This paper investigates the effects of magnetospheric mass loading on the control of dayside magnetic reconnection using global magnetospheric simulations. The study iys motivated by a recent debate on whether the integrated dayside magnetic reconnection rate is solely controlled by local processes (local-control theory) or global merging processes (global-control theory). The local-control theory suggests that the integrated dayside reconnection rate is controlled by the local plasma parameters. The global-control theory argues that the integrated rate is determined by the net force acting on the flow in the magnetosheath rather than the local microphysics. Controlled numerical simulations using idealized ionospheric outflow specifications suggest a possible mixed-control theory, that is, (1) a small amount of mass loading at the dayside magnetopause only redistributes local reconnection rate without a significant change in the integrated reconnection rate and (2) a large amount of mass loading reduces both local reconnection rates and the integrated reconnection rate on the dayside. The transition between global-control- and local-control-dominated regimes depends on (but not limited to) the source region, the amount, the location, and the spatial extension of the mass loading at the dayside magnetopause.
- Dayside reconnection may be controlled by mixed local/global processes
- A small amount of mass loading only redistributes the local reconnection rate
- A large amount of mass loading reduces both the local and the integrated reconnection rate
Dayside magnetic reconnection at the magnetopause is a key process in the interaction between the solar wind and magnetosphere-ionosphere system, which drives the circulation of magnetospheric plasmas and regulates the transport of electromagnetic energy. Thus, a clear understanding of what controls dayside reconnection is crucial to both magnetospheric physics and space weather forecasting.
Rlocal is a function of the plasma mass density (ρs, ρm) and the magnetic fields (Bs,Bm) in the magnetosheath (s) and in the magnetosphere (m). While the properties in the magnetosheath are affected by the solar wind, it was argued in Borovsky et al. that the dayside reconnection does not significantly impact the magnetosheath flow pattern. As a consequence, reconnection should not significantly alter the local parameters on either side of the reconnection site that control the local reconnection rate, and therefore, magnetic reconnection on the dayside magnetopause is purely a local phenomenon. This theory is called the “local-control theory.” In this scenario, each local “reconnection zone” acts independently, and changes in one reconnection zone do not affect its neighbors.
The two theories are not necessarily inconsistent with each other since the local reconnection rate may be governed by the local conditions in the sheath and in the magnetosphere, which clearly depend on the upstream solar wind conditions. However, the two theories diverge on predicting the response of the integrated dayside reconnection rate when the local reconnection rate is altered through mass loading via ionospheric-/plasmaspheric-sourced ions. With localized mass loading, the local-control theory predicts that the reduction in the local reconnection rate will result in a decrease in the integrated reconnection rate. In contrast, the global-control theory predicts that although the local rate of magnetic reconnection is decreased at locations with mass loading, the amount of solar wind flux transported across the dayside X line is unchanged. As a consequence, the integrated dayside reconnection rate will remain unmodified. The questions therefore arise, Which theory is correct? Does mass loading change the integrated dayside reconnection rate?
Mass loading processes have been shown to be important in the interaction between solar wind and magnetosphere, especially during storm times. Results from the IMAGE satellite have shown that plasmaspheric drainage plumes occur during times of enhanced convection whereby the outer region of cold, dense plasmasphere is eroded and transported sunward to the dayside magnetopause [e.g., Chandler and Moore, 2003; Spasojević et al., 2003]. Observations show that the mass loading from very cold plasma drawn from the plasmasphere are by light ions H+ and He+, with number densities that can be greater than 50 cm−3 [Chandler and Moore, 2003]. Recent analysis of Time History of Events and Macroscale Interactions during Substorms data provides evidence that these plumes mass load the dayside magnetosphere reducing the Alfvén speed inside the magnetopause resulting in a decrease in the local dayside reconnection rate [Walsh et al., 2014]. During enhanced geomagnetic activity, ionospheric O+ outflow becomes an important source of magnetospheric plasma [e.g., Chappell et al., 1987; Nosé et al., 2003]. Wang et al. [2015a, 2015b] used observational data from the Cluster spacecraft to demonstrate that despite the increased Larmor radii for O+ ions compared to H+, magnetospheric O+ ions almost fully participate in reconnection exhaust flows and therefore do affect the Alfvén speed and local reconnection rate.
Due to the fact that satellite observations provide only a local determination of reconnection properties at the magnetopause, it is not possible for current satellite observations to determine whether this local change affects the integrated rate. Multifluid global magnetohydrodynamics (MHD) simulations are useful tools to investigate the response of dayside reconnection to magnetospheric mass loading processes. The response of the local reconnection rate, the integrated reconnection rate and plasma properties, and forces in the sheath due to localized mass loading can be investigated to determine which theory best describes the control of dayside reconnection. In global simulations, controlled changes to the local reconnection rate can be made through the addition of mass loading, and the response of the dayside solar wind-magnetosphere (SW-M) interaction can be evaluated. Previous studies showed that when ionospheric outflow-related mass loading is included in global MHD simulations, reductions of ionospheric cross polar cap potential (CPCP) and modifications of magnetospheric reconnection are observed [e.g., Winglee et al., 2002; Glocer et al., 2009a, 2009b; Brambles et al., 2010; Wiltberger et al., 2010; Welling and Zaharia, 2012; Brambles et al., 2011; Yu and Ridley, 2013; Zhang et al., 2015]. However, the mechanism(s) that causes the changes in magnetospheric reconnection and ionospheric CPCP has yet to be adequately resolved.
This paper explores how well the local- and global-control theories predict the response of the dayside reconnection rate to ionospheric-sourced mass loading using idealized, controlled global MHD simulations. Section 2 details the global model, simulation considerations, and the idealized ionospheric outflow specifications. Section 3 describes the effect of mass loading on the dayside reconnection in the simulation and compares the results to the predictions from the local-control and global-control theories. Section 4 summarizes the results.
2 Simulation Information
The Lyon-Fedder-Mobarry (LFM) global MHD model has been used extensively to study the solar wind-magnetosphere-ionosphere interaction. The numerical methods of the LFM model are described by Lyon et al.  and Merkin and Lyon . The multifluid adaption of the LFM code can solve for multiple ion fluids allowing distinct ionospheric-sourced ion fluids to be tracked. The finite volume techniques allow the high-resolution multi-fluid LFM model to complete the calculation on a nonorthogonal grid that is adapted to magnetosheath studies such as dayside reconnection. Reconnection in the code occurs via numerical resistivity [Lyon et al., 2004], which is switched on when the scales of magnetic gradient reach the grid size. The computational grid is approximately 0.125 × 0.2 × 0.2 RE (radial × azimuthal × meridional) near the dayside magnetopause. The reconnection electric field is typically of the order of 0.1vABin, where vA and Bin are the inflow Alfvén speed and magnetic field strength, respectively, indicating that simulated reconnection in LFM is Petschek like [Ouellette et al., 2010, 2013]. Fedder et al.  have also shown that the rate of reconnection in the magnetosphere is controlled by the solar wind conditions and magnetosphere-ionosphere coupling rather than the simulation cell sizes.
To simplify the analysis, all the simulations in this paper use the same idealized solar wind (SW) and interplanetary magnetic field (IMF) conditions. In the simulations, after the magnetosphere is preconditioned for 4 h (IMF Bz=−5 nT between 00:00 and 02:00 and Bz=+5 nT between 02:00 and 04:00 simulation time, ST), IMF Bz is set to be −5 nT for 12 h (04:00–16:00 ST). The IMF Bx and By components are set to 0, and the SW Vx=−400 km/s, Vy=Vz=0. The SW number density and sound speed are set to be 5 cm−3 and 40 km/s, respectively. The dipole tilt is set to 0 in order to remove hemispheric asymmetries, and the ionospheric Pedersen conductance is set to be spatially uniform at 5 mho in both hemispheres in order to remove possible dawn-dusk asymmetries in the magnetospheric reconnection caused by ionospheric electrodynamics [Zhang et al., 2012; Lotko et al., 2014].
The impacts of magnetospheric mass loading on dayside reconnection are investigated using two sets of simulations with idealized ionospheric outflow. The fidelity of the ionospheric outflow model is not important to this study because the outflow specifications are only used to introduce additional mass to the magnetosphere for the 3-D numerical reconnection experiments. The effects of mass loading that originated from plasmaspheric plumes are investigated in a separate study. In the first set, starting from 06:00 ST when the system is in a steady magnetosphere convection state, outflow fluxes are introduced at the low-altitude computational boundary surface (2 RE geocentric) between 58° and 68° magnetic latitude (MLAT), 1930–2130 magnetic local time (MLT) on the duskside and 0230–0430 MLT on the dawnside. In this set of simulations, seven runs with different outflow ion atomic mass (mi=2,4,8,16,32,64,128 AMU) are performed, while the parallel number flux (niv∥i), the parallel kinetic energy flux ( ), and the parallel thermal energy flux ( ) of the outflow introduced at the low-altitude boundary surface are kept the same. This outflow scaling method is developed to minimize the influences from various solar wind-magnetosphere-ionosphere coupling processes other than mass loading. The purpose of using a nonphysical atomic mass of the outflow is to investigate the effects of magnetospheric mass on dayside reconnection rather than represent realistic ionospheric outflow populations. For the oxygen outflow run (mi=16), the outflow number density ni is set to be 102 cm−3, the parallel velocity v∥i is set to be 40 km/s, and the sound speed ci is set to be 20 km/s, which gives a parallel number flux of 4 × 108 cm−2s−1 and a constant hemispheric outflow rate of 0.8 × 1026 s−1. This set of test simulations is named as “both-side runs” in the following sections. In the second set of test simulations, the outflow fluxes are only introduced on the duskside between 58° and 68° MLAT and at 1930–2130 MLT, with the same parallel number flux of 4 × 108 cm−2s−1 and a constant hemispheric outflow rate of 0.4 × 1026 s−1. This set of simulations is named as “duskside runs” in the following sections. The spatial distributions of the outflow number flux mapped to 100 km altitude are shown in Figure 1 (left column). Note that the corotation electric field is set to 0 in both sets of simulations in order to maintain the dawn-dusk symmetry of the simulated mass loading.
The two sets of simulations are designed to isolate the effects of magnetospheric mass on dayside reconnection without changing the global configurations of the dayside magnetosphere significantly. As shown in Figure 1 (left column), in the 14 simulations, the shape of the dayside magnetopause remains approximately unchanged (with a maximum difference around 0.18RE near 1000 MLT). The unmodified magnetopause locations indicate that the direct impact of outflow dynamic or thermal pressure on the dayside SW-M interaction is minimal. Figure 1 (middle column) shows the average outflow number density (calculated between 08:00 and 16:00 ST) in the equatorial magnetosphere a distance of 0.5 RE from the magnetopause. The distribution of density indicates the location of mass loading. In the both-side runs, similar peak number densities occur near 1000 and 1400 MLT, with magnitudes around 1.4 ions/cm−3. In the duskside runs, only one peak occurs near 1400 MLT with similar magnitude as in the both-side runs, indicating that mass loading only occurs on the duskside magnetopause. Figure 1 (right column) shows the temporal evolutions of the simulated Dst indices. In a more realistic outflow simulation, ionospheric outflow contributes significantly to the inner magnetospheric dynamics [e.g., Welling and Ridley, 2010; Yu and Ridley, 2013; Welling et al., 2015]. However, in these test simulations, the outflow is introduced in the closed field line region of the nightside magnetosphere (equatorward of the open-closed boundaries shown in Figure 1 (left column)), and the cold outflow ions populate the inner magnetosphere (approximately inside 7 RE) without being heated in the plasma sheet. Thus, without a drift kinetic ring current model, this cold outflow fluid contributes little to the total pressure of the inner magnetosphere. The Dst indices suggest that the impact of the ring current on the dayside SW-M coupling is also minimal in these simulations.
These idealized simulations with different outflow mass may be used to test the local-control and global-control theories based on the response of the simulated dayside reconnection. If dayside reconnection is purely controlled by local processes (local control), in the both-side runs, the local reconnection rate will decrease at the locations where mass is loaded, as will the integrated reconnection rate as described by equation 2. In the duskside runs, the reconnection rate will show similar decreases on the duskside compared to the corresponding both-side runs and will remain approximately unmodified on the dawnside. If the dayside reconnection rates are purely controlled by the upstream solar wind merging (global control), in all the mass loading simulations, the local reconnection rate may be modified by outflow, but the integrated dayside reconnection rate will remain approximately the same.
3 Simulation Results
Figure 2 shows the MLT distributions of average dayside reconnection rates (08:00–12:00 ST) derived from the two sets of simulations, together with the corresponding distributions predicted by the C-S formula (equation 2). The simulation data are recorded at a 1 min cadence, and the reconnection rate is averaged from the 241 instantaneous snapshots from 08:00 to 12:00 ST. The parameters used in the C-S formula are calculated from the reconnection inflow regions, which are 0.5 RE away from the dayside X line. Since the dipole tilt, IMF Bx and By, SW Vy, and Vz are set to 0 in the simulations, the dayside X line is simply determined using the Bz = 0 contour at the equatorial plane. For figure clarity, only the results from four runs with mi = 2, 8, 32, 128 are shown in Figure 2.
Figure 2a shows the MLT distributions of dayside reconnection rate derived from the both-side runs. The corresponding predictions of dayside reconnection rate from the C-S formula are shown in Figure 2b. In the simulation with mi=2, the distribution of dayside reconnection rate is symmetric about 1200 MLT, with peak rates around 1100 and 1300 MLT. When the ion mass is increased to mi=8, the reconnection rates are reduced by approximately 5% between 1000 and 1400 MLT and enhanced by approximately 5% between 0920–1000 and 1400–1440 MLT. In the mi=32 and mi=128 runs, the dayside reconnection rates are further reduced at all MLT locations and no enhancement occurs in the simulation compared to the mi=2 run. This reduction of dayside reconnection rate with increasing mi is also evident in the C-S predictions shown in Figure 2b, although the spatial distributions between the simulation and theoretical predictions are different. The differences are likely due to the effects of magnetosheath flow shearing which is not included in the C-S formula. The decrease of magnetopause reconnection rate indicates a local response to the mass loading at the magnetopause; that is, the dayside reconnection is not purely controlled by the upstream solar wind.
Figure 2c shows the distributions of average dayside reconnection rate derived from the duskside runs. In the simulation with mi=2, the distribution of dayside reconnection rate shows a dawn-dusk asymmetry with a peak rate near 1320 MLT, while the integrated dayside reconnection rates on the dawnside (0800–1200 MLT) and duskside (1200–1600 MLT) are approximately the same (80 kV versus 81 kV). This asymmetry in the distribution of dayside reconnection rate suggests that the coupled system is redistributing the reconnection rate without changing the integrated rates. In the mi=8 run, the reconnection rate on the duskside is decreased by ≈10% between 1250 and 1350 MLT but enhanced by ≈15% between 1350 and 1450 MLT compared to the mi = 2 run, indicating a similar redistribution of reconnection rate on the duskside. On the dawnside, the reconnection rate is increased by ≈5% between 1000 and 1100 MLT suggesting that more flux is reconnected compared to the mi = 2 case. Similar behaviors are also seen in the run with mi = 32. In the mi = 128 run, the duskside reconnection rates are reduced at most MLT locations between 1200 and 1600 MLT, and a significant enhancement of reconnection rate occurs on the dawnside between 0840 and 1040 MLT. Figure 2d shows similar reductions of duskside reconnection rate with increasing ion mass predicted by the C-S formula. The enhancements of reconnection rate on the dawnside are also evident in the C-S prediction. These enhancements of dawnside reconnection rate suggest a nonlocal response to the duskside mass loading given that the mass loading only occurs at the duskside magnetosphere. If the control of dayside reconnection were purely local, one would expect to see (1) a monotonic decrease of duskside reconnection rate with increasing mi and (2) unmodified dawnside reconnection rate with all mi. However, the duskside runs show that while mass loading is important to the local reconnection rates on the duskside, the system is capable of redistributing the reconnection rates, at least to some extent, in the regions without mass loading.
Further comparisons between the simulation pairs with the same mi in Figures 2a and 2c suggest that the response of the dayside reconnection rate to the local mass loading process is not only determined by local mass loading. For example, in the two runs with mi=128, even though the spatial distribution of mass loading is similar on the duskside as shown in Figure 1 (the peak number density in the duskside run is about 0.05 cm−3 higher than that in the both-side runs), the reconnection rate on the duskside is approximately 10% higher in the duskside run than that in the both-side run. Moreover, the distribution of reconnection rate in duskside run has a plateau region near 1300 MLT, while the both-side run shows a monotonic decreasing of reconnection rate from 1200 to 1600 MLT. The difference between the two runs with mi = 128 is mainly a consequence of different magnetosheath properties. For example, in the duskside run with mi = 128, the average inflow Alfvén speed in the magnetosheath is approximately 4% larger on the duskside between 1200 and 1300 MLT and 11% larger on the dawnside between 0900 and 1100 MLT. This difference suggests that the changes in the local reconnection rate may impact magnetosheath properties. Whether the change of magnetosheath property is a direct local process or a system-level response to mass loading is unknown, and further investigations are needed.
Figure 3 shows the variations of the dawnside and duskside reconnection potentials derived from the two sets of simulations. In order to remove the uncertainties associated with the Kelvin-Helmholtz instabilities near 0600 and 1800 MLT, the dawnside reconnection potential is integrated from 0800 to 1200 MLT and the duskside reconnection potential is integrated from 1200 to 1600 MLT. In the both-side runs, the variations of dawnside and duskside reconnection potentials are approximately the same. For the simulations with mi=16,32,64,128, the reconnection potential decreases with mi. However, for the runs with mi=2,4, the reconnection potentials on the dawnside and duskside remain approximately the same around 81 kV. The integrated dayside reconnection potentials derived from the both-side runs suggest that when mi≤4, the integrated dayside reconnection potential is not significantly modified due to the redistribution of local reconnection rate as shown in Figure 2a. The physical process associated with the redistribution of the dayside reconnection rate is unclear and requires further investigation, which is out of the scope of this paper.
In the duskside runs with mi=2,4, the response of the duskside reconnection potential is approximately the same as those calculated from the both-side runs, with differences less than 1%. However, in the mi=16,32,64,128 runs, the duskside reconnection potential is approximately 5–20% higher than those derived from the corresponding both-side runs, which is consistent with Figures 2a and 2c. On the dawnside, the reconnection potential increases with mi. The enhancement of dawnside reconnection potential is a consequence of enhanced local reconnection rate between 1000 and 1100 MLT as shown in Figure 2c, suggesting that more flux is reconnected on the dawnside without mass loading. This enhanced dawnside reconnection potential is not expected from the local-control theory but is consistent with the global-control theory to some extent. On the other hand, the decrease of total reconnection potential with increasing mi is not consistent with the global-control theory but is expected from the local-control theory. The two sets of simulations show that dayside reconnection may not be exclusively explained by either the local-control or the global-control theory. Rather, the simulation results suggest the possibility of a mixed scenario: when the mass loading exceeds a threshold, the total dayside reconnection rate starts to be dominated by local reconnection processes rather than global SW merging. Further investigations are needed to quantitatively determine the threshold(s), including (but not limited to) the amount, the location, and the spatial extension of the mass loading at the dayside magnetopause.
In this study, we investigate the local- and global-control theories of the dayside reconnection through ionospheric-sourced mass loading using controlled multifluid global simulations. The purpose of using ionospheric outflow specifications in the global model is to introduce approximately the same number density of magnetospheric plasma at the dayside magnetopause. The controlled simulations focus on the role of magnetospheric mass loading by possibly minimizing the effects of direct outflow dynamic/thermal pressure, outflow-enhanced inner magnetosphere pressure/ring current, and ionospheric conductance on the dayside reconnection. The simulation results suggest that the dayside reconnection may be controlled by both local and global processes. With a small amount of mass loading (e.g., quiet time with effective Mi<8), the SW-M system may redistribute the spatial distribution of dayside reconnection rate so that the integrated reconnection rate remains unchanged. While the amount of mass loading is large (e.g., with dense plumes measured by Chandler and Moore ), the local changes of dayside reconnection may cause a system-level reconfiguration of the coupled SW-M and may control the integrated reconnection rate. The quantitative description of the transition between global-control and local-control regimes depends on (but not limited to) the source region (ionospheric and plasmaspheric), the amount, the location, and the spatial extension of the mass loading at the dayside magnetopause.
The research was supported by the following projects: NASA grants NNX11AO59G and NNX11AJ10G and NSFAGS-1404599. Computing resources were provided by the CISL at the National Center for Atmospheric Research (NCAR) under project UDRT0006. NCAR is sponsored by the NSF. Simulation data are being preserved on the NCAR High Performance Storage System (HPSS) and are available upon request.
- 2008), What determines the reconnection rate at the dayside magnetosphere?, J. Geophys. Res., 113, A07210, doi:10.1029/2007JA012645.
- 2013), Physical improvements to the solar wind reconnection control function for the Earth's magnetosphere, J. Geophys. Res. Space Physics, 118, 2113–2121, doi:10.1002/jgra.50110.
- 2010), Effects of causally driven cusp O+ outflow on the storm time magnetosphere-ionosphere system using a multifluid global simulation, J. Geophys. Res, 115, A00J04, doi:10.1029/2010JA015469.
- 2011), Magnetosphere sawtooth oscillations induced by ionospheric outflow, Science, 332(6034), 1183–1186, doi:10.1126/science.1202869.
- 2013), The effects of ionospheric outflow on ICME and SIR driven sawtooth events, J. Geophys. Res. Space Physics, 118, 6026–6041, doi:10.1002/jgra.50522.
- 2007), Scaling of asymmetric magnetic reconnection: General theory and collisional simulations, Phys. Plasmas, 14, 102–114, doi:10.1063/1.2795630.
- 2003), Observations of the geopause at the equatorial magnetopause: Density and temperature, Geophys. Res. Lett., 30(16), 1869, doi:10.1029/2003GL017611.
- 1987), The ionosphere as a fully adequate source of plasma for the Earth's magnetosphere, J. Geophys. Res., 92(A6), 5896–5910, doi:10.1029/JA092iA06p05896.
- 1995), Global numerical simulation of the growth phase and expansion onset for a substorm observed by Viking, J. Geophys. Res., 100, 19,083–19,094.
- 2009a), Modeling ionospheric outflows and their impact on the magnetosphere, initial results, J. Geophys. Res., 114, A05216, doi:10.1029/2009JA014053.
- 2009b), Multifluid block-adaptive-tree solar wind Roe-type upwind scheme: Magnetospheric composition and dynamics during geomagnetic storms-Initial results, J. Geophys. Res., 114, A12203, doi:10.1029/2009JA014418.
- 2010), Role of magnetosheath force balance in regulating the dayside reconnection potential, J. Geophys. Res., 115, A12216, doi:10.1029/2009JA014597.
- 2014), Ionospheric control of magnetotail reconnection, Science, 345(6193), 184–187, doi:10.1126/science.1252907.
- 2004), The Lyon-Fedder-Mobarry (LFM) global MHD magnetospheric simulation code, J. Atmos. Sol. Terr. Phys., 66(15–16), 1333–1350, doi:10.1016/j.jastp.2004.03.020.
- 2010), Effects of the low-latitude ionospheric boundary condition on the global magnetosphere, J. Geophys. Res., 115(A10), 1–15, doi:10.1029/2010JA015461.
- 2003), Change of the plasma sheet ion composition during magnetic storm development observed by the Geotail spacecraft, J. Geophys. Res., 108(A5), 1201, doi:10.1029/2002JA009660.
- 2010), Magnetic reconnection at the dayside magnetopause in global Lyon-Fedder-Mobarry simulations, J. Geophys. Res., 115, A08222, doi:10.1029/2009JA014886.
- 2013), Properties of outflow-driven sawtooth substorms, J. Geophys. Res. Space Physics, 118, 3223–3232, doi:10.1002/jgra.50309.
- 2003), Global response of the plasmasphere to a geomagnetic disturbance, J. Geophys. Res., 108(A9), 1340, doi:10.1029/2003JA009987.
- 2014), Simultaneous ground- and space-based observations of the plasmaspheric plume and reconnection, Science, 343(6175), 1122–1125, doi:10.1126/science.1247212.
- 2015a), Hot magnetospheric O+ and cold ion behavior in magnetopause reconnection: Cluster observations, J. Geophys. Res. Space Physics, 119, 9601–9623, doi:10.1002/2014JA020402.
- 2015b), Dependence of the dayside magnetopause reconnection rate on local conditions, J. Geophys. Res. Space Physics, 120, 6386–6408, doi:10.1002/2015JA021524.
- 2010), Exploring sources of magnetospheric plasma using multispecies MHD, J. Geophys. Res., 115, A04201, doi:10.1029/2009JA014596.
- 2012), Ionospheric outflow and cross polar cap potential: What is the role of magnetospheric inflation?, Geophys. Res. Lett., 39, L23101, doi:10.1029/2012GL054228.
- 2015), The two-way relationship between ionospheric outflow and the ring current, J. Geophys. Res. Space Physics, 120, 4338–4353, doi:10.1002/2015JA021231.
- 2010), Influence of cusp O+ outflow on magnetotail dynamics in a multifluid MHD model of the magnetosphere, J. Geophys. Res., 115, A00J05, doi:10.1029/2010JA015579.
- 2002), Global impact of ionospheric outflows on the dynamics of the magnetosphere and cross-polar cap potential, J. Geophys. Res., 107(A9), 1237, doi:10.1029/2001JA000214.
- 2013), Exploring the influence of ionospheric O+ outflow on magnetospheric dynamics: Dependence on the source location, J. Geophys. Res. Space Physics, 118, 1711–1722, doi:10.1029/2012JA018411.
- 2012), Magnetotail origins of auroral Alfvénic power, J. Geophys. Res., 117, A09205, doi:10.1029/2012JA017680.
- 2015), Role of ionospheric O+ ions on the generation of earthward plasmoids, J. Geophys. Res., 120, doi:10.1002/2015JA021667.