Ja Soon Shim

The Geospace Dynamics Constellation (GDC) mission aims to investigate the dynamic coupling between the magnetosphere, ionosphere, and thermosphere by resolving key spatiotemporal processes at scales ranging from local to global. A key aspect is GDC's ability to reconstruct hemispheric‐scale high‐latitude electrodynamics with comprehensive measurements at multiple local times. This study evaluates the accuracy of GDC reconstructions of electric potential (PHI) and Joule Heating (JH) derived from the AMPERE‐derived electrodynamic properties of the high‐latitude ionosphere (ADELPHI) and Weimer 2005 models by comparing them to the original model outputs used as ground truth. Accuracy is assessed across four selected geomagnetic storm events with GDC reconstructions spanning from Day 91 after GDC launch to the end of the mission using a 20‐day cadence. As the mission progresses, the constellation evolves into three well‐separated orbital pairs with increasing local time separation, significantly improving the PHI and JH reconstruction accuracy. To assess accuracy, we compute several performance metrics. Results from both models show that for the specific application of reconstructing large‐scale high‐latitude electrodynamics, performance metrics improve over mission time as the constellation evolves, reflecting better sampling and constraint for the reconstructions. Notably, however, even during the early mission stage, reconstructions can sometimes be as accurate as in the later stage, depending on the fortuitous sampling of appropriate local time sectors. This quantitative assessment underscores the critical role of orbital geometry and sampling diversity in fulfilling GDC's science objectives and advancing the understanding of space weather dynamics. It also provides a tool for optimizing the GDC orbital characteristics.

Impacts of lower atmosphere forecast uncertainties on the Ionosphere‐Thermosphere (IT) system are investigated using the Whole Atmosphere Community Climate Model with Thermosphere and Ionosphere eXtension (WACCM‐X) for April 2010 and March 2013 geomagnetic storms. For each storm, a specified‐dynamics simulation (analysis run) is carried out by constraining the model dynamics using reanalysis data. Results of the analysis runs are used as initial conditions for forecast runs initialized on 20, 10, 5, 2, and 1 day before the storm onset time. The forecast runs show that errors in TEC compared to the analysis run appear in the equatorial region within 1–2 days after forecast starts with differences of about 10%. These discrepancies gradually expand to high‐latitudes after 10 days. These errors in TECs could be due to the deviations in the semidiurnal (SW2) and non‐migrating (DE3) tides that also occur within 1–2 days after forecast starts. SW2 and DE3 tides could modify the E‐region wind driven dynamo at low latitudes, affecting the vertical plasma drift in the F‐region, leading to the forecast errors in TEC. The TEC forecast errors at high‐latitudes could be due to the change in the column integrated O/N₂, associated with tidal wind variations and resultant delayed change in vertical motions. The SW2 and DE3 tides can be affected by uncertainties in winds in the mesosphere and lower thermosphere (MLT) in the mid‐to‐high latitudes. The MLT wind uncertainties are correlated with gravity wave drag (GWD), suggesting that the uncertainties in GWD can be one of the major sources of IT forecast errors.

Prediction of ionospheric state is a critical space weather problem. We expand on our previous research of medium‐range ionospheric forecasts and present new results on evaluating prediction capabilities of three physics‐based ionosphere‐thermosphere models (Thermosphere Ionosphere Electrodynamics General Circulation Model, TIE‐GCM; Coupled Thermosphere Ionosphere Plasmasphere Electrodynamics Model, CTIPe; and Global Ionosphere Thermosphere Model, GITM). The focus of our study is understanding how current modeling approaches may predict the global ionosphere for geomagnetic storms (as studied through 35 storms during 2000–2016). Prediction approach uses physics‐based modeling without any manual model adjustment, quality control, or selection of the results. Our goal is to understand to what extent current physics‐based modeling can be used in total electron content (TEC) prediction and explore uncertainties of these prediction efforts with multiday lead times. The ionosphere‐thermosphere model runs are driven by actual interplanetary conditions, whether those data come from real‐time measurements or predicted values themselves. These model runs were performed by the Community Coordinated Modeling Center (CCMC). Jet Propulsion Laboratory (JPL)‐produced global ionospheric maps (GIMs) were used to validate model TEC estimates. We utilize the True Skill Statistic (TSS) metric for the TEC prediction evaluation, noting that this is but one metric to assess predictive skill and that complete evaluations require combinations of such metrics. The meanings of contingency table elements for the prediction performance are analyzed in the context of ionosphere modeling. Prediction success is between about 0.2 and 0.5 for weak ionospheric disturbances and decreases for strong disturbances. We evaluate the prediction of TEC decreases and increases. Our results indicate that physics‐based modeling during storms shows promise in TEC prediction with multiday lead time.

The Earth's ionosphere/thermosphere (I/T) system exhibits complicated weather variability that can have adverse effects on human operations and systems, and consequently, there is a need for both accurate and reliable specifications and forecasts for this region. As part of the international effort to evaluate and assess the predictive capabilities of space weather models, four working groups for the I/T system have been created with the goal to devise a concerted model validation effort for the I/T environment. This paper presents an overview of the team efforts and reports on the progress made. As a first step, the working teams have selected to limit the validation efforts to critical I/T parameters that include total electron content, the peak density and height of the ionospheric F region, ionospheric scintillations, and thermospheric neutral densities. As part of this effort, initial lists of participating models and events have been constructed and validation data sets have been identified. In the future appropriate metrics will be selected for the various user and scientific needs.

Accurate determination of thermospheric neutral density holds crucial importance for satellite drag calculations. The problem is twofold and involves the correct estimation of the quiet time climatology and storm time variations. In this work, neutral density estimations from two empirical and three physics‐based models of the ionosphere‐thermosphere are compared with the neutral densities along the Challenging Micro‐Satellite Payload satellite track for six geomagnetic storms. Storm time variations are extracted from neutral density by (1) subtracting the mean difference between model and observation (bias), (2) setting climatological variations to zero, and (3) multiplying model data with the quiet time ratio between the model and observation. Several metrics are employed to evaluate the model performances. We find that the removal of bias or climatology reveals actual performance of the model in simulating the storm time variations. When bias is removed, depending on event and model, storm time errors in neutral density can decrease by an amount of 113% or can increase by an amount of 12% with respect to error in models with quiet time bias. It is shown that using only average and maximum values of neutral density to determine the model performances can be misleading since a model can estimate the averages fairly well but may not capture the maximum value or vice versa. Since each of the metrics used for determining model performances provides different aspects of the error, among these, we suggest employing mean absolute error, prediction efficiency, and normalized root mean square error together as a standard set of metrics for the neutral density.

An objective of the solar and space physics communities has been to predict the behavior of the interconnected physical systems that bring space weather to Earth. One approach is to use first-principles models that may predict behavior of the various space plasma regimes from the magnetized solar corona to Earth’s upper atmosphere. We focus on space weather forecasts in the thermosphere-ionosphere (T-I), with lead time based on the period following a solar eruption. There are generally 1-4 days lead time before the interplanetary coronal mass ejection (ICME) reaches the Earth’s magnetopause. Forecasting the behavior of the T-I with such multi-day lead times requires new ways of using and assessing first principles models, which are capable of predicting many details of the T-I response, including the time history of the global electron density distribution, neutral densities and neutral winds. All facets of the complex T-I system response must be predicted based on input solar and interplanetary parameters. Another influence on the forecast is the condition of the T-I at the time a forecast is produced (e.g. shortly after the CME eruption epoch). However, the role of such pre-conditioning is not well understood for lead times of a few days. To improve our understanding of these forecasts, we have submitted more than 120 multi-day simulation periods to NASA’s Community Coordinated Modeling Center, spanning three coupled T-I models. Approximately 40 T-I storms have been simulated, driven by solar wind and EUV parameters alone. We will present an analysis that characterizes how T-I models respond to the information content of the solar wind, mediated through climatological models of high latitude forcing, and the possible influence of pre-existing conditions. Smoothing across mesoscale variability is inevitable in this scenario. Analyzing the response across events and across models reveals critical information about the predictability of the T-I system as an ICME approaches.

This paper presents preliminary results of a metrics‐based assessment of current modeling capabilities in predicting the ionospheric climatology for two ionospheric characteristics, the foF2 and hmF2. This work is part of the activities of the Ionosphere Working Group of the International Forum for Space Weather Modeling Capabilities Assessment that is facilitated by the Community Coordinated Modeling Center. The results are based on the comparison between modeled and observed monthly medians of the foF2 and hmF2. Modeling formulations considered here include two empirical models (the International Reference Ionosphere [IRI] and Massachusetts Institute of Technology [MIT] Empirical model), as well as two physics‐based, coupled ionosphere‐thermosphere models (the Coupled Thermosphere Ionosphere Plasmasphere Electrodynamics [CTIPe] and Thermosphere‐Ionosphere‐Electrodynamics General Circulation Model [TIE‐GCM]). The comparisons were carried out in year 2012 over seven ionospheric locations distributed worldwide in the Northern Hemisphere. Based on our findings, although significant errors are occasionally obtained from all modeling approaches, empirical models tend to provide systematically better correlation with the observed medians and follow the observed distributions more successfully, offering smaller prediction errors than the physics‐based ones. In case of foF2, the averaged prediction accuracy in terms of root‐mean‐square error (RMSE) ranges around 1.5 MHz for physics‐based models and 0.5 MHz for empirical models. In case of hmF2, it ranges around 30 and 20 km, respectively. The prediction accuracy presents strong seasonal and local time dependence, which differs for different models and ionospheric characteristics (i.e., foF2 or hmF2): it is strong for physics‐based models, weaker for IRI, and very weak for MIT empirical model. This latter outcome may provide suggestive input for future improvements.

To address challenges of assessing space weather modeling capabilities, the Community Coordinated Modeling Center is leading a newly established International Forum for Space Weather Modeling Capabilities Assessment. This paper presents preliminary results of validation of modeled foF2 (F₂ layer critical frequency) and TEC (total electron content) during the first selected 2013 March storm event (17 March 2013). In this study, we used eight ionospheric models ranging from empirical to physics‐based, coupled ionosphere‐thermosphere and data assimilation models. The quantities we considered are TEC and foF2 changes and percentage changes compared to quiet time background, and the maximum and minimum percentage changes. In addition, we considered normalized percentage changes of TEC. We compared the modeled quantities with ground‐based observations of vertical Global Navigation Satellite System TEC (provided by Massachusetts Institute of Technology Haystack Observatory) and foF2 data (provided by Global Ionospheric Radio Observatory) at the 12 locations selected in middle latitudes of the American and European‐African longitude sectors. To quantitatively evaluate the models' performance, we calculated skill scores including correlation coefficient, root‐mean square error (RMSE), ratio of the modeled to observed maximum percentage changes (yield), and timing error. Our study indicates that average RMSEs of foF2 range from about 1 MHz to 1.5 MHz. The average RMSEs of TEC are between ~5 and ~10 TECU (1 TEC Unit = 10¹⁶ el/m²). dfoF2[%] RMSEs are between 15% and 25%, which is smaller than RMSE of dTEC[%] ranging from 30% to 60%. The performance of the models varies with the location and metrics considered.

In order to assess current modeling capability of reproducing storm impacts on total electron content (TEC), we considered quantities such as TEC, TEC changes compared to quiet time values, and the maximum value of the TEC and TEC changes during a storm. We compared the quantities obtained from ionospheric models against ground‐based GPS TEC measurements during the 2006 AGU storm event (14–15 December 2006) in the selected eight longitude sectors. We used 15 simulations obtained from eight ionospheric models, including empirical, physics‐based, coupled ionosphere‐thermosphere, and data assimilation models. To quantitatively evaluate performance of the models in TEC prediction during the storm, we calculated skill scores such as RMS error, Normalized RMS error (NRMSE), ratio of the modeled to observed maximum increase (Yield), and the difference between the modeled peak time and observed peak time. Furthermore, to investigate latitudinal dependence of the performance of the models, the skill scores were calculated for five latitude regions. Our study shows that RMSE of TEC and TEC changes of the model simulations range from about 3 TECU (total electron content unit, 1 TECU = 10¹⁶el m⁻²) (in high latitudes) to about 13 TECU (in low latitudes), which is larger than latitudinal average GPS TEC error of about 2 TECU. Most model simulations predict TEC better than TEC changes in terms of NRMSE and the difference in peak time, while the opposite holds true in terms of Yield. Model performance strongly depends on the quantities considered, the type of metrics used, and the latitude considered.

Upper atmospheric neutral winds and temperature have been observed by the Fabry‐Perot interferometer (FPI) which was installed at Jang Bogo Station (JBS), Antarctica, in 2014. Since JBS is mostly located within the polar cap region, the observed thermospheric winds at 250 km show strong diurnal variations with notable antisunward motions due to the effects of plasma convection. The winds at 87 km, on the other hand, show semidiurnal variations due to the lower atmospheric tidal effects. We found that the winds from green line emission show largely diurnal variations, unlike the other independent observations, which might be due to the auroral contamination. The horizontal wind model 2014 winds at 250 km show reasonable agreement with FPI winds, while large discrepancies exist at 87 km in terms of seasonal variations. There is a distinctive asymmetric seasonal variation of the thermospheric zonal wind in the dusk and dawn sectors. FPI temperatures at 250 km show a fairly close correlation with Kp index, especially in 2015 when the geomagnetic activity is stronger. However, the temperatures at 87 km are mostly independent on Kp index. Finally, during the intense geomagnetic storm, thermospheric winds and temperature are significantly disturbed and Thermosphere Ionosphere Electrodynamics General Circulation Model simulations are in surprisingly good agreement with observation for winds but the temperatures are significantly underestimated during the whole storm period.