Meso-Scale Electrodynamic Coupling of the Earth Magnetosphere-Ionosphere System

Within the fully integrated magnetosphere-ionosphere system, many electrodynamic processes interact with each other. We review recent advances in understanding three major meso-scale coupling processes within the system: the transient field-aligned currents (FACs), mid-latitude plasma convection, and auroral particle precipitation. (1) Transient FACs arise due to disturbances from either dayside or nightside magnetosphere. As the interplanetary shocks suddenly compress the dayside magnetosphere, short-lived FACs are induced at high latitudes with their polarity successively changing. Magnetotail dynamics, such as substorm injections, can also disturb the current structures, leading to the formation of substorm current wedges and ring current disruption. (2) The mid-latitude plasma convection is closely associated with electric fields in the system. Recent studies have unraveled some important features and mechanisms of subauroral fast flows. (3) Charged particles, while drifting around the Earth, often experience precipitating loss down to the upper atmosphere, enhancing the auroral conductivity. Recent studies have been devoted to developing more self-consistent geospace circulation models by including a better representation of the auroral conductance. It is expected that including these new advances in geospace circulation models could promisingly strengthen their forecasting capability in space weather applications. The remaining challenges especially in the global modeling of the circulation system are also discussed.

The classic Dungey cycle: interplanetary plasma flow in a plane containing neutral points on both dayside and nightside (adapted from Dungey 1961). (b) The simplified two-cell convection pattern in the high-latitude ionosphere

Introduction
The terrestrial magnetosphere is formed after the upstream solar wind and interplanetary magnetic field (IMF) impinge on the geomagnetic field. The magnetopause, a boundary that separates the magnetosphere from the interplanetary medium, is where the magnetic pressure of the geomagnetic field is balanced by the dynamic pressure in the solar wind. Dungey (1961) proposed the classic global circulation pattern, namely the Dungey cycle as shown in Fig. 1. When the IMF is oriented southward, opposite to the geomagnetic field, magnetic reconnection occurs on the dayside magnetopause, resulting in solar wind energy entry through the cusp region at high latitudes on the dayside. The anti-sunward solar wind convection pushes the newly opened magnetic field lines toward the nightside, causing their footprints on the Earth to move in the same direction. On the nightside, the magnetosphere is further compressed toward the equatorial plane from the north and south. A current sheet thus is formed in-between the oppositely directed magnetic field lines near the equatorial plane. Magnetic reconnection is again triggered, injecting plasma sources sunward. At the footpoints of the magnetic field lines, the plasma in the ionosphere is frozen in and follows the same convection pattern. It moves firstly anti-sunward from dayside to the nightside at high latitudes and then returns at lower latitudes towards dayside. Two convection cells are hence formed in the ionosphere. Such a large-scale convection pattern is a direct manifestation of the solar wind-magnetosphere-ionosphere coupling. Note that the above picture assumes a "voltage generator"; i.e., the electrostatic potential is specified in the generator region in the magnetosphere. The underlying idea is that the electrostatic potential is constant along magnetic field lines and can be mapped from the magnetosphere to the ionosphere. The response of the ionosphere is therefore a consequence of the imposed electric field. It is the frozen-in condition that allows the mapping of electrostatic potential along the field lines. This basically requires a steady-state assumption, an underlying form of the Dungey's circulation model. However, this large-scale slowly varying convection picture gives rise to pressure balance inconsistency (Erickson and Wolf 1980;Garner et al. 2003), that is, as the plasma adiabatically convects towards the Earth, the plasma pressure could be excessively large there, inconsistent with observations. In many circumstances, especially in the magnetotail and inner magnetosphere, where short time scale phenomena often occur or field-aligned potential drops are present, the above simple mapping picture may not be valid any more.

Fig. 2
Statistical spatial distribution of Region-1 and Region-2 field-aligned currents obtained from Triad observations during weakly disturbed conditions (left) and active conditions (right). Black region indicates downward flowing to the ionosphere and white grids denotes the current flowing out of the ionosphere. Adapted from Iijima and Potemra (1978) Fig. 3 Statistical spatial distribution of Region-1 and NBZ currents in the south polar region during By > 0 (left) and By < 0 (right). The NBZ currents are distributed at higher latitudes (approximately > 80 • ) with opposite flowing polarity to the Region-1 field-aligned current. Adapted from Iijima et al. (1984) Besides the linkage via the Dungey convection, the FACs that flow along magnetic field lines provide another bridge in connecting the magnetosphere and ionosphere systems. The statistical study of Iijima and Potemra (1978) obtained the classic FACs pattern projected onto the ionosphere, as shown in Fig. 2. The Region-1 FACs flow into the ionosphere at higher latitudes on the dawnside and out of the ionosphere on the duskside. The Region-2 FACs at lower latitudes behave oppositely, flowing into the ionosphere in the dusk and out on the dawn. Other FACs, such as NBz FACs (Iijima et al. 1984), may also emerge, as illustrated in Fig. 3. The NBz FACs appear in the polar cap region, usually above 80 • , and flow in the opposite sense to the Region-1 FACs. The morphology of this current system highly depends on the IMF By component. While the Region-1 FACs are usually connected with magnetopause currents, the Region-2 FACs are associated with inner magnetosphere partial ring current (see review by Ganushkina et al. 2018) and the NBz FACs are originated from high-latitude reconnection when the IMF Bz is northward (i.e., positive Bz) (Iijima et al. 1984). Furthermore, other transient FACs may be generated when the magnetosphere experiences disturbances. In that case, the ionosphere would also undergo transient disturbances given its coupling nature with the magnetosphere. Such a perspective of viewing the magnetosphere-ionosphere coupling system often assumes a "current generator". The ionosphere and magnetosphere are connected by currents flowing along the magnetic field direction. The currents are sustained by a generator in the magnetosphere, and are closed via ionospheric currents. If the current is constant, then the ionospheric conductivity and electric field are inversely related.
A third coupling process, namely, particle precipitation, serves as a primary energy source for auroral emission and ionospheric ionization. It originates from various magnetospheric regions, such as magnetosheath, plasma sheet, cusp, or inner magnetosphere ring current/plasmasphere/radiation belts zones. Charged particles enter their loss cones after suffering scattering/diffusion/accelerated processes that render the particles to approach the upper atmosphere but unable to return to the magnetosphere. Precipitating electrons carrying energy from a few eV to a few keV are responsible for diffuse aurora, while those at even higher energies could cause discrete aurora (Newell et al. 2009(Newell et al. , 1996 or pulsating aurora (Johnstone 1978;Miyoshi et al. 2015a,b;Kasahara et al. 2018). A detailed review of the origin of diffuse aurora can be found in Ni et al. (2016). Precipitating protons, on the other hand, are found to be another important energy carrier down to the upper atmosphere. They can contribute approximately 15% of the total energy down to the ionosphere (Newell et al. 2009;Hardy et al. 1989). Once atmosphere neutrals or ions collide with the precipitating energetic particles, some are excited and some are ionized. Enhanced ionization and resultant electron density further increase the ionospheric conductivity.
The relationship between FACs (J || ) and conductivity can be derived as follows. According to the current continuity, the divergence of a current is zero. That is, As the perpendicular current J ⊥ flows in the ionosphere, ionospheric conductivity and electric field obey the Ohms law: where E is the electric field, b is the unit vector along the magnetic field line. The 2 × 2 conductivity tensor, σ ⊥ , is the following: σ P and σ H are the Pedersen conductivity and Hall conductivity, respectively, and are determined by collisional effects as follows: where n e is the local electron density, i and e are ion and electron gyrofrequencies and ν in and ν en are the total ion and electron momentum transfer collision frequencies. The Pedersen conductivity σ P is parallel to the electric field but perpendicular to the magnetic field, while the Hall conductivity σ H is perpendicular to both the electric and magnetic fields.
Combining Equation (1) & (2) and integrating along magnetic field lines, the heightintegrated conductivity σ ⊥ dz, or conductance , is related to the FACs (J || ) and electric potential ( ) via the following equation: ∇ · ( · ∇ ) = −J || sin I The change in the conductivity can therefore influence the electric potential/field. With the assumption of equipotential magnetic field lines in space, both the ionosphere and magnetosphere share the same electric potential. When the electrodynamics over the ionospheric altitude varies, the electric drift and energization of magnetospheric plasma consequently change. The latter further influences other processes in the magnetosphere, such as wave excitation, magnetic configuration, and particle scattering. This manifests the feedback effect of the particle precipitation originated from the magnetosphere. Therefore, the two systems are highly connected such that any changes in one system ought to cause changes in the other. Understanding this complicated space environment is thus challenging but crucial. This review article will cover recent advances in observations and modeling of some magnetosphere-ionosphere coupling processes, mainly focusing on meso-scale dynamics: transient FACs, mid-latitude convections, and auroral particle precipitation. Mass coupling between the two systems is another major coupling topic, but we do not include it in this article. Readers can refer to recent literatures for details (e.g. Zhang and Brambles 2021;Welling et al. 2016;Glocer 2016). This review is organized as follows. Section 2 describes the origin of transient FACs and their effects on the ionosphere. Section 3 presents recent findings on the subauroral convections that directly couple the mid-latitude ionosphere with the inner magnetosphere. Section 4 discusses advances in understanding particle precipitation and their subsequent influences on the coupled system. We summarize in Sect. 5 with a discussion on some remaining challenges, especially in the global modeling of the magnetosphere-ionosphere coupled system.

Coupling Through Transient Field-Aligned Currents (FACs)
The classic Region-1 and Region-2 FACs patterns in Fig. 2 usually develop when the magnetosphere encounters a southward directed IMF. In addition to this large-scale, sustained FACs pattern, transient FACs can also emerge when the magnetosphere is greatly disturbed at short time scales. In this section, we review some recent progress in understanding the transient FACs after the magnetosphere-ionosphere system undergoes disturbances due to either rapid solar wind pressure pulses or enhanced flow bursts traveling from the magnetotail. These two types of disturbances can result in the appearance of transient FACs on either dayside or nightside.

Dayside Origin of Transient FACs
Following a sudden solar wind pressure enhancement, ground-based magnetometers on the morning side often detected a rapid depression of the north component of the magnetic field Bx at high latitudes but an enhancement at low latitudes. Similar variations were discovered on the afternoon side but in opposite polarity. That is, the magnetic field increases at higher latitudes but decreases at lower latitudes (e.g. Engebretson et al. 1999). It is further found that the above changes can only last temporarily for 1-2 minutes, after which the magnetic fields on both morning and afternoon sides and at high and low latitudes change into an opposite sense. The later phase lasts for a slightly longer time. These successive perturbations of the surface magnetic fields have been categorized into preliminary impulse (PI) and main impulse (MI) (Araki 1994).
After calculating the equivalent ionospheric convections from all available magnetometer data, it was found that these magnetic impulsive variations on the ground are related to paired traveling convection vortices (TCVs) (Engebretson et al. 1999) in the ionosphere. Vortices rotate clockwise on the morning side and counter clockwise on the afternoon side in the PI phase. Subsequently, a second pair of TCVs rotate in the opposite sense in the MI phase. As the ion convection vortex rotates oppositely to the ionosphere Hall current that flows in the direction of electron convection, the clockwise convection vortex corresponds to counter-clockwise Hall current. The current at higher latitudes on the morning side would induce southward magnetic perturbation and that at lower latitudes would induce northward magnetic perturbation, resulting in a depression and an enhancement in the total surface magnetic field at high and low latitudes respectively.
Both modeling and observational studies found that these magnetic field perturbations and equivalent ionospheric convections are driven by FACs after the sudden compression of the magnetosphere. Modeling studies revealed the formation process of the FACs pairs in the ionosphere and associated magnetic perturbations on the ground (Fujita et al. 2003a(Fujita et al. , 2005Yu and Ridley 2009;Samsonov et al. 2010;Fujita 2019). Figure 4 shows one modeled example from Samsonov et al. (2010) of the evolution of residual FACs following the solar wind pressure enhancement. Shortly after the magnetosphere compression, a pair of Region-2 sense FACs emerges on the dayside around 70 • (at 1:30), propagating antisunward. About 1 minute later, another pair of FACs with opposite polarity (in the Region-1 sense) appears at lower latitudes near the same local times at 2:15 and gradually enhances while propagating antisunward.
Several studies have proposed the generation mechanisms for these transient FACs following the compression of the magnetosphere. For example, Glassmeier and Heppner (1992) speculated that the FACs are generated at the magnetopause due to the indention of the magnetosphere where pressure gradient exists. Lysak and Dh (1992) suggested that the excited compressional wave converts to a shear mode Alfvén waves in an inhomogeneous magnetosphere, carrying the FACs to the ionosphere. Lühr et al. (1996) proposed that FACs are formed near the magnetospheric boundary layer where local pressure perturbations are triggered. Kivelson and Southwood (1991) also proposed that the FACs are generated by shear Alfvén waves at the boundary. Sibeck et al. (2003) suggested that FACs are generated where transient azimuthal pressure gradients appear after the compression.
These above studies do not consider the FACs in the two phases separately. Later studies found that the two sets of FACs in the PI and MI phases are generated by different processes. For the FACs in the PI phase, some studies proposed that they are generated by an inductive electric field (e.g. Araki 1994;Moretto et al. 2000;Yu and Ridley 2009) but some suggested that the FAC is generated through the wave mode conversion (e.g. Tamao 1964;Lysak and Dh 1992;Lysak et al. 1994;Kivelson and Southwood 1991;Fujita et al. 2003a,b). By conducting global MHD simulations after a sudden solar wind pressure enhancement, Yu and Ridley (2009) found induced dusk-to-dawn electric field following the sudden compression in the dayside magnetosphere cavity, as shown in the first column of Fig. 5. The large dusk-to-dawn electric field (in blue color) propagates earthward and fades away in about two minutes. During this period, the inductive electric field initially generates a duskto-dawn displacement current just inside the dayside magnetosphere, which is closed by the Chapman-Ferraro magnetopause current (see the blue lines in the second row in Fig. 5). They further found that the displacement current is then diverted towards the Earth on the afternoon side and out of the ionosphere back to the circuit on the morning side, resulting in Fig. 4 The temporal evolution of the residual field-aligned current in the polar region following a sudden solar wind density increase. Red indicates radially outward current and blue indicates radially inward current in µA/m 2 . Vectors represent the convection velocity. Adapted from Samsonov et al. (2010) a current loop with the magnetopause current. The pair of FACs connecting with the ionosphere thus is consistent with the PI phase. On the other hand, the wave mode conversion from the compressional wave to the Alfvénic wave could occur in a nonuniform magnetosphere when V A = 0. That is, a nonuniform Alfvén speed V A is necessary for mode conversion. As shown in Fig. 6, from Fujita et al. (2003a), within an inhomogeneous magnetosphere, the contours show a steep gradient of the Alfvén speed V A near L = 7. When Left column: the azimuthal electric field in the 12h meridian after a sudden solar wind pressure enhancement impinges on the magnetopause. Red and blue indicate dusk-to-dawn and dawn-to-dusk electric field. Arrows are the flow velocity. Right column: The 3D view of the electric currents (by blue lines) in the dayside magnetosphere. The contours represent plasma thermal pressure. Adapted from Yu and Ridley (2009) the wavefront of the compressional wave, indicated by the large electric field in color, propagates into this steep region, the Alfvén wave is excited and intensified FACs indicated by the vectors are induced (at t = 3.0 min).
Following the PI phase, the MI phase lasts for a longer time. Fujita and Tanaka (2006) revealed that the transient plasma vortex that appeared near the flank of the magnetosphere is associated with the MI-phase FACs in the ionosphere. Figure 7 (a) illustrates the perturbation in the magnetosphere after the gradual compression of the magnetospheric flank, including a plasma vortex and a small-scale pressure enhancement, which then affect the FACs. The magnetospheric flow vortex and the ionospheric TCVs are connected via the FACs and thus they are in a consistent rotation sense. According to a theoretical study of Ogino (1986), the dependence of the FACs (J || ) on field-aligned vorticity ( || ) and the pressure gradient (∇p) in a low-beta plasma can be expressed as follows: Since the viscosity μ and the resistivity η are generally small quantities in the magnetosphere, the second term on the left-hand side of each equation can be omitted. Therefore, the FACs are closely related with the plasma vorticity and pressure gradient. Note that under a steady state, these two equations can be reduced to the Vasyliunas equation (Vasyliunas 1970). If the vorticity and FACs vary slowly in time, the FACs are proportional to the field- Fig. 6 Evolution of the azimuthal electric field (color contours) and electric currents (vector arrows) projected in the 13.4 h meridian after the shock passage. The equi-contour lines indicate the log(V A /V A0 ). The red and blue colors suggest dawn-to-dusk and dusk-to-dawn electric field respectively. The electric currents are only shown when they are almost parallel to the magnetic field. The vertical black bar indicates the location of the solar wind impulse in the magnetosheath. Adapted from Fujita et al. (2003a) aligned vorticity and square root of plasma density in a highly conductive plasma (see Equation (12) in Ogino (1986)). By comparing the field-aligned vorticity and pressure gradient in the equatorial plane, Yu and Ridley (2009) suggested that the FACs in the MI phase originate from the magnetospheric vortices where pressure gradient is high following the compressional wave passes through. Figure 7 (b) shows the time variation of vorticity d /dt in color and plasma convection in black streamlines. The intense red/blue region pointed by a red arrow indicates an growing vortex rotating in a clockwise/counter-clockwise direction, meaning that the FACs are being induced. This scenario is applicable in the northward IMF conditions. However, the pressure gradient under southward IMF conditions is not large enough to form the vortex. Instead, the vortex is formed as the system recovers from the fast mode wave in the PI phase (Yu and Ridley 2009). It should be noted that not only does the increase of the solar wind pressure induce these short-lived FACs, but a negative solar wind impulse can also invoke similar transients of the FACs. The only difference is that the latter produces the current transients in a reversed order. That is, Region-1 and Region-2 sense FACs appear alternatively. Some studies (e.g. Takeuchi et al. 2000Takeuchi et al. , 2002Hori et al. 2012) suggested that the negative solar wind pressure impulse produces the mirror-image variations on the ground and in the magnetosphere against the positive case, but Fujita et al. (2004) found that the first phase of these variations (i.e., the PI phase) is not completely mirror-imaging in terms of its generation mechanism. dt . The red and blue colors suggest enhanced vorticity in clockwise and counter-clockwise directions respectively (adapted from Yu and Ridley 2009) In particular, the FACs in the PI phase after the solar wind pressure increases is closed by the dawn-to-dusk Chapman-Ferraro magnetopause current, but they cannot hold as the FACs on the ionosphere reverse their polarity after the solar wind pressure decreases. Fujita et al. (2012) revisited the negative responses with a more intense decrease in the solar wind dynamic pressure and found that the magnetospheric current system in the PI phase is an afternoon-to-prenoon current, opposite to the Chapman-Ferraro current. So it is indeed a mirror-image of that in the positive case. For the MI phase, unlike the case with a positive solar wind impulse that features the appearance of Region-1 sense FACs, the sudden solar wind pressure decrease produces Region-2 sense FACs and an additional Region-1 sense FACs, meaning that the two cases are not exactly in a mirror-image mode .
Studies further unveiled that the successive appearance of the oppositely sensed convection vortices or FACs can repeatedly continue, resulting in wave-like oscillations in the ground-based magnetic field perturbations or convection velocity (Hori et al. 2012;Fujita et al. 2012;Yu and Ridley 2011;Samsonov et al. 2010). Preceded by the first two primary pairs of FACs that alternatively emerge on the dayside and then propagate toward the nightside (i.e., the first PI-MI sequence), the same sequence can follow. Figure 8 (a) demonstrated the variations of the line-of-sight velocity observed by the King Salmon HF radar at three regions distributed across high-latitudes (from A to C: latitudes change from low to higher latitude) and magnetic perturbations observed nearby. After a sudden solar wind density decrease, the eastward flow propagates poleward from A to B and C. Three "waves" of the velocity enhancement are detected, although the third one is barely recognized. After bandpass filtering of the ground-based magnetic perturbations, triple sequences of the PI-MI pairs are found. The second and possibly third sequences are however in a much weaker intensity. These oscillations in the magnetic perturbation suggest that the repeated appearance of the vortex or FAC pairs was likely generated by the same mechanism. Simulation results from Yu and Ridley (2011), as shown in Fig. 8 (b), demonstrated that the transient FACs appear four times in alternative senses, as pointed by the black arrows (each pair of arrows indicates a new appearance). This suggests two PI-MI sequences. These variations in the FACs cause magnetic perturbations similar to the observations in the left panel, but the third "wave" seems not to be reproduced. Such periodic variations are attributed to a compressional wave that propagates and bounces between the magnetopause and near-Earth plasmasphere/ionospheric boundary since its launch after the compression (Yu and Ridley 2011;Samsonov et al. 2010).
As the ionospheric plasma is tied to the footpoints of magnetic field lines near the surface, ionospheric dynamics are bound to respond accordingly. Some earlier studies (e.g. Collis and Häggström 1991) analyzed the ionospheric responses after the sudden magnetospheric compression and found an evident electron density depletion and ion temperature increase in the F region. A recent study by Zou et al. (2017), using Poker Flat Incoherent Scatter Radar (PFISR) measurements and global magnetosphere-ionosphere coupled simulations synthetically examined the localized plasma responses in the ionosphere during the passage of these transient FACs. Figure 9 shows that during the transition from the PI phase to the MI phase (i.e., when the horizontal component of the ground magnetic field perturbation changes from a negative impulse to a positive increase, as denoted by the blue period), the F region ionosphere experiences a reduction in the electron density and lifts. The ion temperature increases temporarily due to the frictional heating, and the heated ions upwell along magnetic field lines. However, the enhancement in the electron temperature sustains. By utilizing the University of Michigan Global Ionosphere Thermosphere Model (GITM) (Ridley et al. 2006), Ozturk et al. (2018) performed a modeling study to examine the global thermosphere-ionosphere dynamics in response to the dynamic solar wind pressure increase. They found similar responses as in measurements but the magnitude of the responses is underestimated. Other studies also reported simultaneous ionospheric TEC modulation in response to the sudden compression of the magnetosphere (Hao et al. 2017), but the observed enhancement of TEC was suggested to be resulted from the compression of plasmasphere instead of ionosphere.

Nightside Origin of Transient FACs
On the nightside, transient FACs are also frequently generated by various processes related to magnetotail dynamics. Earlier ground-based observations of the surface magnetic field at mid-latitudes showed that during the substorm expansion phase, the northward component of the nightside geomagnetic field perturbation appears to maximize where the eastward component is zero (see Fig. 10 (a)). To interpret such magnetic signatures on the ground, McPherron et al. (1973) proposed a classic phenomenological model of currents, later named as the substorm current wedge (SCW) (Pytte et al. 1976). This current wedge model, displayed in Fig. 10 (b), is simply a current loop that constitutes a pair of FACs, flowing down into the ionosphere in the early morning local times and out of the ionosphere in the pre-midnight local times, a cross-tail current in the distant magnetotail, and a westward electrojet current in the ionosphere. This SCW model is capable of explaining the observed magnetic perturbation at mid-latitude and has ever since played a prominent role in under-  McPherron et al. (1973) standing the magnetospheric substorm physics and coupling between the magnetotail and the ionosphere. The formation of SCW has been attributed to magnetotail instability (Lui 1996) or the azimuthal diversion and braking of earthward flows that generate flow shears and vortices (Shiokawa et al. 1998;Birn et al. 1999;Cao et al. 2008Cao et al. , 2010. Shiokawa et al. (1998) studied the braking effects of flow bursts and found that although the inertial currents contribute to the initial current reduction and diversion, the dominant and more permanent contribution stems from the pressure gradient terms. Cao et al. (2010) showed that whether a current system produced by the braking of fast flows can evolve into a SCW depends mainly on the conditions of the current disruption regions (i.e., braking regions) before the onset of fast flows. Based on THEMIS observations, Yao et al. (2012) inferred the formation mechanism of SCW in detail. They suggested that a vortex forms after the earthward flow burst from the magnetotail is diverted azimuthally and pressure gradient builds up in the X and Y direction, driving the FACs to flow downward/upward. Their analysis indicated that both pressure gradient and flow vorticity variation help generate the FACs of SCW, but the former contributes more at the dipolarizing region. Such pressure variations as the generator of the FACs in association with substorm auroral arc brightening and extension were further discussed by Yao et al. (2013b).

Fig. 11
An updated picture of the substorm current wedge by Kepko et al. (2015). The original Region-1 type FAC of the SCW is shaded in grey. A new Region-2 type FAC flow at the earthward edge of the Region-1 sense FAC After the concept of large-scale SCW, efforts were also devoted to producing a more realistic representation of the currents, which has expanded our knowledge of the SCW. For example, some studies incorporated a stretched magnetotail configuration, rather than a dipole field, included more electric currents into the model, or parameterized the wedge shape from statistical data. Recent satellite observations and simulations revealed a second current wedge equatorward of the original one, with an opposite FAC sense (Birn et al. 1999;Ritter and Lühr 2008;Birn et al. 2011;Birn and Hesse 2014;Sergeev et al. 2011Sergeev et al. , 2014Yang et al. 2012;Yao et al. 2013a;Sun et al. 2013). MHD simulations suggested that fast earthward plasma flow from the magnetotail can twist magnetic field lines along their paths. As the flow approaches the transition region between dipolar and stretched magnetic configurations, it is diverted azimuthally. The shear flow in azimuth brings up another twist of magnetic field lines in a different orientation. This dynamic process forms two pairs of FACs in different polarities (Birn et al. 2011;Birn and Hesse 2013). In the review article of Kepko et al. (2015), these new insights were incorporated and an updated SCW picture was brought forth. As demonstrated in Fig. 11, in addition to the original wedge with a Region-1 sense FACs, another pair of FACs in Region-2 sense appears at lower L-shell and closes to the high-latitude FACs through the meridian ionosphere.
As more high-resolution in-situ satellite observations become unprecedentedly available, some new features are further discovered. For instance, it is suggested that the large-scale SCW is composed of/embedded with multiple small-scale wedgelets (Liu et al. , 2015Palin et al. 2015), each of which is related to a dipolar flux bundle/transient BBFs (e.g. Angelopoulos et al. 1992Angelopoulos et al. , 1994Liu et al. 2013) moving earthward. As demonstrated in Fig. 12, a small-scale flux tube (or called dipolarizing flux bundle) with a L-shell width of 1 ∼ 3 Re and a local-time span of ∼ 1 hour contains a duskward current sheet as well as a pair of region-1 sense FACs connecting with the Earth. Studies of dipolarization fronts associated with small-scale dipolarization flux bundles revealed 2-dimensional FACs near Fig. 12 Illustration of small-scale wedgelets that constitute the large-scale SCW. The blue and red curves indicate FACs flowing toward and away from the ionosphere. The dipolarization front bundle (DFB) are in close association with the formation of the wedgelets FACs. Adapted from Liu et al. (2015) the DF layer (Yao et al. 2013a;Sun et al. 2013), showing two-loop FACs in association with the dipolarization flux bundle. That is, a pair of Region-1 sense FACs shows up within the DF while a pair of Region-2 sense FACs appears ahead of the DF, closer to the Earth. Such a localized current wedge is basically analogous to or a miniature of the SCW. Multiple such small-scale wedgelets together could result in the buildup of the large SCW (Birn et al. 2011;Liu et al. 2015Liu et al. , 2018aPalin et al. 2016). Nishimura et al. (2020) found that the two types of currents with different spatial scales actually co-exist as a hybrid during substorms. The large-scale wedge spans a few 1000 km in the ionosphere, corresponding to ∼ 10 Re in the magnetotail, while the transient wedgelets in the dawn-dusk direction are 600 km wide in the ionosphere and 3.0 Re in the magnetotail. The latter could recurrently intensify at various longitudes in the nightside auroral oval with a lifetime of ∼ 10 minutes.
Many recent numerical studies have reproduced the small-scale localized wedgelets from multiple fast magnetotail flows/BBFs (e.g., Birn and Hesse 2014;Cramer et al. 2017;Birn et al. 2019;Merkin et al. 2019). These upward and downward FACs forming the transient wedgelets are mostly connected to the outer magnetosphere. It is generally believed that the tail current disruption is related to the diversion of FACs down to the ionosphere. Birn et al. (2020) recently examined the current generator after the bursty flows and suggested that the FACs in the wedge are converted from inside and underneath the Region-1 and Region-2 currents, well toward the equatorial plane. These current conversions occur in the dipolar field near the transition region, about 10-15 Re away from the Earth (e.g. Birn et al. 2019). However, these global or regional simulations do not consider kinetic ring current dynamics, either because the kinetic physics is not coupled into the model or the model is designed in particular for the tail dynamics. The effects of transient BBFs on the kinetic ring current system are not investigated until Yu et al. (2017a). With a kinetic ring current model incorporated into a global MHD model, Yu et al. (2017a) discovered a new current closure of wedgelets driven by BBFs, as shown in Fig. 13. Besides the SCW, the inner magnetospheric westward ring current (shown in green lines near the Earth) could also be considerably disturbed due to the impact of BBFs from the magnetotail. It is suggested that when the fast flows are significantly braked and diverted at the transition region (L = 10), the cross-tail duskward current is converted to be field-aligned, forming the well-known largescale SCW. The FACs appear in a region where the "vortex type I" is formed (Vorticity z is indicated by the green-purple color in the equator). However, several flow vortices are further induced inside L = 6 across the nightside local times (pointed by the blue arrow as Fig. 13 Left panel shows the 3D view of the current system in the magnetosphere after BBFs arrive at the transition region, including the partial ring current around the Earth, the cross-tail current, and fieldaligned currents connecting the ionosphere with the magnetosphere. The red/blue on the current streamlines represent parallel/anti-parallel current along magnetic field lines. The purple/green color in the equatorial plane suggests the vorticity z , rotating clock-wisely/counter-clockwisely. The red/blue color on the Earth sphere represents downward/upward FACs. Two types of vortex are pointed by the blue arrows. The crosstail current in the mid-tail is connected by the FACs in the region where the "vortex type I" is formed. The westward ring current near the Earth is closed by the Region-2 FACs in the dusk and dawn sectors. On the nightside, around the region with "vortex type II", the ring current arches upward toward the high latitudes and eventually closes with the ionosphere via FACs. Right panel shows the residual FACs emerged at the ionosphere altitude after subtracting the FAC pattern before the impact of BBFs. The higher-latitude FACs around 70 • is enhanced as the SCW is formed, and two pairs of FACs show up at lower-latitudes around 60 • , connecting with the westward ring current on the nightside. Adapted from Yu et al. (2017a) "Vortex Type II" on z ). These vortices generate Region 1-sense FACs flowing into/out of the ionosphere. These FACs are surprisingly connected to the westward ring current. It is found that at least two pairs of such FACs emerge out of the ring current after the impact of BBFs. The ionospheric signature is manifested by two pairs of Region-1 FACs at latitudes around 65 • , distributed azimuthally.
As an interconnected system, the ionosphere is destined to exhibit some lower-boundary signatures of these small-scale, transient dynamics in the magnetotail. It is recognized that the north-south aligned auroral streamers are the auroral signature of longitudinally localized flow channels moving earthward (e.g. Henderson et al. 1998;Sergeev et al. 1999;Zesta et al. 2000). Sergeev et al. (1996) presented multi-instrument observations of high-speed Earthward plasma flow events in the midtail and argued that they are consistent with theoretically predicted signatures of plasma-depleted flux tubes or "bubbles". Ahead of the bubbles are flow and magnetic field shears. The azimuthal diversion of the fast flows may be related to auroral arcs equatorward of the auroral oval. Yang et al. (2014) simulated the formation of bright thin aurora arcs preceded by an equatorward moving streamer. As the low-entropy plasma bubble moves equatorward and arrives at the transition region, the azimuthal drift of plasma stretches the bubble in the east-west direction. Region-2 sense FACs are subsequently generated in this direction, in association with pressure and P V 5/3 gradients near the transition region. Such stretched FAC sheets manifest as an azimuthal thin arc near the auroral boundary. These numerical results are consistent with the observational study in Yao et al. (2013b). Ground-based magnetic perturbations can further demonstrate the net effects of these substorm currents. Wei et al. (2021) analyzed a BBF event pene-trated well into the inner magnetosphere and found a close association with the suddenly enhanced magnetic field perturbations on the ground. Observations from Swarm satellites that fly poleward in the night sector show that the large-scale FACs experience successive opposite variations at mid-latitudes, providing evidences for the meridional distribution of FACs as predicted in the updated SCW picture. Nakamura et al. (2005) carried out a conjugate analysis following a small-scale localized flow channel in the plasma sheet, detected by Cluster, and ionospheric disturbances observed by ground-based radars. The ionospheric equivalent currents and possible upward field-aligned currents are found to be consistent with the magnetospheric observations, implying the close association between the isolated flow burst and ionospheric signatures. Based on a statistical study of the flow bursts, a good correlation was clearly found between earthward flow bursts and auroral features (Nakamura et al. 2001).

Coupling Through Mid-Latitude Convection
In this section, we review three types of meso-scale convection at mid-latitudes that play important roles in coupling the inner magnetosphere with the ionosphere. They are the subauroral polarization streams (SAPS), Double Subauroral Ion Drifts (DSAIDs), and Dawn auroral polarization streams (DAPS). These mid-latitude meso-scale convections develop mostly during geomagnetically active times, and all appear to be associated with local Region-2 FACs. But their morphology and occurring locations differ considerably. Details of each type are described below.

Subauroral Polarization Streams (SAPS)
SAPS are enhanced westward plasma flows in the dusk-to-midnight local times equatorward of the auroral oval (Anderson et al. 2001;Foster and Vo 2002;Foster and Burke 2002;Kunduri et al. 2017;Yu et al. 2015;He et al. 2018;Wei et al. 2019b). They are usually characterized by a large flow speed within an elongated flow channel along the longitude. The latitudinal width of SAPS is about 3-5 • and varies with magnetic local time (MLT). SAPS usually occur during geomagnetically active times and are more intense as geomagnetic activity increases (Foster and Vo 2002). Statistical studies based on measurements from the Super Dual Auroral Radar Network radars by Kunduri et al. (2017) showed that 15% SAPS are observed in relatively quiet conditions while 87% are during moderately disturbed conditions (< −75 nT Dst < −50 nT). SAPS could also develop in quiet time substorms (i.e., no evidence of storms in the Dst index) (He et al. 2017). The subauroral ion drifts (SAID), a latitudinally narrower channel with a higher speed that often exceeds 1000 m/s (Spiro et al. 1979), can be occasionally embedded within the SAPS (Foster and Burke 2002), and sometimes regarded as a subset of SAPS. By utilizing multiple DMSP satellites across a broad range of local times, He et al. (2018) investigated the initiation and lifetime evolution of SAPS in detail during an intense geomagnetic storm and found that they originate from the dusk sector and later expand toward the midnight and lower latitudes. The lifetime of SAPS developed during storms are generally longer than the storm main phase but shorter if occurred in quiet time substorms (He et al. 2017). In association with the SAPS, the ionosphere F region often experiences electron density decrease, or a colocated density trough, suggesting the impact of the intense ion flows on the ionosphere composition (Spiro et al. 1978;Foster et al. 2007). Besides the effects on the ionosphere, thermospheric dynamics is also influenced through the ion-neutral coupling processes. Zhang et al. (2015b) reported an event where the neutral winds in the northern hemisphere travel poleward after the passage of the westward SAPS. The strong westward ion flow causes strong westward neutral winds in the trough region, which then last for a few hours before decrease. Subsequently, a poleward neutral wind surge of ∼ 100 m/s occurs due to the poleward Coriolis force. The enhanced collision with the neutrals also enhances the thermospheric heating.
Despite the rich knowledge of the SAPS on their characteristics and association with the thermosphere-ionosphere-magnetosphere system, the formation mechanism of the SAPS is still controversial without a clear consensus. The debate lies in the formation of the fast flow or the enhanced electric field in the narrow channel. One widely accepted mechanism is the "current generator" (Anderson et al. 2001;Karlsson et al. 1998). It suggests that when strong Region-2 FACs flow downward to the ionosphere, they are diverted to high latitudes via horizontal Pedersen currents. If the local Pedersen conductivity is low, then a large poleward electric field is required to satisfy the current divergence and close with higher-latitude Region-1 FACs. The enhanced poleward electric field thereafter induces a westward ion drift. The responsible Region-2 FACs are generated when the ring current ion pressure is highly inhomogeneous and its pressure gradient is not parallel to the magnetic gradient (Vasyliunas 1970). This mechanism suggests the brackets of the SAPS flow channels by the Region-1 and Region-2 FACs, which, however, is not often true. The Region-2 FACs are observed either at the poleward edge of the SAPS or mimic the flow channel shape (He et al. 2016;Mishin 2013). Another mechanism is the "voltage generator" based on the singleparticle approach (Southwood and Wolf 1978). It suggests that the poleward electric field forms when the plasma sheet inner boundaries of ions and electrons separate during injections. As the ion plasma sheet inner boundary is closer to the Earth in the dusk-midnight sector, a radially outward electric field is produced in the magnetosphere, which is directed poleward if mapped into the ionosphere. However, this approach disregards charge neutrality in slow plasma processes. Unlike these two traditional mechanisms, a new scenario has been recently proposed (e.g. Mishin 2013; Mishin et al. 2017). It is suggested that the poleward electric fields or Pedersen currents connecting the Region-1 and Region-2 FACs are inherent in the two-loop circuit of the SCW (see Fig. 11 for the newly updated SCW picture). In this case, the current closed over the meridional ionosphere on the westside of the wedge is the ultimate cause of SAPS on the duskside. Nevertheless, this paradigm is probably only valid during substorms and future validation on such correlation is still needed.
Many numerical modeling attempts were made to capture the SAPS dynamics for a better understanding of their spatial and temporal evolution in a global context. Due to the complicated nature of the SAPS within the ionosphere-thermosphere, a successful simulation of SAPS requires a global model that treats the coupling processes in a self-consistent manner and therefore the feedback effects or circulation dynamics within the coupled system can be captured. For example, Yu et al. (2015) simulated the March 17, 2013 storm event with a global MHD model BATS-R-US (Powell et al. 1999) coupled with a kinetic ring current model RAM-SCB (Jordanova et al. 2006(Jordanova et al. , 2010b and an ionospheric potential solver (Ridley et al. 2004). The model is capable of capturing the location of the SAPS observed by the DMSP satellites, the SAPS electric field near L = 3, and the global dynamics of the Region-2 FACs. But the large intensity of the SAPS is substantially underestimated. Raeder et al. (2016) incorporated a thermosphere model CTIM into the magnetosphere-ionosphere coupled circulation model OpenGGCM (Raeder et al. 2008) to embrace the feedback effects from the neutrals. Their simulation of the same storm event as in Yu et al. (2015) reproduced many statistical features of the SAPS, such as the MLT dependence of the SAPS during storm time, the SAPS appearance near the equatorward boundary of the precipitation where the conductivity is low, as well as an electron density trough in the ionosphere in association with the SAPS flow. A detailed one-to-one comparison was not carried out in their study. Zheng et al. (2008) conducted several numerical experiments to examine the effect of the ionospheric trough conductance on the formation of SAPS and confirmed the important role that conductivity plays in causing a sharper flow channel. Wei et al. (2019b) further analyzed several SAPS events in detail through multi-point observations in combination with simulations, and concluded that the SAPS appears to be related to the Region-2 FACs flowing into low conductance regions, and suggested that the "current generator" seems to be a more consistent scenario. Huba et al. (2017) (Lyon et al. 2004;Roble et al. 1988;Richmond et al. 1992;Toffoletto et al. 2003;Merkin and Lyon 2010), and reported the temporal evolution and spatial distribution of the SAPS.
The above various modeling studies of the SAPS demonstrated a problem that although some important features of SAPS are captured to some extent, there are still gaps between observations and model results. For example, the intensity of SAPS was underestimated in Yu et al. (2015), which is probably attributed to the lack of IT feedback mechanism in their model and a simple specification of ionospheric conductance. The location of SAPS was not well captured in Lin et al. (2019). They suggested that the displacement of SPAS locations was likely a result of inner magnetosphere effects since the ring current pressure distribution is related with the magnetic field topology. The less realistic magnetic field and the missing of auroral diffuse precipitation in their model were speculated as the main cause. It is certain hat more improvements are further needed in modeling studies. Exactly what physics, assumptions, or parameterization are imperfect demands more exploration.

Some Newly Identified Subauroral Convections: DSAIDs, DAPS
Recently a new phenomenon called double-peak subauroral ion drift (DSAIDs) is recognized (He et al. 2016;Horvath and Lovell 2017) and regarded as a subset of SAPS. DSAIDs appear with two velocity peaks in the subauroral zone and preferentially occur in geomagnetically disturbed times. The two velocity peaks are not persistently comparable in magnitude, and they only occur within a limited MLT range. He et al. (2016) examined several cases and statistically concluded the correlation of DSAIDs with the region-2 FACs. That is, the two velocity peaks are usually colocated with two downward flowing region-2 FACs, as shown in Fig. 14 (i). Such statistical association with the Region-2 FACs led the authors to propose that the formation mechanism of DSAIDs is probably the same as SAPS, but driven by two layers of Region-2 FACs. The two FACs flow into the ionosphere in the subauroral region, separated by a few degrees in latitude. If the local conductivity is relatively low from the ambient plasma environment (illustrated in the right panel in Fig. 14), two poleward Pedersen currents are needed to satisfy the current continuity. The low conductance requires an enhanced electric field to close the current circuit, and hence two large westward flow channels appear.
Motivated by the results in He et al. (2016), in order to identify the driver for the formation of the two-layer Region-2 FACs in the magnetosphere, Wei et al. (2019a) examined the ionospheric subauroral convection, currents, and particle precipitation based on low-Earth orbit DMSP satellite measurements during the March 17, 2015 storm. Van Allen Probes Fig. 14 The two panels on the left show statistical comparison between the SAIDs and DSAIDs. From top to bottom are the horizontal/vertical ion drift velocity (positive for eastward/upward velocity and negative for westward/downward velocity), ion and electron density, ion and electron temperature, Pedersen conductance, FACs (positive for downward region-2 FAC), average energy of precipitating ion and electron, and total number flux of precipitating ion and electron. The panel on the right demonstrate the formation mechanism of DSAIDs conceptually. Adapted from He et al. (2016) were further utilized to explore the magnetospheric dynamics. They found that prior to the incident of DSAIDs, intense proton flux injections recurrently occur. With a global magnetosphere circulation model, their simulations reproduced the plasma injections and further revealed that as the recurrent injections bring tail plasma sources directly into the inner magnetosphere, the ring current pressure is enhanced and exhibits two pressure peaks across L shells, as illustrated in Fig. 15. While the enhanced ring current pressure bulge at smaller L shells tends to be steady during the storm time, the one at larger L shells is more dynamic, depending on the frequency of injections. The two ring current pressure peaks then create two sets of Region-2 FACs according to the Vasyliunas theory (Vasyliunas 1970), resulting in the subsequent formation of DSAIDs.
Although the statistics reported in He et al. (2016) suggested the correlation of the twolayer Region-2 FACs with the DSAIDs, Horvath and Lovell (2017) found that in some DSAIDs cases, the two-layer Region-2 FACs are not necessarily available, controversial to the statistical results. Their study implied that the DSAIDs may be driven by some different processes other than the above double-layer Region-2 FACs mechanism. More investigation is therefore desired.
A third class of the mid-latitude fast flows is the dawnside auroral polarization streams (DAPS), termed by Liu et al. (2020). This flow is different from the intense Birkeland current boundary flows (BCBF) (Archer et al. 2017). The latter peaks near the boundary between Fig. 15 A schematic illustration of the formation mechanism of the DSAIDs based on numerical simulations (Wei et al. 2019a). The left panel shows normal current closure following substorm injections that a downward region-2 FAC is closed by the partial ring current. The right panel shows two region-2 FAC layers are formed as the partial ring current exhibits two pressure peaks. The recurrent substorm injections bring in continuous plasma source to dynamically enhance the pressure peak in the outer zone, while the pressure peak in the inner zone is more steady the upward Region-2 and downward Region-1 FACs in the auroral zone. When the two sets of FACs close through the horizontal Pedersen current across the boundary, the horizontal electric field is directed equatorward, which then drives an eastward flow, or called BCBF. According to Liu et al. (2020), the DAPS arises in the dawn sector and directed eastward. However, unlike SAPS that occur equatorward of the auroral zone, DAPS appear as fast ionospheric flows within the returning flow of the auroral convection cell and the flow peak is colocated with the Region-1 FACs at higher latitudes, as displayed in Fig. 16 (top panel). The most prominent feature of the DAPS, thus distinguished themselves from the BCBF, is that there is a steep flow gradient equatorward of the DAPS, i.e., near the boundary between Region-1 and Region-2 FACs. The flow equatorward of the steep gradient is much slower than that poleward. Liu et al. (2020) suggested that some flows that were reported in previous studies probably belong to DAPS (e.g. Aikio et al. 2018;Jiang et al. 2015;Liu et al. 2018b;Zou et al. 2009;Gkioulidou et al. 2009). It is also found that DAPS appears along with a bright, discrete auroral arc and an inverted-V electron precipitation structure within the higher-latitude part of the upward Region-2 FACs. The electron precipitation enhances the local Region-2 FACs. Therefore, it is suggested that DAPS is likely originated from the same process that gives rise to SAPS.
Based on simulations and observations, Liu et al. (2020) put forward the generation mechanism of DAPS. As illustrated in Fig. 16 (bottom panel), when the upward Region-2 FACs on the dawnside is enhanced, it is supplied by the enhanced downward Region-1 FAC and equatorward Pedersen current (light blue arrows). The enhanced Pedersen current requires an increased equatorward electric field (purple arrows) to satisfy the current continuity. This electric field is further modified by the conductivity distribution. The conductivity at lower latitudes where the upward Region-2 FACs are located is enhanced due to downward electron precipitation. The conductivity at higher latitudes where the downward Region-1 FACs are located is relatively low but not absolutely low. The electric field at the lower-latitude part of the aurora is thus suppressed by the high conductivity and that at the higher-latitude part is strengthened to complete the potential drop across the convection zone. This explains the fast eastward flows parallel to the boundary between Region-1 and Region-2 FACs (green arrow), i.e., DAPS. The steep flow gradient equatorward of DAPS is likely to be related to the steep poleward boundary in the Region-2 FACs, a result of the thermal pressure buildup in the magnetosphere due to some magnetospheric processes.

Coupling Through Conductance/Particle Precipitation
In addition to the transient FACs and mid-latitude plasma convection, the magnetosphere and ionosphere are also united via particle precipitation, an energy carrier originated from the magnetosphere and deposited onto the ionosphere. A classic circulation picture of the magnetosphere-ionosphere system involving particle precipitation is illustrated in Fig. 17. Two primary ionization sources in the upper atmosphere are photo-ionization due to solar extreme ultraviolet (EUV) and impact due to incident precipitating electrons. The former is dominantly responsible for the ionization on the dayside, while the latter is mainly occurring in the auroral oval. As electrons precipitate to the upper atmosphere, they collide with atmospheric neutrals that can be either excited or ionized. The ionospheric conductivity is thus enhanced as the electron density increases. The height-integrated conductivity, or conductance, is an essential element within the circulation. As indicated in Equation (5), together with the FACs, the conductance can determine the electric potential pattern and hence the plasma convection/drift in the magnetosphere. The plasma drift alters the velocity distribution, which could carry free energy source for the excitation of various waves in the magnetosphere. Electrons can resonate with these waves and some are lost into loss cones, precipitating down to the atmosphere. As such, the circulation is closed.
Clearly, the conductance plays a crucial role in the above large-scale circulation. How it influences the system is of great interest and has gained considerable attention in the community. Numerical studies have shown that the distribution of the ionospheric conductance could effectively control the magnetotail configuration and plasma convection (Lotko et al. 2014). A conductance pattern with different spatial distributions, such as homogeneous, asymmetric, or large-gradient distributions, leads to different X-line morphology in the magnetotail. The earthward convection speed is also affected. Other studies also suggested that the magnetospheric dynamics are strongly dependent on the specification of auroral conductance (e.g. Raeder et al. 2001;Ridley et al. 2004). Merkin et al. (2005) found that the shape and location of both magnetopause and bow shock appear to depend on the intensity of ionospheric conductance, implying that the conductance could influence the dayside solar wind-magnetosphere coupling. The depth of earthward penetration of fast flows is also affected by the conductance (Ream et al. 2015). A smaller ionospheric conductance allows for deeper penetration of the flow channels from the magnetotail. It also tends to enable more perturbations, such as Pi2 pulsation, to propagate more freely into the inner magnetosphere when the flows slow down near the braking region.

Specification of Ionospheric Conductance in Global Models
As mentioned above, the ionospheric conductance is important in regulating the magnetospheric dynamics and influencing the energy transfer between the magnetosphere and ionosphere, but unfortunately, it is a quantity that cannot be directly measured or derived straightforwardly. Let alone its two-dimensional distribution over the ionosphere. In order to determine the conductance, local conductivity as a function of altitude is needed first. As shown in Equation (4), both Pedersen and Hall conductivities (σ P , σ H ) depend on various ionospheric parameters, including the local electron density n e , collision frequency between ions and neutrals or between electrons and neutrals (i.e., ν in , ν en ), and gyrofrequencies . To resolve these ionospheric parameters, we could use first-principle ionosphere-thermosphere coupled models that include both the ionospheric and thermospheric processes and their couplings too. The NCAR/HAO Thermosphere Ionosphere Electrodynamics General Circulation Model (TIEGCM) (Richmond et al. 1992;Roble et al. 1988), the University of Michigan Global Ionosphere Thermosphere Model (GITM) (Ridley et al. 2006), and the Coupled Thermosphere Ionosphere Model (CTIM) (Fuller-Rowell and Rees 1980;Rees and Fuller-Rowell 1988) are among the most well-known physics-based global models. Driven by topside precipitation, these models self-consistently determine the spatial and temporal evolution of various characteristics in the 3D ionosphere-thermosphere system. The global distribution of the conductivity can be subsequently calculated based on Equations (4).
Given their high complexity and computational expense, simpler models become better candidates if one prefer to determine the ionospheric conductivity efficiently. Electron transport codes, such as Boltzmann 3-Constituent (B3C) (Strickland et al. 1993) and GLobal Air-gLOW (GLOW) model (Solomon et al. 1988;Solomon 2001;Bailey et al. 2002;Solomon 2017), resolve various physical processes in the ionosphere and output the height-dependent ionization rates. Fang et al. (2008) even parameterized the ionization rates based on different precipitating energies, providing a more efficient approach. Then, together with neutral atmosphere models like NRLMSIS (Picone et al. 2002) that provide neutral composition and temperature, the local conductivity profile can also be found. In this type of method, we note that the ionosphere and thermosphere systems are not self-consistently resolved. The efficiency is gained while the physical self-consistency is lost.
The most simple empirical functions were established by Robinson et al. (1987) and have been widely used in the space physics community, especially in global models due to its easy implementation and high efficiency (e.g. Chen et al. 2015a;Fok et al. 2001;Raeder et al. 2001;Yu et al. 2016Yu et al. , 2017bZhang et al. 2015a). The Robinson formulas relate both the Hall and Pedersen conductance with the energy flux and mean energy of the auroral precipitating electrons as follows: where F E is the energy flux of precipitating electrons (in ergs/cm 2 /s), and < E > is their average energy (in keV). It has been a great success in using these simple and efficient formulas for specifying the conductance and subsequently understanding the magnetosphereionosphere coupling, but they also have shortcomings. The formulas were derived only based on very limited data sets (Vondrak and Robinson 1985), namely, only three satellite passes over the Chatanika radar under a moderate activity level. Such database implies that extrapolation could occur if one considers their usage for very intense storm times (Liemohn 2020). Furthermore, the applicable energy range of the above formulas is between 500 eV and 30 keV. Although the majority of precipitating electrons are below the threshold of 30 keV, energetic electrons above 30 keV often show up in the spectrum (for example when pulsating aurora occurs), and make a non-negligible contribution to the ionization and conductivity at lower altitudes (e.g., E/D region) (Hosokawa and Ogawa 2010;Turunen et al. 2016;Yu et al. 2018). Given these limitations, even though the Robinson formulas have been proven to be a simple yet powerful tool to the space physics community, an update of the relation is required, especially when more ground-based measurements of the ionosphere and low-Earth orbiting (LEO) satellite observations are increasingly available (Liemohn 2020).
In fact, numerous efforts have been recently made to develop new relationships between the conductance and topside forcing into the ionosphere, including particle precipitating flux and high-latitude FACs. For example, Kaeppler et al. (2015) reported the modification to the Robinson formulas after fitting the radar measurement-based conductance by the GLOW model (Solomon et al. 1988;Solomon 2001Solomon , 2017 to infer the associated precipitating flux on the top of the ionosphere. McGranaghan et al. (2016) used empirical orthogonal functions (EOFs) to produce the ionospheric conductance map based on DMSP measurements of particle flux and GLOW model calculation. Robinson et al. (2020) developed linear relations between high-latitude FACs and auroral conductance based on Poker Flat Incoherent Scatter Radar (PFISR) measurements and AMPERE-derived FACs. Mukhopadhyay et al. (2020) updated the empirical method from Ridley et al. (2004) who related the auroral conductance to AMIE-driven FACs. Their newly developed model included a full year results of AMIE, and was extended to encompass extreme storm conditions. Wang and Zou (2022) recently reported a statistical conductance model trained from the ground-based PFISR data and in-situ Swarm observations of FACs, providing another option for the community to specify the high-latitude conductance distribution based on FAC information. Studies also suggested that secondary electrons produced by the interaction between precipitating electrons and the neutral atmosphere can reflect multiple times between hemispheres and further contribute to the diffuse aurora (Khazanov et al. 2016). Therefore, Khazanov et al. (2018) modified the Robinson formulas (Robinson et al. 1987) with a correction factor to account for the contribution made by the secondary electrons.
Overall, many progresses have been made towards better estimating globally the auroral conductance. These models range from a more self-consistent but computational expensive method to a more efficient way with insufficient self-consistency. It is suggested that one is aware of the limitations and advantages of these approaches while applying them. For example, as indicated in Robinson et al. (2020) and Wang and Zou (2022), larger FACs under more disturbed conditions are less sampled, calling for cautions when applying these models to severe storms.

Specification of Electron Precipitation in Global Models
Regardless of their different levels of complexity, the above conductance models, either empirical or physics-based, are all driven by the electron precipitation from the topside ionosphere. In principle, the global distribution of electron precipitation can be specified in two ways: using empirical models derived based on satellite data sets or using physics-based model output.
Unlike the conductivity, the precipitation flux is measurable and thus can be directly obtained from in-situ satellite-borne particle detectors. Satellite missions like the Defense Meteorological Satellite Program (DMSP) and NOAA/POES, can monitor the loss-cone precipitating flux over a large MLT range with multiple spacecraft flying across different MLT sectors. A long-term coverage allows for quasi-global maps of precipitating flux under various solar wind and geomagnetic conditions. Such statistical precipitating flux models were developed decades ago and have been widely used in the community, such as Hardy-Kp model (Hardy et al. 1985(Hardy et al. , 1989, and Ovation Prime model (Newell et al. 2009(Newell et al. , 2014 that separates diffuse, monoenergetic, and broadband aurora precipitation. These empirical precipitating models have the superiority of easy access to the driving conditions (mostly geomagnetic indices or solar wind/IMF conditions) and straightforward implementation to other models.
On the other hand, first-principle magnetospheric models are capable of computing the global distribution of electron precipitation based on physical processes, such as global MHD models and kinetic models. As the MHD models are incapable of resolving kinetic physics of the electron distribution functions, researchers utilized the adiabatic kinetic theory to parameterize the electron precipitation (Knight 1973;Lyons et al. 1979;Fridman and Lemaire 1980). The electron thermal flux j 0 in the magnetospheric source region is determined by electron number density n e and temperature T e , assuming that the electron number density is equal to the proton number density and the electron temperature is about seventh of proton temperature. These proton density and temperature are resolved in the global MHD models based on MHD equations. Zhang et al. (2015a) developed an electron precipitation module within a global geospace circulation model by including both diffuse and mono-energetic precipitation. For the diffusive electron precipitation in their module, the energy flux j E and averaged energy < E > are formulated as follows.
where α is loss cone filling factor allowing for certain adjustment of the electron flux. For mono-energetic auroral precipitation, the energy flux j E and averaged energy < E > are expressed as follows.
where J || is the upward FAC, and R m is the magnetic mirror ratio between B i and B s , the magnetic fields at the ionospheric altitude and source region respectively. V represents the potential drop along magnetic field lines. For diffusive electron precipitation, no potential difference is necessary for electrons to be accelerated toward the upper atmosphere (i.e. V = 0). But for mono-energetic electron precipitation, V = 0 and is dependent on the upward FACs J || . Details of the similar methodology can be found in the literature (e.g., Raeder et al. 2001;Tanaka 2000;Zhang et al. 2015a;Connor et al. 2016;Yu et al. 2016;Mukhopadhyay et al. 2020).
Unlike the global MHD models, global kinetic models are capable of determining the phase space distribution functions and accounting for the kinetic physics, such as collisions or diffusions, along the particle drift paths (Jordanova et al. 1997(Jordanova et al. , 2006(Jordanova et al. , 2010aFok et al. 2014;Chen et al. 2015aChen et al. ,b, 2019Yu et al. 2016Yu et al. , 2018Yu et al. , 2020Perlongo et al. 2017;Grandin et al. 2019). The loss cone flux, i.e., the precipitating flux can be tracked if the pitch angle dimension is resolved in the model. In general, the Fokker-Planck equation is the governing equation, based on which the temporal variation of the phase space density f s of a particle species, s, is solved.
The term on the right-hand side represents loss processes that the particle experiences while drifting around the Earth. Among the most important loss processes for electrons is precipitation loss. The electrons in the plasmasheet or ring current are scattered in pitch angle owing to processes like wave-particle interactions, during which some of the electrons are diffused into the loss cone. Recent kinetic models usually include the whistler-mode wave induced scattering processes using either electron lifetimes (Jordanova et al. 2010a;Yu et al. 2016) or diffusion coefficients Yu et al. 2016Yu et al. , 2018 to represent the electron loss rates. If the pitch angle-dependent phase space density distribution in the equatorial plane f o is known, we can determine the loss cone precipitating flux down to the ionosphere j iono (E) as follows: where p is the electron momentum, α c is the critical pitch angle corresponding to the edge of the loss cone. j lc (E) is the averaged flux in the equator within the loss cone and is equal to the precipitating flux at the ionospheric altitude in light of the Liouville's theorem. Figure 18 shows the RAM-SCB simulated global distribution of precipitating electron flux in the equator after considering the scattering by whistler-mode waves. It clearly demonstrates that the precipitating flux is energy dependent. The low-energy electrons precipitate globally outside L = 4, but those more energetic electrons appear to dominantly precipitate in the dawn-to-dayside sector inside the plasmapause. This is because the hiss waves inside the plasmapause can efficiently resonate with tens of keV electrons while chorus waves outside the plasmapause are more effective in scattering a few to tens of keV electrons. Beside the reason in the diffusion loss rates, more energetic electrons transported from the plasmasheet have larger eastward-drifting velocities and thus experience the scattering loss more towards the dawn and noon sectors.
Using these electron precipitating models as the driver, the global distribution of auroral conductance can be determined and the following chain effects and feedback effects, as illustrated in Fig. 17, can be explored. Many studies have been dedicated to modeling the global ring current dynamics in association with the conductance (e.g., Gkioulidou et al. 2012;Chen et al. 2015a;Yu et al. 2016Yu et al. , 2018. For example, Chen et al. (2015a) used two Fig. 18 Equatorial distribution of (a) precipitating electron flux and (b) pitch angle diffusion coefficients associated with whistler-mode waves at about 6, 50, and 164 keV, obtained with RAM-SCB during the 17 March, 2013 storm period. The black dots mark the plasmapause boundary different ways in specifying the electron loss rates in the kinetic model to obtain the precipitating flux and then produce the conductance from the Robinson formulas. The electron loss rates in their study represent the lifetime of electrons before being lost in the atmosphere. Different assumptions were made in the two loss rate models (Chen et al. 2015b) and comparisons with DMSP observations helped identify the best electron loss rate model. Yu et al. (2016) also calculated the conductance using the Robinson formulas, with the precipitating flux obtained from physics-based methods (either from the global MHD model BATS-R-US (Powell et al. 1999;Tóth et al. 2012) or the kinetic model RAM-SCB (Jordanova et al. 2006(Jordanova et al. , 2010bZaharia et al. 2010;Yu et al. 2012)). Figure 19 demonstrated that the MHD method shows a wide coverage of the aurora precipitation, mostly diffuse aurora, and results in large conductance. In contrast, the kinetic method calculates the loss cone electron flux from wave-particle interactions and uses quasi-linear diffusion coefficients rather than lifetime loss rates. The result shows regional electron precipitation above 60 • in the midnight-to-dawn sector, a region where whistler-mode chorus waves are observed to be active and can diffuse electrons of tens of keV (Meredith et al. 2009;Ni et al. 2016). The auroral conductance in the latter case is much smaller than in the MHD calculation. Yu et al. (2018) made a step further by two-way coupling the kinetically calculated precipitating electron flux from RAM-SCBE model (Yu et al. 2017b) with the ionospheric electron transport model like GLOW (Solomon 2017), to bypass the simple empirical Robinson formulas. The GLOW model demonstrated dynamic variations in the altitudinal profile of conductivity in response to short time scale substorm injections. The injections bring in energetic plasma sources, enhancing the precipitating flux at higher-energies that could penetrate deeper and impact the low-altitude D region ionosphere, as shown in Fig. 20. Following the eastward drift, the energetic tail in the precipitating flux spectrum initially appeared at MLT = 0 at 16:30 UT propagates to MLT = 9 at 17:00 UT (see the indication by the arrow).

Fig. 19
Two different methods of specifying the ionospheric particle precipitation: MHD parameterization and kinetic calculation. Adapted from Yu et al. (2016). From left to right are electric potential , field-aligned currents (FACs), precipitating energy flux, and Hall conductance H The low-altitude ionosphere below 100 km experiences large ionization as these energetic precipitating electrons impact the upper atmosphere. A sub-layer in the Pedersen conductivity profile is subsequently generated, resulting in a double-layer height profile, as seen in Fig. 20 (e). Due to quasi-periodic substorm injections, the low-altitude ionosphere also experiences recurrent enhancement in electron density and conductivity. This may complicate the current structure in the ionosphere as suggested in Hosokawa and Ogawa (2010). Such a two-layer structure would not be revealed if a traditional Maxwellian spectrum is assumed or the empirical conductance model is used because neither the energetic tail nor the height profile can be captured.

Ion Precipitation as Another Important Energy Source
Although precipitating electrons carry the majority of energy deposition down to the auroral zone, ions originated from the magnetosphere cannot be neglected. Statistical studies using many years of NOAA/POES measurements suggest that energetic protons of tens of keV carry a similar amount of precipitating flux, especially in the dusk and night sectors . This means precipitating ions also contribute significantly to the ionospheric conductance in some regions (Galand and Richmond 2001;Zou et al. 2014) and need to be included/recognized. Ions in the plasma sheet are subject to charge exchange, coulomb collision, as well as resonant interactions with plasma waves (Jordanova et al. 1997). The last process diffuses ions in pitch angles and further scatter them into loss cones. The electromagnetic ion cyclotron (EMIC) waves are believed to be one primary driver of the precipitating ions at a few to tens of keV. Many studies have established the relation of EMIC waves with the proton precipitation, proton aurora, or proton isotropy boundary (e.g. Jordanova et al. 2001Jordanova et al. , 2007Nishimura et al. 2014;Yahnina and Yahnin 2014;Fuselier 2004;Chen et al. 2014;Miyoshi et al. 2008;Yuan et al. 2010;Yahnin et al. 2021;Tian et al. 2022). Jordanova et al. (2001) simulated the role of He-band EMIC waves in driving the ring current ions into the loss cone using the RAM-SCB model and calculating the generation of EMIC waves self-consistently Fig. 20 (a, b) The precipitating flux spectrum at different MLTs at 60 • at 16:30 UT and 17:00 UT respectively. (c-f) The simulated height profiles of ionization rate, electron density, Pedersen conductivity, and Hall conductivity at MLT = 6 and MLAT = 60 • . Substorm injections bring in energetic electron sources that propagate eastward and enhance the precipitating flux (as indicated by the arrow). The low-altitude region below 100 km shows enhancements in both the ionization and conductivity. Recurrent substorm injections results in recurrent energization in the low-altitude ionosphere. Adapted from Yu et al. (2018) with the evolving ring current distributions. They found that the global patterns of proton precipitation are very dynamic with the most intense proton precipitation along the duskside plasmapause. Later, Jordanova et al. (2006) reported that the gyroresonant interaction between the EMIC waves and the ring current protons results in a reduction of the total ring current proton energy by ∼ 10% in the storm recovery phase. Shreedevi et al. (2021) included the effects of another wave band, the H-band EMIC wave, to comprehend the scattering processes between the EMIC waves and ring current ions. In their study, a statistical EMIC wave model derived by Saikin et al. (2015) based on many years of Van Allen Probes observations was incorporated. This statistical EMIC wave model was found to facilitate global ion precipitation patterns that are in much better agreement with NOAA/POES observations than other earlier empirical EMIC wave models, such as the one using Combine Release and Radiation Effects (CRRES) satellite measurement (Kersten et al. 2014) or the one deduced from the statistical EMIC wave frequency spectra .
The EMIC wave-driven ion precipitation, as another energy source to the upper atmosphere, also influences the ionosphere-thermosphere dynamic processes. Yuan et al. (2014) reported that the enhancement in the ionospheric TEC observed by a Finland GPS receiver on 8 March 2008 is a result of proton precipitation induced by EMIC waves in the magnetosphere. Their calculations of the ionization rates and electron density in the ionospheric E region indicated that the addition of EMIC wave-induced proton precipitation can lift the electron density by ∼ 2.5 orders of magnitude in the E region, consistent with the increase in the TEC. Tian et al. (2022) simulated an event associated with EMIC wave-driven proton precipitation and found a significant enhancement of the ionospheric conductance by over 10 mhos after incorporating the proton precipitation.
In addition to the EMIC waves, field line curvature scattering (FLC) can also diffuse tens of keV protons chaotically in pitch angles (Sergeev et al. 1983(Sergeev et al. , 1993. When a magnetic field line is highly stretched, its curvature radius decreases. The particle's first adiabatic invariant is violated when its gyroradius is comparable to the field line curvature radius. Nightside magnetic field lines are often in a stretching state and therefore the night magnetosphere is naturally a scattering pool for ring current ions of tens of keV (e.g. Ebihara et al. 2011;Yue et al. 2014;Eshetu et al. 2021) or radiation belt electrons over MeV (Tu et al. 2014). Given its significance in scattering ring current ions and contributing to the ring current decay, many global ring current simulations implemented such a mechanism. For example, Ebihara et al. (2011) quantified the pitch angle diffusion of protons due to the FLC scattering and found more rapid ring current decay when the FLC scattering is included. Chen et al. (2019) found that the FLC scattering around 4 < L < 7, where the magnetic field are less dipolar, causes protons to precipitate in localized regions, resulting in sporadic enhancement in the ionospheric conductance. Yu et al. (2020) also quantitatively investigated the role of FLC scattering in scattering tens of keV ions by comparing with in-situ satellite measurements. It is suggested that this scattering mechanism alone cannot completely explain the proton precipitation that occurred across wide L shells, especially in the inner magnetosphere (L < 4). Zhu et al. (2021) compared the roles played by both EMIC wave scattering and FLC scattering in producing proton precipitation and found that the pitch angle scattering induced by the EMIC waves can predominantly account for the globally observed precipitation flux for L < 5, while the FLC scattering appears to be more important in the outer region. As displayed in Fig. 21, when the EMIC waves are taken into account (panel (d)), the agreement with the NOAA/POES observations of precipitating energetic proton flux inside L = 6.5 is qualitatively better, although overestimation is found in MLT = 9-15 and more studies are still desired.

Summary and Discussion
The magnetosphere-ionosphere coupled system has been long considered as one of the most complicated components within the geospace environment. The convection in the magnetosphere drives the ionospheric convection through the electric field connection, the fieldaligned currents (FACs) flowing parallel/anti-parallel to the magnetic field lines transport energy between the two, and the precipitating electrons/ions of magnetospheric origin is closely related to the auroral physics as well as substorm dynamics. The auroral conduc-tance, together with the FACs, governs the convection pattern in the entire system. That means these elements are intrinsically connected. This article reviewed some recent advances in understanding the magnetosphere-ionosphere system, with the focus primarily on the meso-scale electromagnetic coupling processes, including the transient and localized FACs, subauroral convections, and particle precipitation. These coupling interfaces append extra effects on the large-scale patterns and bring in more dynamic and rich physics. We briefly summarize some advances as follows.
In addition to the classic large-scale pattern of Region-1 and Region-2 FACs as reported in the statistical study of Iijima and Potemra (1978), transient FACs are frequently generated on top of the classic picture. Impulsive responses often occur after the interplanetary shock compresses the dayside magnetosphere. High-latitude ground-based magnetometers on the dayside surface exhibit bipolar variations in the magnetic field perturbations, namely, preliminary impulse (PI) and main impulse (MI). These variations help reveal the existence of two pairs of ionospheric vortices, emerging in sequence and rotating in opposite directions. Such phenomena further indicate that during the PI phase, a pair of Region-2 sense FACs emerges on the dayside across the longitude and then fades. It is followed by a new pair of Region-1 sense FACs in the MI phase. The first pair of Region-2 sense FACs are connected with the magnetopause current, but the second pair of Region-1 sense FACs is connected to the magnetospheric vortices in the equatorial plane. The successive appearance of the FACs can repeatedly continue as the compressional fast-mode wave propagates between the magnetopause and plasmasphere/ionosphere boundaries. This has been proven by the wave-like oscillations from the magnetometer data.
Transient FACs also frequently appear on the nightside, in association with magnetotail dynamics. The famous concept of substorm current wedge (SCW), since its proposal in the early '70s, has been substantially extended. The most notable change is the addition of a second current wedge equatorward of the original one. Beside the Region-1 sense FACs in the original SCW picture, another pair of FACs in Region-2 sense is located closer to the Earth and closed by the high-latitude FACs along the meridian plane in the ionosphere. Recent observations further suggest that the large-scale SCW is actually a net result of several smaller-scale wedgelets, which are produced by localized fast flows from the magnetotail, such as bursty bulk flows (BBFs). Global simulations suggest that as the BBFs penetrate deeper, disturbances are found not only in the tail current but also in the near-Earth ring current, causing current disruption. The tail current disruption converts the cross-tail current to be field-aligned, forming the SCW structure. The westward ring current is also disrupted to be field-aligned in localized regions, creating a few pairs of small-scale but transient FACs. Such localized FAC disturbances brought by the BBFs are recently reported from a multi-point observational study.
Subauroral convections often enhance during geomagnetic active times, bringing in tremendous disturbances in the ionosphere and thermosphere. The subauroral polarization streams (SAPS) have been extensively investigated. Some recent advances consist of revealing their whole-life evolution, proposing new formation mechanisms, discovering their impact on the thermospheric neutral winds, and capturing the SAPS in global geospace circulation models. Despite large improvement in those global models and that some important features of the SAPS are captured, we are also alerted by the discrepancies between the observations and simulations, implying that more exploration needs to be done. Besides the well-studied SAPS, recent studies begin to report some other new types of meso-scale convections, including the DSAIDs that have two fast flow channels and DAPS that appear as an intense flow on the dawnside auroral zone. DSAIDs are thought to be a subset of SAPS and are mostly colocated with two-layer Region-2 FACs in the subauroral region, as indicated by statistical studies. Simulations further revealed that the two-layer Region-2 FACs are related to multiple substorm injections that not only enhance the primary partial ring current but also form a new dynamic partial ring current at larger L-shells. The two ring currents are connected to the ionosphere through two Region-2 FACs on the dusk side, creating the DSAIDs. On the other hand, DAPS is featured by a large eastward flow on the dawn side with a sharp gradient equatorward of it. It is proposed that when the upward Region-2 FACs on the dawnside are enhanced, they carry more electron precipitation and result in larger conductivity than the higher-latitude part of the aurora where the downward Region-1 FACs are located. The higher-latitude part has a smaller conductivity and thus requires a larger equatorward electric field in order to satisfy the current continuity. This larger electric field is responsible for the formation of DAPS.
Conductance, or height-integrated conductivity, as a central element in the magnetosphere-ionosphere circulation, has been widely acknowledged to play an essential role in regulating many kinds of dynamics, such as the tail configuration, tail flows, as well as dayside solar wind-magnetosphere coupling. The auroral conductance is especially dynamic and important in hight-latitude electrodynamics. However, the local conductivity is difficult to measure as it is determined by various variables in both the ionosphere and thermosphere (e.g., electron density, neutral density, temperature, collision frequencies, etc), causing the conductance to be even harder to be quantified. A famous empirical model that relate the auroral conductance to the precipitating flux was derived by Robinson et al. (1987) in the '80s, completely avoiding the complexity in computing these parameters for the conductivity. Other physics-based resolutions of the conductance include first-principle physics-based 3D ionosphere-thermosphere circulation models and ionospheric electron transport codes. These conductance models at different levels of complexity have been applied in geospace circulation models. Furthermore, all these models require the topside ionospheric driversthe particle precipitation. Apart from empirical precipitating models by Hardy et al, MHD parameterizations and kinetic approaches have been developed to determine the particle precipitation to drive the ionosphere-thermosphere system, closing the whole system in a more self-consistent fashion. It is also increasingly noted that in addition to electron precipitation, ion precipitation cannot be neglected as before. Studies found it to be a dominant source in the dusk-to-midnight sector at mid-latitudes, and could directly influence the ionospheric responses locally or globally.
It has been acknowledged that the magnetosphere-ionosphere coupling processes are being much better understood with the aid of growing measurements from spaceborne/ground-based/remote sensing instruments, more developed models, as well as theoretical studies. However, we should also pay attention to various challenges or obstacles on the way toward establishing a more comprehensive and precise picture of the coupled environment and developing a more robust forecasting capability of related space weather effects. To name a few, several challenging issues are discussed below.
1. Within global circulation models, the ionosphere is often treated as an energy reservoir of the magnetosphere system. The ionosphere itself is a complicated system. It is not only coupled with the distant magnetosphere but also the thermosphere overlapped in altitudes. But a region between the topside of the ionosphere and the lower-altitude zone of the magnetosphere (i.e., from hundreds of kilometers to a few Earth radii) is left unresolved, neither by the magnetosphere model nor by the ionosphere-thermosphere model. It is thus called the "gap" region. This region in fact contains a range of physics, such as plasma outflow, Alfvénic waves, field-aligned currents, particle precipitation or acceleration, transverse heating, and so on. These physical processes bridge the ionosphere and Fig. 22 The PWOM result after imposing a cusp region on top of the typical two-cell convection pattern. Top three panels show the global distribution of oxygen ion and proton flux and their ratio at 6000 km. Bottom three panels show the meridian distribution of oxygen ion velocity, proton velocity, and electron temperature through the cusp region (Glocer et al. 2018) magnetosphere by transferring energy in and out and are hence of importance for achieving a more comprehensive and self-consistent understanding of the system. Glocer et al. (2009) implemented a first-principle model called Polar Wind Outflow Model (PWOM) into global circulation models, providing the plasma source of ionospheric origin to the magnetosphere as well as receiving the feedback effects from the magnetosphere. Glocer et al. (2018) further incorporated kinetic effects within the PWOM to account for waveparticle interactions at higher altitudes, allowing the kinetic outflow solution to be used as the mass provider to the magnetosphere from the inner boundary. Figure 22 shows the PWOM results of the global distributions of ion fluxes at 6000 km and the cusp outflow velocity and temperature, driven by the wave-particle interactions and monoenergetic precipitating electrons above the cusp. There is a clear enhancement of O + flux at the cusp region, constituted as a major ion source over there. The altitudinal distribution of the outflow velocity over the cusp indicates an enhanced upwelling of the ionospheric ions, providing a mass source to the magnetosphere. Clearly, these temporal and spatial characteristics of the ions in the "gap" region are bound to influence the magnetospheric dynamics. Therefore, incorporating the "gap" region physics in global circulation models is highly demanded. Similar coupling efforts have been carried out in Varney et al. (2016). 2. Another challenge is to understand the multi-scale dynamics in the ionospheremagnetosphere system and their roles in various cross-regional/cross-energy couplings. For instance, wave-particle interactions are represented by microscopic physics as the excitation and damping of the waves are usually within milliseconds. But meanwhile, they could influence the macro-scale dynamics like the ring current decay, radiation belt energization, and auroral precipitation that last for hours. Some studies have made important progress in incorporating micro-scale wave dynamics, such as EMIC waves, with macro-scale models (e.g., Jordanova et al. 1997;Khazanov et al. 2003;Gamayunov et al. 2014;Bortnik et al. 2011;Fu et al. 2016). More studies are still needed to selfconsistently understand the excitation of various types of magnetospheric plasma waves and their impact on the global dynamics. Recently, Jordanova et al. (2018) developed an end-to-end model of the magnetosphere, the SHIELDS framework, by integrating several physics-based models/techniques across multiple scales. This framework aims to address various fundamental problems across different regimes, ranging from the large-scale global magnetosphere to meso-scale inner magnetosphere and small-scale kinetic zone. Such a framework significantly strengthens the capability of representing the complicated geospace environment and allows for a more complete investigation of the system, including particle injections associated with substorms, plasma accelerations, as well as spacecraft surface charging environment that could impose severe threats to the onboard satellite instrument in the magnetosphere. A system-level consideration of the geospace system by integrating as many processes as possibles in different temporal or spatial scales will be a necessary tool in the future, not only to achieve a comprehensive understanding of the entire system, but also for an operational purpose in space weather application. 3. A third challenge is that field-aligned electric potential drop has long been ignored in global models. As the frozen-in condition is generally satisfied in the magnetosphere, equipotential along magnetic field lines is a common assumption. However, this is not realistic or appropriate for regions where mono-energetic precipitation occurs, such as the auroral zone and dayside cusp. The most significant potential drop appears in regions of upward field-aligned current where electromagnetic energy is converted to plasma energy and dissipation loads in the ionosphere. With the simplified potential distribution in the magnetosphere-ionosphere coupling model, the dissipative effects may be underrepresented in the coupled system. For example, Xi et al. (2016) initiated a step forward by including a electric potential drop into global circulation models and investigated the subsequent effects of the dissipative load. They reported a remarkable difference in the tail reconnection. The nightside X-line moves closer to Earth with the reconnection rate raised by 12%. Furthermore, the ionospheric joule dissipation is reduced by 8%. Their study demonstrated the important role of the potential drop, even only in the auroral zone, in regulating the large-scale dynamics. Future magnetosphere-ionosphere modeling studies should consider this long-neglected field-aligned physics, especially in the auroral and cusp regions. Yet, revealing how the electric potential drop distributes on the global scale and how it varies in time is a near-future task.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.