OpenGGCM: Global Geospace MHD Model

• OpenGGCM is a global magnetohydrodynamic (MHD) model that simulates Earth’s geospace system using a stretched Cartesian grid and a second-order predictor–corrector scheme.
• It couples the magnetosphere, ionosphere–thermosphere, and inner magnetosphere’s kinetic ring current via the Rice Convection Model to yield model-derived geomagnetic indices and ground magnetic perturbations.
• Its constrained transport method enforces negligible magnetic divergence while accounting for finite-domain boundary effects in ground magnetic field estimation.

Searching arXiv for the specified OpenGGCM-related papers to ground the article in current literature. The Open Geospace General Circulation Model (OpenGGCM) is a global magnetohydrodynamic (MHD) model of the geospace system that, in the configurations described in recent work, represents the coupled response of Earth’s magnetosphere, ionosphere, and thermosphere to time-dependent solar wind forcing, and in one coupled framework also includes the inner magnetosphere’s kinetic ring current via the Rice Convection Model (RCM) (Maharana et al., 2024). In the literature summarized here, OpenGGCM appears in three closely related roles: as a physics-based geospace model driven by heliospheric or spacecraft solar wind inputs to predict geomagnetic activity and indices (Maharana et al., 2024); as one of several global MHD models used to compare modeled and observed horizontal ground magnetic perturbations during the May 2024 geomagnetic storm (Wilkerson et al., 9 Jul 2025); and as a reference case for examining how magnetic divergence control and finite-domain boundary terms affect Biot–Savart estimates of the magnetic field at Earth derived from MHD simulations (Thomas et al., 25 Aug 2025). Across these uses, a recurring theme is that OpenGGCM’s constrained-transport treatment of the magnetic field keeps $\nabla \cdot \mathbf{B}$ extremely small, while finite-domain outer-boundary effects remain non-negligible in ground magnetic field estimation (Thomas et al., 25 Aug 2025).

1. Model definition and coupled geospace architecture

OpenGGCM is described as a global magnetosphere model that solves a semi-conservative form of the MHD equations on a three-dimensional, stretched Cartesian grid using a second-order predictor–corrector finite-difference scheme (Thomas et al., 25 Aug 2025). In the coupled EUHFORIA–OpenGGCM framework, the outer magnetosphere is solved as a single-fluid, semi-conservative MHD system on a 3D stretched Cartesian grid in GSE coordinates, and the model is electrodynamically coupled to the Coupled Thermosphere Ionosphere Model (CTIM) and, crucially, to the inner magnetosphere’s kinetic ring current via the Rice Convection Model (Maharana et al., 2024).

The three subsystems identified in that configuration are the magnetosphere, the ionosphere–thermosphere system, and the inner magnetosphere ring current (Maharana et al., 2024). The ionosphere–thermosphere coupling interface is at $2.1\,R_E$, where field-aligned currents computed by the magnetosphere close through the ionosphere, while magnetospheric precipitation parameters and electric fields drive ionospheric conductances and dynamo currents (Maharana et al., 2024). CTIM returns height-integrated conductances and dynamo currents to solve the ionospheric electric potential, which feeds back to the magnetosphere (Maharana et al., 2024). The RCM is embedded to represent kinetic ring-current physics, with a dayside boundary of approximately $10\,R_E$, a nightside boundary of approximately $13\,R_E$, and flanks at $\pm 12\,R_E$; it is initialized from OpenGGCM’s MHD density and pressure and returns corrected ring-current pressure and density to the MHD solution (Maharana et al., 2024).

This coupled architecture matters because the recent literature does not treat OpenGGCM merely as a magnetopause-scale MHD solver. Instead, it is used as a system model whose outputs include both magnetospheric state variables and ionospheric electrodynamic quantities, and whose ring-current treatment is sufficiently explicit to support model-derived geomagnetic indices such as Dst and auroral indices (Maharana et al., 2024). A plausible implication is that OpenGGCM’s operational and research value depends not only on the outer-MHD numerics but also on the fidelity of its electrodynamic closures and inner-magnetosphere coupling.

2. Governing equations, numerics, and divergence control

In the cited work, OpenGGCM solves the ideal MHD equations in semi-conservative form with source terms and numerical divergence control via constrained transport (CT) (Maharana et al., 2024). The equations reported include continuity,

$$\frac{\partial \rho}{\partial t} + \nabla \cdot (\rho \mathbf{v}) = 0,$$

momentum in Maxwell-stress form, induction,

$$\frac{\partial \mathbf{B}}{\partial t} = \nabla \times \left(\mathbf{v} \times \mathbf{B} - \eta \,\nabla \times \mathbf{B}\right),$$

and total-energy evolution with adiabatic closure, with $\gamma$ typically $5/3$ (Maharana et al., 2024). The model uses an explicit second-order predictor–corrector scheme and a stretched Cartesian mesh (Maharana et al., 2024).

A central numerical property emphasized in the literature is OpenGGCM’s treatment of the solenoidal constraint. To control magnetic divergence, OpenGGCM employs the Constrained Transport method on a staggered grid, and CT “maintains $\nabla \cdot \mathbf{B} = 0$ to round-off error” (Thomas et al., 25 Aug 2025). No ad hoc divergence cleaning parameters, such as diffusive cleaning coefficients or hyperbolic/parabolic cleaning speeds, are used; rather, the algorithmic choice of CT with staggered storage and edge-centered updates by the curl of the electromotive force enforces the solenoidal constraint intrinsically (Thomas et al., 25 Aug 2025). Historically, the main disadvantage of CT is identified as the difficulty of extending it to generalized grids, with Evans and Hawley, Gombosi et al., and Mignone et al. cited in the paper summarized by (Thomas et al., 25 Aug 2025).

In the specific coupled configuration described in (Maharana et al., 2024), the computational domain contains $2.1\,R_E$0 cells in $2.1\,R_E$1 spanning $2.1\,R_E$2 to $2.1\,R_E$3, and $2.1\,R_E$4 cells each in $2.1\,R_E$5 and $2.1\,R_E$6 spanning $2.1\,R_E$7 to $2.1\,R_E$8, with minimum cell sizes in the inner magnetosphere of approximately $2.1\,R_E$9 in $10\,R_E$0 and $10\,R_E$1 in $10\,R_E$2. Output is produced as 3D MHD fields at hourly cadence, 2D magnetosphere cuts at one-minute cadence, and ionosphere electrodynamic fields at one-minute cadence (Maharana et al., 2024).

The significance of these details is clearest in the magnetic-field reconstruction study. There, OpenGGCM’s CT scheme is directly linked to the finding that divergence-related corrections to ground magnetic perturbation estimates are very small (Thomas et al., 25 Aug 2025). This suggests that, among errors arising in finite-volume magnetic-field inference from MHD simulations, OpenGGCM suppresses non-solenoidal contamination more effectively than schemes that rely on approximate divergence transport or cleaning.

3. Finite-volume magnetic-field decomposition and ground perturbation estimation

A substantial recent analysis of OpenGGCM concerns the mathematical consistency of estimating the magnetic field at Earth from simulation currents using the Biot–Savart law (Thomas et al., 25 Aug 2025). Maxwell’s equation enforces

$10\,R_E$3

but numerical error may violate this condition in simulations (Thomas et al., 25 Aug 2025). The magnetospheric contribution to the magnetic field at a point $10\,R_E$4 on Earth from current density $10\,R_E$5 is estimated as

$10\,R_E$6

where $10\,R_E$7 is the magnetospheric simulation volume and $10\,R_E$8 is the permeability of free space (Thomas et al., 25 Aug 2025).

The study emphasizes that Helmholtz decomposition on a finite domain requires both volume and surface terms (Thomas et al., 25 Aug 2025). For a point $10\,R_E$9 in the inner gap region $13\,R_E$0, the paper writes the Biot–Savart contribution compactly as $13\,R_E$1 and derives

$13\,R_E$2

showing that the Biot–Savart volume integral is balanced by a volume integral involving $13\,R_E$3 and surface integrals on the outer and inner boundaries (Thomas et al., 25 Aug 2025). The complementary formula for the gap region is

$13\,R_E$4

which the authors note is a second way to compute the magnetospheric contribution to $13\,R_E$5 on Earth when the inner region is treated as dipolar (Thomas et al., 25 Aug 2025). The paper further defines

$13\,R_E$6

and states the consistency condition

$13\,R_E$7

so that Biot–Savart alone is exact only when both the divergence volume term and outer-boundary surface term vanish (Thomas et al., 25 Aug 2025).

Within this framework, the outer-boundary surface integral has a specific physical interpretation. $13\,R_E$8 is the contribution arising from the finite extent of the simulation; when the integration domain is not all-space, Helmholtz decomposition introduces a surface term over $13\,R_E$9 that encodes how the boundary fields and their normal derivatives, through the Green’s function for the Poisson operator, transmit information from outside the magnetospheric volume into the interior solution for $\pm 12\,R_E$0 (Thomas et al., 25 Aug 2025). In the reported calculations, this term is evaluated numerically by interpolating the boundary integrands over the outer surface and integrating (Thomas et al., 25 Aug 2025).

For OpenGGCM specifically, the current density used in these integrals is the model-provided $\pm 12\,R_E$1 from the MHD outputs, and the analysis verifies that $\pm 12\,R_E$2 in the quasi-steady MHD regime (Thomas et al., 25 Aug 2025). The inner boundary is a sphere of radius $\pm 12\,R_E$3 centered at Earth, and the outer boundary is a rectangular prism in GSE coordinates spanning $\pm 12\,R_E$4 from $\pm 12\,R_E$5 to $\pm 12\,R_E$6 and $\pm 12\,R_E$7 from $\pm 12\,R_E$8 to $\pm 12\,R_E$9 (Thomas et al., 25 Aug 2025). For the $$\frac{\partial \rho}{\partial t} + \nabla \cdot (\rho \mathbf{v}) = 0,$$0 volume term, second-order finite-difference stencils are used to compute $$\frac{\partial \rho}{\partial t} + \nabla \cdot (\rho \mathbf{v}) = 0,$$1 in the magnetospheric volume (Thomas et al., 25 Aug 2025).

The broader significance is methodological. The study shows that magnetic perturbations at Earth inferred from MHD simulations are not determined by magnetospheric current density alone unless the finite-domain correction terms are negligible. In OpenGGCM, the divergence term is nearly suppressed by construction, but the outer-boundary term is not, so finite-domain consistency remains essential (Thomas et al., 25 Aug 2025).

4. Quantitative behavior of divergence and boundary terms in OpenGGCM

The quantitative findings reported for OpenGGCM are unusually specific. In the simulations examined, when $$\frac{\partial \rho}{\partial t} + \nabla \cdot (\rho \mathbf{v}) = 0,$$2 is large, such as after a southward turning of the interplanetary magnetic field, surface-averaged ratios over Earth’s surface show $$\frac{\partial \rho}{\partial t} + \nabla \cdot (\rho \mathbf{v}) = 0,$$3 at about $$\frac{\partial \rho}{\partial t} + \nabla \cdot (\rho \mathbf{v}) = 0,$$4 of $$\frac{\partial \rho}{\partial t} + \nabla \cdot (\rho \mathbf{v}) = 0,$$5 and $$\frac{\partial \rho}{\partial t} + \nabla \cdot (\rho \mathbf{v}) = 0,$$6 of $$\frac{\partial \rho}{\partial t} + \nabla \cdot (\rho \mathbf{v}) = 0,$$7 (Thomas et al., 25 Aug 2025). When $$\frac{\partial \rho}{\partial t} + \nabla \cdot (\rho \mathbf{v}) = 0,$$8 is small, both ratios increase in percentage terms, although OpenGGCM’s divergence contribution remains very small because of CT (Thomas et al., 25 Aug 2025).

The analyzed solar wind driving for the OpenGGCM run uses GSM conditions at $$\frac{\partial \rho}{\partial t} + \nabla \cdot (\rho \mathbf{v}) = 0,$$9 sunward with density $$\frac{\partial \mathbf{B}}{\partial t} = \nabla \times \left(\mathbf{v} \times \mathbf{B} - \eta \,\nabla \times \mathbf{B}\right),$$0, temperature $$\frac{\partial \mathbf{B}}{\partial t} = \nabla \times \left(\mathbf{v} \times \mathbf{B} - \eta \,\nabla \times \mathbf{B}\right),$$1, $$\frac{\partial \mathbf{B}}{\partial t} = \nabla \times \left(\mathbf{v} \times \mathbf{B} - \eta \,\nabla \times \mathbf{B}\right),$$2, and $$\frac{\partial \mathbf{B}}{\partial t} = \nabla \times \left(\mathbf{v} \times \mathbf{B} - \eta \,\nabla \times \mathbf{B}\right),$$3 except that $$\frac{\partial \mathbf{B}}{\partial t} = \nabla \times \left(\mathbf{v} \times \mathbf{B} - \eta \,\nabla \times \mathbf{B}\right),$$4 is $$\frac{\partial \mathbf{B}}{\partial t} = \nabla \times \left(\mathbf{v} \times \mathbf{B} - \eta \,\nabla \times \mathbf{B}\right),$$5 from 00:00–06:00 and flips to $$\frac{\partial \mathbf{B}}{\partial t} = \nabla \times \left(\mathbf{v} \times \mathbf{B} - \eta \,\nabla \times \mathbf{B}\right),$$6 at 06:00 and remains until 20:00 (Thomas et al., 25 Aug 2025). Dipole tilt is fixed at $$\frac{\partial \mathbf{B}}{\partial t} = \nabla \times \left(\mathbf{v} \times \mathbf{B} - \eta \,\nabla \times \mathbf{B}\right),$$7, the run spans 00:00–20:00 UTC, and snapshots are examined at 01:00, 04:00, 07:00, 10:00, and 16:00 UTC (Thomas et al., 25 Aug 2025). Before 06:00, magnetospheric contributions at the ground are small and the relative fractions of $$\frac{\partial \mathbf{B}}{\partial t} = \nabla \times \left(\mathbf{v} \times \mathbf{B} - \eta \,\nabla \times \mathbf{B}\right),$$8 and $$\frac{\partial \mathbf{B}}{\partial t} = \nabla \times \left(\mathbf{v} \times \mathbf{B} - \eta \,\nabla \times \mathbf{B}\right),$$9 are higher; after the southward turning, magnetospheric currents intensify, $\gamma$0 becomes the dominant term, and the fractional contributions fall (Thomas et al., 25 Aug 2025).

Spatial structure is also reported. Both $\gamma$1 and $\gamma$2 are spatially correlated, so nearby locations on Earth show integrals of similar magnitudes (Thomas et al., 25 Aug 2025). Scatter plots colored by geolatitude and geolongitude reveal coherent bands, reflecting the coherence of ground magnetic perturbations (Thomas et al., 25 Aug 2025). In the OpenGGCM analysis, $\gamma$3 versus $\gamma$4 shows the closest match to the “desired” behavior for justifying Biot–Savart, but there is observable anti-correlated behavior after the IMF flip, indicating systematic boundary contributions that partially offset or modify the interior current contributions (Thomas et al., 25 Aug 2025).

These results are important because they separate two distinct uncertainties in Biot–Savart-only analyses. In OpenGGCM, the self-consistency uncertainty associated with nonzero $\gamma$5 is negligible in the cases examined, whereas the approximation uncertainty from omitting the outer boundary term remains at roughly the $\gamma$6 level in surface average during strong activity and becomes relatively larger when the Biot–Savart estimate is small (Thomas et al., 25 Aug 2025). The paper therefore recommends including the outer boundary term or using the inner boundary formulation $\gamma$7 when possible (Thomas et al., 25 Aug 2025).

5. Comparisons with other MHD frameworks and implications for model interpretation

OpenGGCM is compared directly with SWMF/BATS-R-US in the boundary-integral study and with MAGE and SWMF in the May 2024 geomagnetic storm study (Thomas et al., 25 Aug 2025, Wilkerson et al., 9 Jul 2025). In the divergence-control comparison, OpenGGCM’s CT keeps $\gamma$8 at round-off, and the measured $\gamma$9 is always very small, below $5/3$0 for large $5/3$1 (Thomas et al., 25 Aug 2025). By contrast, BATS-R-US uses the eight-wave method, and the paper notes its known limitation that $5/3$2 can grow where sources are large, with measured $5/3$3 reaching $5/3$4–$5/3$5 for large $5/3$6 (Thomas et al., 25 Aug 2025). In both models, however, the outer boundary contribution is typically larger than the divergence term; in OpenGGCM it is approximately $5/3$7, whereas in BATS-R-US it is $5/3$8–$5/3$9 at times of strong activity (Thomas et al., 25 Aug 2025).

The event-type sensitivity reported in (Thomas et al., 25 Aug 2025) is qualitative but consistent across scenarios. In the simple IMF-flip scenario, both OpenGGCM and BATS-R-US show smaller fractional correction terms when $\nabla \cdot \mathbf{B} = 0$0 is large and larger fractions when $\nabla \cdot \mathbf{B} = 0$1 is small (Thomas et al., 25 Aug 2025). In a separate Carrington superstorm case analyzed with BATS-R-US, the same pattern holds, and the paper concludes that the trends are robust across a simple IMF step and a complex storm (Thomas et al., 25 Aug 2025). OpenGGCM’s specific superstorm results are not provided there.

A separate caution concerns inter-model spread in ground magnetic estimates. The paper notes that differences between MHD models can reach factors of approximately $\nabla \cdot \mathbf{B} = 0$2 for specific values of $\nabla \cdot \mathbf{B} = 0$3 at ground, although trends are similar and cross-model agreement improves when $\nabla \cdot \mathbf{B} = 0$4 is large (Thomas et al., 25 Aug 2025). It also states that Biot–Savart has known limitations in strong, non-symmetric, and time-dependent situations, and that MHD solutions exhibit finite signal propagation times, so these physical limitations should be considered when interpreting fast ground magnetic changes (Thomas et al., 25 Aug 2025).

These comparisons support an interpretation in which OpenGGCM’s main advantage in this context is not the elimination of all magnetic-field inference errors, but the strong suppression of divergence-related error. This suggests that model-to-model discrepancies in ground perturbation products may persist even when solenoidality is well controlled, because finite-volume boundary effects and broader physical-model differences remain.

6. Forecasting workflows, geomagnetic indices, and heliospheric coupling

OpenGGCM is also used as the geospace component of a forecasting chain in which EUHFORIA provides time-dependent solar wind and IMF forcing at Earth (Maharana et al., 2024). In that framework, OpenGGCM is driven by plasma velocity $\nabla \cdot \mathbf{B} = 0$5, magnetic field $\nabla \cdot \mathbf{B} = 0$6, proton number density $\nabla \cdot \mathbf{B} = 0$7, and thermal pressure $\nabla \cdot \mathbf{B} = 0$8 obtained from EUHFORIA simulations, with EUHFORIA providing 1D time series at Earth with 10-minute cadence (Maharana et al., 2024). A relaxation window of approximately $\nabla \cdot \mathbf{B} = 0$9–$2.1\,R_E$00 hours is required before the main phase, and in practice ingestion starts approximately $2.1\,R_E$01–$2.1\,R_E$02 hours before the CME shock arrival so that the ejecta arrives after the relaxation period (Maharana et al., 2024). To avoid numerical instabilities from abrupt IMF changes, the simulation is initialized with a low but non-zero $2.1\,R_E$03 of about $2.1\,R_E$04 (Maharana et al., 2024).

The coupled framework validates two observed geoeffective CME events. The first is the 2012-07-12 CME, simulated in EUHFORIA v2.0 with the magnetized spheromak flux-rope CME model using insertion time 2012-07-12 19:24 UT, speed $2.1\,R_E$05, longitude $2.1\,R_E$06, latitude $2.1\,R_E$07, radius $2.1\,R_E$08, density $2.1\,R_E$09, temperature $2.1\,R_E$10, helicity $2.1\,R_E$11, tilt $2.1\,R_E$12, and toroidal magnetic flux $2.1\,R_E$13 (Maharana et al., 2024). The second consists of three interacting CMEs in September 2017, also represented with spheromaks and explicitly parameterized in speed, direction, size, helicity, tilt, and toroidal flux (Maharana et al., 2024). The paper states that these parameters control the modeled IMF structure and solar wind forcing at Earth, especially the sign and duration of $2.1\,R_E$14 and the dynamic pressure, both of which strongly influence OpenGGCM’s geospace response (Maharana et al., 2024).

Within this workflow, OpenGGCM computes a model Dst by integrating currents in the inner magnetosphere via Biot–Savart to infer ground magnetic perturbations; this quantity represents the ring-current contribution, or “Dst*” in Burton-type terminology (Maharana et al., 2024). The magnetopause current contribution is then reintroduced by inverting the correction

$2.1\,R_E$15

with coefficients $2.1\,R_E$16, $2.1\,R_E$17, and $2.1\,R_E$18 from the O’Brien–McPherron AK2 model (Maharana et al., 2024). Auroral indices AU and AL are constructed from model ionospheric currents by Biot–Savart integration over the electrojet region, with $2.1\,R_E$19 (Maharana et al., 2024).

Model performance is assessed with dynamic time warping (DTW), using

$2.1\,R_E$20

where $2.1\,R_E$21, and a Sequence Similarity Factor

$2.1\,R_E$22

Lower SSF indicates better alignment than a null reference (Maharana et al., 2024). For the 2012 event, DTW scores are $2.1\,R_E$23 for the observed-input run and $2.1\,R_E$24 for the EUHFORIA-driven run, with $2.1\,R_E$25, giving $2.1\,R_E$26 and $2.1\,R_E$27, respectively (Maharana et al., 2024). For the 2017 event, the scores are $2.1\,R_E$28 and $2.1\,R_E$29 against $2.1\,R_E$30, giving $2.1\,R_E$31 for the observed-input run and $2.1\,R_E$32 for the EUHFORIA-driven run (Maharana et al., 2024).

Extremes reported in the event study show that the coupled framework can reproduce storm morphology while still exhibiting amplitude biases. For the 2012 event, observed Dst is $2.1\,R_E$33, while Event1-obs gives $2.1\,R_E$34 and Event1-euh gives $2.1\,R_E$35 (Maharana et al., 2024). For the 2017 event, observed Dst is $2.1\,R_E$36, Event2-obs gives $2.1\,R_E$37, and Event2-euh gives $2.1\,R_E$38 (Maharana et al., 2024). The study’s main highlight is the use of EUHFORIA simulated time series to predict Dst and auroral indices $2.1\,R_E$39 to $2.1\,R_E$40 days in advance, compared with the $2.1\,R_E$41 to $2.1\,R_E$42 hours available from L1 observations (Maharana et al., 2024).

The importance of this result is practical as well as methodological. It shows OpenGGCM functioning as the terminal geospace response model in a heliosphere-to-magnetosphere forecasting chain. At the same time, the reported overestimation tendencies and timing offsets indicate that predictive skill depends strongly on upstream $2.1\,R_E$43 and dynamic pressure accuracy and on careful initialization of the geospace simulation (Maharana et al., 2024).

7. Ground magnetic perturbations, GIC context, and operational limits

During the May 10–12, 2024 geomagnetic storm, OpenGGCM was one of three global magnetosphere MHD models used to simulate horizontal ground magnetic perturbations $2.1\,R_E$44 for comparison with magnetometer observations across the contiguous United States (Wilkerson et al., 9 Jul 2025). In that study, OpenGGCM outputs were not used to drive geoelectric fields or compute geomagnetically induced currents (GIC); instead, the comparison focused on how well model-predicted $2.1\,R_E$45 tracked measured $2.1\,R_E$46 at mid-latitude sites during the most disturbed phases of the storm (Wilkerson et al., 9 Jul 2025).

The model-predicted horizontal magnetic perturbation is defined explicitly as

$2.1\,R_E$47

and the validation uses model outputs at one-minute cadence, with observed one-second magnetometer data averaged to one minute to match (Wilkerson et al., 9 Jul 2025). The time interval analyzed for skill metrics is 2024-05-10T15Z to 2024-05-12T06Z, and the model–data comparison covers $2.1\,R_E$48 unique magnetometer sites where simulation results were available, consisting of all $2.1\,R_E$49 NERC magnetometer sites plus magnetometers near two TVA sites (Wilkerson et al., 9 Jul 2025). The upstream drivers for the simulation models are the interplanetary magnetic field and solar wind parameters from the NOAA DSCOVR spacecraft (Wilkerson et al., 9 Jul 2025).

The skill metrics are correlation coefficient $2.1\,R_E$50 and prediction efficiency (PE), defined in the same form used for GIC,

$2.1\,R_E$51

or, in the GIC implementation,

$2.1\,R_E$52

When means and variances are equal, $2.1\,R_E$53 (Wilkerson et al., 9 Jul 2025). The abstract states that $2.1\,R_E$54 predicted by MAGE, SWMF, and OpenGGCM had correlations ranging from $2.1\,R_E$55 to $2.1\,R_E$56 (Wilkerson et al., 9 Jul 2025). Mean metrics are reported explicitly for MAGE and SWMF only, not for OpenGGCM: MAGE has mean $2.1\,R_E$57 and mean $2.1\,R_E$58, whereas SWMF has mean $2.1\,R_E$59 and mean $2.1\,R_E$60 (Wilkerson et al., 9 Jul 2025). By implication and from the abstract’s range, OpenGGCM correlations fell between $2.1\,R_E$61 and $2.1\,R_E$62, but the study does not report OpenGGCM’s mean PE, RMSE, or bias (Wilkerson et al., 9 Jul 2025).

The paper’s operational conclusion is explicit: the poor prediction efficiency seen for global MHD $2.1\,R_E$63 implies that GIC derived from model $2.1\,R_E$64 would be significantly degraded relative to GIC driven by measured $2.1\,R_E$65 convolved with magnetotelluric transfer functions (Wilkerson et al., 9 Jul 2025). In the same study, measured and modeled GIC by the Tennessee Valley Authority at four sites achieved correlation coefficients above $2.1\,R_E$66 and prediction efficiencies between approximately $2.1\,R_E$67 and $2.1\,R_E$68, demonstrating the stronger performance of measured-$2.1\,R_E$69-driven methods (Wilkerson et al., 9 Jul 2025).

This establishes an important boundary on OpenGGCM’s present operational use. As deployed in that event, it provides physics-based geospace context and qualitative timing and relative magnitude of $2.1\,R_E$70, but its quantitative variance capture for ground perturbations in the contiguous United States was insufficient for reliable operational GIC forecasting when compared with methods based on measured ground magnetic fields (Wilkerson et al., 9 Jul 2025). A plausible implication is that OpenGGCM is better suited, in its current evaluated form, to geospace state interpretation and scenario forecasting than to stand-alone induction-hazard prediction at the level required by power-system operations.