Importance of Coordinate-Invariant Contribution Functions for Thermospheric Emission Analysis: Application to Data-Model comparisons during April 2023 Storm

Geomagnetic storms strongly perturb the thermosphere, but the resulting temperature changes at 120–200 km altitudes remain poorly characterized. Neutral temperatures measured by far-ultraviolet (FUV) imaging spectrographs such as Global-scale Observations of the Limb and Disk (GOLD) are not fixed-altitude measurements; instead, they represent a broad, geometry- and altitude-dependent weighting of the N2 Lyman–Birge–Hopfield (LBH) emission layer within this altitude range. In contrast, models such as the Global Ionosphere–Thermosphere Model (GITM) provide temperature at fixed altitudes. To enable retrieval-consistent data–model comparison, we use contribution functions as solar zenith angle (SZA)- and altitude-dependent weighting kernels. These kernels are transformed into pressure space, combined with line-of-sight (LOS) attenuation, and extended across changing solar EUV conditions using Wasserstein optimal transport interpolation. We apply this method to GITM simulations and GOLD observations during the 23–24 April 2023 geomagnetic storm to illustrate the importance of contribution functions in data–model comparison. The storm reached SYM/H near -233 nT, and GOLD observed strong spatially varying neutral temperature perturbations. Using this weighting approach, we compute an effective temperature (Teff) that is significantly more consistent with the GOLD disk neutral temperature (Tdisk). Relative to a conventional fixed 160 km method, contribution-function weighting shifts modeled temperature toward the GOLD retrieval, reduces spread, lowers RMSE across latitude, and improves point-by-point consistency. These results show that storm-time uplift of constant-pressure surfaces significantly changes the N2 LBH emitting layer sampled by GOLD and must be accounted for in data–model comparisons. By comparing multiple GITM simulations, we further isolate the contributions of geomagnetic forcing and EUV radiation to the temperature response. Our results show that geomagnetic forcing was the dominant driver, producing perturbations locally up to about 20% of the quiet-time baseline, while the concurrent EUV decrease contributed only a small fraction, about 1–3%, of the change. Overall, this study demonstrates that retrieval-consistent temperature diagnostics are essential for robust data–model comparison.