Modeling of Planetary Wave Influences on the Pre‐reversal Enhancement of the Equatorial F Region Vertical Plasma Drift

Temporal and longitudinal variations of the pre‐reversal enhancement (PRE) in the equatorial F region vertical plasma drift are examined based on idealized simulations by the thermosphere‐ionosphere‐electrodynamics general circulation model performed under geomagnetically quiet (Kp = 1) and high solar‐flux (F10.7 = 200) conditions. The model takes into account forcing by large‐scale waves from the middle and lower atmosphere, which leads to day‐to‐day variations of PRE. Simulations are performed under different wave forcing in order to separate contributions of various types of waves. It is shown that the simulated day‐to‐day variability of the PRE intensity is predominantly due to forcing by waves with periods less than 2 days, that is, tides and their modulation. Planetary‐wave forcing (periods of 2–20 days) makes contributions to periodic oscillations in the PRE intensity. Especially, the westward‐propagating quasi‐6‐day wave (Q6DW) with zonal wavenumber 1 is found to be an important source of ∼6‐day oscillations of PRE. Not only the Q6DW from below but also the Q6DW generated within the thermosphere, as well as the secondary waves due to the nonlinear interaction between the Q6DW and migrating tides, is at play. The zonal wavenumber 1 nature of the ∼6‐day oscillations could contribute to longitudinal differences in the appearance of equatorial spread F and plasma bubbles, which are strongly controlled by PRE.


Introduction
The E ×B drift dominates the plasma motion across magnetic field lines in the F region ionosphere. In the equatorial region, the quiet-time dynamo electric field is usually eastward during the day and westward during the night (e.g., A. D. , and thus the F-region vertical plasma drift is upward during the day and downward during the night (e.g., Fejer et al., 2008a;. Around the sunset time, the upward drift velocity (V Z ) often shows a rapid increase before the reversal to downward flow (Woodman, 1970). This phenomenon is known as the "pre-reversal enhancement (PRE)" in the equatorial vertical plasma drift. PRE is frequently observed during solar maximum while it is small or absent during solar minimum (e.g., Fejer et al., 1991;Madhav Haridas et al., 2015;Stoneback et al., 2011). The strong upward drift due to PRE has been identified as a determining factor for the appearance of the equatorial spread F Hysell & Burcham, 2002) and equatorial plasma bubbles (Huang & Hairston, 2015;Kil et al., 2009;Stolle et al., 2008), which can cause radio wave scintillations and hence are major space weather concerns.
The production mechanism of PRE has been explained in terms of the F-region dynamo and its electrodynamic coupling to the E region (e.g., Eccles, 1998;Farley et al., 1986;Heelis et al., 1974;Rishbeth, 1971). Eccles et al. (2015), reviewing various contributing mechanisms, concluded that PRE is primarily due to the curl-free nature of the electric field. Below is a brief summary of the curl-free mechanism. The polarization electric field due to the F-region dynamo can be approximated in the following formula (e.g., Heelis, 2004): where E is the electric field, U F is the F-region wind velocity, B is the Earth's main magnetic field, and Σ F This study is also motivated by renewed interest in planetary-wave influences on the ionosphere in general owing to recent progress in the subject based on global satellite observations (e.g., Gan et al., 2020;Gu et al., 2018;Yamazaki et al., 2018) as well as advances in the understanding of the underlying processes by dedicated modeling work (e.g., Forbes et al., 2018b;Forbes et al., 2018a;Gan et al., 2017;. Most modeling studies on the planetary-wave effects on the ionosphere have been conducted under solar minimum conditions, for which PRE is negligibly small. The present study is unique in examining the ionospheric response to planetary waves for high solar-flux conditions with a special focus on PRE.

Model and Numerical Experiments
The TIE-GCM is a physics-based global three-dimensional model of the coupled ionosphere-thermosphere system (Qian et al., 2014;Richmond et al., 1992). For the present study, we use version 2.0 of the TIE-GCM with horizontal resolution of 2.5 × 2.5° in geographic longitude and latitude. The vertical coordinate of the TIE-GCM is defined as Z = ln (p 0 /p), where p is the pressure and p 0 (=5 × 10 −7 hPa) is the reference pressure. The model extends from Z = −7 (∼97 km altitude) to Z = 7 (∼700 km for solar maximum conditions) with vertical resolution of ΔZ = 0.25. The electrodynamics is calculated in the Magnetic Apex coordinate system (A.  with a realistic geomagnetic field configuration specified by the International Geomagnetic Reference Field (IGRF-12) (Thébault et al., 2015). Geomagnetic activity and solar activity can be specified using the Kp and F 10.7 indices, respectively. All simulations presented in this study were performed under geomagnetically quiet (Kp = 1) and high solar-flux (F 10.7 = 200) conditions.
Effects of upward propagating waves from the middle and lower atmosphere can be introduced by specifying the lower boundary. For this study, the lower boundary of the TIE-GCM was constrained using horizontal winds, geopotential height, and temperature at 97 km from another model called the thermosphere ionosphere mesosphere electrodynamics general circulation model (TIME-GCM) (Roble & Ridley, 1994), which itself is constrained at 30 km with 3-hourly outputs from the Modern Era Retrospective-analysis for Research and Applications (MERRA) (Rienecker et al., 2011). The method of constraining the TIE-GCM lower boundary with TIME-GCM/MERRA data is described in detail by Maute (2017), and we used the same lower-boundary fields as Maute (2017), which are based on the MERRA data for the year 2009. The use of the MERRA data for the solar minimum year of 2009 in our TIE-GCM simulations for high solar-flux conditions is justified by the fact that planetary waves and tides observed in the middle atmosphere below 97 km show little dependence on solar activity (Oberheide et al., 2009;Yamazaki, 2018). Koval et al. (2018) numerically showed that solar activity can affect planetary waves with periods of 4-16 days only above 100 km.
The lower boundary fields given by the TIME-GCM/MERRA data are separated into two parts. One is the daily zonal mean and the other is perturbations from the daily zonal mean. The former contains zonally symmetric (S0) oscillations at planetary-wave periods (e.g., Pancheva et al., 2009;Forbes et al., 2018b) while the latter contains eastward-and westward-propagating planetary waves and tides. We ran three cases of simulations with slightly different lower-boundary conditions. The first case is referred to as "reference" and uses the original values of the zonal mean and perturbation fields. The model results presented in this paper are from the reference simulation unless otherwise noted. In the second case, which is referred to as "no_PW," planetary waves are excluded from the lower-boundary fields by applying a high-pass filter with a cut-off period of 2 days only to the perturbation fields. In this way, eastward-and westward-propagating waves with periods longer than 2 days are removed from the perturbation fields while multi-day oscillations in the zonal mean fields are retained. The third case, referred to as "tides_only," uses the same perturbation fields as the no_PW simulation but additionally, a low-pass filter is used to remove the S0 oscillations with periods less than 100 days. In this way, variations at planetary-wave periods are removed from both the daily zonal mean fields and perturbation fields while retaining seasonal variations, which are dominated by the annual (365 days) and semiannual (182.5 days) components. Waves with periods less than 2 days in the lower thermosphere can be largely attributed to tides and their modulation through the interaction with other large-scale waves (e.g., Nystrom et al., 2018). The lower boundary conditions for the three cases are summarized in Table 1.
Differences in the results obtained from a selected pair of cases can highlight contributions of certain wave forcing mechanisms. For instance, the differences between the reference and no_PW simulations isolate the effect of planetary-wave forcing at 97 km, and hence are referred to as "PW_contribution." Similarly, the differences between the no_PW and tides_only simulations isolate the effect of S0 forcing at 97 km, and are referred to as "S0_contribution." Detailed discussion will be presented later regarding what waves are included in each case listed in Table 1 and how they could contribute to planetary-wave oscillations in PRE.
The three cases of simulations were run for the entire year from January 1 to December 31. The simulation time step was set to 15 s, and the vertical plasma drift velocity V Z was output every 5 min in model time, along with other diagnostic fields. The high sampling rate of the vertical plasma drift velocity helps to accurately determine the time and intensity of PRE. Our PRE detecting routine involves three steps. First, the V Z data at the magnetic equator at an altitude of 346.5 km were plotted as a function of local time for each day. Second, local maxima were searched in V Z between 1 h before and after the time of the E-region sunset at 110 km over the magnetic equator. Finally, the local maximum with the largest V Z was selected as PRE. PRE was sometimes detected after the vertical drift turns downward in the afternoon. Although, strictly speaking, these cases are not "pre-reversal" enhancement, they are kept in our PRE data. Accordingly, the PRE intensity can be a negative value, although such cases are not common. Also sometimes, no local maximum was detected within the said local-time range. In this case, no values were assigned to the intensity and local time of PRE. The PRE data from the reference, no_PW and tides_only simulations are archived at GFZ Data Services (Yamazaki & Dieval, 2021). Figure 1 presents two examples of V Z data plotted as a function of local time. In Figure 1a, a local maximum is found around the time of the E-region sunset. The purple vertical lines indicate ±1 h from the E-region sunset time at the magnetic equator. The intensity and local time of PRE are 18.7 m/s and 19.08 h, respectively. The pattern of the daily variation of V Z is in good agreement with those typically observed during solar maximum (e.g., Fejer et al., 1991;. Previous studies also found that the TIE-GCM is capable of reproducing observed features of the daily variation in V Z including PRE (e.g., Fesen et al., 2000;Heelis et al., 2012;Vichare & Richmond, 2005). Figure 1b shows a case where no PRE was detected.

Seasonal and Longitudinal Dependence
Daily values of the PRE intensity are plotted in Figure 2a as a function of longitude and day of year (DoY). The PRE intensity exhibits both day-to-day and seasonal variations at a given longitude. At a fixed DoY, the PRE intensity is longitudinally dependent. The seasonal and longitudinal dependence is more clearly visible in Figure 2b where temporal and spatial averaging is applied to the PRE intensity with a 30-day time and 15° longitude window. The obtained seasonal and longitudinal pattern resembles those derived from satellite measurements (e.g., Huang & Hairston, 2015;Kil et al., 2009). As mentioned earlier, the seasonal and longitudinal dependence of the PRE intensity is generally attributed to changes in the α angle between the sunset terminator and horizontal magnetic field. Figure  (1) reference TIME-GCM/MERRA TIME-GCM/MERRA Abbreviation: LB, Lower Boundary Table 1 Simulated Cases and Differences the PRE intensity and α angle. It is noted that the PRE intensity at a fixed DoY often shows a longitudinal variation with 4-8 peaks. Such variations are not necessarily associated with the α angle, which shows smoother longitudinal variations. Atmospheric waves with different zonal wavenumbers can contribute to longitudinal variations in ionospheric parameters. For example, the eastward-propagating diurnal tide with zonal wavenumber 3 is known to be a source of the four-peak longitudinal structure of the low-latitude ionosphere (e.g., Lühr & Manoj, 2013;Pancheva & Mukhtarov, 2010). Figure   It is seen that the local time of PRE also varies with DoY and longitude. It exhibits both day-to-day and seasonal variations at a given longitude. To our knowledge, this is the first time that day-to-day variations of the PRE local time due to atmospheric forcing are reproduced by a global model. At a fixed DoY, the PRE local time is longitudinally dependent. The pattern of the seasonal and longitudinal variations shown in Figure 2f, obtained by temporal and spatial averaging as in Figure 2b is different from that of the PRE intensity ( Figure 2b). The mean value of the PRE local time is 18.8 h. Figure 2g shows the seasonal and longitudinal dependence of the sunset time at 110 km at the magnetic equator. The PRE local time is seen to follow the seasonal and longitudinal variations of the E-region sunset time. Earlier, Kil and Oh (2011), using satellite observations, reported that the longitudinal variation of the PRE local time depends on the longitudinal variation of the E-region sunset time. Our simulation results confirm their finding. Figure 2h shows the standard deviation of the PRE local time calculated in the same way as the standard deviation of the PRE intensity in Figure 2d. The mean value of the standard deviation is 7.9 (±3.6) min, which is rather small given that the time resolution of our simulation output is 5 min. Thus, for the remainder of the paper, we focus on the behavior of the PRE intensity.

Planetary-Wave Influences: Ground Observer Perspective
We now examine the day-to-day variability of the PRE intensity from the perspective of ground observers.  Table 1). The tides_only results represent the contributions of tides (24,  reference and tides_only results indicate that the variability of PRE V Z is primarily due to tidal-wave forcing. The day-to-day variability associated with planetary-wave forcing is isolated in the PW_contribution results, which are indicated by the red lines in Figures 3a and 3d. The changes in the PRE intensity due to planetary waves are generally smaller than those associated with tidal waves, but occasionally, they become comparable when the planetary-wave contribution is enhanced (e.g., DoY from 120 to 140 in Figure 3a).

Figures 3b and 3e
show Morlet wavelet power spectra of the PRE intensity at the two locations as derived from the reference simulation. In previous observational studies, wavelet spectra have frequently been used to detect planetary-wave signatures in PRE (e.g., Abdu, Batista, et al., 2006). At 60°E longitude (Figure 3b), the strongest wave signal is seen at a period of 5-6 days around DoY = 120-130. At 180°E longitude (Figure 3e), the 5-6-day oscillation around DoY = 120-130 is not as prominent but a strong wave signal is seen at ∼6 day around DoY = 270-280. Figures 3c and 3f show similar results but for the no_PW simulation, which excludes traveling planetary waves with periods longer than 2 days at the lower boundary of the model (97 km Table 1 for the description of different simulation cases. In panels (b, c, e, and f), the 95% confidence level is indicated by the white dashed line.
to periodic oscillations (in particular those with periods of ∼6 days) in PRE V Z observed from the ground.
The results also suggest that ∼6-day oscillations result not only from planetary waves but also from other processes such as tidal modulation by planetary waves. Considering all longitudes, the amplitudes of ∼6-day oscillations in PRE intensity are 5.2 ± 1.8 m/s around DoY = 120-130 and 6.0 ± 2.4 m/s around DoY = 270-280 from the ground-observer perspective.

Planetary-Wave Influences: Global Perspective
The zonal structure of planetary-wave oscillations in the PRE intensity is examined using simulation results, which cover all longitudes. Fourier spectra (e.g., Wu et al., 1993) of the PRE intensity are presented in Figures 4a-4e for W2, W1, S0, E1, and E2, where "W" and "E" represent westward-and eastward-propagating components, respectively, and the number following W or E indicates the zonal wavenumber. As mentioned earlier, S0 signifies the zonal-mean oscillation. The rich spectra of W2, W1, and E1 contribute to the complex day-to-day variability of the PRE intensity. The largest contribution to the PRE intensity oscillations at periods of 2-20 days comes from the W1 component (i.e., westward-propagating oscillations with zonal wavenumber 1). Especially, ∼6-day oscillations are prominent around DoY = 120-140 and DoY = 270-290, during which wavelet spectra at fixed longitudes show ∼6-day peaks (Figures 3b and 3e). The maximum amplitude of the ∼6-day oscillation in the W1 component exceeds 4 m/s.
Fourier spectra of the zonal wind over the equator at the TIE-GCM lower boundary (97 km) are presented in Figures 4f-4j for W2, W1, S0, E1, and E2. Enhanced W1 activity at periods of ∼6 days is seen around DoY = 120-140 and DoY = 270-290, which is likely associated with the so-called quasi-6-day wave (Q6DW) (e.g., Gan et al., 2018;Liu et al., 2004). Gu et al. (2018) reported the Q6DW with the zonal wind amplitude of up to ∼25 m/s over the equator based on satellite data at altitudes of 90-100 km for the year 2009. Our TIE-GCM results, which are constrained with the TIME-GCM/MERRA output for the same year, are in agreement with the observations of Gu et al. (2018). Q6DW signatures are also found at other latitudes but with smaller amplitudes (not shown here). The meridional structure of the Q6DW is known to be similar to that of the (1,1) Rossby normal mode, which has the maximum zonal wind perturbation over the equator (e.g., Gu et al., 2018;Wu et al., 1994). It is noted that the Q6DW in 2009 is not particularly strong, and a similar level of Q6DW activity has been observed in the mesosphere and lower thermosphere (MLT) region in other years including both solar minimum and solar maximum (e.g., Qin et al., 2019;Yamazaki, 2018).
A comparison of Figure 4g with Figure 4b suggests a link between the Q6DW and ∼6-day oscillations in the W1 component of the PRE intensity. Apart from the Q6DW, there is some S0 activity in the equatorial zonal wind at periods of 8-9 days around DoY = 260-270 and 14-15 days around DoY = 340-350 ( Figure 4h), but no similar S0 oscillations are seen in the PRE intensity ( Figure 4c). W2 and E1 wave activities are relatively weak for the equatorial zonal wind (Figures 4f and 4i), unlike those of the PRE intensity, for which the amplitude frequently exceeds 2 m/s (Figures 4a and 4d). Figure 5 shows W1 spectra of the PRE intensity similar to Figure 4b but for different simulation cases (see Table 1). The tides_only simulation (Figure 5a) reproduces most W1 activities in the reference simulation (Figure 4b), except for ∼6-day oscillations around DoY = 120-140 and DoY = 270-290, which are stronger in the PW_contribution case (Figure 5b). The W1 amplitude is small (<1 m/s) in the S0_contribution case.

Discussion
Our simulation results reveal westward-propagating ∼6-day oscillations with zonal wavenumber 1 (hereafter referred to as " ∼6-day oscillations" for brevity) in the PRE intensity during times of enhanced Q6DW activity. The Q6DW is one of the planetary waves that are commonly observed in the MLT region around equinoctial months (e.g., Forbes & Zhang, 2017;Talaat et al., 2001;Wu et al., 1994). The wave is often regarded as the manifestation of the (1,1) Rossby normal mode of classical wave theory. It is known that the Q6DW can lead to ∼6-day oscillations in the equatorial electrojet  and low-latitude F-region plasma density (Gu et al., 2018;Lin et al., 2020) by modulating the daytime dynamo electric field (Gan et al., 2016). The present study provides evidence that the Q6DW is also important for the evening ionospheric electrodynamics. Below we discuss the connection between Q6DW activity in the MLT region and ∼6-day oscillations in PRE.
We address the issue of possible aliasing into planetary-wave spectra of PRE V Z (Figures 4a-4e). Given the small range of local time within which PRE occurs (Figure 2e), it can be safely assumed that the local time of PRE (t LT in units of days) is a fixed value and does not vary with DoY or longitude. The local time can be expressed using the universal time (t UT in days) and longitude (λ in radians). That is, where Ω is the rotation rate of the Earth (=2π per day where A and ϕ are the amplitude and phase of the wave, respectively. The wave propagates westward if s > 0 and eastward if s < 0. For the Q6DW, s = s 6 (= + 1) and ω = ω 6 (= ∼1/6). Thus, eliminating λ in (3) with Equation (2), the Q6DW is Since t LT is a fixed value, (4) is a function of t UT only and is a simple cosine wave with the frequency of s 6 − ω 6 cycles per day. Therefore, in the fixed local time frame, any wave with the zonal wavenumber s and frequency ω that satisfies the following relationship is a potential source of aliasing into ∼6-day oscillations of PRE: One can find a series of pairs of s and ω that meet (6), which are listed in Table 2. Under the constrain that s is an integer and ω is a positive value, there is no eastward-propagating wave (s < 0) that satisfies (6), which is the reason why Table 2 shows only s ≥ 0.
The waves listed in Table 2 can result from the nonlinear interaction between the Q6DW and migrating tides. It is known that a nonlinear interaction of two global-scale waves, say (s = s w1 , ω = ω w1 ) and (s = s w2 , ω = ω w2 ), leads to the secondary waves with frequencies and zonal wavenumbers that are the sums (+) and differences (−) of those of the interacting waves (Teitelbaum and Vial, 1991). These secondary waves can be expressed as where A (+) and A (−) are the amplitude of the secondary waves, while ϕ (+) and ϕ (−) are the phase. Using Equation (2), these waves are  Table 1  Note. They can be secondary waves resulting from the interaction of the Q6DW and a migrating tide, which are listed as "possible source" (fourth Column). The " + " ("−") sign in the fourth column indicates that the wavenumber and frequency of the secondary wave is the sum (difference) of the wavenumbers and frequencies of the interacting waves. Only the waves with |s| < 4 are Listed here. Q6DW = quasi-6-day wave. DW1 = Migrating diurnal tide. SW2 = Migrating semidiurnal tide. TW3 = Migrating terdiurnal tide. QW4 = Migrating quatradiurnal tide. By definition, s = ω for migrating tides. Thus, in the fixed local time frame, the secondary waves due to the nonlinear interaction of the Q6DW and a migrating tide are: It is obvious that these waves cannot be distinguished from the Q6DW itself, which is expressed as (4) in the fixed local time frame. As a matter of fact, for each wave listed in Table 2, it is possible to find a pair of the Q6DW and a migrating tide that could be the source, as indicated in Table 2. In our TIE-GCM simulations, these secondary waves are included in the TIME-GCM/MERRA lower boundary in all three cases (Table 1). Additionally, these waves can be locally generated above the lower boundary (97 km) through the Q6DW interaction with migrating tides in the reference simulation but not in the no_PW and tides_only simulations.
The Q6DW at ionospheric heights is not only by those which propagate from the MLT region but also by those locally generated in the thermosphere through the second-stage nonlinear interaction (Forbes et al., 2018a(Forbes et al., , 2018b. For example, the nonlinear interaction between the Q6DW (s = + 1, ω = ω 6 ) and eastward-propagating diurnal tide with zonal wavenumber 3 (s = −3, ω = 1), DE3, would produce the secondary waves (s = 1 ∓ 3, ω = ω 6 ±1). If these secondary waves generated below the MLT region enter the thermosphere and go through the nonlinear interaction with DE3 therein, the Q6DW would be locally generated. In this way, the Q6DW could be produced within the thermosphere without the need for the Q6DW spectrum.
To summarize the argument above, ∼6-day oscillations of PRE ( Figure 4b) can arise from the Q6DW or the secondary waves generated by the nonlinear interaction between the Q6DW and migrating tides. In both cases, the source wave could either propagate from below or be generated within the thermosphere. Table 3 shows which source waves are included in our simulation results. In the PW_contribution case, which showed the largest contribution to the Q6DW signature of PRE in Figure 5, ∼6-day oscillations can result from the propagation of the Q6DW from the model lower boundary (97 km) or from the secondary waves generated above the lower boundary. In the tides_only simulation, ∼6-day oscillations of PRE can be due to the Q6DW generated in the thermosphere or due to the propagation of the secondary waves from the lower boundary. In the S0_contribution case, no source wave is expected for ∼6-day oscillations of PRE from the mechanisms discussed here. Indeed, ∼6-day oscillations of PRE are small in the S0_contribution case ( Figure 5c). Figure 6 shows the height versus latitude distributions of the amplitudes of the Q6DW and secondary waves for DoY = 267-284, when ∼6-day oscillations of PRE are most prominent (Figure 4b). The first row of Figure 6 compares the zonal wind components of the Q6DW obtained from the reference, PW_contribution, and tides_only cases; the second row compares the meridional wind components of the westward-propagating ∼21h wave with zonal wavenumber 2 (21hW2), which is a secondary wave due to the nonlinear interaction of the Q6DW and migrating diurnal tide (DW1); and the third row compares the meridional wind components of the westward-propagating ∼11h wave with zonal wavenumber 3 (11hW3), which is a secondary wave due to the nonlinear interaction of the Q6DW and migrating semidiurnal tide (SW2). The zonal wind components of 21hW2 and 11hW3 are in the same order of magnitude but tend to be smaller than the meridional wind components.
The comparison of Figures 6a, 6d, and 6g suggests that below 120 km or so, the Q6DW is dominated by the wave that propagate from the lower boundary of the model (as represented in the PW_contribution case), and above these heights, the Q6DW locally generated in the thermosphere (as represented in the tides_only YAMAZAKI AND DIÉVAL Note. "Secondary waves" here mean those generated by the nonlinear interaction between the quasi-6-day wave and migrating Tides (see also Table 2). The waves in "propagation" are those generated Below the model Lower boundary (97 km), while the waves in "in situ" are those generated locally within the thermosphere above the lower boundary. simulation) is equally important. For the secondary waves, the waves generated above the model lower boundary (Figures 6e and 6f) do not have as large amplitudes as those which propagate from the lower boundary (Figures 6h and 6i). Similar results are obtained for other secondary waves resulting from the nonlinear interaction between the Q6DW and migrating tides (not shown here).
The primary source of the ∼6-day oscillations in PRE during DoY = 267-284 is likely to be the Q6DW that propagate from 97 km based on the fact that (1) the PW_contribution case shows larger ∼6-day oscillations in PRE than the tides_only simulation ( Figure 5) and that (2) secondary waves due to Q6DW-tidal interactions are small in the PW_contribution case (Figures 6e and 6f). This is unexpected as previous studies (e.g., Abdu et al., 2006a) often suspected that planetary-wave oscillations in PRE are due to tidal modulation by planetary waves rather than due to planetary waves themselves. Nonetheless, the fact that the removal of planetary-wave forcing at the lower boundary does not completely eliminate ∼6-day oscillations of PRE (Figures 3c, 3f and 5a) suggests that the Q6DW locally generated in the thermosphere and the upward-propagation of the secondary waves from 97 km also make contributions to Q6DW signatures of PRE.
YAMAZAKI AND DIÉVAL 10.1029/2020SW002685 12 of 16 Figure 6. Latitude versus height plots for (a) the zonal wind amplitude of the westward-propagating quasi-6-day wave (Q6DW) with zonal wavenumber 1, (b) the meridional wind amplitude of the westward-propagating ∼21 h wave with zonal wavenumber 2 (21hW2), and (c) the meridional wind amplitude of the westward-propagating ∼11h wave with zonal wavenumber 3 (11hW3). Note that 21hW2 is a secondary wave due to the nonlinear interaction between the Q6DW and migrating diurnal tide (DW1), while 11hW3 is a secondary wave due to the nonlinear interaction between the Q6DW and migrating semidiurnal tide (SW2). See Table 2 for secondary waves due to the nonlinear interaction between the Q6DW and migrating tides. Panels (d-f) are the same as (a-c) but for the PW_contribution case, while panels (g-i) are for the tides_only case. See Table 1 for the description of different simulation cases.
In summary, the ∼6-day variations in PRE intensity around DoY = 270-280, whose amplitudes are 6.0 ± 2.4 m/s from the ground-observer perspective, are dominated by the W1 component, which accounts for up to 4.3 m/s. The rest is mainly due to the W2 and E1 components. Within the W1 component, the Q6DW that propagate from the lower thermosphere is dominant, accounting for up to 3.0 m/s, and the rest is due to the Q6DW locally generated in the thermosphere and the secondary waves associated with the Q6DW-tidal interaction that propagate from below.
Although this study predicts Q6DW effects on the intensity of PRE, the mechanism by which the wave alters PRE V Z remains to be investigated in future studies. The main driver of PRE is the zonal wind in the F region, but the F-region meridional wind and E-region zonal wind are also known to make nonnegligible contributions to PRE (Du & Stening, 1999;Maute et al., 2012;Millward et al., 2001). Richmond et al. (2015) demonstrated that daytime winds, which are the main driver of the daytime dynamo electric field, have little impact on PRE. Thus, the driving mechanism for the Q6DW effect on PRE can be different from that for the Q6DW effect on the daytime equatorial electrojet intensity and low-latitude F-region plasma density, which has been previously studied (e.g., . More numerical work is needed.

Conclusions
We have examined temporal and longitudinal variations of the PRE in the equatorial F-region vertical plasma drift using idealized simulations by the TIE-GCM. Simulations were run under geomagnetically quiet (Kp = 1) and high solar-flux (F 10.7 = 200) conditions, with realistic atmospheric forcing by large-scale waves based on the TIME-GCM/MERRA output for the year 2009. The main focus of the study was on planetary-wave influences on PRE, which have been previously inferred from local observations (e.g., Abdu et al., 2006a) but have not been evaluated on a global basis. Results and conclusions are as follows: 1. In response to forcing by large-scale waves, PRE shows day-to-day variability in both its intensity (5.6 ± 1.0 m/s in standard deviation) and local time (7.9 ± 3.6 min in standard deviation). 2. The seasonal and longitudinal dependence of the PRE intensity is consistent with that of the angle between the sunset terminator and magnetic meridian, as reported in previous studies. The seasonal and longitudinal dependence of the PRE local time is controlled by the E-region sunset time, in agreement with observations. 3. Day-to-day variations of the PRE intensity at a fixed longitude is predominantly due to tidal-wave forcing (periods less than 2 days). Planetary-wave forcing (periods longer than 2 days) plays a role for periodic oscillations in the PRE intensity, especially those at a period of ∼6 days. 4. Zonal-wavenumber analysis of planetary-wave oscillations (2-20 days) in the PRE intensity reveals westward-propagating ∼6-day oscillations with zonal wavenumber 1 during times of enhanced quasi-6-day wave (Q6DW) activity. 5. The ∼6-day oscillations in the PRE intensity are mainly due to the Q6DW that are generated below the lower boundary of the model (97 km) and propagate into ionospheric heights. 6. The ∼6-day oscillations in the PRE intensity are also due in part to the Q6DW locally generated in the thermosphere above 97 km and the secondary waves generated through the nonlinear interaction between the Q6DW and migrating tides. 7. In the example of the Q6DW event around DoY = 270-280, the amplitude of ∼6-day variation in the PRE intensity is 6.0 ± 2.4 m/s at a fixed longitude. The westward-propagating zonal wavenumber 1 component explains up to 4.3 m/s, within which the contribution of the Q6DW that propagate from the lower thermosphere accounts for up to 3.0 m/s.
Since the PRE intensity has a strong impact on the occurrence of equatorial spread F and plasma bubbles, planetary-wave oscillations of PRE should be taken into account in the prediction of these space weather phenomena. The zonal wavenumber 1 nature of the ∼6-day oscillations could contribute to longitudinal differences in the appearance of equatorial spread F and plasma bubbles.