GIC--Related Observations During the May 2024 Geomagnetic Storm in the United States

TL;DR

The paper investigates how the May 2024 geomagnetic storm induced GICs across the contiguous U.S. by leveraging a large observational dataset (47 GIC sites and 17 magnetometers) and comparing them with TVA-driven and reference GIC models as well as three ΔB_H models. It demonstrates that GIC predictions using TVA's network-based approach achieve strong agreement with measurements (r>0.8; predictive skill $\text{pe}$ in the $0.4$–$0.7$ range), while the reference model is notably less accurate. The study also reveals that three global magnetosphere ΔB_H models have moderate correlations with measurements but generally negative predictive skill, indicating limited use for GIC forecasting without scaling. Empirically, the maximum GIC magnitude correlates with geomagnetic latitude and local ground conductivity through factors $\alpha$ and $\beta$, with the product $\alpha\beta$ providing the strongest, albeit storm-dependent, linear relationship for hazard assessment. The results support using $\alpha\beta$-based scaling as a practical proxy for estimating GIC risk in planning and resilience efforts, and they provide a comprehensive data suite and regression framework for future storm analyses.

Abstract

The May 2024 geomagnetic storm was one of the most severe in the past 20~years. Understanding how large geomagnetic disturbances (GMDs) impact geomagnetically induced currents (GICs) within electrical power grid networks is key to ensuring their resilience. We have assembled and synthesized a large and unique set of GMD-related data, compared model predictions with measurements, and identified empirical relationships for GICs in the contiguous United States for this storm. Measurement data include GIC data from $47$ sites and magnetometer data from $17$ sites. Model data include GIC computed by the Tennessee Valley Authority (TVA) power system operators at $4$ sites, GIC computed using a reference model at $47$ sites, and the difference in the surface magnetic field from a baseline ($Δ\mathbf{B}$) computed at $12$ magnetometer sites from three global magnetospheric models -- the Multiscale Atmosphere-Geospace Environment Model (MAGE), Space Weather Modeling Framework (SWMF), and Open Geospace General Circulation Model (OpenGGCM). GIC measured and computed by TVA had a correlation coefficient $\text{r}>0.8$ and a prediction efficiency between 0.4 and 0.7. The horizontal magnetic field perturbation from a baseline, $ΔB_H$, computed by MAGE, SWMF, and OpenGGCM had a correlation r from $0.21$ to $0.65$. Two empirical relationships were considered: (1) how the correlation between measured GIC site pairs depended on differences in site separation distance, $β$ scaling factor (related to ground conductivity), and geomagnetic latitude; and (2) a regression model for the maximum $\mbox{GIC}$ magnitude at each site given the product of $α$ (related to magnetic latitude) and $β$.

Figures (10)

• Figure 1: Solar wind conditions from the DSCOVR spacecraft and select geomagnetic indices during the May 2024 storm event. AL (AE) is the Auroral electrojet index, Lower (Upper). The K$_p$ index is based on the surface $\mathbf{B}$ field variation over 3 h intervals, with values ranging from 0--9. The SYM-H index, related to Dst, is a measure of the deviation of the $\mathbf{B}$ field from its quiet-time reference. T is the temperature of the proton component of the thermal solar wind. Mag Mach (magnetosonic Mach number) is a dimensionless quantity representing the ratio of the solar wind velocity to the local Alfvén speed. V$_x$ is the $x$ component of the solar wind velocity. B$_y$ and B$_z$ are the $\mathbf{B}$ field components in the $y$ and $z$ directions, respectively.
• Figure 2: GIC and magnetometer locations and locations where model estimates are available. The approximate TVA operating region is shown in the yellow box. *Note that $\Delta\hbox{t}$ for NERC GIC and $\Delta\mathbf{B}$ measured sites varied, with GIC sites having cadences of 1 s, 2 s, 4 s, 5 s, 1 min, 5 min, and 1 h and $\Delta\mathbf{B}$ sites having cadences of 1 s, 10 s, and 1 min.
• Figure 3: GIC times series from 15 TVA sites shown with 40 A baseline offsets, sorted by geomagnetic latitude, and labeled with the site name and its geomagnetic latitude and longitude in degrees. Data from nine sites that were also found in the NERC database are indicated with an asterisk.
• Figure 4: GIC data from 32 sites from NERC's ERO portal with 40 A baseline offsets sorted by geomagnetic latitude and labeled with the NERC site ID and its geomagnetic latitude and longitude in degrees. All NERC GIC data were either measured from a three-phase transformer or scaled to be equivalent.
• Figure 5: $\Delta B_H$ measurements from 15 sites from NERC's ERO portal and 7 sites from TVA with 400 nT baseline offsets sorted by geomagnetic latitude and labeled with the site ID and its geomagnetic latitude and longitude in degrees.