Effects of space weather on the electric power network and mining operations in Alberta, Canada during the October 10–11, 2024 geomagnetic storm

Abstract

Geomagnetically induced currents (GICs) due to space weather can impact electric power networks via transformer saturation, reactive power consumption, and increased total harmonic distortion. In the worst case, these can lead to widespread outages and damage to transformers. Investigating large geomagnetic storm events allows the power industry to better understand and mitigate the associated risks. Here we focus on the impacts in Alberta, Canada during the October 10–12, 2024 geomagnetic storm by modelling GICs in the ≥240 kV power network validated with transformer neutral-to-ground (TNG) GIC measurements. Despite the October 2024 storm being smaller than the May 2024 storm according to global storm disturbance indices, measured GIC exceeded 25 A/phase at three transformers in central Alberta and modelling suggests that the TNG GIC at some transformers in northeastern Alberta exceeded 30 A/phase, similar to the magnitude of the GICs seen during the May 2024 storm. For the October 2024 event, the cause of the largest geoelectric fields and largest GICs is spatially and temporally variable. Large events in northeastern Alberta were linked to the sudden storm commencement, with some large dusk-side events in southern Alberta being linked to a low-latitude substorm. The most spatially-extensive large geoelectric field event was associated with a large nightside substorm which developed over Alberta. In addition to GICs flowing at high voltage levels, the October 2024 storm also resulted in adverse technological impacts in northeastern Alberta at lower voltage levels. These included 34.5 kV capacitor banks and large 13.8 kV mine shovels tripping offline at industrial mining operations. The capacitor bank trips were likely due to increased total harmonic distortion due to transformer saturation from GIC at higher voltage levels, while the mine shovel trips may have been due to a unique failure mode related to induced current on long (∼10 km) low-current 30 V DC ground check circuits. To our knowledge, this is the first time that direct space weather impacts on mining operations have been reported in the literature.

1 Introduction

Space weather resulting from solar-terrestrial coupling can cause large disturbances to the Earth’s geomagnetic field. The resulting geomagnetic storms induce geoelectric fields in the ground which can generate a potential difference between grounded terminals of large man-made infrastructure such as electric power transmission networks, long-distance pipelines, and undersea cables [1]. In power transmission networks, the electromotive force from the geoelectric field in transmission lines results in low-frequency (<1 Hz) geomagnetically induced currents (GICs) which flow in transmission lines and transformer windings via the transformer neutral-to-ground (TNG) connection. Since the temporal variations of GICs are much less than the 50–60 Hz AC power system carrier frequency, they act as quasi-DC currents which can bias transformers into saturation leading to increased reactive power consumption, generation of harmonics, and transformer heating, among other problems [2]. If the GICs are large and sustained, they can overload networks resulting in voltage instability, transformer damage, and/or power outages.

On October 10–12, 2024 there was a large geomagnetic storm with a Dst index of −335 nT, the second largest storm of the current solar cycle after the May 2024 storm (e.g., [3, 4]). Although it was a smaller geomagnetic storm than the May 2024 storm in terms of Dst, and has therefore garnered less attention, it still had some notable global effects including strong ionospheric disturbances [5–7], red aurora at lower latitudes [8], and potentially caused a premature satellite re-entry [3]. This storm also had characteristics which made it susceptible to generating large ground magnetic disturbances and large geoelectric fields (GEFs) [4, 9, 10]. Alberta is located at high geomagnetic latitude near the auroral oval and has significant datasets to characterize the ground geomagnetic disturbance using the CARISMA magnetometer array [11] as well as extensive magnetotelluric (MT) measurements to characterize the ground conductivity [12, 13] (Figure 1). These both influence the resulting geoelectric field behaviour and therefore the GIC risk [14–16]. Via an ongoing industry-academia collaboration, we have developed an accurate DC-equivalent network model of the ≥240 kV transmission network. This model builds upon earlier work reported by Cordell et al. [17], incorporates transformer-level details and expanded network resistance data based on information provided by the Alberta Electric System Operator (AESO) and transmission facility owners (AltaLink Management Ltd. and ATCO Electric Ltd.). As with the earlier model, the updated model can be used to compute data-driven GIC across the entire high-voltage (≥240 kV) electric power network in Alberta. Ongoing industry TNG GIC measurements at six transformer neutrals across the province allow for assessment of model performance and data validation.

FIGURE 1

12].

During the storm, there were several key events with large geoelectric fields and GICs and, critically, these resulted in space weather impacts including 34.5 kV capacitor bank trips at one substation, and multiple DC ground check circuit trips on 13.8 kV industrial mine shovels in northeastern Alberta. These effects are notable in that they both occurred at voltage levels <240 kV and, to our knowledge, is the first documented case of mine shovels tripping offline due to space weather. Here, we present an analysis of this storm to highlight where and when large GIC were observed and modelled across the province and highlight the relationship between power network effects and space weather. We first focus on the methodology (Section 2) and then present specific impacts observed in northeastern Alberta (Section 3.1). We then zoom out to examine GIC observed and modelled at the province-scale (Section 3.2 and 3.3) and identify specific spatiotemporal patterns in the observed hazard. This observation motivates further discussion of the broader space weather conditions (Section 3.4) which led to these specific times of increased GIC.

2 Modeling methods

GIC modelling is often broken into two distinct steps. The first involves the calculation of the induced geoelectric field using magnetometer data and Earth impedance information. The second involves the calculation of the network level distribution of GICs in the power network using the modelled geoelectric field as input. The broad workflow is summarized in Figure 2.

FIGURE 2

2.1 Geophysical step: geoelectric field estimation

In order to estimate the geoelectric field and GICs in the provincial power network during this storm, we use existing 1 Hz horizontal geomagnetic data from 11 sites from the CARISMA magnetometer array [11] as well as horizontal geomagnetic data from 4 NRCan magnetic observatories and 3 USGS magnetic observatories [18] (Figure 3). North () and east () magnetic data for a 48-h period from October 10–11, 2024, are considered, resulting in a dataset denoted , where is a function of time, , at samples and are the locations of the magnetometers. At each time step, the horizontal geomagnetic data are spatially interpolated to query points, . In this case, we use where are the locations of = 411 MT impedance tensor measurements across the province (see Figures 1, 3 for locations or and ). The interpolation is done rapidly by creating a sparse () matrix, , of Delaunay triangulation weights which is constant for all time steps (i.e., depends on geometry only). Then an () matrix, is constructed containing all the original data at all time steps. The interpolated data can then be found using which maps to . The result is ∼70 million interpolated time series data points computed in <0.25 s on a standard MacBook Pro laptop. Each horizontal component is converted to the frequency domain by Fourier transform of each time series after pre-whitening by detrending, applying a 10% Tukey window, and padding with = 10,000 samples at each end following Kelbert et al. [19]. This converts the magnetic time series, to magnetic spectra, , defined at each MT site with frequencies between the Nyquist frequency (0.5 Hz) and the inverse of the duration of the padded time series. The entire magnetic data preparation for 48 h of data takes <3.5 s on a standard laptop.

FIGURE 3

The raw impedance tensor data at each location are not used directly because of noise and inconsistent bandwidth (e.g., some MT sites have valid data from 1 s to 20,000 s periods, while others have poor data quality and/or edited data points at periods >5000 s). Instead, the 3-D conductivity model from Wang and Unsworth [12] is used which is based on a best fit of the raw MT impedance tensors using 3-D geophysical inversion techniques (see [12, 20]). The conductance of the existing model from Wang and Unsworth [12] from the base of the sedimentary basin to 50 km depth is shown in Figure 1. ModEM frequency-domain electromagnetic modelling software is used to compute new impedance tensors at the same locations as the measurements. The computed impedances are smooth and noise-free, and all site locations have impedances computed at identical frequencies (e.g., identical bandwidth with = 50 periods from 1 s to 31,000 s). The computed impedance, ) is then interpolated in the frequency domain to match the magnetic data frequency bins (i.e., ) interpolated to )), and no extrapolation is performed. In other words, the 48-h magnetic data set in the Fourier domain has a longest period of 172,800 s, but the maximum period for the impedance is only 31,000 s, and so the magnetic field spectra are effectively truncated to periods of 31,000 s. The impedance decreases in magnitude as a function of period (e.g., in effect a high-pass filter), so there is very little induction at periods >10,000 s.

After this procedure, the result is a set of spatially interpolated horizontal magnetic spectra and impedance tensors defined at each of the 411 MT site locations () on the same set of frequencies () (see Figure 2). The horizontal magnetic spectra and impedance tensor at each site are multiplied together in the frequency domain to solve for the horizontal electric field spectra at each location:where is the electric field spectra in V/m, is the impedance tensor in Ω, is the magnetic flux density spectra in Tesla, is the magnetic permeability of free space and , , and are all functions of and . Equations 1, 2 are optimized using page-wise matrix multiplication for each site where Z is and B is . The horizontal electric field can then be converted back to the time domain using the inverse Fourier transform to give horizontal electric field time series at each MT site location, in V/m. The entire procedure to compute the 2-day of electric data from the 2-day of magnetic data takes approximately 15 s on a standard laptop.

Note that there are several choices made during the implementation of this methodology. For example, we could compute the impedance anywhere in the 3-D model space, including on a regular grid, or along transmission lines. In other words, we can choose any , but here we choose . We also make choices about how to interpolate the magnetic field data (e.g., choice of specific interpolation method, as well as whether to interpolate in the time- or frequency-domain). We use the above methodology based on tests indicating that it appears to be very stable and results in the best fit between measured and modelled GICs while also allowing for decreased computational time. Specific inter-comparisons between other implementations of these methods is beyond the scope of the current paper.

2.2 Engineering step: GIC modelling

Once we have the geoelectric field components at various locations, we can include power transmission network information to compute GICs in the power network. First, we must compute the driving potential difference (i.e., the source e.m.f. that pushes current in the network elements). The geoelectric fields are spatially interpolated onto the 281 transmission line routings (L) for all transmission lines with nominal voltage at or above 240 kV in Alberta using 2 km segments (e.g., one interpolated geoelectric field every 2 km along the transmission line). This interpolation is done using the same q = Ad method as detailed above, with a total of >6500 transmission line query points. Following this, the potential difference (i.e., induced voltage, ) for each transmission line at each time step is computed by numerically integrating the interpolated horizontal electric field along each transmission line:where is the potential difference for the th transmission line at a given time step. Each transmission line routing is broken into segments of 2 km length, and are the north-south and east-west distances for the th segment, respectively, and and are the electric field components interpolated at the +1 segment nodes. The total calculation time for the potential difference simulation takes ∼15 s on a standard laptop.

After the potential difference is solved for, then Kirchoff’s Circuit Law (KCL) is used to solve the amount of current flowing in the ≥240 kV Alberta power transmission network, including to ground through TNG connections [21, 22]. The DC-equivalent network model includes nodes and edges. The nodes represent buses and neutrals in substations, while the edges connect buses via either transmission lines (substation-to-substation), or via transformer windings (bus to neutral connections). In total, the network model included 161 substations with 305 transformers, as well as 13 tap junctions or switching stations. These were connected by 281 transmission lines. The basic network configuration was modified to account for scheduled maintenance during the geomagnetic storm days, according to a list provided by the AESO. This resulted in the exclusion of 4 transmission lines and 12 transformers, which were out of service during the storm interval. Transmission line resistances (), transformer winding resistances (), and transformer types were provided by the AESO, AltaLink Management Ltd., and ATCO Electric Ltd. Earthing resistances () for 120 of the 161 substations were provided by AltaLink and ATCO Electric. Most of the missing earthing resistances were at substations owned by ENMAX and EPCOR, two smaller transmission facility owners serving Calgary and Edmonton, respectively. If earthing resistance values were missing, the median value of the known earthing resistance values was used (0.13 Ω). Note that, because we only model the ≥240 kV network, all transformers with low-side <240 kV nominal voltage were modelled as ungrounded Δ connections. Thus, the majority of transformers are modelled as Yg-Δ with 240 kV high side and <240 kV low side. The network also consists of 35 grounded autotransformers, all of which either step down from 500 kV to 240 kV, or from 240 kV to 138 kV. Finally, there are 4 Yg-Yg phase-shifting transformers which have 240 kV voltage on both high and low side. In total, the DC-equivalent model uses 621 nodes (buses and neutrals) connected by 666 unique edges (transmission lines and transformer windings).

The current source for the ith edge is given bywhere is the resistance of the edge elements from and . Note that we set for all because the transformer winding distance is considered to be negligible relative to the geoelectric field. The nodal voltages can be calculated using the Modified Lehtinen-Pirjola method or Nodal Admittance Matrix (NAM) method [21, 22]:where is the nodal admittance matrix, is the earthing admittance matrix, and is the nodal current source vector from Equation 4. The GIC per phase, flowing along each network element from th node to th node is given bywhere is the admittance of the segment, is the nodal voltage across the segment ( to ) using nodal voltages solved for in Equation 5, and is the induced voltage of the segment due to the electric field from Equation 3. The GIC calculation described in Equations 4–6 treats each time step as independent and is parallelized for post-event analysis; 2 days of simulation time runs in approximately 200 s on a standard MacBook Pro laptop. The simulation computes per phase current values and assumes that quasi-DC current is equal in all phases. Note that the Modified Lehtinen-Pirjola method is mathematically and conceptually identical to the Nodal Admittance Matrix method [23].

3 Results and discussion

3.1 Space weather impacts

During the storm, there were several notable trips at capacitor banks in a substation in northeastern Alberta and at mine shovels in northeastern Alberta associated with large-scale industrial mining activities. Capacitor banks are located within substations and are designed to improve the power factor and reduce harmonics in the power network. They are particularly important in areas with large intermittent loads related to industrial operations. The mine shovels operate using 13.8 kV feeder cables up to 10 km long and use a 30 V DC ground check circuit to ensure adequate grounding of mobile mining equipment. During the storm, the DC ground check circuit tripped multiple times at multiple shovels, resulting in significant interruptions to operations. Figure 4 shows the ground magnetometer data from the CARISMA station at Fort McMurray (MCMU) [11], the modelled geoelectric field at MT sites within 150 km of MCMU, and the timing of the observed capacitor bank trips (∼80 km north of MCMU) and the observed mine shovel trips (∼70 km north of MCMU). Note that there is significant spatial variability in the calculated geoelectric field values shown in the bottom panel of Figure 4 which depends on the MT site location used to calculate it; within 150 km of MCMU, there are 18 MT sites and peak geoelectric field at those locations varies from <0.5 V/km up to 16 V/km, despite the fact that the interpolated driving geomagnetic field does not vary significantly on this spatial scale. The geoelectric field calculated at the MT site closest to the trip locations had peak values of approximately 2 V/km. The spatial variability is largely due to the geological heterogeneity of the Precambrian Canadian shield in this region and suggests that denser MT site coverage may be required to more accurately capture the geoelectric field behaviour in this critical region. Similar rapid variations in geoelectric field magnitude have been noted elsewhere and denser coverage is likely needed in geologically complex areas [24].

FIGURE 4

The substation capacitor banks experienced several trips between 15:15 UT on October 10 and 08:00 UT on October 11, coinciding exactly with spikes in geomagnetic activity and large, rapid spikes in the geoelectric field. This includes significant geoelectric field spikes exceeding 2 V/km at around 15:15 UT, 17:00 UT, and 23:00 UT on October 10, and around 05:00 UT and 08:00 UT on October 11. Two trips did not coincide with large spikes in the geoelectric field although they did coincide with a general period of increased geomagnetic activity between 19:00 and 20:30 UT on October 10. Note that the trips occurred on medium voltage (34.5 kV) capacitor banks which were not modelled in our existing GIC model. These trips likely occurred due to increased total harmonic distortion, which exceeded 40% during the storm. Quasi-DC GIC at 240 kV voltage level cannot flow to lower voltage levels for standard Yg-Yg windings, but harmonic distortion on the high side can be transferred across voltage levels [25].

The mine shovel DC ground check circuit tripped multiple times throughout the course of the storm, and each trip resulted in mine shovels being non-operational until reset. Nearly all mine shovel trips coincide closely with periods of increased geomagnetic activity and/or large geoelectric field spikes. This includes trips that coincide with the capacitor bank trips discussed above, such as a flurry of trips between 04:30 UT and 06:00 UT on October 11 during one of the largest 1000 nT deviations in the magnetic field observation at MCMU (Figure 4). However, there are also some trips which occurred during relatively less active periods during the storm (with associated geoelectric fields <2 V/km) between 00:00 UT and 03:00 UT on October 11. Despite the geoelectric fields being relatively small during this period, there was still quite significant geomagnetic activity at MCMU that may have resulted in localized geoelectric fields not well captured by the current MT site distribution and/or magnetic field interpolation. The mine shovels operate using 13.8 kV feeder lines which were also not modelled in the ≥240 kV GIC network model. The cause of these trips may be unrelated to GIC flowing at higher transmission voltage levels and may instead be caused by a direct trip on the low-current 30 V DC ground check circuit which is independent of transmission or distribution voltage level. Other mine shovel trips have been noted during other recent geomagnetic storm events (e.g., Nov 11–12, 2025, January 19–21, 2026), but not as many as occurred during the October 10–12, 2024, event. The storm-specific impact may be due to the length and orientation of cables which changes with time as the shovels and their cables are moved based on the requirements of the mine operations. Understanding the precise cause of these trips is beyond the scope of the current paper, but the data shows that they occur as a clear response to the local space weather conditions. As far as we are aware, this is the first documented case of long-cable mine shovel trips due to space weather reported in the literature and suggests an additional space weather impact which has not been previously discussed. This space weather impact could also affect other mining operations in other jurisdiction, but the causal connection between space weather and the trips may not have yet been made.

3.2 GIC measurements and model results

Beyond the specific impacts in northeastern Alberta, there was also notable GIC flowing in the transmission network across the whole province as indicated by both model results and the TNG GIC observations. Amongst the TNG GIC data being collected in Alberta, data are measured using a Hall probe device on the transformer neutral at a substation near Edmonton, AB, and at a substation ∼100 km west of Edmonton, AB (Figure 1) [17]. At Substation #1, which contains two 500/240 kV autotransformers, GIC exceeded 150 A total to ground (i.e. 25 A/phase in each transformer, assuming equal quasi-DC current in each phase) (Figure 5A). At Substation #2, which contains one 500/240 kV autotransformer, GIC exceeded 75 A to ground (∼25 A/phase) (Figure 5B). The modelled and observed GIC values are about half of that observed during the May 2024 storm [26], but they are still notable in terms of their magnitude, especially in the context of other GIC studies which often report GIC values < 10 A/phase in other jurisdictions, even during large storms (e.g., [27–29]). There were three other transformer banks with TNG GIC measuring devices: one in central Alberta and one in southern Alberta were measuring at transformers that were not operational at the time of the storm due to maintenance (and thus had 0 A readings), and the other was located in southern Alberta and measured much smaller magnitude GICs <5 A/phase with degraded data quality due to noise and industry-applied threshold algorithm. The model took into account the disconnected network elements and so the model also had 0 A modelled at the two disconnected transformers.

FIGURE 5

Outside of the region in central Alberta where there are large TNG GIC measurements, there were also other notable large GIC events captured by the model during the course of the geomagnetic storm using the observed geomagnetic field time series as data input. Figure 6 shows the peak modelled GIC across the network during the entire storm interval. Figure 6A shows this in map view where the largest modelled total |GIC| flowing to ground at each substation during the entire storm period from 12:00 UT on October 10 to 23:59 UT on October 12 is plotted as colored circles scaled by their magnitude. This plot indicates regions that tended to have clusters of large GIC in central Alberta (a hub for 500 kV connections and significant power generation, and where TNG GIC measurements exist), and in northeastern Alberta (primarily 240 kV connections, but in a region without any TNG GIC instrumentation is installed). This pattern has been noted in previous publications for other geomagnetic storms [17, 26] and can be attributed to fixed characteristics such as power network topology and ground conductivity [14]. Note that if a substation has multiple transformers, then the total GIC flowing to ground in the substations as shown in Figure 6 will be split between the different transformers.

FIGURE 6

Figure 6B shows the same peak GIC values flowed to ground at each substation during the entire storm interval as a function of time for each substation and colored by substation latitude. This clearly shows that there are distinct intervals of peak GIC activity across the province grouped into three clusters of time. The first occurs at ∼15:15 UT on October 10 and primarily impacts higher latitude substations, reaching >30 A/phase total to ground at the same substation where the 34.5 kV capacitor bank trips occurred. The second occurs at ∼02:00 UT on October 11 and primarily impacts lower latitude substations, but with lower GIC magnitudes <10 A/phase concentrated in southern Alberta. The third cluster occurs at ∼08:00 UT on October 11 and has a larger spatial scale which spans the entire province and impacts the majority of substations, reaching modelled peak values > 40 A/phase at Substation #1 where the measured TNG GIC was also largest in the province during this interval. These clustered times also correspond to peaks in the observed GIC (Figure 5). Capacitor bank trips in Figure 4 also occurred at ∼15:15 UT on October 10 and ∼08:00 UT on October 11, coinciding with the two distinct GIC clusters that impacted northern Alberta and all of Alberta, respectively. There were no capacitor bank trips that occurred around 02:00 UT as this cluster of peak GIC values primarily impacted substations in southern Alberta with lower overall GIC <10 A/phase.

Overall, these results show that data-driven GIC modelling results in relatively large GIC in the province with key clusters of activity both spatially and temporally. It is notable that, while the modelled GIC exceeded 30 A/phase in northeastern Alberta, these values are still relatively modest, and the storm itself was not extreme in terms of global indices, especially when compared to the May 2024 storm [4]. And yet, there were still impacts seen on the power network in the form of tripped components and other network anomalies. This highlights that impacts due to space weather can occur even during relatively moderate storm events, and even when the GIC is much less than 75 A/phase benchmark NERC values commonly used to assess transformer risk [30]. The primary impact of concern in the case of 75 A/phase is that there will be damage to transformers due to heating, but clearly other important space weather impacts can occur when GIC in the high voltage transformers is much less than 75 A/phase. Indeed, the Hydro-Québec outage of March 1989 was initiated by trips on static var compensators (similar to capacitor banks in that they manage reactive power), which then cascaded to a full blackout, and did not result in damage to transformers specifically due to excessive heating from GIC [31].

3.3 Model performance

The present version of the province-wide, high-voltage GIC model used here is seen to be a further improvement as compared to the results generated by the previous model reported by Cordell et al. [17] when compared to the observed TNG GIC measurements for a previous geomagnetic storm. The prior model from Cordell et al. [17] used a more rudimentary node-based approach which did not include transformer-specific information and resulted in GIC being underestimated by a factor of two during the April 23–24, 2023 geomagnetic storm at some substations, and good agreement at other substations. The new GIC model presented here still tends to generally underestimate the absolute value at which the GIC peaks at the Edmonton substation (top panel Figure 5), with a linear regression coefficient (i.e., slope of best-fit regression line) of 0.68 (where >1 indicates overestimate and <1 indicates underestimate). Substation #1 contains two 500/240 autotransformers each with a TNG Hall sensor, and the observed peak GIC magnitude reached 160 A total to ground in the substation (i.e., 26 A/phase in each common autotransformer winding) while the model only predicted 124 A total to ground (i.e., 20 A/phase in each common autotransformer winding). However, the correlation between the measured and modelled GIC was excellent with Pearson correlation coefficient R = 0.80. The GIC model performs better in terms of magnitude at the substation west of Edmonton (bottom panel Figure 5) with a linear regression coefficient of 0.86. Substation #2 contains one 500/240 autotransformer with TNG Hall sensor and the peak observed value of 78 A (26 A/phase in common autotransformer winding) matches the peak modelled value within ±1 A. The Pearson correlation coefficient is R = 0.73. Note that model performance and statistical summaries are also significantly influenced by an inconsistent sampling rate of the TNG GIC sensor which uses a threshold algorithm that is common in industry to limit data storage, and which leads to non-uniformly sampled data, while the GIC model has a consistent 1 Hz sample rate determined from the input geomagnetic time series (see also [17]). As a result, sometimes the TNG GIC sensor reads a constant value near-zero while the model has small magnitude fluctuations. This degrades the correlation coefficient as it is insensitive to magnitude.

It also appears that the present GIC model underestimates may not be solely related to inaccuracies of the network model since the same network model performed better for the May 10–12, 2024 geomagnetic storm at Substation #1 [26]. Thus, some of the variations seen may be due to storm-specific features, for example, such as could result from small and meso-scale geomagnetic disturbance structures which may still not be well-captured by the existing magnetometer coverage in Alberta, even though the coverage is far denser than many other jurisdictions.

To further elucidate the causes of the discrepancies between the GIC model and measured TNG GIC discussed above in Figure 5, we show a comparison between the modelled geoelectric field and a measurement of the geoelectric field in Figure 7. The measured geoelectric field has been collected continuously at the Athabasca University Geophysical Observatory (AUGSO) located at (54.608°N, 113.642°W) using a magnetotelluric data logging instrument from Phoenix Geophysics. The modelled geoelectric field is taken from the nearest MT site (abt100) approximately 16 km away, calculated using geomagnetic data interpolated at this location using the magnetometer data from the sites in Figure 3. The measured geoelectric field at this location does not exceed 1 V/km, which is much smaller than the >2 V/km geoelectric fields seen in northeastern Alberta. This is likely because the measurement location is in a region with relatively thick (>1.5 km) sedimentary cover, whereas northeastern Alberta is located on the resistive Precambrian Canadian Shield. Such large variations in geoelectric field are expected based on local geology (e.g., [32, 33]).

FIGURE 7

The geoelectric field dataset from AUGSO can be used to validate the accuracy of the geoelectric field model at this location, and the direct geoelectric field comparison here improved agreement as compared to the comparison between the GIC model and TNG measurement. The geoelectric field model has correlation coefficients >0.80 and linear regression coefficients within the range 1.0 ± 0.2, when compared to the geoelectric field measurement. While this is only one location for an electric field comparison, the fact that the geoelectric field model agrees so well with the geoelectric field data could indicate that the discrepancies in the GIC model are mostly due to power network topology effects (e.g., exclusion of lower voltage levels), or perhaps arising from localized spatial variations in the electric field along power line segments.

Regardless of the specific causes of the discrepancies, further analysis of which is beyond the scope of the present study, the available TNG GIC measurements agree reasonably well with the GIC model given the large number of physical processes at play and the long chain of events from sparse ground magnetic field data, to calculated geoelectric field from a geophysical conductivity model, to network parameters to GIC. The modelling challenge is evidenced by the overall accuracy and prediction efficiency which other GIC studies have been able to achieve (e.g., [34, 35]), some of which require scaling factors >2 to adjust modelled magnitudes (e.g., [36, 37]). Based on the agreement between the modelled and observed TNG GIC during this storm, similar comparisons for the May 2024 superstorm [26], and the large amount of accurate network data provided by industry partners, we have increased confidence in the accuracy of the model results (shown in Section 3.2), even at locations in the Alberta power grid where TNG GIC measurements are not available.

3.4 Space weather drivers

Understanding the broader space weather drivers for these key events which caused clusters of more intense GIC and coincident with trips on capacitor banks and mine shovels is important to better understand the causes of specific GIC risk during the October 10–11, 2024 storm. Although the connection to the solar wind-magnetosphere coupling processes is not the primary focus of this paper, we present a brief overview of some of the key aspects in this subsection. The solar wind conditions (interplanetary magnetic field and proton density), mapped to the nose of the magnetosphere from the OMNI dataset (https://omniweb.gsfc.nasa.gov), and its connection to ground magnetic field data and magnetic field variations at selected ground magnetometers in Alberta are shown in Figure 8 over the entire storm interval from October 10–11, 2024. The early part of the storm resulted in sustained, large ground magnetic field variations and rapid dB/dt which were primarily concentrated at northern Alberta sites (FSMI, FCHP, MCMU), a period which also coincided with low-voltage capacitor bank trips and mine shovel trips in northeastern Alberta. This period was associated with rapid variations in the interplanetary magnetic field (IMF), as well as an interval of enhanced solar wind density [4]. The related enhanced solar wind dynamic pressure, as well as its fluctuations, can act as a direct compressional driver for perturbations and waves which propagate from the magnetosphere through the ionosphere, generating geomagnetic disturbances on the ground. Similarly, the fluctuations in the z-component of the IMF could leave to bursts of magnetic reconnection which further drive geomagnetic disturbances in the magnetosphere-ionosphere-Earth system. Around 22:00 UT on October 10, the interplanetary magnetic field Bz component becomes consistently negative, with an even larger magnitude and much smaller level of fluctuations. The solar wind density is also seen to transition to much lower values at this time. This is followed by a sequence of three substorms as indicated by bays in the ground magnetometer data, and perturbations in the resulting modelled geoelectric field model at approximately 02:00 UT, 05:00 UT and 08:00 UT, each with increasing intensity and spatial footprint. Substorms during the October 2024 event were primarily confined to the main phase on October 10–11, 2024 with less activity during the recovery phase on October 12 [4].

FIGURE 8

3.4.1 Sudden storm commencement (15:15 UT Oct 10)

Figure 9 shows the geoelectric field and GIC modelled at 15:15:42 UT on October 10, 2024. This corresponds to the time of the sudden storm commencement (SSC) seen clearly in the solar wind data and associated with the arrival of the sharp density increase and increase in magnetic field seen in Figure 8. The SSC did not induce especially large geoelectric fields or GICs at latitudes <55°N in Alberta, but it did induce large geoelectric fields and GICs primarily concentrated at higher latitudes. A similar pattern was seen in Europe, with SSC driving the largest geoelectric fields at around 65° MLAT in Scandinavia (a similar magnetic latitude as seen here) [38]. The SSC for this storm was also much larger than the SSC for the May 2024 storm, and which has been attributed to an oblique angle of incidence for the incoming interplanetary coronal mass ejection as it interacted with the magnetopause [3, 38].

FIGURE 9

From a space weather hazards perspective, the SSC event was very rapid and, as observed in ground magnetometers, lasted less than 2 minutes. We use 1 s cadence magnetometer data in the GIC modelling presented here which is able to accurately capture this rapid SSC. It is worth mentioning that 1-min sampled magnetometer data would not accurately capture the induced geoelectric field or GIC resulting from this SSC event [39]. This SSC event resulted in the largest calculated geoelectric fields in the province throughout the entire storm interval, exceeding 16 V/km at 1 MT site location, and 10 MT site locations exceeding 2 V/km in northeastern Alberta. However, because it was more localized, it did not drive significant GIC across much of the province, with the model showing only a handful of substations having GIC >10 A/phase, all located in northeastern Alberta. Note that there were also six sites with large geoelectric fields in southwestern Alberta and southeastern British Columbia >2 V/km due to extremely resistive Cordilleran geology, but these were not near the ≥240 kV Alberta power network and do not impact the space weather hazard in Alberta, although they may have importance for the BCHydro power network.

3.4.2 Localized GIC event (01:50 UT Oct 11)

A second event resulting in peak GIC in southern Alberta was associated with a localized substorm at approximately 01:50 UT on October 11 (19:50 MLT on October 10). This was the first in a sequence of substorm events each approximately 3 h apart at ∼02:00, 05:00 and 08:00 UT. This event is notable because it initiated at lower latitudes <52°N in Alberta, as shown in Figures 8, 10. The more northerly magnetometers did not show any notable geomagnetic disturbances for this event, while more southerly stations did, and this results in more pronounced geoelectric fields and GIC in southern Alberta only. The event was also notable as it occurred in the local dusk sector. While this event did not result in particularly large GIC (<10 A/phase), it did correspond to the time of the largest peak GIC at many substations in southern Alberta (Figure 6). This event highlights that large storms which push the auroral oval to lower latitudes can result in substorms with onsets at lower latitudes. This has been previously noted for the May 2024 storm in Alberta [26] and has also been noted for other large geomagnetic storms (e.g., [40]). Such low-latitude substorm onsets can mean that higher latitude substations (i.e., those usually deemed to be at higher risk) may instead actually be at lower risk for substorm-related large GICs during major geomagnetic storms because they can be situated on open field lines above the auroral oval. Since Alberta was in the local dusk sector at the time of this event, it is likely that the impacts seen in southern Alberta were associated with the westward expansion of the substorm disturbances from an onset which itself was located further east and closer to local midnight.

FIGURE 10

3.4.3 Regional GIC event (08:00 UT Oct 11)

Finally, the third event was much larger in scale, with GIC impacts spanning the entire province. This event occurred around 08:00 UT on October 11 (02:00 MLT) as shown in Figure 11. This event was a more typical nightside substorm event in Alberta at this time, and resulted in some of the largest GICs across the province with four substations exceeding 30 A/phase. This event also coincided with both 13.8 kV mine shovel trips and 34.5 kV capacitor bank trips in northeastern Alberta (Figure 4). The largest geoelectric fields during this event exceeded 1 V/km across much of the province with a maximum calculated geoelectric field magnitude of 6.8 V/km in northeastern Alberta. It is also notable that the geoelectric fields were oriented mostly unidirectionally, primarily north-south across much of the province during the sustained electric field pulse associated with this event. The north-south geoelectric field direction during this pulse was oriented such that it resulted in strong coupling with the longest transmission lines in Alberta which tend to have a similar geographic orientation. This highlights that the common perception that the overhead electrojet during geomagnetic storms is dominated by east-west oriented ionospheric current systems leading to predominantly east-west aligned geoelectric fields is an oversimplification of real storm-time dynamics. As discussed in detail by Parry et al. [16], the overall electric field response can be very strongly influenced by both complex ionospheric source and Earth conductivity gradients.

FIGURE 11

The analysis presented here highlights that different space weather drivers can result in multiple, distinct intervals characterized by large GICs throughout a single geomagnetic storm, and that these drivers can have complex spatiotemporal patterns. No single event or driver is responsible for the largest GIC at every location in the province during this geomagnetic storm. Furthermore, relatively localized and variable geoelectric fields are capable of driving large GICs only in specific parts of the network (e.g., the SSC only strongly influenced northern Alberta, and the lower latitude substorm only influenced southern Alberta). Only one of the events considered here involved large, relatively unidirectional geoelectric fields across the whole grid, and even then, the direction of the geoelectric field was not as expected from simplistic assumptions of drivers arising from predominantly east-west ionospheric current systems. GIC risk assessments which make a variety of such simplifying assumptions may therefore not adequately capture the full complexity of the associated hazard, and may be unable to accurately quantify the associated risk.

4 Conclusion

This work uses the example of the October 10–11, 2024 geomagnetic storm and a model of the high-voltage Alberta power network to assess the GIC impacts associated with this large storm. Significantly, we also report on a number of operational anomalies which occurred during this storm, including capacitor bank trips which occurred on the lower voltage (34.5 kV) side of a substation in northeastern Alberta, as well as operational anomalies which occurred at large-scale mining operations in northeastern Alberta related to long (∼10 km) DC ground check circuits for large 13.8 kV mine shovels, and resulted in significant interruptions to mine operations. The capacitor bank trips and mine shovel trips coincided with periods of intense geomagnetic activity and large induced geoelectric fields >2 V/km, demonstrating a clear correlation with space weather. To our knowledge, this is the first reported case of space weather impacting DC ground check circuits on mine shovels and may be of relevance to operators of industrial equipment in other jurisdictions, especially those operating in areas prone to large geomagnetic disturbances and large geoelectric fields. Both the capacitor banks and mine shovel trips occurred at lower voltage levels <40 kV, highlighting the need for more integration between HV and UHV GIC modelling and LV distribution and end-users.

In terms of the GIC response in the Alberta high voltage network, Hall effect transformer neutral-to-ground (TNG) measurements of GIC indicate that GIC reached magnitudes >25 A/phase in transformers in central Alberta. The data-driven GIC model uses dense magnetometer and magnetotelluric datasets to estimate GIC in network elements and suggests GIC values > 30 A/phase in northeastern Alberta although no TNG GIC measurements were present in that area at the time. While the discrepancies between measured and modelled GICs at locations where measurements were available are minimal, we show how continuous geoelectric field measurements can be used to further validate the model. In our case, the geoelectric field model agrees very well with measured geoelectric field from the Athabasca University Geophysical Observatory (AUGSO) in central Alberta. We hypothesize that the largest source of error in the present model is that it is assumes no GIC in the <240 kV elements of the network. The continuous geoelectric field measurement from AUGSO did not exceed 1 V/km, but modelled geoelectric fields exceeded 10 V/km in some locations, particularly in northeastern Alberta where the underlying ground conductivity acts to enhance the induced geoelectric field. Such discrepancies highlight the value of additional geoelectric field measurements, especially in northeastern Alberta where predicted geoelectric fields are largest but where conductivity models are more poorly-constrained by sparse magnetotelluric measurements.

Finally, we highlight the spatial and temporal variability of real geomagnetic storms by analyzing discrete intervals of time associated with clusters of large GIC. The October 10–12, 2024 storm had an unusually strong sudden storm commencement at ∼15:15 UT on October 10 which resulted in large calculated geoelectric fields, concentrated in northeastern Alberta, and which exceed 15 V/km. This SSC event also coincided precisely with the times when the first 34.5 kV capacitor bank and 13.8 kV mine shovel trips were noted. Another event at 01:50 UT on October 11 impacted only southern Alberta and which we associate her with a lower latitude substorm, even though Alberta was located on the dusk flank at the time. Finally, the largest substorm at ∼08:00 UT occurred when Alberta was just post-midnight local time, and which was seen to impact the entire province. Although it did not generate the largest localized geoelectric fields, it did generate the most spatially-extensive period of large geoelectric fields which spanned the entire province, resulting in large GICs at many substations across the province within a short 10 min interval.

References

• 1.

  BotelerDH. Geomagnetic hazards to conducting networks. Nat Hazards (2003) 28(2):537–61. 10.1023/A:1022902713136

  • CrossRef
  • Google Scholar
  • View reference in article

• 2.

  IEEE. (2024) IEEE guide for establishing power transformer capability while under geomagnetic disturbances. 10.1109/IEEESTD.2023.10431702

  • CrossRef
  • Google Scholar
  • View reference in article

• 3.

  OliveiraDMZestaENandyD. The 10 October 2024 geomagnetic storm may have caused the premature reentry of a starlink satellite. Front Astron Space Sci (2025) 11(January):1522139. 10.3389/fspas.2024.1522139

  • CrossRef
  • Google Scholar
  • View reference in article

• 4.

  De MichelisPConsoliniEGiannattasioFConsoliniG. Intense ground magnetic perturbations during the 2024 may and October geomagnetic storms. J Geophys Res Space Phys (2026) 131(3):e2025JA034945. 10.1029/2025JA034945

  • CrossRef
  • Google Scholar
  • View reference in article

• 5.

  PaulKSHaralambousHMosesMTripathiSC. Effects of the October 2024 storm over the global ionosphere. Remote Sensing (2025) 17(3):2329. 10.3390/rs17132329

  • CrossRef
  • Google Scholar
  • View reference in article

• 6.

  SinghRScipiónDEKarimKCondorPFloresRPachecoEet alIonospheric responses to an extreme (G5‐Level) geomagnetic storm using multi‐instrument measurements at the jicamarca radio observatory on 10–11 October 2024. J Geophys Res Space Phys (2025) 130(4):e2024JA033642. 10.1029/2024JA033642

  • CrossRef
  • Google Scholar
  • View reference in article

• 7.

  AaEZhangS-RCosterAJHairstonMRKerrRBSouzaJRet alSuper equatorial plasma bubbles and strong longitudinal variability during the 10–11 October 2024 geomagnetic storm. J Geophys Res Space Phys (2025) 130(8):e2025JA034224. 10.1029/2025JA034224

  • CrossRef
  • Google Scholar
  • View reference in article

• 8.

  KataokaRNakanoSUchinoSReddySA. Extended red Aurora associated with super substorm igniting the October 10, 2024 magnetic storm as revealed by citizen science. Earth, Planets and Space (2025) 77(1):64. 10.1186/s40623-025-02178-w

  • CrossRef
  • Google Scholar
  • View reference in article

• 9.

  GrayverAMinJOlsenNSmithA. Assimilation of ground and satellite magnetic observations unravels the ionospheric, magnetospheric, and induced fields during the may and October 2024 geomagnetic storms. Geophys Res Lett (2025) 52(18):e2025GL116964. 10.1029/2025GL116964

  • CrossRef
  • Google Scholar
  • View reference in article

• 10.

  WatersCLMarshallRA. The relationship between the magnetometer data derived GIC index and measured GIC in high voltage transformers in Australia. Space Weather (2026) 24(1):e2025SW004665. 10.1029/2025SW004665

  • CrossRef
  • Google Scholar
  • View reference in article

• 11.

  MannIRMillingDKRaeIJOzekeLGKaleAKaleZCet alThe upgraded CARISMA magnetometer array in the THEMIS era. Space Sci Rev (2008) 141(1):413–51. 10.1007/s11214-008-9457-6

  • CrossRef
  • Google Scholar
  • View reference in article

• 12.

  WangEUnsworthM. Three-dimensional crustal and upper-mantle resistivity structure of Alberta, Canada: implications for Precambrian tectonics. Geophys J Int (2022) 230(3):1679–98. 10.1093/gji/ggac128

  • CrossRef
  • Google Scholar
  • View reference in article

• 13.

  HannesonCUnsworthMJ. Regional-scale resistivity structure of the middle and lower crust and uppermost mantle beneath the southeastern Canadian Cordillera and insights into its causes. Geophys J Int (2023) 234(3):2032–52. 10.1093/gji/ggad183

  • CrossRef
  • Google Scholar
  • View reference in article

• 14.

  CordellDUnsworthMJLeeBHannesonCMillingDKMannIR. Estimating the geoelectric field and electric power transmission line voltage during a geomagnetic storm in Alberta, Canada using measured magnetotelluric impedance data: the influence of three-dimensional electrical structures in the lithosphere. Space Weather (2021) 19(10):e2021SW002803. 10.1029/2021SW002803

  • CrossRef
  • Google Scholar
  • View reference in article

• 15.

  CordellDMannIRDimitrakoudisSParryHUnsworthMJ. Long-term peak geoelectric field behavior for space weather hazard assessment in Alberta, Canada using geomagnetic and magnetotelluric measurements. Space Weather (2025) 23(3):e2024SW004305. 10.1029/2024SW004305

  • CrossRef
  • Google Scholar
  • View reference in article

• 16.

  ParryHGCordellDRUnsworthMJMacMullinRMannIR. The influence of 3-D Earth conductivity, geoelectric field polarization, and power grid topology on GIC risk. Space Weather (2025) 23(11):e2025SW004544. 10.1029/2025SW004544

  • CrossRef
  • Google Scholar
  • View reference in article

• 17.

  CordellDMannIRParryHUnsworthMJCuiRClarkCet alModeling geomagnetically induced currents in the Alberta power network: Comparison and validation using hall probe measurements during a magnetic storm. Space Weather (2024) 22(4):e2023SW003813. 10.1029/2023SW003813

  • CrossRef
  • Google Scholar
  • View reference in article

• 18.

  GjerloevJW. The SuperMAG data processing technique. J Geophys Res Space Phys (2012) 117(A9). 10.1029/2012JA017683

  • CrossRef
  • Google Scholar
  • View reference in article

• 19.

  KelbertABalchCCPulkkinenAEgbertGDLoveJJRiglerEJet alMethodology for time-domain estimation of storm time geoelectric fields using the 3-D magnetotelluric response tensors. Space Weather (2017) 15(7):874–94. 10.1002/2017SW001594

  • CrossRef
  • Google Scholar
  • View reference in article

• 20.

  KelbertAMeqbelNEgbertGDTandonK. ModEM: a modular system for inversion of electromagnetic geophysical data. Comput and Geosciences (2014) 66(May):40–53. 10.1016/j.cageo.2014.01.010

  • CrossRef
  • Google Scholar
  • View reference in article

• 21.

  PirjolaRJBotelerDHTuckLMarsalS. The lehtinen–pirjola method modified for efficient modelling of geomagnetically induced currents in multiple voltage levels of a power network. Ann Geophysicae (2022) 40(2):205–15. 10.5194/angeo-40-205-2022

  • CrossRef
  • Google Scholar
  • View reference in article

• 22.

  BotelerDHPirjolaRJ. Comparison of methods for modelling geomagnetically induced currents. Ann Geophysicae (2014) 32(9):1177–87. 10.5194/angeo-32-1177-2014

  • CrossRef
  • Google Scholar
  • View reference in article

• 23.

  BotelerD. Methodology for simulation of geomagnetically induced currents in power systems. J Space Weather Space Clim (2014) 4:A21. 10.1051/swsc/2014018

  • CrossRef
  • Google Scholar
  • View reference in article

• 24.

  MurphyBSLucasGMLoveJJKelbertABedrosianPAJoshua RiglerE. Magnetotelluric sampling and geoelectric hazard estimation: are national-scale surveys sufficient?Space Weather (2021) 19(7):e2020SW002693. 10.1029/2020SW002693

  • CrossRef
  • Google Scholar
  • View reference in article

• 25.

  SalehSAZundelEWYoung-MorrisGJewettDBrownSHillEFSet alInvestigating harmonic distortion in power transformers due to geomagnetically induced current flows. IEEE Trans Industry Appl (2023) 59:1–11. 10.1109/TIA.2023.3287471

  • CrossRef
  • Google Scholar
  • View reference in article

• 26.

  ParryHGCordellDRBabuSSMacMullinRMannIR. GP44A-08 Comparing solar wind drivers of large geomagnetically induced currents in Alberta, Canada during the intense May and October geomagnetic storms of 2024. American Geophysical Union Fall Meeting (2025).

  • Google Scholar
  • View reference in article

• 27.

  BalchCCJingCKelbertAAronsPRichardsonK. Geoelectric field model validation in the southern California Edison system: case study. In: 2023 IEEE Energy Conversion Congress and Exposition (ECCE), October (2023). p. 6107–14. 10.1109/ECCE53617.2023.10362883

  • CrossRef
  • Google Scholar
  • View reference in article

• 28.

  DimmockAPLanabereVJohlanderARosenqvistLYordanovaEBuchertSet alInvestigating the trip of a transformer in Sweden during the 24 April 2023 storm. Space Weather (2024) 22(11):e2024SW003948. 10.1029/2024SW003948

  • CrossRef
  • Google Scholar
  • View reference in article

• 29.

  BaileyRLHalbedlTSSchattauerIAchleitnerGLeonhardtR. Validating GIC models with measurements in Austria: evaluation of accuracy and sensitivity to input parameters. Space Weather (2018) 16(7):887–902. 10.1029/2018SW001842

  • CrossRef
  • Google Scholar
  • View reference in article

• 30.

  NERC. TPL-007-4 transmission system planned performance for geomagnetic disturbance events (2019). Available online at: https://www.nerc.com/pa/Stand/Reliability%20Standards/TPL-007-4.pdf (Accessed November 24, 2023).

  • Google Scholar
  • View reference in article

• 31.

  BolducL. GIC observations and studies in the hydro-québec power system. J of Atmos and Solar-Terrestrial Phys Space Weather Effects Technol Syst (2002) 64(16):1793–802. 10.1016/S1364-6826(02)00128-1

  • CrossRef
  • Google Scholar

• 32.

  LucasGMLoveJJKelbertA. Calculation of voltages in electric power transmission lines during historic geomagnetic storms: an investigation using realistic Earth impedances. Space Weather (2018) 16(2):185–95. 10.1002/2017SW001779

  • CrossRef
  • Google Scholar

• 33.

  LoveJJLucasGMBedrosianPAKelbertA. Extreme-value geoelectric amplitude and polarization across the northeast United States. Space Weather (2019) 17(3):379–95. 10.1029/2018SW002068

  • CrossRef
  • Google Scholar

• 34.

  TortaJMMarcuelloACampanyàJMarsalSQueraltPLedoJ. Improving the modeling of geomagnetically induced currents in Spain. Space Weather (2017) 15(5):691–703. 10.1002/2017SW001628

  • CrossRef
  • Google Scholar

• 35.

  ButalaMDKazerooniMMakelaJJKamalabadiFGannonJLZhuHet alModeling geomagnetically induced currents from magnetometer measurements: Spatial scale assessed with reference measurements. Space Weather (2017) 15(10):1357–72. 10.1002/2017SW001602

  • CrossRef
  • Google Scholar

• 36.

  MarshallRAWangLPaskosGAOlivares‐PulidoGVan Der WaltTOngCet alModeling geomagnetically induced currents in Australian power networks using different conductivity models. Space Weather (2019) 17(5):727–56. 10.1029/2018SW002047

  • CrossRef
  • Google Scholar

• 37.

  Mac ManusDHRodgerCJInghamMClilverdMADalzellMDivettTet alGeomagnetically induced current model in New Zealand across multiple disturbances: validation and extension to non-monitored transformers. Space Weather (2022) 20(2):e2021SW002955. 10.1029/2021SW002955

  • CrossRef
  • Google Scholar

• 38.

  PiersantiMZurzoloS. Evaluation of geomagnetically induced currents during the SSC of the October 2024 in Europe. In: GP41A. New Orleans, USA: American Geophysical Union Fall Meeting (2025). Available online at: https://agu.confex.com/agu/agu25/meetingapp.cgi/Paper/1907722 (Accessed November 24, 2023).

  • Google Scholar

• 39.

  TrichtchenkoL. Frequency considerations in GIC applications. Space Weather (2021) 19(8):e2020SW002694. 10.1029/2020SW002694

  • CrossRef
  • Google Scholar

• 40.

  AkasofuS-I. Auroral substorms: search for processes causing the expansion phase in terms of the electric current approach. Space Sci Rev (2017) 212(1):341–81. 10.1007/s11214-017-0363-7

  • CrossRef
  • Google Scholar

• 41.

  CordellD. Darcycordell/ABGIC: aBGIC_SpaceWeather_2024_Repository. GitHub (2024). Version v2. 10.5281/zenodo.11122309

  • CrossRef
  • Google Scholar