First Synoptic Observations of Geomagnetic Storm Effects on the Global‐Scale OI 135.6‐nm Dayglow in the Thermosphere by the GOLD Mission

The Global‐scale Observations of the Limb and Disk (GOLD) mission provides, for the first time, a synoptic view of the thermospheric response to a moderate geomagnetic storm at a 30‐min cadence. The observations showed that the OI 135.6‐nm emission substantially decreased at middle and high latitudes, but it increased at low latitudes. The depletion pattern between 4 and 6 November 2018 was likely due to the strong upwelling of the high‐latitude thermosphere. The deepest depletion in brightness that occurred near local noon was of ~300–400 R (~35% of the quiet time background level) and displayed a striking westward movement. First‐principles modeling reasonably reproduced the overall morphology of brightness response to the geomagnetic storm despite the depletion magnitude in brightness being underestimated. The postprocessing of the simulation results revealed that the westward movement of the deepest depletion was attributable to the horizontal wind changes driven by the geomagnetic storm.


Introduction
The Earth's thermosphere is an externally driven system that accommodates complex processes of energy and momentum deposition, dissipation, and conversion. Neutral density and composition are of particular interest and importance since they affect the orbits of satellites, space debris, and plasma densities in the Earth's ionosphere (Burns et al., 2013;Crowley, 1991;Forbes, 2007;Roble, 1983;Sarris, 2019). The sources of significant thermospheric changes in neutral density and composition include geomagnetic storms and solar radiation from above (Prolss, 2011) and atmospheric waves of various scales from below (Akmaev, 2011;Liu, 2016;Oberheide et al., 2015;Vincent, 2015). Geomagnetic storms cause large changes in thermospheric densities at high latitudes via Joule heating, and the changes are then extended into the middle and low latitudes via the resulting changes in global circulation (Burns et al., 2007;Prolss, 2011;Richmond & Lu, 2000). The pioneering studies on this subject date back to Allan (1972) and references therein. Over the past two decades, considerable advances have been made in understanding the effects of geomagnetic storms on the global thermosphere by instruments in low-Earth orbit (LEO) (Bruinsma et al., 2006;Burns ©2020. The Authors. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

10.1029/2019GL085400
Key Points: • GOLD provided, for the first time, synoptic observations of the thermospheric response to a geomagnetic storm with a 30-min cadence • Dayglow brightness depletions occurred at high latitudes and were extended into midlatitudes with evident interhemispheric asymmetry • The deepest depletion occurred near local noon and moved westward during and after the geomagnetic storm Supporting Information: • Movie S1 • Figure S1 Correspondence to: Q. Gan, blondganquan@gmail.com
However, previous observational studies, especially those based on polar-orbiting LEO satellites, were incapable of separating temporospatial changes and providing a synoptic, global-scale view of geomagnetic storm effects. This is because, for instance, LEO satellites usually sample in an asynoptic way. On the other hand, synoptic measurements, which mean simultaneous global-scale measurements at uniform increments in time, can be achieved from the geostationary orbit.
In addition, thermospheric observations remain rather rare. The routine thermosphere monitoring we have is made by either ground-based instruments with limited spatial coverage at high temporal resolution or LEO satellite-based instruments with global-scale coverage at limited temporal resolution. Having continuous and global-scale observations at high time cadence enables advances in understanding of the physical processes that change the thermosphere-ionosphere and improvements in the ability of general circulation models to simulate the changes. With this objective, the National Aeronautics and Space Administration GOLD (Global-scale Observations of the Limb and Disk) mission successfully launched a far ultraviolet (FUV) imager into the geostationary orbit on 25 January 2018. The imager is a hosted payload on the SES-14 satellite.
The FUV imager has two channels; each has a plane scan mirror capable of scanning the entire disk of the Earth. The GOLD mission data products include daytime brightness of the FUV emissions, neutral temperatures, O/N 2 column density ratios, and the nighttime peak electron densities on the disk. In this study, we use disk dayglow emission measurements to describe the storm time behavior of thermospheric composition in a synoptic way for the first time. The readers are referred to Eastes et al. (2017) for the details of mission objectives, instrument, and data products. Most recently, the nighttime disk scans have been used to depict the morphology of the equatorial ionization anomaly (see Eastes et al., 2019).
This study presents the first observational evidence of the large OI 135.6-nm emission changes that occurred on the Days 309-310 in 2018 as a consequence of a geomagnetic storm. Brightness change is proportional to the change in atomic oxygen mass mixing ratio and is a proxy for the composition response to geomagnetic storms. As it is shown in Figure 1, Kp (top panel) was 5 at 1800-2400 UT on Day 308 (4 November 2018) and reached 6 at 0300-0600 UT on Day 309. The Dst index (middle panel) shows that the main phase of the geomagnetic storm occurred around 1000 UT of Day 308. The strong energy and momentum inputs due to the substorms primarily occurred between 1500 UT on Day 308 and 1400 UT on Day 309 as the AL index (bottom panel) demonstrated sharp jumps, consistent with Kp, which was greater than 4. In the OI 135.6-nm emissions, GOLD observed a substantial depletion at middle and high latitudes relative to the quiet time level. First-principles model calculation reproduced the general morphology of the geomagnetic storm effect on the brightness, but it underestimated the magnitude of the brightness depletions and overestimated their latitudinal extent. The paper is organized as follows. Section 2 overviews the GOLD data and the models used in this study. Section 3 presents the major findings. Section 4 gives the conclusions.

Data and Model
In this study, we use the atomic oxygen emission at 135.6 nm from GOLD L1C disk images measured by Channel A, which has the most extensive dayside coverage. During a scan, the 10.8°tall image of the spectrometer entrance slit crosses the disk in two swaths; one covers the northern latitudes and the other covers the southern latitudes. All disk scans are performed east to west. The L1C disk images are binned at a constant angular resolution that is equivalent to approximately 125 km × 125 km at nadir. The atomic oxygen 135.6-nm emission was integrated from 133 to 137 nm. Note that the 135.6-nm brightness persists at night due to photon production by recombination of the ionosphere. The entrance slits of each channel can provide spectral resolution of 0.21 nm or 0.35 nm, depending on the mission objectives. The former was employed during the geomagnetic storm observation we discussed here. While a band of the N 2 Lyman-Birge-Hopfield (LBH) emission (3,0 at 135.4 nm) overlaps the OI 135.6-nm peak, the contribution is small. Based on the LBH bands at~138.4 nm (5,2 and 2,0), the relative vibrational populations observed by GOLD and laboratory data (Franck-Condon factors of Benesch et al., 1966), the N 2 LBH contribution to the emission observed at~135.6 nm is only 9-12% of the total. The L1C data and documentation guide are available online (http://gold.cs.ucf.edu).
The neutral atmosphere is simulated by the TIE-GCM (Thermosphere Ionosphere Electrodynamics General Circulation Model), a first-principles general circulation model developed at National Center of Atmospheric Research. It simulates the circulation, temperature, compositions, and electrodynamics by self-consistently solving the fully coupled hydrodynamics, thermodynamics, and mass continuity equations of the neutral atmosphere along with ion energy, ion momentum, and ion continuity equations from the mesopause (~100 km) through the thermosphere (Roble et al., 1988). A recent description of the model is given by Qian et al. (2014). Lower atmospheric diurnal and semidiurnal tides are prescribed by the global-scale wave model (Hagan et al., 1999) at the model lower boundary. Low-latitude ionospheric dynamo electric fields are self-consistently calculated in the TIE-GCM using the modeled ionospheric conductivity and neutral winds (Richmond et al., 1992). Magnetospheric forcing is driven by the Heelis et al. (1982) model that gives an empirical specification of electric potential parameterized by the geomagnetic Kp index. The auroral precipitation is specified using the formulation described by Roble and Ridley (1987), which is also parameterized by the Kp index. The solar F10.7 cm index is used as a proxy of the solar extreme ultraviolet radiation over the Days 305-311 (1-7 November) in 2018. The model is run at a horizontal resolution of 2.5°(latitude) × 2.5°(longitude).
The neutral atmosphere and ionosphere from the TIE-GCM are then used by the GLOW (Global airgLOW) model to calculate the OI 135.6-nm emission. GLOW is a single-column, single-time model that is run repeatedly to generate global maps that are coincident in time with the GOLD images. It computes ionization, dissociation, and excitation rates; performs energetic electron transport calculations; applies ion-neutral chemistry; and outputs the airglow emission rates. The readers are referred to Solomon (2017) for details.  , the brightness pattern was almost symmetric about 45°W longitude since the local time of the subsatellite point was near local noon. A slightly higher brightness occurred in the Southern Hemisphere (SH) due to the smaller solar zenith angles. Note that the brightness on the edge of the disk could be the greatest due to the large emission angle, but this was not an issue in quantifying the storm-induced changes that were calculated on each individual universal time between different days, as brightness distribution due to either solar zenith angle or emission angle is not changed significantly over short periods. On Day 308 (second row, 4 November 2018), which was geomagnetically quiet before 1800 UT, the brightness relative to that of Day 307 appeared as a patch of enhancement over the Atlantic. Since the pattern is persistent and the brightness is greater than the random noise (~25 R), it may indicate a possible connection to the large-scale waves from below (Chang et al., 2010;Forbes et al., 2018;Gan et al., 2015Gan et al., , 2017Oberheide et al., 2009;Yue & Wang, 2014). A possible pathway through which atmospheric waves produce signatures in composition was suggested by Ward (1999). A detailed study is needed to establish this connection.

Results and Discussion
We now turn to the Geomagnetic Storm Day 309 (third row, 5 November 2018). The observations showed great changes in the OI 135.6-nm emission on the disk. Specifically, the depletion attained 350 R poleward of 30°N in the Northern Hemisphere (NH), equivalent to~35% of the quiet time background level. The SH middle and high latitudes showed a somewhat weaker depletion of~250 R. A boundary between low-latitude enhancements and higher-latitude depletions was situated at~20-40°( the geographic grid interval was 15°as marked in the left column, and the map was centered at 45°W over the equator) depending on longitudes in both hemispheres. For instance, the boundary in the SH moved to~20°S in the 75-90°W longitude sector, whereas the boundary was poleward of 30°S at other longitudes. Interestingly, the boundary in the NH at~90°W extended to a higher latitude of~45°N forming a mesoscale "hook" structure in all the images on Day 309. We postulate that such a localized feature might be related to the thermospheric dynamics including the Joule heating and horizontal wind divergence/convergence. This feature will be further explored in a separate paper. On Day 310 (fourth row, 6 November 2018), the geomagnetic storm effect was fading away; however, a~150 R brightness depletion can still be seen at middle and high latitudes of both hemispheres in the geomagnetic storm recovery phase. Previous studies suggested that the depletion of atomic oxygen at middle and high latitudes during a geomagnetic storm is mainly attributable to the strong upwelling as a result of enhanced Joule heating and ion drag. The low-latitude enhancement of atomic oxygen could be caused by the resulting equatorward and downward circulation (Burns et al., 1995). Overall, the GOLD observations delineated an asymmetry of the storm-induced brightness depletion and thereby atomic oxygen mass mixing ratio reduction between the two hemispheres. Namely, the higher-latitude depletion of OI 135.6-nm emission brightness in the NH was greater than in the SH, whereas the latter covered a more extensive region in the GOLD field of view. Several mechanisms have been proposed to account for such interhemispheric asymmetry in the literature. For instance, there is an asymmetry in the energy input of Joule heating and momentum inputs of ion drag in the GOLD field of view due to the geographic asymmetry of the magnetic poles. Then, the storm-induced circulation is hemisphere dependent, and the resulting boundary between low-latitude enhancement and higher-latitude depletion could be morphologically different. Meanwhile, the prevailing summer-to-winter circulation in the thermosphere (from the SH to the NH in this case) due to the solar extreme ultraviolet heating also affects the extent of the middle-and high-latitude depletion region Prolss & Zahn, 1977Qian et al., 2016. Furthermore, the B y polarity (y component of the interplanetary magnetic field) is another possible candidate to explain the magnitude and the spatial extent of depletions (Goncharenko et al., 2006;Immel et al., 1997). We would like to point out that the significant geomagnetic storm effect in brightness has been observed since the first disk scan on 0630 LT of Day 309 (see supporting information Movie S1 that shows the evolution of brightness changes on a 30-min basis from Day 308 to 310). On Day 310 when the Kp index was less than 4 (recovering phase as indicated by Dst in Figure 1), storm-driven brightness depletions still existed in the GOLD images, despite these depletions in middle and high latitudes being considerably weakened. Figure 3 illustrates a comparison of the brightness changes between GOLD observations and TIE-GCM+GLOW calculations. We selected three GOLD images for a detailed comparison, including the prestorm onset (left column) and storm-driven (right two columns) moments. All the UTs given are near local noon at the subsatellite point (less than 3-hr difference). This ensures that dayglow observations are available on the whole disk. Also, the contour level is reduced to 25 R to aid quantitative comparison of fine structures. At 1600 UT on Day 308 (left column), which was just before the storm onset, GOLD (upper left panel) observed the patches of~50 R brighter emission in both hemispheres, approximately symmetric with respect to the equator. Even though it was weak, the brightness was enhanced over an extensive area, unambiguously indicating a small enhancement of atomic oxygen densities. The model (bottom left panel) does not, however, reproduce a coincident signature.
At 1200 UT on Day 309, when the geomagnetic storm effect was dominant, the simulation (bottom middle panel) compares well with the GOLD observation (top middle panel), in particular regarding to the depletion of OI 135.6-nm emission at higher latitudes. Both the observations and the simulations exhibited the greatest brightness change (either depletion or enhancement) near local noon at those UTs, and the brightness change tended to be smaller toward the early morning sector. However, we note some discrepancies between data and model. First, the maximum depletion in middle and high latitudes was underestimated by~100-150 R by the model. The difference of the overall shape of the depletion region was evident between the observation and the simulation in the SH. The observed depletion region was extended toward the equator near the west coast of the South America, but it was not reproduced by the TIE-GCM. The possible reasons include, but are not limited to, that the empirical high-latitude inputs of convection pattern and auroral precipitation in the TIE-GCM may differ from the actual energy and momentum inputs, and the model thus would not accurately simulate the strength and shape of the depletions. A similar scenario is found when comparing data and model at 1800 UT of Day 309 (right column). Merging the results (on Day 309) in Figure 2, it is seen that the region of deepest depletion, especially in the NH, appeared to move westward between 1200 and 1800 UT. This phenomenon was qualitatively reproduced by the TIE-GCM+GLOW (see supporting information Figure S2). It thus merits a closer look at the simulation results to elaborate on the driver of this behavior. As it is seen in Figure 4, the atomic oxygen showed the deepest depletion in the region of 10-30°W and 55-65°N at 0600 UT on Day 309, and the deepest depletion moved westward with time (note that the region poleward of 70°N is not covered by the GOLD's field of view). The simulated horizontal winds were strongly southward at 0600 UT but turned westward in the following hours. A possible scenario is that Joule heating produced temperature enhancement, which set up a pressure gradient force driving equatorward winds at 0600 UT in the region where the deepest depletion occurred. Subsequently, the equatorward wind turned westward due to the balance between the Coriolis force and the pressure gradient force. Considering the high correlation between the strengthened westward winds and the atomic oxygen depletions in the same region, we thus suggest that the horizontal winds, advecting the air parcels, played a major role in the westward motion of the greatest depletion area. This indicates a strong dynamical control of the storm time thermospheric composition variability.

Conclusions
In this study, we describe the National Aeronautics and Space Administration GOLD observations of the thermospheric response to a moderate geomagnetic storm occurring on Days 309-310, 2018. This is the first time that a global-scale synoptic view of thermospheric evolution during a geomagnetic storm is achieved at this high 30-min cadence. The GOLD images demonstrated substantial depletions of~35% (300-400 R) in OI 135.6-nm emission relative to the quiet time background level at middle and high latitudes of both hemispheres. The shape of the depletion region displayed asymmetry between the NH and the SH. The deepest depletion tended to occur near local noon and displayed a westward motion. The first-principles modeling (TIE-GCM+GLOW) reproduced the overall morphology of the OI 135.6-nm emission response to the geomagnetic storm, but the model calculation underestimated the depletion strength by up to 40-50%. A further analysis of the simulation results revealed that the westward horizontal winds were the crucial driver of the westward motion of the depletion area in atomic oxygen. GOLD is providing new information of thermosphere-ionosphere dynamics with unprecedented, synoptic observations of daytime airglow emission, and a good opportunity to advance our capability to forecast space weather.