New Observations of Large‐Scale Waves Coupling With the Ionosphere Made by the GOLD Mission: Quasi‐16‐Day Wave Signatures in the F‐Region OI 135.6‐nm Nightglow During Sudden Stratospheric Warmings

The recently launched Global‐scale Observations of the Limb and Disk (GOLD) Mission observed prominent quasi‐16‐day oscillations (Q16DOs) in the OI 135.6‐nm nightglow from the equatorial ionization anomaly (EIA) region during 26 November 2018 to 25 January 2019. Meanwhile, the enhanced and concurrent quasi‐16‐day wave (Q16DW) activity was observed in the middle atmosphere by the Sounding of the Atmosphere using Broadband Emission Radiometry (SABER) instrument. This implies that the F‐region ionosphere couples with the lower atmosphere through large‐scale waves. A noteworthy phenomenon was that the Q16DOs displayed an asymmetric behavior between the northern and southern EIA crests. This asymmetry indicates that the ionospheric Q16DOs might be driven by the Q16DW‐modulated tides, which not only modulated the E‐region dynamo winds but also modulated the F‐region field‐aligned neutral winds via vertical penetration, producing the oscillations in plasmas densities.


Introduction
It is understood that F-region ionosphere is regulated by the forcing from the Sun, the magnetosphere (e.g., Forbes et al., 2000;Lei et al., 2008;Rishbeth & Mendillo, 2001), and the lower atmosphere (e.g., Akmaev, 2011;Liu, 2016;Oberheide et al., 2015;Sassi et al., 2019;Vincent, 2015). During solar and geomagnetically quiet times, a considerable amount of day-to-day ionospheric variability is due to traveling planetary waves (PWs) in the lower atmosphere, including eastward-propagating equatorial Kelvin waves and westward-propagating Rossby waves (Altadill, 2000;Borries & Hoffmann, 2010;Forbes et al., 2018;Lastovicka, 2006;Yi & Chen, 1993; and references therein). The latter are also called "normal modes (NM)", typically having periods of 2-20 days and zonal wavenumbers of 1-4, which are the global solutions of the Laplace's tidal equation (Salby, 1981(Salby, , 1984. In the recent literature, several well-known PWs were widely referred to as "Q2DW," "Q6DW," "Q10DW," and "Q16DW" (the numbers stand for the wave periods; Q, D, and W are the acronym of "quasi-," "day," and "wave," respectively). Over the past decade, ©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.

10.1029/2020JA027880
Special Section: Early results from the Global-scale Observations of the Limb and Disk (GOLD) mission considerable advances have been made in delineating the global morphology of PW-induced signatures in the ionosphere and the understanding of physical mechanisms by which the PW signatures are transmitted into the F-region heights (Chang et al., 2010(Chang et al., , 2014Gan et al., 2015Gan et al., , 2016Gan et al., , 2017Gan et al., , 2018Pedatella & Forbes, 2012;Yamazaki et al., 2018;Yue et al., 2013Yue et al., , 2016; and references therein). However, those studies are primarily focused on the wave components with relatively shorter wave periods (i.e., 2, 3-4, and 6 days) but larger phase speeds and vertical wavelengths.
The quasi-16-day wave (Q16DW) is recognized as the second symmetric mode (1, −4) of zonal wavenumber 1 Rossby waves. It is usually enhanced at midlatitude and high latitude in the winter hemisphere (Day et al., 2011). The observed 16-day wave can, to some extent, depart from the theoretical prediction because the latter assumes the ideal atmosphere, that is, isothermal, windless, and dissipationless. Other than the significant role in the middle atmosphere dynamics, the Q16DW also induces periodic variations in the ionosphere. Early studies have used ionosonde measurements, establishing a connection between the quasi-16-day oscillations (Q16DOs) in the F 2 -region ionosphere and the Q16DWs in the mesospheric winds (Forbes, 1996;Forbes et al., 1995;Forbes & Leveroni, 1992;Mendillo et al., 2002;Rishbeth & Mendillo, 2001;Yi & Chen, 1993). From a satellite observation perspective, Pedatella and Forbes (2009) showed a global picture of the Q16DW-induced oscillations in electron densities using CHAMP (Challenging Minisatellite Payload) data. Based on a linear spectral model, Forbes et al. (1995) suggested that the Q16DW is not able to penetrate to altitudes above 100 km, and the coupling is more than likely indirect. The probable candidates include the upward propagating tides and dissipated gravity waves (Meyer, 1999) that are modulated by the Q16DWs. The recently launched Global-scale Observations of the Limb and Disk (GOLD) mission provides an excellent opportunity to revisit this significant and recurrent but not fully understood phenomenon from a new perspective.
This study aimed at looking for the periodicity in the F-region equatorial ionization anomaly (EIA) during boreal winter in 2018/2019, when a sudden stratospheric warming (SSW) event occurred in the Northern Hemisphere and PWs were rather active. This thus provided an excellent opportunity to explore the wave coupling in the atmosphere-ionosphere system. Section 2 overviews the data used in this study. Section 3 presents the major findings. Section 4 is the conclusions.

Data Sets
The NASA (National Aeronautics and Space Administration) GOLD mission flies a far ultraviolet imager onboard the SES-14 satellite in the geostationary orbit. It has two identical channels, and each one scans one third of the Earth (roughly 120°W-20°E and 60°S-60°N) every 30 min. The imager is a spectrometer which monitors the OI 135.6 nm and the molecular nitrogen LBH bands emissions. These are used to retrieve the thermospheric daytime O/N 2 ratio, temperature, and nighttime peak electron density (Nmax) on the disk. In this study, we used the nighttime OI 135.6-nm emission data from GOLD disk scans measured by two channels simultaneously. The nighttime 135.6-nm emission produced by the ionosphere recombination peaks in the F 2 -region where the electron density maximizes. The L1C data (version 3) and documentation guide are available at http://gold.cs.ucf.edu. We refer readers to Eastes et al. (2017Eastes et al. ( , 2019 for a detailed description of GOLD instruments and the nighttime data product. Employing the daytime disk scans, Gan et al. (2020) showed a remarkable thermospheric response to a moderate geomagnetic storm.
Sounding of the Atmosphere using Broadband Emission Radiometry (SABER) onboard the Thermosphere Ionosphere Mesosphere Energetics and Dynamics (TIMED) satellite is a 10-channel broadband radiometer covering the spectral range from 1.27 to 17 μm. 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. Remsberg et al. (2008) assessed the quality of version 1.07 temperature data and suggested that the systematic error is no more than 2 K below 70 km, while in the upper mesosphere and lower thermosphere region, the error increases with altitude from 1.8 K at 80 km to 6.7 K at 100 km. The effective vertical resolution of SABER temperature is approximately 2 km. SABER is in a slowly precessing polar orbit. The sampled local time shifts~12 min on a daily basis, allowing approximately 24-hr local time coverage every 60 days at low to middle latitudes when both ascending and descending samples are combined. Due to the line of sight of SABER and the yaw maneuver of the TIMED satellite, the latitude coverage of SABER data is from 53°1 latitude in one hemisphere to 83°in the other. In the current work, the latest version 2.0 temperature data were utilized to trace PW activity in the stratosphere and mesosphere. Figure 1a displays the time series of the Kp (blue) and F10.7 (red) indices between 26 November 2018 and 24 February 2019. The time span covers the boreal winter, during which wave activity was significantly enhanced, to facilitate an investigation of wave impacts from the lower atmosphere on the thermosphere-ionosphere system. Wavelet analyses were performed on Kp and F10.7 indices to examine the periodic forcing by geomagnetic field and solar extreme ultraviolet band radiation, which may also induce ionospheric oscillations at the PW periods (Lei et al., 2008;Pedatella et al., 2010). It turns out that Kp (Figure 1b) was dominated by the 24-day and 8-9-day oscillations, whereas a~24-day oscillation was evident in F10.7 (Figure 1c). This means that ionospheric oscillations associated with PWs, particularly those having a period of 10-20 days, are distinguishable during the given period. Figure 1d displays the zonally averaged zonal winds (red curve) and temperature (blue curve) at 30 hPa and 60°N. Note that the westerly jet of~30 m/s decreased rapidly after day 25 (21 December) and the zonal winds reversed to weak easterlies (wind velocities <10 m/s) between day 35 and day 52 (31 December to 16 January 2019). Then, the zonal winds tended to return to their normal state. Such a disturbance in zonal winds is consistent with the well-defined SSW phenomenon, which is characterized by the polar vortex weakening due to enhanced PW activity. Moreover, the zonally averaged temperature increased by~10-15 K during days 22-40. SSW events provide a suitable condition for the vertical propagation and amplification of PWs. As such, potential impacts of PWs on the ionosphere are achievable, for example, through the nonlinear interaction with tides or altering the tidal vertical wavelength (Chang et al., 2011;Gan et al., 2017;Liu et al., 2010). The spatial coverage was roughly over the eastern America and western Atlantic Ocean. The EIA bands stood out with increased brightness toward the terminator. The peak brightness was of~40 Rayleigh (R) over the South American sector, consistent with that reported by Eastes et al. (2019). The same GOLD image is illustrated in Figure 2b but smoothed to 3°(longitude) × 3°(latitude) in the quasi-dipole magnetic coordinate system (Richmond, 1995). The two crests of the EIA were located at around 10°and −15°magnetic latitudes, indicating a hemispheric asymmetry about the dip equator. In order to determine the periodicities in the OI 135.6-nm emission, the time series of brightness at 20°1 magnetic longitude (averaged over the longitudinal bin centered at 20°with 5°offset) is shown in Figure 2c. The brightness in the northern crest region demonstrated an oscillating pattern; that is, the higher brightness occurred on~day 7, 22, and 39, while the brightness was relatively lower between those days in the same region.

Results and Discussion
Next, an empirical orthogonal function (EOF) analysis was performed on the time series to identify the predominant oscillations and their latitudinal dependence. The first 10 EOF modes were derived simultaneously, but only the EOF1 and EOF2 are illustrated here since they account for 56.6% and 21.0% of the total variability. The rest of the EOFs had an individual contribution to the total variability of less than 10% and did not show any periodicities connected to the PWs of interest. Figure 2d displays the latitudinal structure of the EOF1, describing the symmetric brightness changes about the dip equator between the two crests. The amplitude in the crest regions is of 0.2-0.25, greater than that over the magnetic equator by a factor of~4. The temporal behavior of the EOF1 is described in Figure 2e, which showed an oscillation of 60 days during the given period. Specifically, the brightness was enhanced (weakened) before (after) day 25. Several reasons are likely responsible for this tendency in brightness. First, the F 2 -layer peak electron density has a strong seasonal dependence in the EIA region, maximizing at around December solstice and decreasing afterward (Qian et al., 2013). Since the nighttime OI 135.6-nm brightness is produced by recombination, the brightness change is intimately related to the change of electron densities. Second, we note that the moment when the brightness change in Figure 2e turned over from positive to negative was aligned with the onset of the weakened polar vortex (day 25). This implies that the brightness depletion after day 25 (21 December 2018) may also be associated with the SSW. Pedatella et al. (2016) suggested that an enhanced migrating semidiurnal tide (SW2) during SSW events is able to induce a global-scale upward circulation and thereby a reduction of zonal mean O/N 2 ratio and electron densities (Liu & Roble, 2002;Oberheide et al., 2020;Shepherd et al., 1999). In addition, the intraseasonal variability (e.g., Sassi et al., 2019; 10.1029/2020JA027880 Journal of Geophysical Research: Space Physics Vergados et al., 2018) in the lower atmosphere could be responsible for the observed 60-day oscillation in brightness as well. However, the 60-day periodicity is beyond the scope of the current study. It is noticeable that the wavelet spectrum of the EOF1 time sequence, as shown in Figure 2f, suggested an 8-9-day oscillation on days~10-30 (6-26 December 2018). We postulate that this periodicity was possibly due to the coincident fluctuations in geomagnetic activity, as illustrated in Figure 1b. Figure 2g displays the latitudinal structure of the EOF2 in the GOLD data. Similar to the EOF1, the second largest principal component represented significant brightness changes in the EIA crest regions, but the variability was roughly opposite between the northern and southern hemispheres. It is noteworthy that the EOF2 (Figure 2h) showed a regular oscillating pattern with a period of~16 days, as suggested by the wavelet spectrum in Figure 2i. Within the given time span, the Q16DO roughly had four cycles, and the absolute magnitude of the oscillation was greater between days 10-40 and subsequently decayed by~60% after day 40. Since the quasi-16-day periodicity was not detected in the Kp and F10.7 indices, it was thus possibly tied to the waves arising in the troposphere and middle atmosphere. A closer look at the Q16DOs in the EIA crest regions is shown in Figure 3. The brightness anomaly (solid curves) was derived by subtracting the 20-day running means from the daily brightness in the northern (orange) and southern (blue) crests, which correspond to the bins of 5°-15°and −20°-10°magnetic latitudes, respectively. The percentage anomaly (dotted curves) was also calculated by normalizing the absolute anomaly to the 20-day running mean. The error bars indicate the random noise level. As expected, the anomalies exhibited 3-4 peaks and valleys, and the maxima brightness of the northern crest (solid orange curve) occurred near day 7, 22, 38, and 52, demonstrating a~16-day oscillation which was aligned with the EOF2 in Figure 2. Also, the magnitude of the anomalies in the northern crest decreased from a peak value of 10 R on~day 22 to 3 R on~day 38, while the corresponding percentage anomalies increased from~60% to 75%. The oscillation in the southern crest showed a similar overall decrease, but its magnitude appeared to be somewhat weaker than that in the northern counterpart. The significant decay in the absolute anomalies was possibly due to the overall deletion of brightness after day 25 (21 December 2018) as shown in Figure 2c.
A noteworthy point in Figure 3 is that the brightness anomalies were asymmetric (in antiphase between day 20 and day 50) between the southern and northern crests. Presumably, the Q16DO was induced by the E-region dynamo modulation due to tides and PWs, which has been widely proposed to account for the F-region ionosphere periodicities during the daytime in the context of geomagnetically and solar quiet conditions (e.g., England et al., 2010;Immel et al., 2006;Jin et al., 2008). If the coupling process is governed by symmetric E-region dynamo modulation, the resulting oscillations between the northern and southern

10.1029/2020JA027880
Journal of Geophysical Research: Space Physics crests would be more in-phase. The asymmetry, however, implies that the E-region dynamo modulation was not the exhaustive scenario in this case, which is not unexpected since the E-region dynamo effect (ExB) gradually fades away after the sunset. In other words, the E-region dynamo modulation may play a secondary role in transmitting the quasi-16-day periodicity into the F-region heights. Another candidate mechanism is the F 2 -region in situ neutral winds, associated with the vertical penetration of tides and PWs, which drive the field-aligned motion of the plasma and produce the oscillations in electron densities. Using the first-principle modeling, Forbes et al. (2018) suggested this new mechanism to interpret the large-scale wave impacts on the ionosphere. However, to our best knowledge, there has been no observational proof to confirm this mechanism. Note that the EOF analysis also indicated the asymmetric oscillations in brightness between the northern and southern EIA crests (see Figure 2g).
Thus far, the Q16DOs in airglow have been established. We now turn to wave diagnostics in the mesosphere. Figure 4a displays the temperature anomalies as functions of longitude and time at 40°N and 80 km from SABER measurements. The time span is the same as that in Figures 2 and 3. The temperature anomalies were computed by subtracting the zonal mean temperature on a daily basis for ascending and descending portions of orbits separately (Gan et al., 2014). This algorithm effectively removes migrating tides and avoids tidal aliasing. In the mesosphere, the temperatures anomalies ( Figure 4a) are characterized by a zonal wavenumber 1 structure. The wave pattern showed an obvious westward propagation during day 10-50 (6 December 2018 to 15 January 2019), after which the wave appeared to be stationary. The propagating wave component was significantly strengthened before the zonal wind reversal around day 35 (31 December 2018; see Figure 1d) since the dramatic change of zonal mean winds resulted in the changes to the refractive index and meridional gradient of potential vorticity, in favor of PW propagation and amplification (Charney & Drazin, 1961;Smith, 1983). Figure 4b illustrates the time series of temperature anomalies (blue curve) at 285°E longitude and its wavelet spectrum. This longitude sector is in the GOLD field of view. The red curve denotes the temperature anomalies derived by a 14-18-day bandpass filtering. In principle, the quasi-16-day perturbations were salient, and the peak wave magnitude of 10 K occurred between day 25-35 (21-31 December). To further characterize the wave property, 2-D spectral analysis proposed by Salby (1982) was performed on the temperature anomalies (Figure 4c). It turns out that the quasi-16-day fluctuations had zonal wavenumber 1 and propagated westward, establishing a connection with the (1, −4) Rossby normal mode. Figure 4d further illustrates altitude versus latitude Q16DW magnitudes on day 25 (21 December 2018). From a global-scale perspective, the Q16DW showed a double-peak structure with a maximum magnitude of 12-14 K in the boreal stratosphere and mesosphere, as well as a localized peak of~8 K at~100 km in the Southern Hemisphere.
In regard to vertical coupling with the ionosphere, the Q16DW showed considerable complexity. Despite the observed Q16DW having a large vertical wavelength of approximately 70 km (the orange line in Figure 4d denotes the Q16DW phase structure at 50°N, where the magnitude maximized) in the boreal middle latitudes, its slow zonal phase speed (~18 m/s at 50°N) would result in quick wave dissipation, and thereby, a direct vertical penetration into the F-region heights was not anticipated. However, the

10.1029/2020JA027880
Journal of Geophysical Research: Space Physics large wave magnitude in both the mesosphere and stratosphere enabled the Q16DW to modulate tides, for example, the migrating semidiurnal tides (SW2). Those tidal components have been revealed to affect the ionosphere via direct penetration to the F-region heights (Pedatella et al., 2016;Pedatella & Maute, 2015). Also, as it is seen, the Q16DW may cross the equatorial region in the mesopause and lower thermosphere, accounting for the Q16DW in the Southern Hemisphere (Chandran et al., 2013;Day et al., 2011). Such a cross-equatorial behavior may lead to the wave momentum deposition into mean flows on a global scale, as well as the changes of circulation and constituent transport. All the above speculation merits a comprehensive investigation using the first-principle whole atmosphere models (for instance, the Whole Atmosphere Community Climate Model extended) in the near future.

Conclusions
The GOLD observations revealed prominent Q16DOs in the nighttime OI 135.6-nm emission over the EIA region during 26 November 2018 to 25 January 2019. Concurrently, SABER observed a Q16DW event at middle latitudes in the Northern Hemisphere, indicating that the Q16DW in the middle atmosphere was possibly the leading cause of the Q16DOs in the F-region ionosphere during this period. EOF analysis, along with a detailed examination of the temporal evolution of the brightness in the EIA crests, showed that the Q16DOs accounted for~20% of the total brightness variability at around midnight. More importantly, the Q16DOs in the F-region ionosphere were asymmetric between the northern and southern crests. This asymmetry sheds light on the possible coupling mechanism. That is, other than modulating the E-region dynamo winds, the Q16DW-modulated tides possibly penetrate to the F-region heights and produce the Q16DOs in the field-aligned motion of the plasmas and thereby in electron densities during SSWs.