Longitudinal variations of noontime thermospheric winds in response to IMF Bz temporal oscillations: broken mean circulation and standing feature

By using the coupled magnetosphere-thermosphere-ionosphere model, we explore the longitudinal/UT differences of the dayside neutral wind in response to the 60 min periodic oscillation of the interplanetary magnetic eld (IMF) Bz. The southward propagation of the traveling atmospheric disturbances (TADs) in meridional wind stands at about 20º MLat, which is related to the geomagnetic eld conguration, neutral temperature, and electron density changes. The meridional wind travels continuously from high to low latitude in the western southern hemisphere, while they are broken into pieces in the eastern southern hemisphere. The broken mean circulation that is a combined effect of TADs and simultaneous responses of the meridional winds driven by oscillating solar wind conditions is induced by the stronger roles of the ion drag than the pressure gradient. Note here that the mean circulation is the background meridional winds in the base case. The ion drag shows obvious longitudinal differences associated with the penetration of the ionospheric electric eld during the oscillation of IMF Bz.


Introduction
The energy carried by the solar wind can be fully deposited into the Earth's upper thermosphere at high latitudes, and produce longitudinally extended waves (Bruinsma et al., 2007Liu et al., 2018b). These waves then propagate away from the source region to other regions of the coupled Ionosphere-Thermosphere (IT) system. These waves are named as traveling atmospheric disturbances (TADs), which are a key factor in understanding the variability of the IT system. Exploring the possible mechanisms of TADs is critical in the modeling and forecasting of near-Earth upper atmosphere environment.
In the past decades, TADs during disturbed periods have attracted much attention (e.g., Bruinsma & Forbes, 2007;Liu and Lühr, 2005;Liu et al., 2010Liu et al., , 2014Liu et al., , 2018bOliveira et al., 2017;Otsuka et al., 2004;Shiokawa et al., 2003;Sutton et al., 2009;Thome et al., 1964;Zhang et al., 2019). Previous studies have reported that the storm time energetic particle precipitation and Joule heating (Rodríguez-Zuluaga et al., 2016), the gravity waves (Liu et al., 2014), and the sunset terminator (Beer, 1973) are critical in the generation of TADs. The storm time energy deposition can signi cantly increase the neutral temperature at high latitudes, changing the global circulation and triggering the TADs via the pressure gradient force.
The upward propagating gravity waves, which are ubiquitous in the mesosphere and lower thermosphere, and atmospheric waves that are associated with the sunset terminator, can deposite energy and momentum at higher altitudes. This energy deposition also plays a role in the neutral temperature changes in the thermosphere, and cause TADs.
As reported by Dungey (1961), the geomagnetic storms can be induced by the interaction between the IMF Bz in southward and the geomagnetic eld. During disturbed periods, a large amount of energy and momentum carried by the solar wind is deposited into the Earth's upper atmosphere, changing the global circulation and producing TADs. TADs have been studied using meridional winds, neutral temperature, composition (O/N2), and air mass density data (e.g., Bruinsma and Forbes, 2007;Bruinsma et al., 2009;Fujiwara et al., 2006;Lei et al., 2008). Typically, TADs propagate with velocities of 400-1000 (100-300) m/s, periods of 0.5-3 (0.25-1) hour, and wavelength of thousands (hundreds) of kilometers in large-scales (medium-scales) (e.g., Bruinsma and Forbes, 2009;Hocke and Schlegel, 1996). Liu and Lühr (2005) found that during storm times the enhancement of the thermospheric air mass density in the noontime showed two different features (see the upper panel of Figure 3 in Liu and Lühr, 2005). One was the enhancement in the density at 0º~40º geomagnetic latitudes (MLat), which originated from the TAD propagating from the auroral region to low latitudes at around 20 UT on 30 October 2003. The other was the almost instant enhancement in the density at aurora and middle latitudes at about 20:00 UT on 30 October 2003. These two features overlapped at 20:00 UT and 40º MLat, forming broken mean circulation. This broken feature and its possible mechanisms have rarely been studied in the literature, which is one of the focus of the study.
Another interesting topic is raised by Sharma et al. (2011). They reported a longitudinal variation of the storm time O/N2 enhancement: it was stronger at Yibal (22.18º Geographic latitude (GLat), 56.11º Geographic longitude (GLon)) than at Kunming (25.03º GLat, 102.79º GLon) and Udaipur (24.67º GLat, 74.69º GLon). This longitudinal variation was attributed to the air upwelling due to the equatorward meridional wind (Sharma et al., 2011). The question is then whether TADs in the meridional wind also show notable longitudinal variations? Based on observations from all-sky airglow imagers, Otsuka et al. (2004) investigated the geomagnetic conjugacy of the medium-scale traveling ionospheric disturbances (TID). The conjugacy structures came from the polarization electric eld that maps along geomagnetic eld lines and moves the F region plasma upward/downward, producing the mirrored structures of plasma perturbations in the Northern and Southern Hemispheres. TID can be treated as the ionospheric counterpart of TAD with phase and time delay compared to TAD. Thus, the hemispheric conjugacy of the TAD is worthy of investigation.
While the TADs in thermospheric winds during storm time have been established in the literature, the exact longitudinal and hemispheric changes of the meridional wind responses to temporal oscillations of IMF Bz is still poorly understood, due to the fact that the thermospheric winds are balanced among many drivers (i.e., pressure gradient, ion drag, Coriolis). The thermospheric winds are important factors in the understanding of changes in the ionosphere-thermosphere coupling system. In addition, the periodic oscillations of IMF Bz are a common phenomenon seen in the solar wind, which can produce periodic oscillations in the energy and momentum deposition in the upper atmosphere and TADs (e.g., Liu et al., 2018a;Zhang et al., 2019). Taking the abovementioned into consideration, the present work is to study the longitudinal and hemispheric patterns of the meridional wind changes at a xed local time, which show some interesting signals of the broken mean circulation, standing feature, and the related longitudinal and hemispheric variations during the period of the oscillating IMF Bz. Furthermore, these thermospheric wind changes have not caught so much attention in the literature. The possible drivers for the noontime wind changes are revealed, including the possible effects of the geomagnetic topology.
Electron density, plasma motion, and neutral temperature changes are also examined.

CHAMP Data
The orbit of the CHAllenging Minisatellite Payload (CHAMP) satellite was near-polar (87.3º inclination) (Reigber et al., 2002). The orbital period of CHAMP is about 93 min. The initial orbital altitude was 460 km in 2000, then decreased to about 400 km in 2005 and about 300 km in 2008 due to atmospheric drag. The neutral air density data is deduced from the measurements by the accelerometer on board of the CHAMP (Sutton et al., 2007). In this work, the observed mass density data at geographic latitudes of -60º~60º are processed according to the orbit-by-orbit method (Häusler et al., 2007).

CMIT Model
The Coupled Magnetosphere Ionosphere Thermosphere (CMIT) model consists of two parts. One is the Thermosphere Ionosphere Electrodynamic General Circulation Model (TIEGCM) v1.95, the other is the Lyon-Fedder-Mobarry (LFM) global magnetohydrodynamic magnetospheric model Wiltberger et al., 2004). When investigating the coupling in the MIT system, the model outputs have achieved good agreements with observations and other model outputs, con rming the reliability and stability of CMIT (e.g., Cnossen & Richmond, 2013;Liu et al., 2018a;Wang et al., 2008). The TIEGCM is a three-dimensional, time-dependent model of the coupled IT system. The drivers of TIEGCM are shown as follow: the particle precipitation and high-latitude electric elds (Heelis et al., 1982), solar extreme ultraviolet and ultraviolet spectral uxes (Richards et al., 1994), the upward-propagating atmospheric tides (Hagan & Forbes., 2002. Note here that the atmospheric tides, including migrating and nonmigrating tides, are either speci ed by the global scale wave model or derived from the Sounding of the Atmosphere using Broadband Emission Radiometry (SABER) and TIDI observations. The TIEGCM has a horizontal resolution of 2.5º GLat × 2.5º GLon, and a vertical resolution of quarter-scale height with 57 levels. The model extends from ~97 km to ~600 km in the vertical direction, with the upper boundary depending on the solar activity. As revealed by Lyon et al. (2004), the drivers of LFM Magnetosphere model are the solar wind and IMF data, and the boundary of LFM extends from -300 RE to 2RE. The RE is the Earth's radius. In the coupled MIT system, the modeled ionospheric conductivity from TIEGCM is imposed into the LFM, and the simulated auroral particle precipitation and high-latitude electric elds from the LFM are used to drive the TIEGCM.
In this study, the changes in the IT system due to the effects of IMF Bz oscillation are explored by performing two CMIT simulations. One is the 60-min case that Bz varies with a maximum amplitude of 10 nT (from -10 nT to 10 nT) and a sine function of a period of 60 min, the other is the CMIT base run with a constant Bz value of 0 nT. The hemispheric power, joule heating, and partial precipitation oscillate as well (Figures not shown). The auroral oval (size and locations) change in a similar way to the particle precipitation. The electric eld penetration also oscillates with a similar amplitude as observed (Figures not shown, Wei et al., 2008). As reported by Wei et al. (2008), the penetrating electric eld oscillates with the observed IMF Bz. The maximum amplitude of the penetrating electric elds at Jicamarca (11.9ºS GLAT, 76.8ºW GLON) is ~1 mv/m in their case. In our simulations, the maximum PPEF is ~3 mv/m (Figure not shown), which might be overestimated (Merkin et al., 2005;Wang et al., 2008). As shown in the previous work of Zhang et al. (2019), the disturbed wind in the oscillation period of 60 minutes in IMF Bz is strongest when compared to the other two cases with an oscillation period of 10 and 30 minutes. Thus, in the present work, we will focus on the 60-min case. In addition, the other input solar wind conditions for CMIT are solar wind density (5 cm -3 ), velocity (400 km/s), IMF Bx and By (both are 0 nT).
To show the thermospheric responses to IMF oscillation, the background winds giving in CMIT base run are removed.

Results
Broken Mean Circulation are the error of CHAMP observations, and are smooth. Due to the lack of meridional winds observations by CHAMP, the neutral densities are used to represent TADs. In Figure 1b, the disturbed neutral density shows two different features. One is the disturbed density occurring at almost the same time at -60º~-30º geographic latitudes (GLat) at 05 UT on Nov. 11, 2003 (see orbit 2). The density changes at -60º~-30º geographic latitudes (GLat) are much stronger at orbit 2 than that at other orbits. Note that the CHAMP satellite has a speed of 7.6 km/s, which is much faster than the propagation of TADs. Thus, the density changes at orbit 2 can be regarded as a response without signi cant time delay at latitudes. The other is the propagation of the density disturbance from -30º GLat to -10º GLat in ~2 hours, with a phase speed of ~300 m/s (magenta arrows in Figure 1b). The density changes at orbit 3 and -30º~-10º GLat are enhanced, which could be propagated from high latitudes as observed at orbit 2. These two features demonstrate the broken mean circulation as shown in Figures 1b and 1c. Figure 1c is the universal time changes of neutral mass density observed by CHAMP, as reported in Figure 3 of Liu and Lühr (2005). The broken mean circulation structure is indicated by the black arrows. The TADs in neutral density propagates equatorward from ~60° GLat at 17 UT to 20° GLat at 20 UT. In addition, the neutral density at -60° ~ 20° GLat seems to be enhanced at almost the same time. Moreover, during the storm period (29-31 Oct, 2003), the electron density also exhibited an evident combined feature of traveling ionospheric disturbances (TIDs) and the simultaneous responses ( Figure 1d). The electron density at 40º ~ 70º (-40õ -60º) GLat in the Northern (Southern) Hemisphere increases at almost the same time, as indicated by the vertical black arrows. Then, the enhancement travelled to low latitudes in ~4 hours (the sloped black arrows).
The broken mean circulation is the combined characters of the traveling atmospheric disturbances and simultaneous changes in meridional winds. The broken mean circulation is shown in the observed neutral density. To disclose this phenomenon, two idealized numerical experiments are performed by using the CMIT model. The reason why idealized cases are used in the following study instead of the realistic case, is that the external driving (e.g., geomagnetic activity) are signi cantly variant in the realistic case. In the idealized case, we can hold the external condition unchanged and only change IMF condition. Figure 2 shows the geographic latitude and longitude variations of the modeled meridional winds (Figures 2a and   2b) and their disturbances (Figure 2c) at 12 LT. Here positive value stands for northward direction. At a xed local time, one longitude corresponds to one universal time. Thus, the plot gives the temporal and longitudinal variations of wind changes in global scale. In Figure 2a, there do not exist signi cant temporal oscillations in the background meridional winds. In Figure 2b, the meridional winds show outstanding periodic variations as that of IMF Bz. The equatorward penetration of TADs in meridional winds seems to be much stronger at above 20° MLat than that at the equator of 0° ~ 20° MLat (the area closed by black and red lines). Thus, to clearly shown the wind disturbances, a differential analysis between modeled meridional winds in Figures 2a and 2b has been carried out, which has been shown in Figure 2c. At ~100º GLon, the notable TADs in a phase speed of ~462 m/s are generated in both hemispheres and propagate from high to low latitudes, which are shown by red and black arrows. In the Southern Hemisphere, the TADs occurring at -30º~120º GLon propagate to lower latitudes, with several sudden changes in the wave phase along the propagation direction. This change in wave phase tends to occur at higher latitudes with UT, for instance, from (-40º GLat, 90º GLon) to (-50º GLat, 30º GLon). At 120º~180º GLon, the magnitude of TADs which is the deviation from the background value is weaker (~8 m/s) than that at other longitudes, but shows similar broken structures as that at -30º~120º GLon. The broken mean circulation, a combined effect of TADs and the simultaneous response of neutral winds, is found in previous observations (Figure 3 of Liu and Lühr., 2005). Liu and Lühr (2005) only showed the neutral density variations derived from CHAMP observations during the storm period in November, 2003, but did not mention the broken men circulation. The broken mean circulation does not catch so much attention in the literature, which is a major new nding in our work. Furthermore, as our previous work disclosed (Zhang et al., 2019), the CMIT model can be used in this work to investigate the ideal cases of IMF Bz oscillations and to understand the physical mechanisms. In our study, the broken mean circulation is well reproduced in our simulations, but not at the same latitudes and events. We understand that the approach is qualitative rather than quantitative. However, this does not con ict to achieve the major conclusion and has little in uence on the reliability of our conclusions.

Standing Features in Meridional Winds
In the Northern Hemisphere (Figure 2c), the TADs in meridional winds can be found at longitudes of -60º~120º GLon due to pressure gradient. The magnitude of TADs at -60º~0º GLon is stronger (~30 m/s) than that at 0º~120º GLon (~20 m/s). However, those notable equatorward TADs stand around 20º MLat (black line). At -180º~-60º GLon and 120º~180º GLon, the periodic wind disturbance originating from high latitudes does not have any notable time delay with respect to latitudes, which are caused by the ion drag. Similar to TADs at -60º~120º GLon, these instances wind responses also stand at around 20º MLat. We will use the model to further investigate the physical mechanism of these standing features.
One can notice that the meridional wind changes that show signi cant hemispheric asymmetry and longitudinal differences, including TADs and simultaneous responses without signi cant time delay with respect to latitudes, are unexpected and seldom investigated during the temporal oscillations of IMF Bz. The broken mean circulation occurs in the Southern Hemisphere, while uninterrupted TADs and standing features appear in the Northern Hemisphere. The longitudinal differences and hemispheric asymmetry will be further explored in the discussion section.

Discussion
As disclosed by previous work, the thermospheric winds are determined by a balance between several forcing (Wang and Lühr, 2016). For instance, the pressure gradient, ion drag, viscosity, Coriolis, and momentum advection. Similar to Hsu et al. (2016), we perform a term analysis of these forcing in meridional momentum equation. In this work, only the ion drag and pressure gradient are shown, because they are dominant over the other drivers.
Broken Mean Circulation Figure 3 shows the differences in forcing terms of meridional winds between the CMIT 60-min oscillation case and the background run. Figure 3a shows the differential total accelerations in the meridional wind, where broken mean circulation appears at -30º~180º GLon in the Southern Hemisphere. A comparison between Figures 3a and 3b (acceleration due to pressure gradient), 3c (acceleration due to ion drag) shows that the broken mean circulation result from the combined roles of ion drag and pressure gradient.
These are associated with both penetration electric elds and the energy input from the solar wind to the upper thermosphere (Dungey, 1961). Wherever the ion drag overwhelms the pressure gradient, the propagation of TADs will be broken into several pieces. Under the northward IMF condition, when the ion drag is weaker than the pressure gradient, the uninterrupted TADs become obvious.
The pressure gradient is the reason for the propagation of wind disturbances from high to low latitudes, which is related to the enhanced Joule heating and related neutral temperature. Under the southward IMF Bz condition, the energy carried in the solar wind can be fully and directly deposited into the upper atmosphere due to the dayside magnetic reconnection (Liu et al., 2018a). Due to the nightside reconnection, the energy is also released into the upper thermosphere from the magneto-tail (Liu et al., 2018a). The energy carried by solar wind accumulates in the magneto-tail, and then under the effects of the solar wind variability or magnetosphere internal processes, it is suddenly released. Both processes contribute to the enhanced Joule heating and neutral temperature in the polar region (as shown in Figure  4d).
The ion drag is vital for the development of the broken mean circulation, which is related to both ion density and velocity. The former is related to the vertical transport caused by both E×B drift and neutral winds, and the latter is related to E×B ion drift (Richmond, 1995). During the rapid temporal change of IMF, the prompt penetration electric elds (PPEF) from the high to the low latitudes occurs (Nishida, 1968). It is well-known that PPEF is induced by an imbalance between Region 1 (R1) and Region 2 (R2) eld-aligned currents (FACs) at high latitudes. The under-shielding can appear when R1 FACs have a stronger amplitude than R2 FACs under southward IMF Bz, and overshielding can occur when R1 FACs have a weaker amplitude than R2 FACs. As disclosed in previous works, PPEF produces the changes of ionospheric plasma density and vertical ion drift at different latitudes at the same time, because it occurs instantaneously in global scale (e.g., Liu et al., 2004;Rodríguez-Zuluaga et al., 2016;Yizengaw et al., 2004). Figure 4 shows the geographic longitude and latitude variations of the disturbed electron density, vertical plasma velocity, meridional plasma velocity, and neutral temperature. The enhancements of wind and temperature at -60°~-180° GLon might be related to the longitudinal patterns of electron density, which have a peak in America sector and causing by the topology of the geomagnetic eld (Wang and Zhang, 2017;Zhang et al., 2019). The wind and temperature can be affected by electron density via ion-neutral collision. The E×B changes in America sector might be due to the offset between geographic latitudes and geomagnetic latitudes which are more notable at -60º ~ 180º GLon. One can notice that both the disturbed plasma density and vertical plasma velocity (Figures 4a and 4b) are stronger at 0º ~ -180º GLon than that at 0º ~ 180º GLon, due to the con gurations of the geomagnetic eld. The geomagnetic eld can in uence the disturbance of plasma, because the plasma can be transport along the geomagnetic eld lines by thermospheric winds (Rishbeth, 1967). The ion drag in the Southern Hemisphere comes mainly from the E×B ion drifts because the plasma density changes are very weak. It can be seen from Figure 4c that the meridional plasma velocity in the Southern Hemisphere is much stronger at -60º ~ 180º GLon (~60 m/s) than that at the remaining longitudes (~15 m/s) corresponding to the geomagnetic eld geometry. As disclosed in Zhang S. et al., (2012) and Zhang K. et al., (2018), the eld-aligned plasma velocity due to thermospheric winds is related to geomagnetic inclination and declination. The equation of wind projection along the magnetic eld lines is: where VV is the plasma velocity in the vertical direction driven by the winds, Vn is the meridional winds, Un is the zonal winds, I is the inclination, D is the declination, the positive (negative) signal stands for that in the Southern (Northern) Hemisphere. Field-aligned winds push the plasma to move along the magnetic eld lines. In addition, the plasma can also be driven upward or downward due to eastward or westward electric elds (E × B drifts), which also depends on the geomagnetic eld con guration. The offset between geographic latitude and geomagnetic latitude is different at different longitudes. Thus, plasma density has a notable longitudinal dependence. The longitudinal distribution of electron density can feed back to further in uence the neutral winds via ion drag, as the two processes that determine the ion drag, the electron density, and the relative velocity between the ions and neutrals all have longitudinal dependence. This further demonstrates that the IT system is a closely coupled, nonlinear one. This is because the offset between geographic latitudes and geomagnetic latitudes is more notable at -60º 1 80º GLon. This is consistent with the longitudinal variation of ion drag as shown in Figure 3c. The longitudinal variation of the ion drag related to the E×B drift can explain the longitudinal variation of the broken mean circulation.
There also exists a signi cant hemispheric asymmetry in the broken mean circulation, caused by the combined roles of pressure gradient and ion drag. In the longitude sector, both the effects of pressure gradient and ion drag are strong enough and cannot overwhelm each other in the Southern Hemisphere.
As shown in Figure 4a, the electron density enhancements are more than 3 × 10 11 m -3 at geomagnetic latitudes above 20º MLat in the Northern Hemisphere, which are signi cantly stronger than those in the Southern Hemisphere (1 × 10 11 m -3 ). Thus, the ion drag effects, which are proportional to the plasma density, show the corresponding patterns. In addition, the plasma density changes in the Northern Hemisphere are much larger at 0º ~ -180º GLon (~5 × 10 11 m -3 ) than those at 0º ~ 180º GLon (~3 × 10 11 m -3 ), thus the ion drag effects are also stronger in the 0º ~ -180º GLon sector. Note here that in Figure 3b, the pressure gradient that is associated with temperature disturbances under the temporal oscillations of IMF Bz conditions has similar magnitudes in the two hemispheres, and does not show outstanding longitudinal patterns as that in the ion drag effects. Then, the role of ion drag overpowers the pressure gradient at this sector and is overwhelmed at the remaining longitudes.

Standing signature
As shown in Figure 2c, the TADs of meridional winds stand at ~20º MLat. A similar standing feature can be found in the plasma density ( Figure 4a) and neutral temperature ( Figure 4d). As shown in Figure 3, the standing feature in the meridional wind is due to the combined roles of ion drag and pressure gradient.
As disclosed in the literature (e.g., Breig, 1987;Fejer et al., 1999), the E×B drifts, ambipolar diffusion, thermospheric neutral winds, and chemistry are vital in the variations of ionospheric electron density at middle and low latitudes. Daytime PPEF is eastward for the undershielding and westward for the overshielding (Peymirat et al., 2000). Then, the ionospheric plasma is driven upward/downward along the magnetic eld line, causing the density enhancement/decrease that is dependent on the longitudes in the LT-xed map, as shown in Figure 4a. The plasma density enhances at almost all longitudes at ~20º MLat (black curve line, Figure 4a) in the Northern Hemisphere. This could be related to the standing feature of TADs (Figure 2c). The density enhances signi cantly at MLat above 20º MLat, introducing strong ion drag effects to the equatorward propagation TADs in meridional winds. Thus, the meridional winds are slowed down in this sector, with a pattern of standing feature. The plasma moves to high altitudes at the equator, then decrease to low altitudes along the magnetic eld line for the ambipolar diffusion, summerwinter thermospheric winds, and gravity force, producing the notable equatorial ion anomaly (EIA) structures that are signi cant at ~20º MLat (black curve line, Figure 4a). At ~20º MLat, the density changes are stronger at -60º ~ -180º GLon and 120º ~ 180º GLon than those at remaining longitudes due to the geomagnetic eld topology, consistent with the results of Immel and Mannucci (2013) and Greer and Ridley (2017). Because ionospheric plasma can move along the geomagnetic eld lines by ambipolar diffusion and neutral winds. The plasma can also be transported upward/downward under the effects of eastward/westward electric elds (E×B drifts). The equatorward/poleward winds can move the plasma along the magnetic eld lines (upward/downward) due to the wind projection as shown in equation (1), producing density changes. The transportation of plasma due to the winds has a close relationship with the geomagnetic con guration. For instance, when the declination is negative (positive), the eastward winds can move the plasma to a lower (higher) altitude, causing a decrease (enhancement) of electron density. In the Northern (Southern) Hemisphere, there are two (one) zones of negative declination and two (one) zones of positive declination. Thus, the electron densities with the modi cation of geomagnetic eld morphology can have large longitudinal variations. Thus, the effects of ion drag overpower the pressure gradient and prevent the propagating of TADs at -60º ~ -180º GLon and 120º 1 80º GLon, which is a key factor in the standing feature in meridional winds at these longitudes.
At -60º ~ 120º GLon, the effect of the pressure gradient on the meridional wind dominates over the ion drag when the electron density is weak (Figure 3a). Note here that the temperature seems to enhance more at latitudes above 20º MLat than that at 0º ~ 20º MLat, indicated by the black curve line ( Figure   4d). Thus, the standing feature at this longitudinal sector seems to be related to the neutral temperature changes, which exhibit a similar structure. At this longitude sector, the offset between the geomagnetic and geographic latitudes is more notable, correspondingly the neutral temperature change can be larger. Because the electron density changes are noticeable in this longitude sector, and thus the corresponding neutral temperature change due to the plasma collisional heating, which is the dominant heating mechanism for the neutrals in the upper thermosphere, is large (Zhang et al., 2018).
In addition, the standing feature appears only in the Northern Hemisphere, showing a signi cant hemispheric asymmetry. This asymmetry is closely related to the electron density changes under the temporal oscillations of IMF Bz conditions. At ~300 km, the electron density ( Figure 4a) changes are much larger in the Northern Hemisphere than those in the opposite hemisphere. The density changes in the Northern Hemisphere are greater than 3 × 1011 m-3 at geomagnetic latitudes along and above 20º MLat, whereas those are ~1 × 1011 m-3 at the conjugated latitudes in the Southern Hemisphere. The stronger the plasma density changes are, the larger in uences of ion drag on the meridional winds are (Figures 3a and 3c). Thus, the equatorward propagation of TADs in the Northern Hemisphere stands at 20º MLat, and does not appear in the Southern Hemisphere.

Conclusion
Using coupled magnetosphere-thermosphere-ionosphere model results, the hemispheric and longitudinal differences of the meridional wind changes during oscillating interplanetary magnetic eld Bz at 60-min are explored in this work. The main results are shown below.
1) The broken mean circulation only exists at 0º~180º geographic longitude (GLon) in the Southern Hemisphere, which is due to the combined roles of the pressure gradient and ion drag forces.
2) The standing features in TADs of meridional winds are found at ~20º magnetic latitude (MLat), which are induced by the geomagnetic eld con gurations, neutral temperature, and electron density changes.

List Of Abbreviations
Interplanetary magnetic eld (IMF).

Competing interest
The authors declare that they have no competing interests.

Funding
The National Center for Atmospheric Research is sponsored by the National Science Foundation. We are really grateful for the sponsor from the National Nature Science Foundation of China (NO. 41974182, 41674153, 41431073, 41521062 and 42004135), and the Spark Project at Wuhan University (2042020gf0024).

Authors' contributions
Kedeng Zhang, Hui Wang, Wenbin Wang, Jing Liu made contributes to the theoretic interpretation of results and drafted the manuscript. Kedeng Zhang and Jie Gao made contributions on the data processing and model work.