Comprehensive and Open Assessment of Thermospheric Models During Geomagnetic Storm Times Within CCMC Framework

Abstract Thermospheric neutral density controls satellite drag in low Earth orbit and varies strongly during geomagnetic storms. We present a multi‐mission, phase‐resolved assessment of empirical and physics‐based thermosphere models using 151 storms from 2001 to 2023, including MSIS, DTM, JB2008, WAM‐IPE, WACCM‐X, GITM, CTIPe, and TIE‐GCM. High‐resolution densities from satellites are compared to model outputs using a pre‐storm debiasing procedure. Skill is quantified using the debiased mean observed‐to‐modeled ratio , standard deviation , and the Pearson correlation , evaluated separately for onset, main and recovery, and post‐storm phases. Across all phases, DTM2020 shows the best performance (ratios near unity, lowest , highest ), followed by JB2008 and DTM2013. MSIS models systematically underestimate density during the main and recovery and post–storm phases by ∼20%–30% and 10%, respectively. Based on these findings, we recommend DTM2020 and JB2008 as empirical models for satellite drag computations during the geomagnetically active periods, replacing MSIS as the standard reference. Among physics‐based models, WACCMX‐Heelis performs most reliably (ratios closest to unity, smallest ). GITM shows the best mean ratio, but its standard deviation of and are nearly twice that of other models, suggesting that its density is wider spread than others. Sensitivity to electrodynamic forcing is evident: Heelis‐driven runs of WACCM‐X and TIE‐GCM generally outperform Weimer‐driven counterparts. Latitude‐local time maps reveal persistent high‐latitude underestimation of neutral density and diurnal structure in several physics‐based models. Results are synthesized into model scorecards and delivered via the CCMC ITMAP platform to support open and reproducible, storm‐time validation and future operational benchmarking.