Tongan Volcanic Eruption Induced Global‐Scale Thermospheric Changes Observed by the GOLD Mission

The 2022 Tongan volcanic eruption released significant energy into the atmosphere. Tropospheric satellite images show that the eruption generated pressure waves that traveled globally. The Global Observation of the Limb and Disk (GOLD) mission observed significant wave‐like thermospheric temperature perturbations (>100 K) from 12 to 16 UT. These temperature perturbations' spatial curvatures and arrival times are initially similar to the tropospheric wave‐fronts but differ significantly with eastward propagation. The perturbations had a phase speed of ∼300–400 m/s and wavelengths greater than 2,400 km. Near‐concurrent Ionospheric Connection Explorer neutral wind measurements suggest that the eruption's effects reversed the direction of the prevailing thermospheric zonal winds around the perturbed regions. The eruption's global and whole atmospheric effects provide a unique opportunity to study how different atmospheric layers exchange energy and momentum during explosive events. GOLD's synoptic observations are uniquely positioned to study these effects in the middle thermosphere.


Tongan Volcanic Eruption Induced Global-Scale
We present the effect of the eruption in the middle thermosphere (near 150 km) as observed by the Global-scale Observations of Limb and Disk (GOLD, Eastes et al., 2017) mission in this paper. Additionally, Ionospheric Connection Explorer (ICON, Immel et al., 2017) mission's neutral wind observations near 150 km are used to trace wind-pattern changes when near concurrent measurements were present. The GOLD mission has been making synoptic disk observations of daytime thermospheric composition and temperature since late 2018. The studies using the GOLD data include temperature changes during various geomagnetic activities (Laskar et al., 2021); composition changes during a solar eclipse , nighttime ionospheric structures Karan et al., 2020), and geomagnetic storms (Cai et al., 2020(Cai et al., , 2021(Cai et al., , 2023Correira et al., 2021;Gan et al., 2020a;Karan et al., 2023). Lower and upper atmosphere coupling via atmospheric waves has also been observed by the GOLD mission (Gan et al., 2020b;Gan, Oberheide, et al., 2023;Krier et al., 2021;Oberheide et al., 2020, etc.). Please refer to Eastes et al., 2020 for details on initial scientific results from the GOLD mission. On 15 January 2022, GOLD observed significant wave-like temperature perturbations (>100 K) that were initially near-concurrent with the arrival of the pressure wave seen in the tropospheric images near 10 km. Observational results presented here and by recent studies (Burt, 2022;Wright et al., 2022;Zhang et al., 2022, etc.) show that the eruption had a global-scale and whole atmospheric impact. This provides the scientific community an opportunity to study how the different atmospheric layers respond to and exchange energy and momentum during a single explosive source on the ground. GOLD observations are well-positioned to study how the wave sources in the lower atmosphere propagate and affect the middle thermosphere near 150 km.

Observation Results
The far-ultraviolet imaging spectrometer in GOLD (∼133-165 nm) has three interchangeable entrance slits with different resolutions (0.21, 0.35, or 2.16 nm). GOLD has two identical channels with separate mirrors that can independently scan the limb or the disk of the Earth. The scans are made from a geostationary orbit above 47.5°W longitude. Scan mirrors reflect light from observed regions onto one of the entrance slits where they get dispersed. The spectral and spatial information of the dispersed light is then recorded and reduced (see Eastes et al., 2017;McClintock, Eastes, Beland, et al., 2020 andMcClintock, Eastes, Hoskins, et al., 2020 for details). The daytime disk scans are made by scanning east to west starting from the north-eastern limb and ending at the south-western limb. Both the northern and southern hemisphere scans (made separately) have ∼15-min cadences each.
For this study, we use the GOLD imaging of the daytime disk neutral temperature (TDISK) and the column atomic oxygen to molecular nitrogen ratio (O/N 2 ). One full disk image is obtained every 2 hours with a total cadence of ∼30 min. Daytime disk neutral temperature is derived using the rotational structure of N 2 Lyman-Birge-Hopfield (LBH) band spectra (138-148 nm;Evans et al., 2018). The column O/N 2 ratio is derived using the brightness ratio of O I 135.6 nm and N 2 LBH band emissions (see Correira et al., 2021). Integration of observed spectra from 135 to 137 nm for the 135.6 nm emission and from 140.5 to 148 nm for the LBH band emission is used to calculate the brightnesses. Figure 1, top row, shows GOLD observed mean disk temperatures at 12:10, 14:10, and 16:10 UT for 11-13 January 2022 while the bottom row shows the temperature on 15 January 2022, at the same UTs. Both sets of images are smoothed by taking a 2 by 2 (latitude by longitude) pixel bin running mean of the Level 2 (L2) TDISK (which has a 250 km by 250 km nadir resolution). Large-scale temperature depletion structures, spanning both hemispheres, are seen on 15 January at each UT, but not on 11-13 January. These large-scale structures are especially evident when subtracting the baseline temperatures from 15 January temperatures at the same UT ( Figure 2a). Baseline temperatures are the mean 11-13 January 2022 temperatures that are multiplied by the median temperature ratio between 15 January and the 11-13 January mean at the same UT. The scaling accounts for the overall higher temperature (by ∼200 K, ∼20%) on 15 January compared to the 11-13 January mean (please compare the top and bottom rows in Figure 1). This temperature difference was most likely due to a higher incident solar flux on 15 January and a minor geomagnetic storm that occurred a day prior. Laskar et al. (2021) reported an overall increase in the disk temperature associated even with minor geomagnetic storms. The QEUV (integrated solar EUV energy flux 1-45 nm, see Correira et al., 2021 for details) measured by GOLD ranged from 2.0 to 2.05 erg cm −2 s −1 during 11-13 January, while it was ∼10% higher (2.2 erg cm −2 s −1 ) on 15 January. Similarly, At 12:10 UT, a large-scale temperature perturbation (deviations ∼ 100 K) with respect to the baseline is seen entering the GOLD's field of view from the west in the southern hemisphere ( Figure 2a). The temperature perturbation feature is initially roughly co-incident and morphologically similar to the tropospheric pressure-wave structure seen in Geostationary Operational Environmental Satellite (GOES) imaging ( Figure 1 in Amores et al., 2022). This suggests that the eruption induced waves in the lower atmosphere acted as the source of the wave-like thermospheric temperature perturbations. These perturbation structures persist at 14:10 and 16:10 UT and are also roughly co-incident with tropospheric pressure-waves near the west coast of South America. But the spatial curvature of the perturbation features changes with eastward propagation, which is especially evident at 16:10 UT ( Figure 2a).

Neutral Temperature Changes
The ICON neutral winds (near 150 km, vectors in Figure 2a) change from eastward to westward at 14:10 UT in the region co-incident with the large-scale temperature depletion. GOLD and ICON measurements were almost concurrent near this region. GOLD temperature measurements were made at 14:17 UT while ICON winds were obtained at 14:05 UT. A recent simulation study using the High Altitude Mechanistic General Circulation Model (HIAMCM) + Model for gravity wavE SOurce, Ray trAcing and reConstruction (MESORAC) also shows a similar change in thermospheric winds within the eruption related wave-fronts (Vadas et al., 2023 and references therein). Thus, the eruption-related effects seem to induce a change in the direction of the prevailing thermospheric zonal winds.

Column O/N 2 Compositional Changes
GOLD's column O/N 2 observations on 15 January 2022, show significant changes with respect to the baseline ( Figure S1 in Supporting Information S1). A climatology of thermosphere composition seen in column O/N 2 was reported by , which could be used as a baseline to identify fluctuations associated with various sources. But, a geomagnetic storm had occurred around 23 UT on 14 January 2022, with a minimum Disturbance Storm-time (Dst) index of −94 nT. Many studies have shown strong column O/N 2 ratio enhancements at low-latitudes and depletions at higher latitudes associated with geomagnetic activity (Cai et al., 2020(Cai et al., , 2021Crowley et al., 2006;Gan et al., 2020a;Meier et al., 2005). The composition changes seen on 15 January show similar strong storm-related changes which could mask any potential eruption-related effects. However, at 12:10 UT, a depletion in the column O/N 2 ratio is seen concurrent with the eruption-induced temperature changes in the west (see Figure 2a and Figure S1 in Supporting Information S1 at 12:10 UT). Additionally, at 14:10 UT, less significant O/N 2 enhancements are seen toward south of the region where the zonal wind changes from eastward to westward ( Figure S1 in Supporting Information S1). This is also the region where the peak temperature depletion occurs at 14:10 UT (Figure 2a). Furthermore, recent studies have shown long duration Total Electron Content and O/N 2 depletion near the eruption-site (Astafyeva et al., 2022;He et al., 2023), which is consistent with less significant O/N 2 changes seen concurrent to regions with temperature perturbation minima in GOLD observations. Thus, the eruption-related effects might have also altered the normal storm time thermospheric composition changes. The wave-like temperature changes seen on 15 January deviate significantly (peak deviation ∼ 100 ± 25 K) with respect to the baselines in both hemispheres. The wave-like features are moving from the west to the east in both hemispheres. These eastward propagating temperature perturbations (shown in Figure 1, bottom row, and Figure 2a) are not associated with the geomagnetic activity because (a) storm-related perturbations propagate mostly in a meridional direction and (b) earlier studies have shown that there is an overall increase in temperature at all latitudes associated with geomagnetic storms (Laskar et al., 2021). There is a hemispheric asymmetry in temperature perturbations where the peak-to-trough southern hemispheric (local summer) deviations are stronger than in the northern hemisphere (local winter). The asymmetry perhaps results from seasonal differences.

Further Analyses and Discussion
The temperature perturbation's peak-to-peak longitudinal distance is ∼40° apart in the southern hemisphere and ∼30° apart in the northern hemisphere (see Figures 2b and 2c at 16:10 UT) corresponding to a spatial distance of ∼4,000 and 3,000 km, respectively. Considering the angle of wave propagation (∼22°, assuming concentric waves from the eruption site) with respect to the zonal direction at 25°S and 25°N average latitude, the wavelength estimates are ∼3,200 and 2,400 km for the southern and northern hemisphere, respectively. However, because of the large longitudinal sampling (∼6°), there could be smaller-scale wave structures that are not observed.
The eruption's ionospheric signatures were observed up to 72 hr later . The differential Total Electron Content (dTEC) observations used by Zhang et al. (2022) observe the topside ionosphere at altitudes greater than 300 km. Since GOLD observations are from a lower altitude (near 150 km), similar long-duration effects were expected. However, no obvious eruption-related temperature or composition changes are observed after 15 January in GOLD observations (see Movie S1 for 16 January temperature changes). In Figure 2a, the simulated pressure wave-front locations from Amores et al., 2022 are overplotted at GOLD observation UTs. These simulated wave-front locations are comparable with wave-fronts seen in GOES tropospheric images (see Figure 1 in Amores et al., 2022) and trace the great circle distance from the eruption site with a Lamb mode phase speed (∼300 m/s). As noted previously, the temperature perturbations morphology observed by GOLD are also similar with the tropospheric pressure waves seen with GOES especially at 12:10 UT. However, the temperature perturbation's spatial morphology alters with eastward propagation and differs significantly from the wave-front morphology of Amores et al., 2022 simulation (and in GOES imaging). This could be due to varying thermospheric winds at different locations that non-uniformly alter the phase velocity of the temperature perturbations. No change in the curvature of the eruption induced zonal wind perturbations was seen in a simulation that included effects of Coriolis force ( Figure 5 in Vadas et al., 2023). However, the simulation did not include effects of the geomagnetic storm that occured the day prior. GOLD's viewing line-of-sight projection effects can also be ruled out as the reason for the change in curvature because the solar terminator location in GOLD's brightness images can be accurately simulated even at oblique angles. Any significant line-of-sight projection effect would have caused a mismatch between observed and simulated terminator locations, especially at oblique angles.
Temperatures (1,000 km averages) at great circle distances away from the eruption center for 13-17 January are shown at different UT in Figure 2d. Coherent temperature depletion structures are observed from ∼12 to 18 UT on 15 January, but not on other days. We fit a linear distance equation, d = v t, where d is the distance from the eruption center, v is the phase speed and t is the time since the volcanic eruption (4:15 UT), to estimate the phase speed. Phase speeds around 300-400 m/s are obtained for the propagating temperature depletion structures. This is similar to the wave speeds estimated at other atmospheric altitudes for the event (Wright et al., 2022;Zhang et al., 2022). Figure 3 shows 15 January and baseline average temperatures as a function of great-circle distance from the Tonga eruption at 12:10 UT and 14:10 UT for northern (left) and southern scans (right), separately. Fitting a linear distance equation to the perturbation minimum from the eruption site, phase speeds of ∼383 and 336 m/s are obtained for the northern and southern scans, respectively, at 14:10 UT. This suggests that the perturbations propagated faster in the northern hemisphere. The effect is also noticeable in Figure 2a at 14:10 UT when the northern temperature depletion structure is ahead of the southern one, although the northern scans start ∼12 min earlier. Seasonal differences and/or stronger eastward thermospheric wind in the northern hemisphere are perhaps responsible for the asymmetry.
A strong equatorial electrojet (EEJ, peaks ∼ 100 km) followed by a strong counter-EEJ was measured by European Space Agency Swarm satellites Le et al., 2022). The Swarm observations were coincident with extreme neutral winds observed with ICON between 90 and 300 km . These strong winds could alter the phase speed of the temperature perturbations seen with GOLD. If there are differences in wind speeds between the northern and southern hemisphere it could cause the change in the perturbation's morphology and explain the hemispherical asymmetry in the propagation phase speed. However, no southern hemispheric ICON wind measurements were available around this period (including the baseline days) within GOLD's field-of-view (FOV). Climatological (quiet-time) studies based on the Horizontal Wind Measurement (HWM) model and NASA's Thermosphere, Ionosphere, Energetics and Dynamics mission (TIMED) Imaging Doppler Interferometer (TIDI) zonal wind measurement near 250 km during the December  Figure S2 in Drob et al., 2015). Wind measurement by the CHAllenging Minisatellite Payload (CHAMP) satellite near 450 km shows strong westward winds during storm times for both northern and southern hemispheres around mid-latitudes (near 30°) at local afternoon (see Figure 2 in Xiong et al., 2015). We are not aware of any storm-time, daytime thermospheric zonal wind measurement studies in the southern hemisphere around GOLD's FOV and observation altitude during the December solstice. Thus, we cannot verify if latitudinal difference in thermospheric wind speed is responsible for hemispheric asymmetry in the propagation phase speed and the change in the temperature perturbation's spatial morphology with eastward propagation.
The wavelength of the temperature perturbation estimated with GOLD observations is ∼2,400-3,200 km from Figures 2b and 2c. The estimation is even larger (>5,000 km) when the temperatures are binned as a function of the great circle distance from the eruption site (see Figure 3). However, this higher estimation is most likely because the perturbations only align with great circle distances in the west. The wavelength estimate of ∼2,400-3,200 km is consistent with the range of wavelength seen in ICON measurements (Vadas et al., 2023). However, it is not consistent with the wavelength of ∼500-1,000 km seen with dTEC measurements . The dTEC measurements have a much higher effective altitude (>300 km) compared to GOLD (∼150 km) and the altitudinal differences in wave propagation/dissipation pattern may account for the difference. We could also be overestimating the wavelength due to limitations in spatial resolution (sampling: ∼6°, 600 km in Figures 2b and 2c and ∼1,000 km in Figure 3).

Conclusions
The Tongan volcanic eruption on 15 January 2022, was an extreme event with global and whole-atmospheric impact. The eruption triggered a spectrum of wave signature seen across multiple atmospheric layers (Wright et al., 2022). Here, we presented the GOLD observation of the eruption-related wave-like thermospheric neutral temperature perturbations (peak changes > 100 K with respect to baselines). Near concurrent ICON neutral winds measurements (around 150 km) show that the thermospheric zonal winds change from eastward to westward around the temperature perturbation structure, suggesting the wave-like perturbation altered prevailing thermospheric winds. These global-scale perturbations arrived in GOLD's field of view about 8 hr after the eruption (Figure 1 bottom row and Figure 2a). We estimate the phase speed of the waves to be ∼300-400 m/s, similar to the wave propagation estimates for other atmospheric layers (∼320 m/s near the surface: Wright et al., 2022; 300-350 m/s in the ionosphere: Zhang et al., 2022). The spatial wavelength is estimated to be ∼2,400-3,200 km. This is not consistent with the wavelength estimation in the ionosphere, but it is consistent with the range of wavelength seen in ICON zonal wind observation (Vadas et al., 2023). The presence of smaller-scale waves that are not observed because of GOLD's spatial resolution, differences in filtering window used or changes in wave behavior at different altitudes could explain the discrepancy. The composition changes (in terms of O/N 2 ) seem to be dominated by effects of a geomagnetic storm that occurred at 23 UT on 14 January with strong O/ N 2 enhancement at low-latitudes and no obvious wave structures. However, less significant O/N 2 enhancement is observed near the peak temperature perturbations where the zonal wind also reversed from eastward to westward, suggesting that the eruptions effects possibly had some effect on compositional changes. But we cannot clearly distinguish between the storm-time and the eruption related composition changes. The spatial morphology of the temperature depletion structure changes with eastward propagation possibly altered by differences in prevailing thermospheric wind at different locations. The observation of large-scale, whole atmospheric, and long-distance waves induced by a single explosive event allows us to study how different atmospheric layers respond to explosive, impulsive events and interlayer couplings. GOLD observations are uniquely positioned to study such effects in the middle thermosphere.