Atmosphere-Ionosphere (A-I) Coupling as Viewed by ICON: Day-to-Day Variability Due to Planetary Wave (PW)-Tide Interactions

Abstract

Coincident Ionospheric Connections Explorer (ICON) measurements of neutral winds, plasma drifts and total ion densities (:=Ne, electron density) are analyzed during January 1–21, 2020 to reveal the relationship between neutral winds and ionospheric variability on a day-to-day basis. Atmosphere-ionosphere (A-I) connectivity inevitably involves a spectrum of planetary waves (PWs), tides and secondary waves due to wave-wave nonlinear interactions. To provide a definitive attribution of dynamical origins, the current study focuses on a time interval when the longitudinal wave-4 component of the E-region winds is dominated by the eastward-propagating diurnal tide with zonal wavenumber s = −3 (DE3). DE3 is identified in winds and ionospheric parameters through its characteristic dependence on local solar time and longitude as ICON’s orbit precesses. Superimposed on this trend are large variations in low-latitude DE3 wave-4 zonal winds (±40 ms⁻¹) and topside F-region equatorial vertical drifts at periods consistent with 2-days and 6-days PWs, and a ~3-day ultra-fast Kelvin wave (UFKW), coexisting during this time interval; the DE3 winds, dynamo electric fields, and drifts are modulated by these waves. Wave-4 variability in Ne is of order 25%–35%, but the origins are more complex, likely additionally reflecting transport by ~20–25 ms⁻¹ wave-4 in-situ winds containing strong signatures of DE3 interactions with ambient diurnal Sun-synchronous winds and ion drag. These results are the first to show a direct link between day-to-day wave-4 variability in contemporaneously measured E-region neutral winds and F-region ionospheric drifts and electron densities.

1. Introduction

It is now widely accepted that tides, planetary waves (PWs) and ultra-fast Kelvin waves (UFKW) propagating upward from the lower atmosphere play an important role in determining the longitudinal and day-to-day variability of the ionosphere. Insofar as PWs and the low-latitude ionosphere is concerned, this realization has its early roots in the work of Ito et al. (1986) and Chen et al. (1992; see also their references to the earlier Chinese literature), who first recognized the connection between quasi-2-days wave (Q2DW) variations in winds, and Q2DW variations in electric fields generated by neutral-plasma interactions in the E-region (i.e., the wind dynamo), ground magnetic perturbations (Ito et al.) and the critical frequency of the F-region, f₀F₂ (Chen et al.). (The E-region electric fields E map along magnetic field lines to the F-region, where they are manifested as E × B drifts that transport plasma from equatorial to tropical latitudes to produce the “equatorial ionization anomaly (EIA).”) These authors moreover confirmed their hypotheses through numerical modeling. Chen et al. (1992; (see also later work by Pancheva et al., 2006) furthermore postulated that the Q2DW exerted its influence through modulation of tidal winds, rather than directly through the Q2DW wind field. This point of view is at the heart of the neutral wind-driven ionospheric variability examined in the present work.

Between about 1990 and 2010, many additional analyses that documented PW variations with periods of order 2–20 days in the ionosphere were published (e.g., see review by Laštovička et al., 2006). Most of these were conducted with ionosonde data, and emphasized middle latitudes. The origins of ionospheric variability on a global scale due to “meteorological influences” also began to be investigated through modeling during the latter half of this period (Mendillo et al., 2002; Rishbeth, 2006; Rishbeth et al., 2009; see also; H.-L. Liu et al., 2013). Around the same period of time, the connection between longitudinal “wave-4” structures in the ionosphere and vertically propagating tides were made from satellite measurements (Immel et al., 2006; Sagawa et al., 2005), leading to subsequent satellite-based studies focused on this subject (e.g., Lin et al., 2007; H. Liu & Watanabe, 2008). Modeling studies (e.g., Fang et al., 2009; Hagan et al., 2007; Jin et al., 2008, 2011) solidified our quantitative understanding of these observations. Analogous to A-I coupling by tides, observational (Abdu et al., 2015; G. Liu et al., 2015; Onohara et al., 2013; Takahashi et al., 2007) and modeling (Chang et al., 2010; Gu et al., 2014; Yamazaki et al., 2020) studies demonstrated that quasi-3-days ultra-fast Kelvin waves (UFKWs) also impart their variability on the ionosphere through the dynamo mechanism.

More recent studies have revealed that PW-modulated tides (Forbes et al., 2018; Forbes, Zhang & Maute, 2020) and UFKW (Forbes, Maute & Zhang, 2020) with sufficiently long vertical wavelengths (≳60 km) penetrate to the F-region and modify field-aligned plasma transport. At a given height in the F-region, the vertical components of both plasma drifts and field-aligned plasma motions give rise to ionospheric variability due to vertical oscillations of the F-layer (England et al., 2010; Forbes et al., 2018; Forbes, Maute & Zhang, 2020). Moreover, nonlinear interaction between PWs and tides, and the secondary waves (SW) that arise from these interactions, also play key roles in driving day-to-day ionospheric variability (Egito et al., 2020; Gan et al., 2017; Gu et al., 2018; Miyoshi & Yamazaki, 2020; Yue et al., 2016). Modeling demonstrates that secondary waves (SWs) arising from Q2DW-tide (Nguyen et al., 2016) and Q6DW-tide (Forbes, Zhang & Maute, 2020) interactions are mainly excited below 100 km, and that their exponential growth with height combined with molecular dissipation leads to SW amplitude maxima in the 110–140 km height range. SWs are in fact themselves capable of carrying PW periodicities into the ionosphere (e.g., Miyoshi & Yamazaki, 2020). Since the secondary waves occur with wave periods and zonal wavenumbers different than the primary interacting waves, the aggregation of primary and secondary waves and their effects results in a degree of longitudinal and day-to-day variability much more complex than that from linear superposition of the primary waves alone.

Much of what we know about A-I coupling by tides, PWs and UFKWs is derived from derived from general circulation model (GCM) simulations of the coupled atmosphere and ionosphere. A major impediment to advancing this area of science is the absence of contemporaneous observations of the physical parameters involved in the A-I coupling mechanisms. The Ionospheric Connection (ICON) Explorer mission (Immel et al., 2018), launched October 11, 2019, into a 27°-inclination 575 × 610-km near-circular orbit (mean altitude 592 km), was designed to help close this measurement gap, and thereby bring us one step closer to quantifying the linkages between lower-atmosphere variability and the “space weather” of the ionosphere. In the present paper, we employ ICON observations to explore the connections between E-region winds, F-region plasma transport and densities, in the context of day-to-day variability connected with PW-tide interactions. The following section describes in physical and mathematical terms the waves that were present during the January 1–21, 2020 study interval. Section 3 describes the ICON data that are utilized; Section 4 presents the results, and Section 5 summarizes the main conclusions.

2. Mathematical and Physical Context

The non-tidal waves most widely studied so far in terms of A-I coupling include the westward-propagating PW oscillations with quasi-periods (zonal wavenumbers) of 2 days (s = +3) (e.g.,Gu et al., 2018; Yue et al., 2016), 6 days (s = +1) (e.g., Gan et al., 2017; Miyoshi & Yamazaki, 2020), and eastward-propagating ultra-fast Kelvin waves with periods between 2 and 4 days (s = −1) (e.g., Chang et al., 2010; Forbes, Maute & Zhang, 2020). The commonly used shorthand notations for these waves are Q2DW, Q6DW and UFKW, respectively. For each of these waves, zonal wavenumbers other than those quoted do exist in nature, but have received comparatively little attention in A-I coupling studies.

The Q2DW, Q6DW and UFKW referred to above in fact play a key role in the present study, which focuses on modulation of the eastward-propagating diurnal tide with s = −3 (DE3) by these waves, and consequences to the F-region ionosphere. A solar tidal variation (such as DE3) in an atmospheric variable (i.e., temperature, density, pressure, horizontal or vertical wind) is expressed mathematically as

$$A_{n,s}\cos \left(n\Omega t+s\lambda -{\varphi }_{n,s}\right)$$ (1)

where A_n,s and ϕ_n,s represent the height (z) and geographic latitude (θ)—dependent amplitude and phase; the integer n = 1(2) refers to diurnal (semidiurnal) periods of 1.0 days (0.5 days); Ω = 2πd⁻¹; integer s is the zonal wavenumber; t is universal time (UT) and λ is longitude. DE3 therefore corresponds to [n, s] = [1,−3]. A non-tidal wave with period T in days = 1/δ can be similarly expressed by replacing n by non-integer δ and s by integer m in 1. For example, the Q2DW, Q6DW, and UFKW referred to above correspond to [0.5, +3], [0.17, +1], [0.25 – 0.50, −1], respectively.

When applying (1) to data in order to determine A_n,s and ϕ_n,s for an oscillation of given n or δ and s, this is done (e.g., through least squares fitting) for various combinations of z and/or θ so that the height and/or latitude structures of A_n,s and ϕ_n,s can obtained. The above furthermore implicitly assumes a steady-state oscillation, although this is most often not true. Moreover, the independent variables t and λ must be sufficiently sampled to fully constrain the fit. In the case of solar tides, measurements must be sampled with respect to t and λ to yield sufficient coverage in local time. For instance, 40 days of data are needed to depict the latitude dependence of DE3 at a given height by applying this method to the full latitude range of ICON wind measurements (Cullens et al., 2020). Below, an alternative method of quantifying the presence of DE3 in ICON data as well as its day-to-day variability due to PW modulation is described, and is applied to ICON data in Section 4.

Setting the quantity in parentheses in Equation 1 equal to a constant and differentiating yields the zonal phase speed of the tide, C_ph = −nΩ/s. If s = n, the tidal oscillation moves synchronously with the westward motion of the Sun to a ground-based observer (−Ω), and is referred to as a sun-synchronous or “migrating” tide; otherwise, for s ≠ n the tide is solar asynchronous or “non-migrating.” By convention, DWs or DEs denote a westward or eastward-propagating diurnal tide, respectively, with zonal wavenumber = s. For semidiurnal oscillations “S” replaces “D”. Tidal oscillations that are zonally symmetric (s = 0) are denoted D0 and S0.

Rewriting (1) in terms of local solar time t_LST where t = t_LST − λ/Ω results in

$$A_{n,s}\cos \left[n\Omega t_{LST}+(s-n)\lambda -{\varphi }_{n,s}\right]$$ (2)

DE3 now appears as wavenumber |s − n| = 4 in this frame of reference, and a non-tidal wave appears with zonal wavenumber m − δ. Expression (2) is applicable to measurements from a slowly precessing satellite such as UARS, TIMED and ICON. In the more general form of Equation 2, the rate of precession α enters in a 1/(1 − α) multiplicative correction to n and δ in these space-based zonal wavenumbers (Forbes & Moudden, 2012); for ICON (α = 0.01852) the 1/(1 − α) = 1.0189 correction is negligible for present purposes.

As shown in earlier papers (e.g., Forbes et al., 2004, 2006: see also Forbes & Zhang, 2018; England et al., 2019, for more recent applications at Mars), setting the derivative of the quantity in brackets in Equation 2 equal to a constant and taking the derivative relative to LST and longitude for a tide with given n, s:

$$\frac{d{t}_{LST}}{d\lambda }=\frac{n-s}{n\Omega }=\frac{n-s}{n}\cdot \frac{4.0h}{{60}^{°}}$$ (3)

where h denotes hours. In Section 4 the slope given by Equation 3 is used to elucidate the consistency of a wave component with given n and s with the data.

As noted above, the present study focuses on January 1–21, 2020 which includes the coincident occurrence of the Q2DW, Q6DW, an UFKW and the DE3 tide, thus providing the opportunity to investigate how these PWs influence the ionosphere through PW-DE3 interactions. These waves are the subjects of numerous observation and modeling studies (e.g., see Introduction), and are understood at more than a rudimentary level. The Q2DW and Q6DW are quasi-resonant oscillations of the atmosphere, likely of stratospheric origin, and likely amplified in unstable regions of the atmosphere and perhaps as a result of reflection at critical levels where their zonal phase speed matches the zonal-mean zonal wind speed. The UFKW and DE3 are forced by latent heating associated with convective activity in the tropical troposphere.

Some depictions of the aforementioned waves based on MIGHTI (Michelson Interferometer for Global High-resolution Thermospheric Imaging) wind measurements are provided in Figure 1. Figure 1a shows 40-days mean DE3 zonal wind (U) amplitudes (window moved forward one day at a time) as a function of latitude and day of year (DOY) ~360–115 during 2019–2020. Depictions similar to Figure 1a for DE3, SPW4 and SE2 zonal and meridional winds at 106 km are provided in Figure S1 in the Supporting Information for later reference. The choice of 40-days means covering the latitude range of 9°S–42°N follows from numerical experiments performed by Cullens et al. (2020), and is based on the need to provide adequate local time and longitude coverage for the extraction of solar tides given the atmosphere sampling patterns of the ICON MIGHTI instrument (see Section 3). Two maxima are seen in Figure 1a, one during January and the other during March–April. Climatologically, the most active months for DE3 are June–November, with secondary activity during March–May (X. Li et al., 2015; Truskowski et al., 2014). The 2020 March–April maximum falls into the latter category. Occasionally (e.g., 2007 and 2012; X. Li et al., 2015) significant DE3 amplitudes occur during January, and 2020 seems to be one of these exceptional years. For later interpretations, it is also important to note that the centroid of the DE3 distribution in Figure 1a occurs in the Northern Hemisphere (NH). Northward (southward) shifts of DE3 with respect to the equator during NH winter (summer) conditions have been tied to asymmetries in the zonal-mean zonal wind distribution through modeling (Gasperini et al., 2017; Zhang et al., 2011).

Figure 1.

Depictions of the major waves referred to in this paper, obtained through analysis of Michelson Interferometer for Global High-resolution Thermospheric Imaging wind measurements. (a) Latitude versus day of year dependence of mean DE3 U wind amplitude at 106 km, obtained by fitting within 40 days windows sliding 1 day at a time. The dashed white line indicates the January 1–21, time period of interest. (b) Same as (a), except for 23 days mean [0.5,+3] amplitude of V at 98 km. (c) Period versus wavenumber spectrum of U at 98 km between 5°N and 15°N during January 1–21, 2020.

Although UFKWs are also generated by latent heating, they do not follow a regular climatology similar to DE3, but rather undergo intra-seasonal oscillations with periods between 20 and 60 days (Forbes et al., 2009; Miyoshi & Fujiwara, 2006). Although DE3 is generally categorized as a solar tide, it is an UFKW too, except with diurnal period. UFKWs and DE3 are confined to low latitudes, have similar zonal wind and temperature structures that are quasi-symmetric about the equator, and relatively small meridional (V) wind amplitudes compared to zonal (U) wind amplitudes. For this reason, we will concentrate in this paper on DE3 zonal wind amplitudes at 106 km, which coincides with the peak height of E-region Hall conductivity (σ_H) (See further discussion supporting this choice in Section 4).

The Q2DW is a summertime phenomenon. A Q2DW was measured using a meteor radar system at Jicamarca, Peru (11.7°S, 76.7°W) during January–February, 2020 (Chau et al., 2021). However, modeling indicates (e.g., Palo et al., 1999) that during January–February Q2DW events, penetration to the equatorial E-region, and into the Northern Hemisphere is possible. The Q2DW as measured by ICON MIGHTI is depicted in Figure 1b, which illustrates the V wind component of the Q2DW at 98 km as a function of latitude and DOY obtained by fitting [0.5, s = +3] using a 23 days moving window, and using the longitude subdivision method of Moudden and Forbes (2014) to improve temporal resolution. Amplitudes up to 15–20 ms⁻¹ are indicated, similar to the wind speeds measured at Jicamarca. There is also indication of an s = −2 Q2DW in the MIGHTI data at about half the amplitude (not shown), which in principle arises through nonlinear interaction between [0.5, s = +3] and DW1. However, it is noted that aliasing can be an issue when viewing the Q2DW from space (e.g., Moudden & Forbes, 2014; Nguyen et al., 2016; Palo et al., 2007; Tunbridge et al., 2011), such that [0.5, s = +3] and [0.5, −2] both appear as zonal wavenumber |s − n| = 2.5 from the space-based perspective. In fact, all SW arising from nonlinear interaction between any migrating tide and either [0.5, s = +3] or [0.5, s = −2] all yield |s − n| = 2.5. Therefore, Figure 1b should be interpreted as no more than an indication that the rising portion of a Q2DW event occurred during January 1–21, 2020.

Figure 1c provides a period versus zonal wavenumber spectral view of PW and UFKW activity during January 1–21, 2020 between 5°N–15°N at 98 km. The UFKW is mainly s = −1, with periods extending from about 2 to 4.5 days, consistent with recent modeling and data analyses (Forbes, He et al., 2020; Forbes, Maute & Zhang, 2020). Also seen is the Q6DW with periods ranging between 5 and 8 days centered mainly at s = +1, but with non-negligible amplitudes at s = 0 and s = +2. A smaller Q6DW occurs at s = −3, but its theoretical categorization is not known to the authors. The amplitude near 2 days period at s = −2 is likely the [0.5, −2] Q2DW mentioned above. The Q2DW, UFKW and Q6DW are depicted here at 98 km, since as noted in the Introduction, the PW modulation of solar tides (i.e., generation of SW and in particular those associated with DE3) primarily occurs below 100 km. In other words, penetration of a PW above 100 km is not a prerequisite for a PW signal to appear in the ionosphere.

In the following, we capitalize on the characteristic longitudinal “wave-4” signature of DE3 to demonstrate how the multi-day periodicities associated with the Q2DW, Q6DW and UFKW are transmitted to the F-region ionosphere through their modulation of DE3 and the A-I coupling mechanisms reviewed above and in the Introduction.

3. ICON Measurements

The wind measurements employed for this study were made by the MIGHTI (Michelson Interferometer for Global High-resolution Thermospheric Imaging) instrument on ICON (Englert et al., 2017). Specific data utilized are Version 4 (V04) U (positive eastward) and V (positive northward) winds between 96 and 106 km altitude and 12°S and 42°N geographic latitude, the latter being the full latitude extent of MIGHTI wind measurements. These are derived from two perpendicular tangent-point line-of-sight (LOS) vector measurements on the limb by observing the Doppler shift of the 557.7 nm “green-line” emission of atomic oxygen. Wind measurements are conducted during both day and night within this height regime. For later reference, the difference in local time between ascending and descending wind measurements for MIGHTI, which range from 0–2 h at the latitude limits quoted above to almost 12 h near 18°N, are shown with a solid line in in Figure 2. For green-line wind measurements, accuracies are better than 5.8 ms⁻¹ 80% of the time (Harding et al., 2017). The exceptions occur near the day/night boundaries and occasionally near the EIA, due to variations of wind and emission rate along the LOS. The precision of wind measurements ranges between 1.2 and 4.7 ms⁻¹ (Harding et al., 2017).

Figure 2.

Local Solar Time difference between ascending and descending measurements of Michelson Interferometer for Global High-resolution Thermospheric Imaging winds (solid line) and Ion velocity meter drift and ion density measurements (dotted line) as a function of latitude.

The Ion Velocity Meter (IVM) on ICON makes in-situ measurements of ion drifts, density, temperature and major ion composition (Heelis et al., 2017). The data utilized here consist of V02 meridional drift (Vm) measurements between about 10.0 and 20.0 LST averaged between ±5° magnetic latitude, which for all practical purposes represent vertical drifts (W_i) at the magnetic equator. IVM drift measurements are accurate to 7.5 ms⁻¹ within ±15° magnetic latitude whenever O⁺ densities are greater than 10⁴ cm⁻³ (Heelis et al., 2017), conditions which prevail throughout the data set considered in this study. Also included in our analysis are V02 total ion (O⁺ + H⁺) concentrations measured by IVM, considered equivalent to electron density and hereafter referred to as Ne.

4. Results

4.1. Vertical Drifts

In E-F-region coupling studies conducted in magnetic coordinates, it is usually the case that field-aligned conductivity-weighted winds are used to diagnose the connection between E-region dynamics and F-region drifts. However, UFKW represent a special type of wave wherein the U wind component dominates over the V component, U winds are concentrated at low latitudes, and the convergence of height-integrated wind-driven zonal current is the main driver of the eastward electric field that produces vertical drifts over the equator. It is also true that UFKWs possess broad latitudinal height and latitude structures, and therefore variability is strongly correlated between different heights and latitudes. And, as noted above, the geographic coordinate system is the more natural and effective one for diagnosing the origins of the tidal dynamics based on expression (3). Therefore in the present study zonal winds at the single height of 106 km (which typically corresponds to the peak height of the Hall conductivity) and ordered in geographic coordinates are used here to diagnose the E-region to F-region connection. This choice is also influenced and supported by the recent study by Forbes, He et al. (2020), which demonstrates that the conductivity-weighted U at the equator, or averaged between 20°S and 20°N, or simply equatorial U at the height of peak Hall conductivity, for a 2–3.5 day period UFKW all yield about the same northward magnetic perturbation on the ground. The GCM-electrodynamics model investigation by Jin et al. (2008) also concludes that the DE3 zonal wind near the Hall conductivity peak plays a major role in determining the daytime wave-4 electric field.

Figure 3 depicts the relationship between the longitudinal wave-4 components of U at 106 km and vertical drifts over the magnetic equator as measured by ICON during January 1–21, 2020. Figures 3a and 3b result from wave-4 fits to daily ascending orbital measurements of U averaged over 10°N–20°N and 5°S–5°N, respectively, moved forward every 6 h. The 6 h increment serves as a quasi-interpolation that enables a smooth depiction of the variability that includes Q2DW components. The y-axes of Figures 3a and 3b only extend between 0 and 90° longitude, since for wave-4 the structures between 90°–180°, 180°–270°, and 270°–360° longitude by definition duplicate those between 0–90°. Two x-axes are provided on each panel; the bottom x-axis labeling is linear in terms of day of year (DOY), and the top x-axis provides the corresponding LSTs which are nearly linearly related to DOY as a result of orbit precession. The bottom panel (c) is constructed in the same way as (a) and (b), except for W_i averaged between 5°S and 5°N magnetic latitude at the mean 592 km height of ICON. Ascending orbit measurements are shown in Figures 3a–3c since they include late morning to early evening hours when dynamo coupling between the E and F regions is active. From Figures 3a–3c we note that amplitudes are roughly of order ±40 ms⁻¹ for U and ±10 ms⁻¹ for W_i.

Figure 3.

(a),(b): Longitude versus day dependence of wave-4 component of ascending U at 106 km averaged between 10°N–20°N and 5°S–5°N, respectively, during January 1–21, 2020. (c): Same as (b) except for vertical drifts W_i at the mean Ionospheric Connections Explorer altitude of 592 km and MagLat = 5°S–5°N. Local solar times (LST) are indicated at the top of (a), (b) and (c). Dashed black lines indicate the expected LST versus longitude slope for DE3; dotted for SPW4 and SE2, which are included for reference only. Note that the longitude scales only extend from 0° to 90°, since for wave-4 the structures between 90° and 180°, 180°–270°, and 270–360° longitude by definition duplicate those between 0 and 90°. (d), (e), (f): Lomb-Scargle periodograms of the wave-4 amplitudes corresponding to (a), (b) and (c), respectively, for day of year 7–21 when the cause-effect relation between DE3 winds and DE3 drifts appears to be active.

Figures 3a–3c also include black lines indicating the theoretical slopes for DE3 (dashed), SPW4 and SE2 (dotted) from expression (3); namely $\frac{d{t}_{LST}}{d\lambda }=16\text{h}/{60}^{°}$, 0 h/60°, 8 h/60°. (Note that slopes downward to the right are positive.) The dashed lines for DE3 are overlaid in Figures 3a–3c since they correspond best with the overall trends in the data; SPW4 and SE2 dotted lines are shown for reference. Note that prior to about DOY 9, the patterns are also plausibly consistent with “no trend,” i.e., SPW4. Evidence for the presence of SPW4 near the equator is provided in Figure S1. Comparing the dashed lines denoted “DE3” in Figures 3a and 3b, the DE3 winds are in phase with respect to longitude between the two latitude regimes, while the DE3 W_i in Figure 3c is roughly in antiphase with the DE3 winds. Both of these features are consistent with the results and interpretation given in the modeling work of Jin et al. (2008). With respect to being in near-antiphase in longitude, wave-4 W_i is shifted almost 15° eastward of DE3 U. Jin et al. (2008) did not mention such a shift in their work, which employed a simple aligned dipole magnetic field. We suspect that the origin of this 15° shift may lie in the distorted way in which a DE3 wind field ordered in geographic coordinates interacts with a magnetic field whose inclination, declination and magnitude varies with longitude (see, e.g., Maute et al., 2012).

It is interesting that W_i is consistent with dynamo generation by DE3 after about DOY 7, but is comparatively muted in character prior to DOY 7. A possible explanation is that electrodynamic coupling may be weak during this post-sunset (LST = 17.8–20.6 h) time period when E-region conductivities are low. However, another factor potentially at play is that the SH wave-4 wind field between 4°S and 12°S (see Figure S2 in Supporting Information) does not show any appreciable amplitudes until about DOY 7. Therefore, the wave-4 E-region wind field appears to be more asymmetric about the equator prior to DOY 7. It was noted above that coincidence of the DE3 peak winds with maximum Hall conductivity is an important factor in making DE3 an efficient generator of electric fields (Jin et al., 2008). These authors also make the point that the symmetry of DE3 about the equator in their model was key to its dynamo efficiency; basically, the induced dynamo currents and electric fields from both hemispheres reinforce each other. This is consistent with the finding of Maeda (1974) that wind fields that are asymmetric about the equator exert their dynamo influence through the generation of field-aligned currents. Our interpretation is that the dynamo generation of electric fields and plasma drifts by DE3 as displayed in Figure 3 is inactive prior to DOY for one or both of the above reasons. By the same token, SE2 and SPW4 do not reveal themselves in the plasma drifts at any time during DOY 1–21 due to their strong hemispheric asymmetries.

The signatures of DE3 in all three panels of Figure 3 are consistent with the interpretation that the variability in wave-4 F-region plasma drifts originates predominantly from the wave-4 variability in electric fields produced by E-region dynamo action of DE3 winds. The nature of the wave-4 variability in Figures 3a–3c is illustrated in the Lomb-Scargle spectra shown in Figures 3d–3f, which are based on the 6-hourly time series of wave-4 amplitudes between DOY 7–21 when the cause-effect relation between DE3 winds and DE3 drifts is active. These spectra indicate that the wave-4 variability occurs over periods of roughly 2–8 days, consistent with the spectrum (Figure 1c) of PW and UFKW oscillations that existed during January 1–21, 2020. We do not expect exact coincidence between these spectra since the drifts represent an integrated response to the conductivity-weighted DE3 wind field over a range of heights and latitudes, and each location may reflect somewhat different PW variability. We contend that the evidence presented above supports the interpretation that the day-to-day variability depicted in Figure 3 originates in the PW and UFKW modulation of DE3.

Results for MIGHTI winds similar to those in Figures 3a–3b were obtained for magnetic latitudes 0°–10°N and 10°–20°N (not shown), the latter being consistent with the footpoint of the field line where ICON IVM observes W_i at the magnetic equator. However, the patterns did not match the DE3 slopes quite as well as in Figure 3, likely because the longitude-LST dependencies underlying the slopes obtained from expression (3) are derived from equations in the geographic coordinate system. The Lomb-Scargle spectra for magnetic latitudes 0°–10°N and 10°–20°N were somewhat different those in Figure 3 too, also reflecting the basic fact that the natural coordinate system for Figure 1c is geographic or geocentric coordinates. It is also noteworthy that the period versus wavenumber spectra similar to Figure 1c differed depending on height and latitude, although containing the same basic periodicities. Thus Figure 1c should only be interpreted as roughly representative of the longer-period wind variability that occurred during January 1–21, 2020.

Figure 4a provides an additional perspective on wave-4 variability in U at 106 km. Here we have applied the same fitting methodology as in Figure 3 at all latitudes between 9°S and 30°N, in order to create the latitude versus day depiction of wave-4 amplitudes in Figure 4a. Since no multi-day binning is performed, only data in this latitude range is utilized in order to avoid data gaps near the extrema of latitudes accessed by the MIGHTI instrument (see Figure 2). The contrast with the 40-days climatology of DE3 is notable. The amplitudes in Figure 4a are often 2–3 times those depicted in Figure 1a, and with much day-to-day variability in both amplitude and latitude structure. There may also be contributions from SE2 and SPW4 in Figure 4a, but as noted above these seem to be making secondary contributions electrodynamically. Figure 4b provides a similar depiction for wave-4 variability in Ne between MagLat = 20°S–20°N. This is the subject of the following subsection.

Figure 4.

Latitude versus day variability of longitudinal wave-4 amplitude, during January 1–21, 2020: (a) U at 106 km; (b) percent variation of Ne with respect to zonal mean, corresponding to a mean satellite altitude of 592 km. The local solar time scales at the top of each figure correspond to θ = 0° and MagLat = 0° for (a) and (b) respectively.

4.2. Total Ion Density, or Ne

In this section we examine the behavior of the longitudinal wave-4 component of Ne during the January 1–21, 2020 time period as measured by the IVM instrument on ICON. The wave-4 component of Ne is expressed here in terms of percent residual with respect to the longitudinal or zonal mean of Ne as depicted in Figure 5 as a function of magnetic latitude and day during January 2020. The ~1100–2100 LSTs of equatorial crossings of the IVM instrument are given at the top of this figure, which are identical to those shown at the top of Figure 3c for W_i. The altitudes of Ne values in Figure 5 correspond to a mean altitude of 592 km. Since F-layer peak heights at low latitudes are typically in the range 275–325 km under low solar activity conditions (e.g., K.-F. Li et al., 2018), the measurements shown in Figure 5 are estimated to be roughly 250–300 km above the F-layer peak (hmF2). Absence of the EIA peaks is expected this far above hmF2 (e.g., K.-F. Li et al., 2018). The skewed shape of the zonal-mean Ne distribution with respect to hemisphere and day (or LST) may due to the change in solar zenith angle as a function of geographic latitude and season.

Figure 5.

Zonal-mean Ne as a function of MagLat and day during January 1–21, 2020 at a mean altitude of 592 km. local solar time scale is given at the top, and corresponds to MagLat = 0°.

The variations in the wave-4 component of Ne are depicted in Figure 6 for the three magnetic latitude ranges (a) 10°N–20°N, (b) 5°S–5°N, and (c) 20°S–10°S. The variability in wave-4 of Ne illustrated in Figure 6 is quite significant, of order ±25–35%, depending on latitude and day during January 2020. Some part of this variability is obviously in the 2–6 days range, consistent with the hypothesis that PWs and UFKWs in this range are modulating tides that impose their variability on the ionosphere. However, there is also a longer-term modulation of the intensity of this variability that possibly originates in the latitude-dependent LST variability of the basic state that is depicted in Figure 5. Our objective below is to concentrate on the shorter-term variability, and to consider separately the Ne response for DOY ≳ 7 when electrodynamic coupling between the E-and F-regions is “active” according to the discussion surrounding Figure 3, and for DOY ≲ 7 when evidence indicates that it is inactive.

Figure 6.

(a), (b), (c): Longitude versus day dependence of wave-4 component of percent residual of Ne from zonal-mean in Figure 5, at Ionospheric Connections Explorer height averaged between 10°N–20°N, 5°S–5°N, and 20°S–10°S, respectively, during January 1–21, 2020. Dashed black dashed and dotted lines indicate the expected local solar times versus longitude slopes for DE3 and SE2, respectively. Local times are indicated at the top of (a), (b) and (c). Note that the longitude scales only extend from 0° to 90°, since for wave-4 the structures between 90°–180°, 180°–270°, and 270°–360° longitude by definition duplicate those between 0 and 90°. (d), (e), (f): Lomb-Scargle periodograms of the wave-4 amplitudes corresponding to (a), (b) and (c), respectively, for day of year 7–21 when the cause-effect relation between DE3 winds and DE3 drifts appears to be active.

Analogous to Figure 3, the theoretical slopes of several tides with wave-4 signatures in this reference frame are superimposed for reference. Also, panels (d), (e) and (f) illustrate the Lomb-Scargle (L-S) spectra of the wave-4 amplitudes corresponding to (a), (b) and (c) in Figure 6 for DOY 7–21, consistent with Figure 3c. To focus on the shorter-term variability of PW/UFKW origin, the L-S spectra are based on residuals from a 9-days running mean. After about DOY 8, the mean trends in Figures 6a and 6c are, on average, consistent with the interpretation that the illustrated wave-4 structures are associated with a DE3 oscillation with day-to-day variability imposed by interactions with a spectrum of PWs. That is, the appearance of DE3-related variability at the locations of 10°S–20°S and 10°N–20°N is consistent with meridional transport (i.e., “fountain effect”) driven by the vertical drifts at the magnetic equator illustrated in Figure 3c. However, this consistency with drifts in Figure 3c does not exist for the equatorial-region variations in Ne as depicted in Figure 6b. Instead, the trend after DOY 9 is more SE2-like, suggesting the possible influence of SE2 meridional winds, which is discussed in the following section.

It is noteworthy that the L-S spectra in Figures 6d–6f all share the same spectral peaks, a dominant one near 2-day period, and another at about 4.0–4.5 days. These Ne spectra are similar to the W_i spectrum, which has peaks near 2.5 and 5.0 days. However, the W_i spectrum has its dominant (secondary) peak near 5.0 (2.5) days. It is visually evident that some part of the 2-days variability in Ne occurs during the DOY 4–8 time frame, which includes the DOY < 7 period when vertical drifts are weak (see Figure 3c). Another interesting feature in Figure 6 is the upward trend (corresponding to $\frac{d{t}_{LST}}{d\lambda }<0$ from expression (3)) in wave-4 Ne amplitudes seen prior to DOY 6 in Figure 6a, prior to DOY 9 in Figure 6b, and prior to DOY 8 in Figure 6c. As indicated in these figures, the theoretical slope from expression (3) for DW5 ($\frac{d{t}_{LST}}{d\lambda }=-16\text{h}/{60}^{°}$ opposite to that of DE3) agrees with these upward trends reasonably well. The slope for SW6 (−8 h/60°) also provides a good match for the upward trends in Figures 6 (as well as forthcoming Figures 7 and 8), but SW6 was assumed of secondary importance for reasons given in subsection 4.4. As explained in the next section, DW5 can arise through nonlinear interaction between SPW4 and DW1 or between DE3 and SW2, but cannot propagate upward from the E-region due to its short vertical wavelength. These features of the Ne response combine to suggest that the observed Ne variability has strong connections to E-region neutral dynamics, but not all connected with DE3-driven vertical drifts, and moreover that F-region neutral dynamics is somehow involved. This prompts us to look at the nature of the wave-4 dynamics in the F-region in the following subsections.

Figure 7.

(a) and (b): Same as (a) and (b) in Figure 3, except for U at 295 km. (c) and (d): same as (d) and (e) in Figure 3, except for U at 295 km.

Figure 8.

Same as Figure 7, except for V.

4.3. F-Region Wave-4 Dynamics: Observations

Figures 7a–7d and 8a–d provide information similar to that in the top two rows of Figure 3, except for U and V averaged between 280 and 310 km, hereafter “295 km.” The gaps correspond to outliers (>3 × median values). We know from theory and modeling that molecular diffusion of momentum is sufficiently fast at these altitudes that a vertical shear in U or V cannot be maintained. Therefore the depictions in Figures 7 and 8 can reasonably be assumed to apply at ICON altitudes as well.

At 10°N–20°N (Figure 7a) the wave-4 variability in U is of order ±20 ms⁻¹, roughly 50% of that at 106 km (Figure 3a), and mainly occurs with a ~3–5 day periodicity (Figure 7c). A transition occurs from DW5-like to DE3-like trends around DOY 7; the possible origin of DW5 is discussed later in this section. It is also conceivable that the structures prior to DOY 9 are consistent with SPW4, i.e., a theoretical slope of zero, or a horizontal straight line. A 40% reduction in wave-4 variability with respect to 106 km occurs at 5°S–5°N (Figure 7b as compared with Figure 3b), with the variability also occurring mainly between 3 and 5 days. Similar to the trend at 106 km between 10°N and 20°N (Figure 3a) the main contributor to U at 295 km at these latitudes appears to be DE3. This suggests that DE3 propagated from 106 to 295 km with a ≲50% reduction in amplitude, carrying the 3–5 days signal with it. This decrease in amplitude with height is in reasonable accord with results from numerical modeling of DE3 under solar minimum conditions (e.g., Gasperini et al., 2017), which indicates a ~40% reduction. Note also the significant phase (longitude of maximum at 0000 UT) difference between DE3 at 106 km (Figure 3a) and at 295 km (Figure 7a). However, phase difference ambiguities in the data between just two heights preclude any definitive comparison with theory/modeling.

Shifting from 10°N to 20°N to the equator (Figures 7a to 7b), DE3 gives way to SE2 as the main contributor to the U wind field, similar to the change in Ne between Figures 6a and 6b. Prior to DOY 8, a weaker presence of DE3 at lower latitudes is consistent with the hemispheric asymmetry in DE3 noted in connection with the discussion around Figure 3. Also similar between Ne and U are the transitions at 5°S–5°N from DW5 to SE2 around DOY 8–11 (compare Figures 6b and 7b), and the corresponding DW5-DE3 transitions at 10°N–20°N, which occur around DOY 7–8 (compare Figures 6a and 7a). Indeed, Figures 6b and 7b are remarkably similar, yet represent neutral and plasma measurements at different altitudes made by different instruments. The only direct relationship that might exist between F-region zonal winds and Ne would be through field-aligned transport in connection with declination of the magnetic field, and the corresponding changes in topside Ne due to vertical movements of the F-layer.

The V wind counterpart to Figure 7 is provided in Figure 8. The V amplitudes in Figure 8 are of order ±20 ms⁻¹, similar to those in Figure 7. In-situ meridional winds can similarly modify topside Ne through field-aligned plasma transport and vertical movements of the F-layer. At 5°S–5°N (Figure 8b) the dynamics is similar to that for U (Figure 7b) in that there is a transition from DW5 to SE2 around DOY 8–12. However, for V the periodicities (Figures 8c and 8d, corresponding to DOY 7–21) are not as well expressed as U in terms of isolated peaks near 2 and 4 days; rather the power is spread across 1.5–8.0 days with no peak rising above the 90% confidence level. The spectrum is not fundamentally changed when calculated over DOY 1–21 (not shown).

Summarizing to this point, our interpretation is that the wave-4 Ne variability depicted in Figure 6 is driven in part by plasma drifts with origins in DE3 E-region dynamo action after DOY 7, and in part by field-aligned transport due to in-situ winds connected with DE3, SE2, SPW4 and DW5 throughout the DOY 1–21 period. We now turn to a discussion of the possible origins of DE3, SE2, SPW4 and DW5 in the F-region.

4.4. F-Region Wave-4 Dynamics: Possible Origins

There are several processes that potentially underly the behavior and appearance of DE3, SE2 SPW4 and DW5 at F-region heights. First, we know from GCM modeling (e.g., Hagan et al., 2009; Pedatella et al., 2012) that SE2 and SPW4 can arise as a result of nonlinear interaction between DE3 and DW1, which results in SE2 and SPW4 as products defined by the sum and difference frequencies and zonal wavenumbers of the interacting waves: DW1 × DE3 = [1,+1] × [1,−3] → [2,−2] + [0,+4] = SE2 + SPW4. These interactions in the above GCM studies were mainly focused on the E-region, but in principle they can take place throughout the vertical column. In a model, the products occur in advective terms in the momentum and thermal energy equations, but also in the ion drag term in the momentum equation where DW1 would represent the diurnal variation in ion drag; they thus serve as momentum and heat sources for SE2 and SPW4. The migrating semidiurnal component (SW2) of in-situ winds and ion drag can play a similar role. By extension, we hypothesize that the only plausible sources for DW5 are SPW4 × DW1 = [0,+4] × [1,+1] → [1,+5] + [1,−3] = DW5 + DE3 and/or DE3 × SW2 = [1,−3] × [2,+2] → [3,−1] + [1,+5] = TE1 + DW5. It is furthermore noted that SW6 can arise from the interaction between SPW4 and SW2: SPW4 × SW2 = [0,+4] × [2,+2] → [2,+6] + [2 − 2] = SW6 + SE2. Given the generally larger value of DW1 compared with SW2 in the thermosphere-ionosphere, the role of SW6 was assumed secondary to that of DW5 in prior subsections even though it provided an equally satisfactory slope as DW5 for the interpretation of structures in Figures 6, 7 and 8.

In addition to DE3, SE2 can also originate in the troposphere (Zhang et al., 2010a,b). While the first symmetric component of DE3 propagates upward with a vertical wavelength of order 56 km according to classical tidal theory (Chapman & Lindzen, 1970; see also Truskowski et al., 2014 for a tabulation), the first symmetric component of SE2 has such a long vertical wavelength (>200 km), that it becomes evanescent in the mesosphere and does not penetrate effectively to the thermosphere. On the other hand, the first antisymmetric and second symmetric modes of SE2 have classical theory vertical wavelengths of order 183 and 77 km (Truskowski et al., 2014), and very effectively propagate into the viscous thermosphere, more so than DE3 (Forbes et al., 2014). In fact, the first antisymmetric mode of DE3 has a classical wavelength of only 30 km, and dissipates in the lower thermosphere; therefore DE3 is expected to be symmetric about the equator in the upper thermosphere and this is borne out by CHAMP and GRACE data (Forbes et al., 2014).

The contrast between DE3 and SE2 behaviors and characteristics in the thermosphere is illustrated in Figure 9, which depicts the height-latitude structures of U and V for the first symmetric component of DE3 and the first antisymmetric component of SE2. Phases corresponding to the amplitudes in Figure 9 can be found in Figures S3 and S4 of the Supporting Information. These depictions are the so-called “Hough Mode Extensions (HMEs)” for DE3 and SE2 (Forbes & Hagan, 1982; see also Svoboda et al., 2005, and Oberheide et al., 2011, for descriptions of how HMEs are used to fit tidal data). The HMEs are basically extensions of classical tides (i.e., in an atmosphere without mean winds and dissipation; Chapman & Lindzen, 1970) into the viscous thermosphere. The DE3 HME amplitudes in Figure 9 are normalized to an equatorial U amplitude at 106 km of 35 ms⁻¹ based on the data in Figure 3b, roughly a factor of two larger than the 40-days climatological values in Figure 1a. The SE2 amplitudes are more arbitrarily normalized to a U amplitude of 15 ms⁻¹ at 106 km and 30° latitude. We note from Figure 9 that DE3 contributes little in the way of meridional winds (consistent with the fact that the symmetric part of DE3 is a Kelvin wave), the influences of DE3 on the F-region are modest compared to the E-region, and that modest SE2 amplitudes in the E-region (~15 ms⁻¹) can grow to quite substantial amplitudes (~50–60 ms⁻¹) in the F-region.

Figure 9.

Hough Mode Extensions U (left) and V (right) amplitudes as a function of height and latitude for the first symmetric component of DE3 (top) and first antisymmetric component of SE2 (bottom) Solar minimum conditions (F10.7 = 60) conditions are assumed.

Figure 9 enables some insights into the observations depicted in Figures 7 and 8. The F-region DE3 U amplitudes contributing to the wave-4 variability in Figure 7 are likely of order 10 ms⁻¹, about half of the displayed amplitudes, and consistent with the inference that other waves such as SPW4 and SE2 are making important contributions. Figure 9 also suggests that the F-region DE3 V amplitudes are less than 5 ms⁻¹, and thus make comparatively small contributions to the ±25 ms⁻¹ amplitudes displayed in Figure 8. The suggestion that SE2 represents a better fit to the trends in Figure 8 also emerges as plausible, given the factor of three amplification from E-region to F-region wind amplitudes indicated in Figure 9 for SE2.

The potentially large amplitudes of SE2 in the F-region also have other implications. The interaction between SE2 and DW1 (the latter in terms of either winds or ion drag) results in the terdiurnal tide with s = −1 (TE1) and DE3: DW1 × SE2 = [1,+1] × [2,−2] → [3,−1] + [1,−3] = TE1 + DE3, and the interaction between SE2 and SW2 results in the quatra-diurnal tide with s = 0 (Q0) and SPW4: SW2 × SE2 = [2, 2] × [2,−2] → [4, 0] + [0, 4] = Q0 + SPW4. These interactions plausibly provide an additional DE3 source for wave-4 U and V, and potentially a source for SPW4 U in Figure 7. In addition, Q0 and TE1 both appear as wave-4 in the satellite frame, and can add to the lack of coherence in Figures 7 and 8 as compared with, e.g., Figure 3b at 106 km. We furthermore note that the interactions between DE3 and SE2 with the Q2DW, Q6DW and UFKW produce secondary waves below the thermosphere, some of which have the capability to propagate to the F-region. It is these secondary waves that likely carry, at least in part, the PW and UFKW periodicities to the F-region and that produce the wave-4 variability in U, V and Ne at these periods, as recently demonstrated in numerical simulations of Q6DW-ionosphere interactions by Miyoshi and Yamazaki (2020). However, we could not identify any such waves that would occur with |m − δ| = 4, and contribute to the dynamics displayed in Figures 7 and 8.

The main point the above discussion is meant to convey is the following. While DE3 may be the main contributor to wave-4 zonal winds in the E-region and perhaps even F-region equatorial vertical drifts, the F-region wave-4 neutral dynamics is a much more complicated matter as a result of tide-tide interactions and their wave products, and this in turn adds further complexity to the F-region Ne response. Modeling is obviously needed to explore the veracity of the above interpretations.

5. Conclusions

A major objective of the ICON mission is to advance our understanding of how troposphere variability (“terrestrial weather”) translates to ionosphere variability (“space weather”) with particular focus on low latitudes. Toward this end, ICON provides for the first time coincident measurements of neutral winds and temperatures in the E-and F-regions, F-region plasma drifts and densities, and chemical composition. In addition, while the low inclination of the ICON orbit limits latitude coverage, it also provides new insights and capabilities through its enhanced LST sampling, a feature that is leveraged in the present work.

The troposphere dynamics pertaining to the present study involves the diurnal cycle of deep tropical convection and the accompanying release of latent heat of evaporation, an excitation source of mesosphere/lower thermosphere solar tides (Williams and Avery, 1996; Forbes et al., 1997; Hagan and Forbes, 2002, 2003). With regard to DE3, these works can be traced back to Yagai (1989) who recognized that excitation of DE3 occurred as the result of the wave-4 component of land-sea difference (or topography) in the tropics. DE3 propagates upward with a vertical wavelength of order 60 km, enabling it to propagate well into the thermosphere, reaching peak amplitudes around 110 km where its growth is terminated by molecular dissipation (Figure 9). Cast in a local time framework, DE3 appears as longitudinal wave-4, and thus essentially carries the wave-4 distribution of land-sea difference and latent heating into the thermosphere and ionosphere as originally recognized by Sagawa et al. (2005) and Immel et al. (2006). In this same framework, The phase lines of DE3 wave-4 have the characteristic slope $\frac{d{t}_{LST}}{d\lambda }=16\text{h}/{60}^{°}$; with the LST precession provided by ICON, DE3 can be identified in the data without the usual 24 h coverage in local time required to extract diurnal tides from the observations. In this paper, similar tidal diagnoses were performed on wave-4 vertical plasma drifts over the magnetic equator, and wave-4 electron density (Ne) variations (percent residuals with respect to zonal mean) between ±20° magnetic latitude, during January 1–21, 2020 with the following results:

1. Low-latitude DE3 zonal winds at 106 km (the nominal height of peak Hall conductivity) varied significantly (±40 ms⁻¹) at periods of order 2–7 days, while other prevailing oscillations during the same time interval consisted of quasi-2-days and quasi-6-days PWs, and a 2–4-days UFKW. Our interpretation is that the DE3 variability originated from nonlinear interaction with (i.e., modulation by) these long-period waves.

2. The wave-4 equatorial vertical drifts were also characterized by DE3-like behavior after DOY 7, with ±10 ms⁻¹ variability at periods 2–7 days. Our interpretation is that the wave-4 vertical drifts were mainly driven by the PW-modulated DE3 zonal winds. While DE3 winds were strong in the NH prior to DOY 7, they were absent in the SH. We attribute the weak plasma drifts prior to DOY to this asymmetry in the wave-4 wind field, which is counter to the efficient generation of electric fields and currents (Jin et al., 2008).

3. Amplitudes of wave-4 zonal and meridional winds at 295 km were of order ±20 ms⁻¹ and also varied with periods of order 2–7 days. However, in addition to DE3, signatures of SE2, SPW4 and DW5 were evident, as determined by their expected slopes $\frac{d{t}_{LST}}{d\lambda }$ based on their individual periods and zonal wavenumbers. DW5 and SPW4 (DE3 and SE2)were confined to the time period prior to (after) DOY 7–11.

4. Wave-4 variability in Ne was of order 25%–35%, but the origins were not as universally associated with DE3 as the zonal winds at 106 km and the equatorial vertical drifts. In the magnetic latitude regimes 10°S–20°S and 10°N–20°N normally associated with the EIA maxima, wave-4 Ne variability was dominated by DE3-like behavior after about DOY 8. This is consistent with transport by DE3 drifts in the meridional plane. However, prior to DOY 8, Ne wave-4 variability occurred with a longitude versus LST slope consistent with DW5.

5. Near the magnetic equator, Ne wave-4 variability was mainly associated with SE2 (DW5) after (before) DOY 11.

The above results represent the first experimental confirmation of A-I coupling due to PW-modulated tides previously anticipated from theory and modeling (Forbes et al., 2018). However, in addition to Ne responding to drifts connected with E-region dynamo action and field-aligned plasma transport associated with PW-modulated tides that propagate into the F-region, there is an additional layer of complexity. That is, nonlinear interactions between DE3 and possibly SE2 and the ambient ionosphere-thermosphere generate additional non-negligible wave components that characterize the F-region wind field and further complicate the Ne variability. As shown here, while DE3 decays with height above about 120 km, SE2 grows with height in the thermosphere, and can originate from modest amplitudes at lower heights. SE2 is also a product of tropospheric heating (Hagan and Forbes, 2003; Zhang et al., 2010a,b). Both SE2 and SPW4 can also arise through DE3 nonlinear interactions with sun-synchronous diurnal (DW1) variations in winds and ion drag, as demonstrated in general circulation models (e.g., Hagan et al., 2009; Pedatella et al., 2012). Moreover, SE2 can in principle interact in-situ with these DW1 variations to serve as a source of TE1 and a secondary source of DE3; and can interact with sun-synchronous semidiurnal (SW2) variations in winds and ion drag to produce SPW4 and the quatra-diurnal tide with s = 0 (Q0). Of particular note in the present study is DW5, which can arise from the SPW4 × DW1 and DE3 × SW2 interactions. All of these secondary products appear as wave-4 in the longitude-LST framework, contributing to the complexity of the wave-4 Ne response.

It is interesting to note that the modeling study of Pedatella et al. (2012) identified SPW4 as an important contributor to wave-4 variations in the ionosphere in addition to DE3, through the generation of vertical drifts and meridional winds, but that SE2 had little impact on the ionosphere. In an analysis of COSMIC hmF2 and NmF2, He et al., (2011) showed that the symmetric part of the ionospheric response was controlled mainly by DE3-generated electric fields, while the antisymmetric part of the response was determined by SE2 trans-equatorial winds, especially insofar as nighttime wave-4 structures were concerned. The current results indicate that SE2 plays a role in determining both the daytime wave-4 F-region zonal and meridional wind field and Ne response near the magnetic equator, but may also significantly impact wave-4 variability through its nonlinear interactions with the ambient ionosphere-thermosphere. The degree to which it exerts its influence likely depends on its symmetric versus asymmetric character, and whether it originates from below or is excited in-situ through DE3-DW1 interaction. Further modeling efforts focused on A-I coupling by DE3 and SE2 are needed to assess the veracity of our interpretations, and to clarify the roles of SPW4 and SE2, and perhaps other secondary waves arising from tide-tide nonlinear interactions (e.g., DW5), in this coupling process.

Key Points:

• Coincident Ionospheric Connections Explorer measurements of neutral winds, plasma drifts and total ion densities (:=Ne) are analyzed during January 1–21, 2020

• We show for the first time that planetary wave winds modulate DE3 and produce longitudinal wave-4 variations in F-region vertical drifts ~±10 ms⁻¹, Ne ~ ±30%

• Measured F-region wave-4 winds suggest that SPW4, SE2 & DW5 arising from tide-tide and tide-ion drag interactions also contribute to Ne variability

Data Availability Statement

The data utilized in this study are available at the ICON data center (https://icon.ssl.berkeley.edu/Data).