illustrated in Table 2. With a recorded peak negative excursion of the Dst index at 406 nT according to World Data Center Kyoto (https://wdc.kugi.kyoto-u.ac.jp/) equivalent to the SYM-H index reaching about − 518 nT, this event stands as the most intense storm of solar cycle 24–25 at the time of this writing. Solar wind energy transferred through multiple channels affected the entire Geospace. Zhang et al. (2025) report of a new proton belt with high energy fluxes was created by the combine effect of inward proton penetration from solar proton events and likely energization due to significant compression of the magnetopause with the southward IMF-Bz during the SSC period. Upon further inspection of the solar wind characteristics listed in Table 2, it is immediately clear that this storm was not extraordinary in its drivers, but rather comparable to other historical events. We must make mention here that the October 2003 storm IMB Bz and density information is estimated due to widely known data challenges experienced during that era.

Here, we focus on the less intense substorm event deep in the main phase of the storm on 11 May. An account by Foster et al. (2024) estimates the onset of this substorm to be between 02:00–02:02 UT ( ∼ 22 MLT) based on images from the Boston University all-sky imager located at Millstone Hill, Massachusetts (42.6 N, 71.3 W). This puts the location of the onset in the vicinity of the Great Lakes and East Coast region of the United States. Interestingly, the substorm event was not captured on any of the well-known substorm lists. The inability to capture some substorm onsets is a widely documented limitation (Newell and Gjerloev, 2011; Chu et al., 2015; Forsythe et al., 2020; Ohtani and Gjerloev, 2020; Lao et al., 2024), but is outside of the scope of this current report. However, it would suffice to say that the extreme equatorward expansion of the auroral oval at the time of this event could be the main reason why this particular substorm onset was not captured. It should be noted that under extreme equatorward expansion, the stations used in the computation of the electrojet indices may have been located outside of the auroral oval coverage.

Figure 3 shows a series of ASI images from the Missouri Skies Observatory (40.25N, 94.32W) taken on 11 May 2024. As seen in Figure 3, before 02:00 UT, the images do not show any auroral activity, but red aurora suddenly appears to the north around 02:05 UT. The red aurora is then observed to extend southwestward with time, as revealed in Figure 3. The western edge of the red aurora is the front of the substorm westward expansion, and it includes an auroral streamer. In a recent report, Foster et al. (2024) investigated the changes in ionospheric total electron content (TEC) triggered by the storm-induced energetic particle precipitation during the Gannon storm. They provide a detailed account of the intense auroral breakup and westward surge occurring at the peak of that storm. Foster et al. (2024) also showed that sudden bursts of significantly elevated vTEC were related to the intense red aurora located on the leading edge of the equatorward and westward TEC enhancement. They attributed this TEC increase to extremely low-energy precipitation following the rapid substorm breakup.

In Figure 4 we examine the horizontal geomagnetic field observations on the surface of the Earth. Exhibited in this figure are the geomagnetic field B x component (top) recorded at selected ground stations and the rate of changes of B x (bottom) for the period 10–11 May, 2024. Details of these ground sites are listed in Table 1. For this investigation, a chain of mid-latitude magnetometers running from East to West across the continental United States was selected. A closer look at Figure 4 shortly after 02:00 UT reveals a clear substorm characteristic commonly seen in ground magnetometer observations. We note that before the marked decrease in the B x component, there was a sudden rapid enhancement of the field. This is a typical signature associated with the mid-latitude positive bay (MPB) phenomena, a frequently observed feature driven by auroral substorm–related activity (Chu et al., 2015; McPherron and Chu, 2017). However, it was only recently that Ngwira et al. (2023) showed, for the first time, a direct relation between MPBs and the occurrence of large GICs using data over the North American region. The presence of auroral streamers in the ASI images suggest that the current system driving the geomagnetic perturbations involves BBFs in the magnetosphere (Lyons et al., 2013). Streamers are considered to extend equatorward within the auroral oval from PBIs, which have been linked to flow bursts (BBFs) that can transport plasma across the nightside magnetic separatrix into the plasma sheet (Lyons et al., 2011) and are therefore related to localized regions of enhanced tail reconnection. While streamers can occur during many different geomagnetic conditions, they are commonly seen as localized structures during the much larger-scale auroral displays of a substorm expansion phase (Henderson et al., 1998).

What is more captivating about this event in Figure 4 is the level of perturbation observed at these mid-latitude locations. Firstly, before the negative excursion, there was a sharp increase of B x from −316 nT to −175 nT ( Δ B x = −141 nT) at the USGS Fredericksburg (FRD) Observatory in Virginia and an extreme increase from −307 nT to 142 nT ( Δ B x = −449 nT) at Missouri magnetometer site (R01). The sharp enhancement was quickly followed by rapid decrease with an absolute value of the geomagnetic perturbation, Δ B, roughly ∼ 641 nT recorded at FRD and ∼ 733 nT at R01. These levels of change are extremely large and rare for these mid-latitude locations. In the American sector, Figure 1 of Foster et al. (2024) also reported very rare magnetometer deviations of −700 nT (Missouri) to −1000 nT (Millstone Hill) using the MagStar observation network. In addition, Waghule and Knipp (2025) presented the temporal correlation of large GICs recorded near Fredericksburg, Virginia, and the ionospheric observations derived from GNSS TEC and Radars from the SuperDARN network in the United States. Their investigation suggests that peak GIC variability occurred deep in the main phase of the storm, during which time, the F-region ionosphere appears to have been significantly elevated and expanded. The present study and other cases from previous studies highlight the complexity and severity of the Gannon storm event.

Notably, the storm’s impact on the Δ B perturbations was observed to propagate westward accompanied by a reduction in the intensity, as seen from Figure 4 for FRD, R01, BOU, and FRN. This westward propagation is consistent with the initiation of the auroral electrojet (AEJ) enhancement reported by Foster et al. (2024). According to Foster et al. (2024), the auroral intensification extended westward through the central United States near 94 W longitude around 02:02 UT, while there was a large expansion and southwestward propagation associated with auroral breakup that appeared near 02:05 UT ( ∼ 21 MLT). Interestingly, these intense perturbations revealed here were confined to certain portions of the North American sector in the vicinity of the substorm onset region around the Great Lakes and East Coast of the United States. This strongly suggests that the impact of the substorm was both spatially and temporally localized in nature. It is also important to note that, similar substorm-related perturbations in ground geomagnetic field data were also seen in the South American sector (not shown).

In contrast to the mid-latitudes (Figure 4), the geomagnetic field characteristics at high-latitudes and the higher mid-latitudes are presented in Figure 5. Firstly, the geomagnetic response at the high-latitudes and the higher mid-latitudes are vastly different. For instance, we clearly see large perturbations at higher mid-latitudes around 02:00 UT on 11 May, whereas some high-latitude stations only show perturbations at 30-min before 02:00 UT. As well, the signature of the supersubstorm on 10 May is more visible at high-latitudes than the mid-latitudes. However, the most notable differences are observed during the recovery phase when significantly large perturbations are captured over the high-latitudes. These perturbations are accompanied by equally large fluctuations in the field, dB x /dt, that correspond to the intervals when SML index was < –2500 nT, as demonstrated in Figure 2.

Turning our attention to the auroral currents and the boundary dynamics, we utilize the EIC and current amplitude maps derived from the SECS technique outlined earlier. Figure 6 shows a series of SECS maps for this event in the interval between 01:55 UT to 02:15 UT on 11 May. As noted before and seen in Figure 2, this interval does not contain any supersubstorms, however, the isolated intense substorm event during the said period caused extreme geomagnetic field deviations at mid-latitudes in the American Sector, as discussed previously.

Thorough examination of Figure 6 shows that before 02:00 UT the EIC was weak (small arrows) over most of the Continental United States, except to the North Western region, while a clear two cell convention pattern existed to the north in Canada and the Arctic region. The top panel (a) shows orange squares for the high-Latitudes, orange circles for higher mid-latitudes, and orange triangles marking mid-latitudes. The Missouri Skies observatory is represented by a mauve star. At 02:05 UT, a sudden westward intensification of the EIC is observed around the East Coast and Great Lakes region. The intensified current then propagates westward with time. This is consistent with the geomagnetic recordings in Figure 4 showing the Δ B x perturbation propagating to the west with decreasing intensity. In addition, Foster et al. (2024) reported similar westward propagation during this event based on the observed GPS TEC perturbations over North America. They attributed the observed Large TEC perturbations to the westward expansion of low-energy electron precipitation captured by direct auroral observations.

Furthermore, a careful look at the EIC currents in Figure 6 indicates that the strongest geomagnetic perturbations are in the region between the upward (red) and downward (blue) current systems. According to literature, this is the region with the most intense auroral currents, as discussed in previous reports (Ngwira et al., 2018, 2023; Engebretson et al., 2024). Another interesting features is the formation of the Harang (red) upward current system between 02:00 UT to 02:10 UT on 11 May. The Harang reversal (HR), sometimes known as the Harang discontinuity, is a longitudinally extended ionospheric signature near midnight where electric fields and plasma flows reverse direction from westward to eastward with decreasing latitude (Harang, 1946; Erickson et al., 1991). It is well understood that the HR poleward extension maps to the plasmasheet, while its equatorward extension maps close to the inner edge of the plasmasheet (Erickson et al., 1991). In the magnetic midnight sector where the HR region maps to the magnetotail, the reversal is viewed as a fault line isolating the inflated and collapsed magnetic fields during substorms (Maynard, 1974), whereas some studies suggest that substorm injection occurs in the vicinity of the HR (Weygand et al., 2008; Despirak et al., 2022).

In a study on the coupling between the HR evolution and substorm dynamics using a combination of SuperDARN, DMSP, and IMAGE observations, Zou et al. (2009) found that the substorm auroral onset appeared quite near the center of the HR flow shear, which they considered to be related to the development of the substorm upward FACs. The location and characteristics of the HR are controlled by the IMF orientation, including the IMF clock angle. Fundamentally, the IMF clock angle, specifically the IMF-By, can determine the dynamics and properties of the HR by modulating the ionospheric convection pattern, magnetotail configuration, and AEJ strength and distribution (Anderson et al., 2008; Grocott et al., 2010). Therefore, the HR appearance reflects enhanced magnetospheric plasma convection in the magnetotail that involves an upward flowing FACs from the ionosphere due to diamagnetic ion drift. Its latitudinal location is associated with the strength of geomagnetic activity, often propagating equatorward as the intensity of activity increases, as established in Figure 6. Whereas the southward IMF-Bz is considered a primary condition for the development of GMDs, IMF-By also plays a major role, particularly it strongly influences the dawn-dusk asymmetry (Yoshizawa, 1986; Ohtani et al., 2018).

The dependence of GMDs on the orientation of the IMF-By component has continued to receive wide attention (Holappa et al., 2021; Engebretson et al., 2022; Ohtani et al., 2023). During the Gannon storm, there were two particularly interesting intervals when IMF-By experienced sharp reversals going from negative to positive orientation. At the time of the first flip on 10 May, IMF-By changed from around −64 nT at 22:27 UT to 69 nT at 22:48 UT. The dramatic IMF-By flip happened simultaneously with the IMF-Bz sudden rotation from about −40 nT to 50 nT under high solar wind velocity ( ∼ 770 km/s) and strong dynamic pressure around 30 nPa, as illustrated in Figure 1. The unique geomagnetic response to this flip has been reported in recent papers (Kleimenova et al., 2025; Vichare and Bagiya, 2024; Ohtani et al., 2025). We see in the present study (Figures 4, 5) that this extraordinary IMF-By flip was also associated with some intense geomagnetic perturbations at mid-latitudes with dB/dt exceeding an absolute value of ∼ 260 nT/min. This level of perturbation is not common at these latitudes during most storms, but appears largely in association with more severe events, such as the Gannon storm.

In contrast, the second IMF-By flip on 11 May turning from −26 nT at ∼ 01:44 UT to 12 nT 2-min later at 01:46 UT has not been reported in previous studies, to our knowledge. During this same period, the IMF-Bz become strongly southward ranging between −8 nT and −32 nT. The formation of the HR in Figure 6 matches up pretty well with the rotation of the IMF-By component near 02:00 UT. Note that prior to about 02:00 UT there is no upward current (red) at Missouri Skies Observatory. However, at 02:05 there is a little bit of red aurora (Figure 3) and the spherical elementary current (Figure 6) shows some upward current but to the east. At 02:10 UT there is upward current (probably the strongest upward current in that region between 01:30 to 02:30 UT) at Missouri Skies Observatory, which is associated with discrete auroral streamer at the observatory (Figure 3). In the all-sky images at 02:15 UT and 02:20 UT there still is strong red aurora and upward current around Missouri Skies Observatory. Figure 1 illustrates that after the second flip around 01:45 UT, the IMF-By remained predominately positive for almost 1-h. At the typical mid-latitudes (Figure 4), the dB/dt were large during the IMF-By flip on 10 May, however, at the higher mid-latitudes (Figure 5d), the dB/dt were much more intense around 02:00 UT on 11 May. These dB/dts were even larger than those seen at high-latitudes (Figure 5b) during the specific time window. This case highlights the complex nature of the solar wind-magnetosphere-ionosphere coupling process. Furthermore, it must be noted here that no Pi2 pulsations are present (not show) in association with the second IMF rotation. However, Pi2 pulsations were present earlier in the storm main phase in association with the dramatic IMF rotation around 22:30 UT as revealed by Kleimenova et al. (2025).

Finally, we turn our focus to the auroral oval equatorward boundary dynamics. This location is estimated using the SECS technique and is depicted by the dashed mauve curve in Figure 6. Understanding the motion of the boundary is important in terms of the location and magnitude of the geomagnetic effects on the ground. Specifically, the auroral oval moving into regions that are considered less prone to intense auroral currents, e.g., the lower 48 States of the United States. However, it is highly possible that the extent of equatorward boundary may have been much further south than depicted here. There were a number of reported aurora sightings in low latitude regions, including Mexico (Gonzalez-Esparza et al., 2024; Hayakawa et al., 2025). One of the limitations to getting a more accurate location of the boundary is the sparse density of magnetometers over the United States and Mexico, as outlined earlier. Obtaining a boundary is further complicated by the intensity of the Gannon storm, as manifested by the messy current patterns produced in Figure 5. Most of the new USGS variometers are closer to the northern United States boarder and really only two stations contribute to the boundary below 35 deg GLat. Therefore, the majority of the time the boundary could be below the “field of view” of the spherical elementary current grid. In general the equatorward boundary sits between 40 deg and 45 deg MLat at magnetic midnight and possibly lower between 01:30 and 01:45 UT ( ∼ 42 deg MLat) and higher between 02:15 and 02:30 UT (44 deg MLat). Lastly, we also note that the Fresno (FRN) magnetometer site sits outside of the estimated auroral boundary location, which could explain why the geomagnetic field response signature is much different to the other mid-latitude stations in Figure 4.