GOLD Synoptic Observations of Quasi‐6‐Day Wave Modulations of Post‐Sunset Equatorial Ionization Anomaly During the September 2019 Antarctic Sudden Stratospheric Warming

Using observations from the Global‐scale Observations of the Limb and Disk (GOLD) mission, we investigate post‐sunset ionospheric responses to the September 2019 Antarctic sudden stratospheric warming –first ever from a synoptic perspective. Observations reveal a prevalent quasi‐6‐day periodicity in the equatorial ionization anomaly region over South America and the Atlantic, coincident with enhanced quasi‐6‐day wave (Q6DW) activity in the mesosphere (Liu et al., 2021, https://doi.org/10.1029/2020JA028909). The atmosphere‐ionosphere coupling via large‐scale waves is rarely studied over the ocean due to the lack of observations. More importantly, further analyses suggest that multiple pathways are involved in transmitting the quasi‐6‐day periodicity from the middle atmosphere into the post‐sunset F‐region ionosphere, including modulation of F‐region field aligned winds and pre‐reversal enhancements by the tides and or Q6DW. A remarkable depletion in electron density, attributable to the overall change in thermosphere composition driven by the dissipative tides and or Q6DWs, is also seen during the period of enhanced Q6DW activity.

• A prominent quasi-6-day periodicity in the post-sunset equatorial ionization anomaly (EIA) is observed during the 2019 Antarctic sudden stratospheric warming, coincident with mesospheric quasi-6-day waves (Q6DWs) • The EIA displays an overall depletion over South America and the Atlantic when the Q6DWs are quite active • Multiple mechanisms play roles in coupling of the Q6DW with the post-sunset EIA

Supporting Information:
Supporting Information may be found in the online version of this article.
Over the past decade or so, following the studies by Goncharenko and Zhang (2008) and Chau et al. (2009), considerable efforts have been made in delineating ionospheric responses to SSWs in a variety of parameters at low-and mid-latitudes, including electron density, vertical ion drift, and so forth (Chau et al., 2009;Fang et al., 2014; L. P. Goncharenko et al., 2018;Yue et al., 2010). Meanwhile, much modeling work has been conducted to understand the observations (e.g., Fang et al., 2014;Pedatella, Fang, et al., 2016). Well established ionospheric anomalies due to SSWs include, but are not limited to, (a) quasi-semidiurnal disturbances in, for example, vertical ion drift, TEC (total electron content), and equatorial electrojets, as a consequence of amplified migrating semidiurnal tides (Oberheide, 2022;Pedatella, Richmond, et al., 2016) and lunar M2 tides (Forbes & Zhang, 2012); (b) ionospheric quasi-periodic variations at the periods of ∼2, 6, 10 and 16-day have been observed, associated with planetary waves in the mesosphere-stratosphere (Liu et al., 2021). Despite substantial advances being made in understanding how ionospheric disturbances are connected to the polar vortex break-up, most of the studies have focused on the Arctic SSW cases.
September 2019 marks the strongest Antarctic minor SSW event in the historical record, and it is also the only Antarctic SSW event since September 2002. This unprecedented opportunity has stimulated multiple studies to link ionospheric anomalies to the break-up of the SH polar vortex. For example, a pronounced quasi-6-day periodicity was seen in the low-latitude electron density with the Swarm satellite )-a Earth observation mission that is a constellation of three satellites launched by European Space Agency. Goncharenko et al. (2020) reported a similar periodicity in the GNSS TEC over the South American sector and suggested that the strength of the quasi-6-day periodicity is longitudinally dependent. These studies have attributed the quasi-6-day periodicity in the ionosphere to the enhanced Q6DW activity in the mesosphere. However, the prior studies have put much effort on the daytime ionospheric responses, a different question raised and discussed in this manuscript is whether or not the Q6DW-induced fluctuations would be sustained in the nighttime ionosphere.
The recently launched GOLD mission is distinctive from previous aeronomy missions . Having a unique observational sampling-also called synoptic view-enables GOLD to monitor the space weather of the ionosphere from meso-scale plasma bubbles (Eastes et al., 2019) to large-scale wave signatures (Gan et al., 2020a;Goel et al., 2022) in the EIA region across the South America, the Atlantic Ocean and the western coast of Europe and Africa. Based on this observational capability and through the combination with middle atmosphere observations made by SABER/TIMED (Sounding of the Atmosphere using Broadband Emission Radiometry/Thermosphere Ionosphere Mesosphere Energetics and Dynamics), the overarching goal of this work is to study the coupling of the post-sunset ionospheric EIA with the Q6DW propagating from the middle atmosphere during the September 2019 SSW. The manuscript is organized as follows. Section 2 overviews the observational data sets. Section 3 presents the morphology of the quasi-6-day periodicity observed by GOLD and the discussion on relevant driving mechanisms. Section 4 presents the conclusions.

Data and Methodology
The NASA (National Aeronautics and Space Administration) GOLD mission flies a Far Ultraviolet (FUV) imager onboard the SES-14 satellite in geostationary orbit. The imager is a spectrometer that monitors the OI 135.6 nm and the molecular nitrogen LBH bands emissions. In this study, we use the peak electron density data from GOLD nighttime scans measured by the two channels, covering the northern and southern hemispheres. The nighttime 135.6 nm emission produced by ionospheric recombination, which peaks in the F 2 -region where the electron density maximizes. We refer readers to Eastes et al. (2017) for a detailed description of GOLD instruments and the nighttime data product. In principle, the GOLD imager observes the EIA approximately between 8°E and 80°W in longitude and from ∼1-hr local time after sunset to later local times. The geographic area scanned by GOLD covers South America, the Atlantic Ocean, and the western coast of Europe and Africa.
SABER onboard the TIMED satellite is a 10-channel broadband radiometer covering the spectral range from 1.27 to 17 μm (Russell III et al., 1999). It provides temperature profiles retrieved from two 15 μm and one 4.3 μm CO 2 radiometer channels in a tangent height range spanning approximately 20-110 km. SABER is operated in a high-inclination low-Earth-orbit and can make measurements at two local times per day at each given latitude. The sampled local time shifts a small amount each day (12 min), allowing 24-hr local time coverage at low to middle latitudes when observations are accumulated within a continuous 60-day window. Due to the line of sight of SABER and the yaw maneuver of the TIMED satellite, the latitude coverage of SABER measurements is from 53° latitude in one hemisphere to 83° in the other. In the current work, the latest version 2.08 temperature data are utilized to trace planetary waves in the middle atmosphere.
GOLD observers the nighttime ionosphere at the same local times and locations on every single day-therefore called synoptic observations in contrast to the observations made by satellites in a low Earth orbit. This allows us to perform principal component analysis and appropriately derive the ionospheric signatures driven by the Q6DW. PCA is a well-established approach that has a unique advantage in deriving the leading spatial structures in state variables as well as their temporal variations (Kutzback, 1967). Gan et al. (2020a) successfully applied PCA to the GOLD nighttime images, where a quasi-16-day modulation in the crests region of the EIA was readily identified over the American sector. In the present study, we further expand the PCA analysis from a given longitude sector to the whole disk, achieving a full characterization of the quasi-6-day wave modulation of the EIA within the GOLD's field-of-view. A short description of applying the principal component analysis (PCA) of GOLD nighttime observations is offered as follows. The GOLD nighttime data are first sorted into the 1-hr local time bins. Subsequently, the 35-day averaged value (August 31-October 4) at each location and local time is subtracted out to derive the time sequence of residuals, which are written in the form of Equation 1.
where ′ ( ) denotes the GOLD Nmax residuals. PC ( ) (principal components) represents spatial variability, and ( ) denotes the temporal evolution of the correspondent PC ( ) , also referred to as an expansion coefficient. PCs are subject to orthogonality. In other words, Equation 1 describes using an orthogonal basis set to reproduce significant variability in the NmF 2 map during a given time period. To derive the PC ( ) and ( ) , we calculate the covariance matrix R of ′ ( ) ; then R is put into an eigenvalues and eigenvectors problem context.
where ⃖ ⃗ eigenvectors, is eigenvalues . Consequently, PC ( ) is the solution of Equation 2. The expansion coefficient ( ) is derived from the projection of ′ ( ) to the PC ( ) at each time step.
Based on the PCA results, we further elaborate on the potential driving mechanisms. In the meantime, we want to point out that the PCA has limitations in connecting "empirical modes" with specific physical processes. One of the major reasons is that the standard principal components are orthogonally constrained, while orthogonality is not mandatory in physical systems. As such, apart from the standard PCA, we apply varimax rotation (Kaiser, 1958;Mestas-Nunez, 2000) to derive rotated principal components, which increase the loadings for larger modes and vice versa. Combining standard PCA and varimax rotation lends us the confidence to interpret relevant physical processes.

Results and Discussion
The occurrence of the Antarctic SSW event is recognized from a rapid temperature surge and a deceleration of prevailing westerly zonal winds in the high-latitude stratosphere, as illustrated in Figure 1. Specifically, the temperature (yellow curve, Figure 1a) averaged over the polar cap (60°-90°S) is increased by ∼40 K in the upper stratosphere from late August through middle September. Meanwhile, a substantial decrease in zonal wind, from ∼80 m/s to 15 m/s, is seen at 60°S and 10 hPa (yellow curve, Figure 1b). The noticeable changes in temperature and zonal wind indicate the occurrence of an SSW event in the southern hemisphere. Given that the zonal wind is not switched over from westerlies to easterlies, this event is further categorized as a minor SSW.
Longitudinally structured pattern in the low-latitude F-region peak electron density (NmF 2 ) averaged at 19 LT over September 2019 is shown in Figure 1c. The two crests of the EIA stretch from the western coast of South America through the western coast of Africa. The NmF 2 in the crest regions increases rapidly toward the continents, and the maximum NmF 2 is approximately 5 × 10 5 /cc. In the middle of the Atlantic Ocean, the NmF 2 shows a significant drop, and the two crests tend to intersect. A striking longitudinal variation of the EIA is more than likely associated with atmospheric tides propagating from the lower atmosphere Immel et al., 2006;Sagawa et al., 2005). A comprehensive study of this interesting phenomenon is beyond the scope of this work. We note that the post-sunset NmF 2 varies significantly from one day to next at 75°W following the commencement of the SSW event, as illustrated in Figure 1d. At this longitude sector, the northern and the southern crests are located at the geographic equator and 15°S, respectively. GNSS (Global Navigation Satellite System) TEC (total electron content) data are given in Figure 1e to compare with the GOLD NmF 2 . Apparently, day-to-day variability agrees well between the NmF 2 and the TEC. The six peaks between August 31 and October 5 are coincident between the two quantities, indicating a 6-8-day periodicity at night. As aforementioned, such a quasi-6-day periodicity is very likely tied to enhanced Q6DW activity in the mesosphere-stratosphere, which has been manifested from the daytime ionospheric observations by Yamazaki et al. (2020). It is also noticeable that the EIA crests demonstrate a prominent asymmetry in amplitude with respect to the magnetic equator, which is likely associated with the summer-to-winter interhemispheric circulation.
SABER observations are employed to trace large-scale wave activity in the mesosphere. Figure 1f displays the geopotential height residues at 40°S and 97 km, which are derived by subtracting the zonal mean from each profile on a day-to-day basis (Gan et al., 2014). Note that planetary waves are typically pronounced in the mid-latitude geopotential heights, and some of them have large amplitudes in zonal winds over the equatorial and low-latitude region, for example, the Q6DW. A westward-propagating pattern stands out with six alternating positive and negative peaks at each longitude over the entire month. Day-to-day variations are highly aligned between the SABER geopotential height and the post-sunset peak electron density, suggesting that a Q6DW modulation of the F-region ionosphere is predominant after sunset.
Thus far, the GOLD NmF 2 data have been analyzed at 75°W longitude. We will now expand our data analysis within the entire GOLD's field-of-regard. Since the NmF 2 (Figure 1c) increases considerably toward the South American and the African continents, the GOLD images at 19LT are first divided into two longitudinal regions (45°-80°W, 40°S-20°N and 10°-30°W, 15°S-30°N). Then, two-dimensional principal component analyses are applied to the observations within each region. The reason we analyze image sequences individually over the two regions is a quasi-6-day modulation of F-region electron density maybe longitudinally dependent, given that the F-region ionosphere is primarily shaped by nonmigrating tides, as exemplified in Figure 1c.
The NmF 2 deviation on a given day, as displayed in Figure 2a, is overall lower than the monthly mean values across the entire spatial domain. On this given day, the NmF 2 decreases by up to 4 × 10 5 /cc. Figures 2b and 2c illustrate the first and second principal components (PC#1 and PC#2), which take up to ∼54.3% and ∼14.2% of the total NmF 2 variance in September, respectively. The variance accounted by each PC is the square of the linear correlation coefficient between the fitted PC and raw data over the given time period and geographic area. We compute the correlation of the NmF 2 deviations and the reconstructed NmF 2 changes based on the principal component analysis. The correlation coefficient is on the order of 0.93 (Figure 2d), implying that PCA analyses well retain the significant temporal and spatial variations during this period of time. Note that the major difference between the first two components is the relative changes between the trough and the crests of the EIA. Namely, the former one delineates the in-phase changes between the crests and the trough, whereas the latter one shows the anti-phase manner. This suggests that physical processes tied to those two components are quite distinct.
Regarding temporal evolution, the NmF 2 PC#1 (green curve in Figure 2e) displays an oscillating pattern, closely correlated with the lower thermospheric fluctuations in geopotential height (yellow dashed curve in Figure 2e). The peaks and valleys are well aligned between the SABER and the GOLD, indicating that the NmF 2 PC#1 largely manifests the Q6DW-induced fluctuations. Moreover, a sustained reduction is seen in the NmF 2 PC #1 during the course of days 10-25, when the Q6DW reaches the maximum amplitude. Such a depletion in NmF 2 is attributable to an overall O/N 2 reduction in the thermosphere, as a result of wave dissipation (Gan et al., 2015(Gan et al., , 2018Oberheide et al., 2020;Pedatella, Richmond, et al., 2016;Yamazaki & Richmond, 2013;J. Yue & Wang, 2014). The PC#2 temporal variation (Figure 2f) also appears a quasi-6-day periodicity within day 8-24, despite the strength being vulnerable compared to the PC#1. The periodogram analysis has not indicated a significant quasi-6-day oscillation (Q6DO) in the Kp index (see Figure S1 in Supporting Information S1); that is, the geomagnetic activity could not be the driver of the ionospheric Q6DO seen by GOLD.
A combined effect of the first two principal components implies that the Q6DWs modulate the post-sunset ionosphere via multiple pathways. Specifically, the PC#1 is likely linked to plasma transport by in-situ neutral winds that are modulated by upward propagating tides; those relevant tidal components interact with the Q6DW at lower altitudes. A similar coupling paradigm has been demonstrated (Forbes et al., 2018;Gan et al., 2017Gan et al., , 2020aYamazaki & Miyoshi, 2021) in prior studies. A recent study by Forbes et al. (2020) also suggested that the Q6DWs can be generated in the thermosphere via a two-step nonlinear wave-wave interaction. As such, not only the modulated tides, but also the Q6DW gives rise to the direct modulation of the F-region neutral winds.
Pertaining to the PC#2 (Figure 2c), one of the possible drivers is associated with pre-reversal enhancement (PRE). The PRE refers to a rapid increase in the F-region vertical ion drifts before the reversal from daytime upward drifts to nighttime downward drifts (Heelis et al., 2012). Prior studies have established that a huge amount of day-to-day variability in PRE are attributable to large-scale wave forcing (H. L. Liu, 2020;Liu & Richmond, 2013). Employing the TIE-GCM (Thermosphere Ionosphere Electrodynamics General Circulation Model), Yamazaki and Diéval (2021) demonstrated a prominent quasi-6-day wave modulation of the PRE during equinoxes. In principle, the PRE displays a higher intensity over the South American and Atlantic sectors around autumn equinoxes and typically peaks around 19LT. Thus, a quasi-6-day modulation of the PRE could occur during this SSW event. Given that the PRE uplifts plasmas from lower to higher altitudes in the magnetic equatorial region and plasmas diffuse to higher latitudes along the magnetic field lines, the quasi-6-day wave driven fluctuations via the PRE would be in anti-phase between the trough and the crests of the EIA (Gan et al., 2016;Lin et al., 2020). We note that the PC#2 temporal evolution is not in agreement with the quasi-6-day wave (yellow dashed curve in Figure 2f) before day 5 and after day 25, but rather two evident negative dips around day 2 and 28 in the NmF 2 (green curve in Figure 2f) coincide with the increased geomagnetic activity (Kp = 5, black curve). This could be an impact of prompt penetration electric fields from the polar region due to geomagnetic activity.
The region over the Atlantic and the western coast of the Africa, as seen in Figure 3a, also demonstrates an overall depletion in the NmF 2 on Sep 13 relative to the monthly mean values. The first two principal components are responsible for more than 60% (estimated from the correlation coefficient of 81% in Figure 3d) of the total NmF 2 variance during the given time period. Again, the first component ( Figure 3b) showcases a quasi-6-day periodicity in the NmF 2 , and a fairly good temporal correlation is seen between the PC#1 (green curve in Figure 3e) and the SABER geopotential height (yellow dashed curve in Figure 3e). Since the PC#1 pattern and the periodicity are much the same as that over South America, we suggest the same driving mechanism producing the quasi-6-day periodicity in the NmF 2 over the Atlantic sector. A recent study by Liu et al. (2021) has found a strong quasi-6-day modulation of tides during this event, which is in favor of our speculation.
Conversely, the PC#2 (Figure 3c) in the Atlantic region is significantly distinct from the corresponding component over South America. Much weaker day-to-day variations of the PC#2 (green curve in Figure 3f) between day 5-25, as opposed to the active quasi-6-day waves (yellow dashed curve in Figure 3f), suggest that the quasi-6-day wave related modulation are insignificant in this component. Instead, the PC#2 time sequence displays a close correlation with the geomagnetic activity. The primary peak in PC#2 around September 2 coincides with the elevated geomagnetic activity (black curve). Attributing the changes represented the PC#2 to the geomagnetic storm has a solid physical basis. Specifically, the Joule heating during a geomagnetic storm could induce the upwelling in the high-latitude thermosphere, and the heated air parcels produce an equatorward pressure gradient that drives significant meridional winds blowing from higher latitudes toward the equator (Fuller-Rowell et al., 1994;Gan et al., 2020b). As a result, one would expect changes in the EIA.
As mentioned in Section 2, a physical interpretation of principal components is not straightforward sometimes. This is because the principal component decomposition is orthogonally constrained, while orthogonality is not necessary in physical systems. To further support our interpretation above, a varimax rotation is applied to further maximize the loadings of major components, as displayed in Figure 4. Over the South American sector, the rotated PC #1 and #2 are quite similar to the standard PC #1 and #2 (top panels in Figure 4); that is, they largely retain the relative changes in the EIA trough and crests. Meanwhile, the temporal evolutions (bottom panels in  Figure 4) are not significantly altered between the standard and rotated ones. We are thus confident that two individual driving mechanisms are involved to couple the Q6DW with the post-sunset EIA and that it is not an artifact of the underlying orthogonality assumption in PCA analysis. The same analysis is performed to the GOLD observations over the Atlantic, and the difference between the standard and rotated PCs is again insignificant.

Conclusions
Employing the GOLD observations, we investigated the post-sunset F-region ionospheric response to the 2019 Antarctic SSW event. GOLD observations reveal that the low-latitude ionosphere at night is primarily shaped by large-scale waves arising in the middle atmosphere during the period of this event. Striking quasi-6-day periodicities were seen in the crest regions of the EIA over the South America and the Atlantic, coincident with the mesospheric quasi-6-day waves. A complex paradigm, transmitting the quasi-6-day periodicity from the mesosphere and lower thermosphere into the nighttime ionosphere, was seen from this case study.
Over South America and the Atlantic Ocean, the prime driver is connected to upward propagating tides, which are modulated by the quasi-6-day waves in the lower atmosphere and induce the in-situ transport of plasmas along the magnetic fields in the F-region ionosphere. Also, the same tidal winds can modulate pre-reversal enhancements of vertical ion drifts and further lead to a modulation of electron densities at a quasi-6-day period. Other than the periodicity, the observed NmF 2 shows an overall depletion when the quasi-6-day waves are most active. Such a depletion is likely linked to the altered residual mean circulation in the lower thermosphere due to wave dissipation, consistent with the scenario evidenced by Gan et al. (2015Gan et al. ( , 2018.