Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
 Simulations with the global ionosphere plasmasphere model driven by whole atmosphere model winds show significant longitudinal and day-to-day variations in the ionospheric parameters. Under fixed solar and geomagnetic activity levels, the contributions of lower atmosphere tides to the longitudinal and day-to-day variability in the upper atmosphere are estimated. Larger relative variability is found in the nighttime than in the daytime, which is consistent with observations. The perturbations from the lower atmosphere contribute about half of the observed variability in the ionospheric F2 peak plasma density under moderate solar activity and geomagnetic quiet conditions. The daily variability of the equatorial vertical plasma drifts is primarily driven by the day-to-day amplitude changes of the migrating semidiurnal tide, while the wave-4 and wave-3 longitudinal variations during September are dominated by the nonmigrating diurnal eastward propagating tides with zonal wave numbers 3 and 2, respectively.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Sagawa et al.  discovered the now well-known “wave-4” longitudinal structure in satellite observations of an airglow emanating from the postsunset equatorial ionization anomaly in March–June 2002. Immel et al.  associated this structure with the modulation of the E region dynamo electric field and corresponding vertical plasma drift by nonmigrating tides, in particular, the diurnal eastward wave number 3 (DE3) tide excited by the diurnal variation of convection and latent heat release over the wave 4 pattern of tropical topography [e.g., Tokioka and Yagai, 1987]. High-inclination satellites provide a global view necessary to reveal the longitudinal structure but typically take up to a few months to accumulate sufficient data as a function of local time. Analyzing satellite observations from an equatorial orbit, Araujo-Pradere et al. [2011; 2012] first confirmed the existence of the wave 4 structure and associated sharp longitudinal gradients in the vertical E × B plasma drift on a day-to-day basis.
 Theoretical studies with coupled upper atmosphere/ionosphere models driven by DE3 climatologically imposed at the lower boundary from a linear tidal model or satellite observations have qualitatively confirmed its connection to the wave 4 variation [Hagan et al., 2007; Pedatella et al., 2011]. Recently, Pedatella et al.  performed similar simulations to investigate the role of other possible drivers such as the semidiurnal eastward wave number 2 (SE2) tide and stationary planetary wave 4 (SPW4). With an interactively coupled whole-atmosphere/ionosphere model Jin et al.  first demonstrated complex day-to-day variability of the ionospheric longitudinal structure, from which the wave 4 modulation emerges as a long-term average feature. The day-to-day variability of the peak F2 layer electron density (NmF2) has been studied by Rishbeth and Mendillo  using long-term ionosonde data. In years of medium solar activity (F10.7 ~ 140), the daily fluctuations of NmF2 have a standard deviation of 20% by day and 33% by night. They concluded that the variability is largely contributed by the geomagnetic activity and partially by meteorological sources from the lower atmosphere.
 The goal of this study is to evaluate the relative contribution of different tidal and planetary waves to the longitudinal structure of the equatorial plasma drift and its day-to-day variability [e.g., Fuller-Rowell et al., 2008]. As in previous modeling studies [e.g., Hagan et al., 2007; Jin et al., 2008; Pedatella et al., 2012], a “one-way” coupling scheme is adopted to drive the Global Ionosphere Plasmasphere (GIP) [Millward et al., 2007] model with winds from the Whole Atmosphere Model (WAM) [Akmaev, 2011].
2 Models and Numerical Experiments
 WAM is a general circulation model of the neutral atmosphere extending from the surface to the exobase at the nominal height of about 600 km [Akmaev, 2011]. Built from the operational weather prediction Global Forecast System model, it has a realistic representation of lower atmosphere sources of tidal waves important for the E layer dynamo, including the migrating diurnal tide with zonal wave number 1 (DW1) and DE3 [Akmaev et al., 2008; Akmaev, 2011].
 GIP is a further development of the ionosphere-plasmasphere component of a coupled thermosphere ionosphere plasmasphere model [Millward et al., 1996]. It utilizes a magnetic apex coordinate system [Richmond, 1995] in which a global three-dimensional grid of magnetic field lines is created by tracing through the full International Geomagnetic Reference Field. The horizontal resolution is about 1° × 4.5° in latitude-longitude. It has been previously coupled with the Thermosphere-Ionosphere-Electrodynamics General Circulation Model and empirical models to study the longitudinal variability in the F region and topside ionosphere [e.g., Fang et al., 2009; Pedatella et al., 2011].
 In this study, the dynamo electric field is calculated self-consistently by the electrodynamic solver of Richmond  using the field-line integrated conductivities from GIP, neutral composition and temperature from the NRLMSISE-00 empirical atmosphere model [Picone et al., 2002], and neutral winds from WAM. The electric fields and neutral winds are then used in GIP in the zonal and meridional plasma transport calculations. The hourly neutral winds are from a free climatological annual run of WAM similar to that analyzed before [Akmaev et al., 2008; Akmaev, 2011]. The winds are analyzed to extract the tides and planetary waves considered important for longitudinal modulation of the ionosphere and to study their daily variability. Unlike in some previous studies [e.g., Pedatella et al., 2011, 2012], no tidal filtering or amplitude and phase adjustment of the neutral winds input into GIP is performed.
 Satellite observations and model simulations indicate that DE3 maximizes in the lower thermosphere in August–October and the results are presented here for the month of September, as in many previous studies. Both WAM and GIP have been run under constant and quiet geomagnetic and moderate solar activity conditions (F10.7 = 120), so that any ionospheric variability, including changes in the longitudinal structure from 1 day to the next, is entirely forced from below. It is in fact the daily variation of the longitudinal structure of the vertical plasma drift that allows us to confidently identify its primary drivers.
3 Results and Discussion
 Figures 1a–1c show the daily values of integrated electron content (IEC) from 90 km to 1000 km in TECu (TEC units; 1016 electron/m2), NmF2, and the vertical E × B drifts at the magnetic equator and longitude 0°E in September, respectively. For IEC and NmF2, low-latitude values (22°N, solid lines) and middle-latitude values (40°N, dashed lines) are shown. The magnitudes of NmF2 and IEC at both latitudes are reasonably consistent with previous observational studies [e. g., Rishbeth and Mendillo, 2001; Jee et al., 2004; Lei et al., 2005]. The IEC and NmF2 values decrease with latitude. The vertical drifts are typically upward in the daytime and downward in the nighttime. The pre-reversal enhancement (PRE) also appears on some of the simulation days. The values of vertical drifts under moderate solar activity are comparable with climatology [e.g., Scherliess and Fejer, 1999]. Interestingly, the magnitudes of vertical drifts in the study of Pedatella et al. , also using GIP as the ionospheric model, are about 5 to 10 m/s smaller than in our results.
 Our simulation also captures a strong day-to-day variability in these ionospheric parameters. Note again that except for daily variations of WAM winds, there is no other source of day-to-day changes in the simulation. It is well established that migrating tides, in particular, the semidiurnal westward wave number 2 (SW2) tide, play an important role in driving the daytime equatorial electrodynamics [e.g., Millward et al., 2001; Fang et al., 2012]. Figure 1d compares the daytime maximum vertical drift from Figure 1c with the SW2 tidal amplitude of zonal wind in the dynamo region near 115 km at the equator and the maximum SW2 amplitude at this height occurring near 30°S (half of the maximum amplitude is shown in the figure). Because other tides and waves also contribute to the electrodynamics, a perfect correlation of the drift and SW2 amplitudes is hardly expected. However, during most of the month, the maximum plasma drifts clearly correlate well with the amplitude of SW2 at the equator. During periods when the maximum SW2 amplitude exceeds twice that at the equator, such as at the beginning of the month, the day-to-day changes in maximum drifts seem to be largely influenced by SW2 tidal winds at other latitudes and appear to correlate well with the maximum amplitude of the tide.
Rishbeth and Mendillo  analyzed ionosonde data from 13 stations (five high-latitude, five midlatitude, and three low-latitude stations) in all seasons under moderate solar activity during1957–1990 to study the day-to-day variability of the NmF2. The daily fluctuations of NmF2 have a standard deviation of 20% by day (10–16 LT) and 33% by night (21–03 LT). They concluded that large part of F2 layer variability is linked to the geomagnetic activity and attributed the rest to meteorological sources from the lower atmosphere. To estimate the day-to-day variability of IEC and NmF2 from our simulations, the percentage standard deviations of these two parameters are calculated at each LT (Figure 2). The percentage standard deviation at each LT is calculated by dividing the 30 day standard deviation by the 30 day mean value and then multiplied by 100. Clearly, the relative variability of both IEC and NmF2 is greater in the nighttime than in the daytime. Also, the variability is greater at low latitudes than at midlatitudes. The low-latitude NmF2 between 3 LT to 4 LT shows the largest variability of 31%. All other parameters also have largest variability at around 5 LT. Table 1 summarizes the percentage standard deviations of IEC and NmF2 averaged over daytime (9–18 LT) and nighttime (20–06 LT) at midlatitude and low-latitude locations. The average values again show a larger variability of IEC and NmF2 in nighttime and at low latitudes. The results of larger variability of NmF2 at nighttime are consistent with the analysis done by Rishbeth and Mendillo  but with smaller amplitude. In our simulation the geomagnetic and solar activity are assumed constant. With only the perturbations of the neutral wind fields from the lower atmosphere, the underestimation of variability in NmF2 is anticipated. Also, the study by Rishbeth and Mendillo  combining data from 13 different locations could possibly capture more variability. To what extent the ionospheric variability can be attributed to tides and other waves from lower atmosphere, solar activity, or geomagnetic parameters will be studied using the coupled WAM-GIP with realistic geomagnetic and solar parameters as inputs in the future.
Table 1. Percentage Standard Deviations of IEC and NmF2 Averaged Over Daytime (9–18 LT) and Nighttime (20–06 LT) at Midlatitude and Low-Latitude Locations.
Standard Deviations (%)
IEC Low Latitude
NmF2 Low Latitude
 Figure 3 illustrates the longitudinal variation of the equatorial vertical plasma drift as a function of local time and its daily changes during two selected 5 day periods out of the 30 day September simulation. A typical daytime upward drift and downward nighttime drift prevail at all locations, while at some longitudes, the upward drifts can also be seen during the night and in the predawn sector. The daytime drift and PRE both show a strong longitudinal variation changing from 1 day to the next. For example, on day 14 only three pronounced longitudinal peaks can be seen around noontime. The longitudinal variation becomes more complex on day 15, turning into a clear wave 4 on day 16, and becoming distorted again by day 18. This daily evolution is clearly associated with an enhancement (in excess of 35 m s−1) and subsequent decrease of the DE3 zonal wind amplitude at the geographic equator in the E layer during the same period (Figure 4). Note that the DE2 zonal wind also increases to comparable magnitudes on day 16 but at southern latitudes off the equator and then at the equator by day 18. This variation of equatorial tidal zonal winds clearly contributes to the variation of the longitudinal structure of plasma drifts from wave 3 to wave 4 and back to wave 3.
 During the second period, the longitudinal structure again changes from predominantly wave 3 on day 22 to wave 4 on day 24 and back to wave 3 on days 25 and 26. This again is associated with the enhancement of the DE3 zonal wind amplitude on day 24 with a subsequent decrease, preceded and followed by enhanced amplitudes of DE2. In longer-term averages, such as observed from high-inclination satellites, DE2 and DE3 appear to complement each other in the lower thermosphere during different seasons with the former primarily maximizing in November–January and the latter in August–October [e.g., Forbes et al., 2008]. This is consistent with long-term satellite observations of integrated plasma density exhibiting primarily a wave 4 modulation during equinoxes and a wave 3 structure around the December solstice [e.g., Scherliess et al., 2008]. Our simulations suggest that both types of longitudinal variation or their superposition may be likely observed on any given day.
 On most days, the longitudinal structure of the daytime drift shows minimum values near longitude 300°E. This indicates that other factors such as the longitudinal structure of the geomagnetic field (i.e., the declination angle and magnitude of the field) also need to be considered in order to fully understand the longitudinal variations in the ionosphere [e.g., Hagan et al., 2007; Fang et al., 2012].
 As in previous studies [e.g., Pedatella et al., 2012], Figure 4 also presents the zonal wind amplitude of other tidal and planetary wave components which may potentially generate a wave 4 (SE2 and SPW4) or wave 3 (SE1) variation at a fixed local time. None of these waves appear to contribute to the variation of the plasma drifts. Even though the semidiurnal (SE2 and SE1) tidal winds reach substantial amplitudes of up to 20–25 m s−1 on some days, they peak at middle and high latitudes and cannot efficiently contribute to the modulation of equatorial electric fields [e.g., Jin et al., 2008], in agreement with the conclusion of Pedatella et al. .
 The amplitude of the SPW4 zonal wind in the E region remains below about 15 m s−1 in agreement with satellite observations [Oberheide et al., 2011] and appears insufficient to make a substantial contribution to the longitudinal variation of the plasma drift. These conclusions are at variance with the claim of Pedatella et al.  that SPW4 is an important contributor to the wave 4 structure in the equatorial ionosphere. This wave is primarily generated in the lower thermosphere by nonlinear interactions of DW1 and DE3. It should be noted that in the simulations of Pedatella et al. [2012, Figure 2], the peak monthly mean zonal wind amplitudes of 35 m s−1 and over 40 m s−1 for DW1 and DE3, respectively, are both overestimated by at least a factor of 1.5–2 compared with satellite observations for comparable conditions [e.g., McLandress et al., 1996; Oberheide et al., 2011]. As a result, the peak amplitude of the SPW4 zonal wind is overestimated by about a factor 3 [e.g., Oberheide et al., 2011].
 Associated with the wave 4 structure, Araujo-Pradere et al. [2011; 2012] observed sharp longitudinal gradients of vertical drifts at the boundaries of the four peaks. The largest gradient is 3 m s−1 deg−1 in the eastern Pacific sector (240°E to 260°E) and the Peruvian sector (285°E to 300°E) during equinox. On day 16 when the wave 4 is pronounced in our simulation, the gradients are 0.26 m s−1 deg−1 from the trough to the peak (252°E) and −0.19 m s−1 deg−1 from the peak to the trough at 10 LT. On day 24, the gradients are 0.12 m s−1 deg−1 and −0.44 m s−1 deg−1 near the peak at 257°E. The gradients from simulations are averaged over larger distances and the values are smaller than the observations. It should be noted that the study in Araujo-Pradere et al. [2011; 2012] did not distinguish the changes of drifts due to changes in local times and latitudes. Also the sharp gradients were only shown at a few longitudes and the global distributions of density/drift at the corresponded time were not provided. Therefore, the observed longitudinal gradients can also be associated with other causes. The spatial scales of these sharp gradients are relatively small compared to the planetary-scale tides such as DE3 and DE2.
 Coupled WAM-GIP simulations exhibit significant day-to-day variability of ionospheric parameters entirely driven by dynamical process from the lower atmosphere. The relative changes are larger during nighttime than in daytime and account for about 50% of observed ionospheric variability during moderate solar and quiet geomagnetic conditions.
 In general agreement with observations and previous modeling studies, the equatorial vertical plasma drifts show pronounced longitudinal variations changing on a daily basis from a wave 4 to wave 3 or even more complicated patterns. Comparing these changes to the day-to-day variations of tidal zonal winds in the dynamo region at 115 km, we conclude that during September equinox, the wave 4 longitudinal structure is dominated by DE3 while the wave 3 structure is primarily driven by DE2.
 Global-scale tidal waves cannot fully account for the sharp longitudinal gradients observed in the equatorial region. These are likely caused by smaller-scale waves. Both WAM and GIP may have to be run at a higher horizontal resolution to reproduce these observations.
 Future studies will also include evaluation of the impact of other sources of ionospheric variability, such as geomagnetic and solar activity, with the coupled model. With a recently developed whole atmosphere data assimilation and forecast system [Wang et al., 2011; Fuller-Rowell et al., 2011], the ionosphere/plasmasphere model can be driven by a more realistic neutral atmosphere to reproduce and study particular events, capture, and eventually forecast the day-to-day variations in the upper atmosphere and ionosphere.
 Funding for this research was provided by NASA Heliosphysics Theory Program grant NNX11A061G.
 The Editor thanks Joseph Huba and an anonymous reviewer for their assistance in evaluating this paper.