Dual-lobe reconnection and cusp-aligned auroral arcs

,

1 Introduction Milan et al. (2020) recently proposed that dual-lobe reconnection (DLR), occurring during periods of prolonged northwards interplanetary magnetic field (IMF), can close significant amounts of the previously open magnetic flux of the magnetosphere.This newly-closed flux is distributed to high latitudes in the dawn and dusk sectors, producing the horse-collar auroras (HCA) morphology and giving the polar cap a teardrop shape (e.g., Murphree et al., 1982;Hones Jr et al., 1989;Elphinstone et al., 1993).In this paper we discuss what happens if this process continues until the magnetosphere approaches total closure and the role of DLR in producing the cusp-aligned auroral arcs that appear across the polar regions during northwards-directed IMF.
Dungey's open model of the magnetosphere (Dungey, 1961), invoking magnetic reconnection at the low latitude dayside magnetopause and in the magnetotail (Figure 1a), has been extremely successful for explaining many aspects of the structure and dynamics of the magnetosphere, including the excitation of magnetospheric and ionospheric convection (the typical twin-cell convection pattern), the dependence of geomagnetic activity on the orientation and strength of the IMF (Fairfield & Cahill Jr, 1966;Fairfield, 1967), the existence of an extended magnetotail (Wolfe et al., 1967;Ness et al., 1967) and regions near the poles largely devoid of auroras (the polar caps).The extension of Dungey's picture to include time-dependent dayside and nightside reconnection rates, the expanding/contracting polar cap model (Cowley & Lockwood, 1992), has allowed it to incorporate phenomena such as substorms (Lockwood & Cowley, 1992;Milan et al., 2007) and geomagnetic storms (Milan et al., 2009).Day-and nightside reconnection rates can also be used to assess the open flux content and length of the magnetotail (Milan, 2004).
During northwards IMF (or NBZ) conditions distorted convection, which can include sunward flows at noon, is observed within the polar ionosphere (e.g., Reiff & Burch, 1985;Huang et al., 2000;Chisham et al., 2004), driven by the occurrence of reconnection between the IMF and the magnetic field of the magnetospheric lobes, tailward of the cusps (Dungey, 1963;Cowley, 1981).If there is a significant IMF B Y component then single lobe reconnection (SLR) occurs, that is an individual IMF field line reconnects in the northern or southern hemisphere only (Figure 1b).In this case an asymmetric pattern of flows is observed, the sunward flow having a significant dawnwards or duskwards component depending on the polarity of B Y , produced by magnetic tension forces on the newly-reconnected field lines.Usually, SLR does not change the open magnetic flux content of the magnetosphere, occurring with an open lobe field line and resulting in a new open field line.On the other hand, if B Y ≈ 0 then dual-lobe reconnection can occur, in which the same IMF field line reconnects in both hemispheres, producing closed field lines (Figure 1c).In this case a relatively symmetrical pattern of ionospheric flows is expected, as discussed by Milan et al. (2020).
Under prolonged NBZ conditions the magnetotail can become unusually short (Fairfield et al., 1996) and display atypical structure and dynamics (e.g., Huang et al., 2002), suggesting that the magnetosphere is entirely closed.In this paper we discuss the possibility that this closure is caused by prolonged DLR.The role of DLR in producing partial flux closure has been investigated (Imber et al., 2006(Imber et al., , 2007;;Marcucci et al., 2008;Milan et al., 2020), but the process is still poorly understood.However, it has been asserted on the basis of magnetospheric simulations, that in the case of purely northward IMF DLR can close the magnetosphere entirely (Figure 1e) and magnetospheric circulation goes into reverse, with sunward ionospheric flows at noon and antisunward return flows at lower latitudes (Song et al., 1999;Siscoe et al., 2011).On the other hand, if SLR occurs in a closed magnetosphere, then flux will be reopened (Figure 1d).In this paper we discuss flux closure by DLR and the complicated dynamics that are driven when the magnetosphere is (nearly) closed.
NBZ auroral features can include a cusp spot (Milan et al., 2000b;Frey et al., 2002;Carter et al., 2020), high latitude detached arcs or HiLDAs (Frey, 2007;Carter et al., 2018;Han et al., 2020), and transpolar arcs (e.g., Frank et al., 1982;Kullen et al., 2002;Cumnock et al., 2002;Cumnock & Blomberg, 2004;Milan et al., 2005;Fear et al., 2014;Carter et al., 2017).Under purely NBZ conditions (B Y ≈ 0), when DLR is expected to occur, the polar regions can become filled with multiple sun-aligned arcs, also known as cuspaligned arcs due to their characteristic pattern (Y.Zhang et al., 2016), which are as yet poorly understood (e.g., Q.-H.Zhang et al., 2020).As they form under similar conditions, we might expect that horse-collar auroras and cusp-aligned arcs are in some way related.In this paper we present observations of auroras, energetic particle precipitation, ionospheric flows, and field-aligned currents that support the suggestion that closure of the magnetosphere can occur through DLR, producing HCA, and that this is also related to the formation of cusp-aligned arcs.

Observations
We employ observations of the auroras, ionospheric flows, and precipitating particles in both northern and southern hemispheres (NH and SH) from the Defense Meteorological Satellite Program (DMSP) F16, F17, and F18 satellites in sun-synchronous orbits near an altitude of 850 km.The Ion Driftmeter (IDM) component of the Special Sensors-Ions, Electrons, and Scintillation thermal plasma analysis package or SSIES (Rich & Hairston, 1994) measured the cross-track ionospheric convection flow at 1 s cadence (approx.7 km spatial resolution).The Special Sensor Ultraviolet Spectrographic Imager or SSUSI experiment (Paxton et al., 1992), measured auroral luminosity in a swath either side of the orbit in the Lyman-Birge-Hopfield short (LBHs) band, 140 to 150 nm (see Paxton and Zhang (2016); Paxton et al. (2017Paxton et al. ( , 2021) ) and the references cited therein for further description of the instrument and data products).The Special Sensor for Precipitating Particles version 4 or SSJ/4 instrument (Hardy, 1984) provided measurements of precipitating ions and electrons between 30 eV and 30 keV, in 19 logarithmically-spaced energy steps, at a cadence of 1 s.
The Super Dual Auroral Radar Network or SuperDARN (Chisham et al., 2007) provided measurements of ionospheric flows in the NH.We also present observations of the NH and SH field-aligned currents (FACs) derived from magnetometer measurements onboard the satellites of the Iridium telecommunications constellation processed using the from the Active Magnetosphere and Planetary Electrodynamics Response Experiment (AMPERE) technique (Anderson et al., 2000;Waters et al., 2001;Coxon et al., 2018).Panels (a) to (d) show "keograms" of auroral observations from the SSUSI instruments on DMSP F16 and F17 in the northern and southern hemispheres.These show the auroral emission in the LBHs band along the dawn-dusk meridian with a latitudinal resolution of 1 • and a cadence of approximately 110 minutes (the orbital period of the spacecraft).Grey regions show missing data: either missing images, or portions of the dawn-dusk meridian that were not sampled (predominantly in the southern hemisphere).Panels (e) and (f) show corresponding keograms of field-aligned current magnitude (red and blue for upwards and downwards FACs, respectively) from AMPERE, with a latitudinal resolution of 1 • and a cadence of 10 minutes.
Prior to interval I, pairs of upwards/downwards FACs seen at dawn and dusk are the region 1 and region 2 (R1/R2) currents first identified by Iijima and Potemra (1976).
A good correspondence between the location of the FACs and auroras is seen.FACs and auroras are mainly located between colatitudes of 16 and 20 • before the arrival of the shock, and between 20 and 35 • during the storm main phase.The enlarged polar cap indicates that the open flux content of the magnetosphere, F P C , was enhanced during the period of strong driving.From the start of interval I until 19 UT on 19 March the auroras and FACs contracted polewards, being mainly located polewards of 18 • colatitude, and were weak.The FACs were no longer R1/R2 currents (which are associated with convection during southward IMF (Milan, 2013;Milan et al., 2017)), but as will be discussed below were produced by dayside processes during NBZ.During intervals I and III, especially as observed by DMSP F16, the auroras extended all the way to the poles such that the polar cap disappeared, and indeed later we will argue that the magnetosphere was probably nearly closed at these times.At the end of the interval, after 20 UT on 19 March, the IMF rotated southwards, the R1/R2 FACs reappeared and the auro-  (Chamberlain, 1961), or 1 erg cm −2 s −1 (1 J m −2 s −1 ).
the DMSP orbit is not able to confirm the convection pattern during this period (though this will be discussed further below).However, the auroral evolution at this time is consistent with the DLR scenario described by Milan et al. (2020) Most other panels in Figure 3 show that the polar regions were filled with cuspaligned arcs for the majority of the time until the end of interval III.Here we will describe just a few particular examples and return to others in the Discussion.From panels (h) to (v) the orbits of the spacecraft allow the dayside convection pattern to be measured, and confirm the flows expected for DLR, that is, sunward near noon and antisunward flows into the high-latitude dawn and dusk sectors, and sometimes sunward flows at latitudes below the main auroral emission.Panels (t) and (w) show periods when the IMF has rotated to |θ| ∼ 45 • , when DLR is no longer expected but single lobe reconnection (SLR) will open flux (see Discussion) and produce lobe stirring.In both cases the polar regions become devoid of auroras suggesting an open polar cap.In the case of panel (w), the ionospheric flows confirm the expected convection pattern for SLR.Between these times, panels (u) and (v) show that when the IMF returns to low clock angles the flow pattern becomes consistent with DLR again, and the polar cap refills with auroras in both the northern and southern hemispheres.
In panel (x) the IMF has rotated to become southwards, the convection pattern becomes twin-celled and the polar cap is open, consistent with the observations at this time in Figure 2.
Figure 4 shows AMPERE-derived field-aligned current density, in both the northern and southern hemispheres, at four selected times.At each time, especially in the northern hemisphere, the FACs are consistent with DLR and the vorticity of the associated convection pattern (often referred to as NBZ FACs).Reverse, lobe cells are expected to be accompanied by a pair of up/down FACs straddling noon at latitudes near or above 80 • , with reversed-polarity FACs at lower latitudes, as discussed by Milan et al. (2020).Such a pattern is observed in all panels (except the SH in panel (a)), but the strength of the FACs is greatest in the northern hemisphere.The strength of NBZ FACs is anticipated to be controlled by the conductance of the ionosphere (Milan et al., 2017), though it is not expected that there will be much hemispherical asymmetry in the conductance in these near-equinox observations.The location of the solar terminator (green) in each hemisphere suggests that the conductance will be somewhat higher in the northern hemisphere, except in panel (a).In Figure 5 (corresponding to Figure 3h), DMSP F17 passes at the sunward edge of the auroras.The DLR-driven flow pattern is very clear in the IDM measurements, with little small-scale structure superimposed on the the antisunwards / sunwards / antisunwards convection.At lower latitudes, hot electrons (up to and exceeding 10 keV) are observed, the signature of trapped plasma on closed field lines produced during the normal Dungey cycle.Polewards of this are intense low energy (up to 1-2 keV) electrons and ions with a broad range of energies; the highest fluxes of ions are observed in the region of sunward flow.We interpret these ions and electrons as magnetosheath plasma precipitating downstream of the lobe reconnection site (i.e., sunwards of the ionospheric projection of the x-line).If we could observe the particle characteristics along the noon meridian we would expect a reverse dispersed ion signature within the sunward convection throat (Woch & Lundin, 1992).
We note an increase in electron energies observed at 09:51 UT, at the duskward edge of the convection pattern.In this region there is a negative gradient in V ⊥ , which implies converging electric field and hence upward FAC, consistent with the AMPERE measurements.We interpret this as a region in which electrons are accelerated downwards to carry the required FAC, producing a region of auroral emission near the 15 MLT meridian (arrow labelled 2).Examination of the panels in Figure 3 reveals that this blob of auroral emission is a consistent feature of the observations.In Figure 6 (corresponding to Figure 3g), DMSP F17 passes further towards the pole and crosses the main region of NBZ FACs, with the down/up FAC pattern clearly observed in the bottom panel.The DLR flow pattern is again clear, but now there is more small-scale structure superimposed on the convection, with ±500 m s −1 flow variations seen around 13:04 and 13:08 UT.As F17 traverses from sunward to antisunward flow at 13:06 UT, a clear, broad inverted-V signature is seen in the electrons, co-located with a region of pre-noon auroral emission (arrow labelled 1); again, this auroral feature is consistently seen in Figure 3. Narrower inverted-V structures are seen to be associated  -12-manuscript submitted to JGR: Space Physics with negative gradients in V ⊥ where the small-scale but large-amplitude fluctuations are seen, both within the sunward flow region and the dawnside antisunward flow region.
The pre-noon auroral blob (1) associated with upward FAC identified in Figure 5 is the high-latitude detached arc (HiLDA) discussed in a case study by Frey (2007) and statistically by Carter et al. (2018).HiLDAs are produced by the upward FAC associated with the clockwise vorticity of the NBZ reverse convection pattern.Re-examination of the observations presented by Carter et al. (2018) reveals that on average a post-noon auroral blob is also observed co-located with the region 1 polarity FAC, similar to the post-noon auroral blob (2) identified in Figure 5.Both are produced by precipitation associated with upward FAC due to the vorticity of the convection pattern, and the same morphology is observed in both NH and SH.Going forwards, we will refer to these as type 1 and 2 HiLDAs.
The pass of F16 in Figure 7 is similar to that in Figure 5, though crosses the noon meridian slightly antisunwards of the type 1 HiLDA.We show this pass to emphasise the consistency of the type 1 and 2 HiLDAs morphology.We also use it to highlight the multiple cusp-aligned arcs and associated flow-shears and inverted-V structures pointing into the cusp region and even merging with the type 1 HiLDA, that is, reaching into the centre of the pre-noon DLR flow vortex.This pass also shows that the cusp-aligned arcs sunwards and antisunwards of the dawn-dusk meridian do not necessarily connect with each other, though this is discussed further below.Patches of ion precipitation are seen across the polar regions, suggesting that significant regions of closed flux exist.
In Figure 8  In Figure 10 we examine the SuperDARN observations in more detail.Four radars of the network -Rankin Inlet, Inuvik, Hankasalmi, and Pykvibaer -contribute most to the measurements, and the line-of-sight velocity derived from the backscatter observed by each are shown in the left-hand panels (separated into two panels to avoid overlap), at five selected times.When combined using the standard SuperDARN "map-potential" fitting procedure (Ruohoniemi & Baker, 1998), these measurements give rise to the convection pattern presented in Figure 4b.The central beam of the Rankin Inlet radar looks almost directly into the convection throat at noon, observing sunward flows up to 1.5 km s −1 (towards the radar, blue); to either side antisunward flows (away, red) are seen,  consistent with the DLR convection pattern.This pattern of flows is observed by Rankin Inlet over the whole 1.5 h period shown, confirmed by the range-time-velocity panel (g).
Flows are strongest and weakest at the farthest and nearest ranges, consistent with the rotation of the flow within the DLR throat region from sunwards to dawnwards and duskwards as latitude decreases.Panel (f) shows the corresponding range-time-power measurements.
Regions of higher backscatter power are seen to quasi-periodically (3-6 min) propagate towards the radar; the curved nature of the striations suggests deceleration along the lineof-sight, consistent with the gradient in the velocity, and we suggest that regions of enhanced backscatter power are entrained within the convection as it circulates within the DLR pattern.Although not obvious with the colour scale employed, these power enhancements are accompanied by fluctuations in the convection speed with an amplitude of several 100 m s −1 superimposed on the background ∼ 1 km s −1 flow.The nature of these sunward-propagating power/velocity fluctuations is similar to poleward-moving radar auroral forms (PMRAFs) observed in the dayside polar cap during southward IMF, which are the counterpart of poleward-moving auroral forms (PMAFs) produced by flux transfer events (FTEs), that is quasi-periodic bursts of low latitude reconnection (e.g., Pinnock et al., 1995;Provan et al., 1998;Milan et al., 2000a;Wild et al., 2001).The observations of Figure 7 come from this interval.It is interesting to speculate if the time-variable flows observed by SuperDARN are associated with the cusp-aligned arcs observed to be reaching into the cusp region.
Flows on the nightside are monitored by the Hankasalmi and Pykvibaer radars.The fitted convection pattern, Figure 4b, shows relative weak and smooth flows on the nightside, in contradiction to the highly structured flows observed by DMSP IDM (Figures 7 and 8).However, both nightside radars do observe towards/away line-of-sight velocity enhancements with speeds of several 100 m s −1 (highlighted by arrows) that come and go and move about.The corresponding range-time-velocity panels, (i) and (j), show variations in the flow with periods close to 20 min.We conclude that structured flows are indeed observed by SuperDARN during this interval, but that the spatial resolution of the map-potential fitting procedure is too coarse to properly characterise this.The DMSP observations also imply that there are many small-scale FACs, accompanied in the case of upwards FACs by cusp-aligned arcs, but these are not resolved by the spatial resolution of the AMPERE technique.

Discussion
Following the main phase of the geomagnetic storm of St. Patrick's Day 2013, the IMF had near-zero clock angle for long periods over the next two days.Observations of auroral emissions, ionospheric flows, and field-aligned currents suggest that the magnetosphere underwent dual-lobe reconnection, closing open magnetic flux (Dungey, 1963;Cowley, 1981).This lead first to a horse-collar auroral configuration (e.g., Murphree et al., 1982;Hones Jr et al., 1989;Elphinstone et al., 1993;Milan et al., 2020) before the whole polar regions became full of cusp-aligned auroral arcs.We interpret this as partial and then almost full closure of the magnetosphere.We anticipate that the poleward edge of the HCAs at dawn and dusk map to the magnetopause of the dawn and dusk flanks of the magnetosphere, encompassing the newly closed flux that is appended to the magnetosphere by DLR.If this process continued to full closure of the magnetosphere, we would expect that the dawn and dusk magnetopause would map to a line running along the noon-midnight meridian, and there are times when such an auroral arc is observed, for instance Figure 4 panels (c) and (d).If the magnetosphere does close entirely, then DLR should drive reverse convection (Song et al., 1999;Siscoe et al., 2011), and in general the observed flows support this.In addition, the observation of (presumably trapped) ion precipitation across most of the traversals of the polar regions by DMSP (Figures 7 and 8) suggest significant flux closure.3x.It is interesting to note that although the IMF magnitude is small (≈ 5 nT) and B Z is only −1 nT at this time, the reopening of the magnetosphere is rapid and the strength of the flows is large.It appears that the magnetosphere "abhors a vacuum" and will reopen readily.
We anticipate that achieving full closure of the magnetosphere through DLR is difficult.As the magnetosphere approaches full closure, any IMF field line that reconnects in one hemisphere but not the other, as in scenario (d), will reopen flux.This failure to That the magnetosphere is nearly closed explains why auroral emission can occur over the entire polar regions: the closed flux will contain plasma that can be accelerated to produce auroras; in contrast, the polar regions are usually devoid of auroras because the open lobes are evacuated.However, we have also shown that the auroral emission is not uniform, but comprises multiple cusp-aligned arcs.Although these arcs have been reported before (e.g., Q.-H.Zhang et al., 2020), their origin is still uncertain.Q.-H.Zhang et al. (2020) showed that such arcs are associated with inverted-V precipitation signatures and filamentary field-aligned currents, associated with shears in the convection flow, just as presented in this paper.They suggested that these convection shears were propagated into the central magnetotail by Kelvin-Helmholtz Instability (KHI) waves on the magnetopause.In their scenario it is necessary for the magnetosphere to be nearly closed, such that closed flux from the central tail maps across the polar regions where the arcs are observed: DLR can provide this closure.Also, if the arcs are formed by the KHI and waves propagating from the magnetopause towards the central tail, we might expect to see a "phase motion" of the arcs in the ionosphere.This phase motion would be expected to emanate from the noon-midnight meridian (where the dawn and dusk flank magnetopause would map if the magnetosphere was closed) and move to lower latitudes both dawnwards and duskwards.Unfortunately, the cadence of the SSUSI images is too coarse to allow this behaviour to be resolved.
Alternatively, we suggest that patchy and bursty lobe reconnection could also lead to flow channels.The interleaving of open and closed flux described above will lead to different reconnection geometries (as exemplified in Figure 1) at different points along the x-line.(See also the discussion in Fear et al. (2015) in the context of transpolar arcs.) The spatially and temporally structured ionospheric flows observed by the Rankin In- be open and connected into the solar wind and will be subject to significant east-west magnetic tension forces; in cases (c) and (e) the field lines will be closed and draped across the dayside magnetopause, with little impetus to move.We might expect quite different evolutions of these differing field lines subsequent to reconnection.In all these cases however, the flows produced will be sunward on both the sunward and antisunward sides of the x-line, shown as green flow streamlines in Figure 11.
In Figure 1d we have envisaged that the single lobe reconnection occurs in the northern lobe, and the discussion above has applied to the northern hemisphere ionosphere.
If, on the other hand, the reconnection occurs in the southern lobe, then in the northern hemisphere the field line will be opened but connected through the tail into the solar wind in the southern hemisphere.It is unclear how this field line will subsequently evolve, but certainly there is no immediate application of a tension force to make it move sunwards; we have shown possible antisunwards flows away from the x-line by blue streamlines in Figure 11.We note that although the Rankin Inlet radar observes predominantly sunwards flows in the noon sector, at far ranges (around 1400 km in Figure 10), patches of antisunwards velocities are also seen: these could be the source of the antisunward flow chanels seen by DMSP on the nightside, with the boundary between anti/sunwards flows seen by the radar mapping to the ionospheric projection of the x-line.
The result of the patchy and episodic reconnection will be bursty sunwards flows sunward of the x-line, with more complicated flows on the nightside, as suggested by the observations: it was for this reason that Figures 5 to 8 were shown in order of dayside and auroral emissions, including both type 1 and 2 HiLDAs and cusp-aligned arcs.We expect that this complicated pattern of flows would evolve with time, and indeed the flow channels and auroral arcs are seen to move about between passes of the DMSP spacecraft.It is probable that the auroral emissions are even evolving during the spacecraft traversal of the polar regions (over approximately 20 mins), so the images should not be treated as a "snapshot" of the auroral pattern.There is some indication (e.g., Figure 7) that the dayside and nightside portions of the cusp-aligned arcs may not necessarily join, in which case they could be produced by different mechanisms, maybe the mechanism of Q.-H.Zhang et al. (2020) on the nightside and lobe reconnection on the dayside.On the other hand, this apparent disconnection could be a consequence of motions of the arcs during the traversal of DMSP.Higher temporal resolution auroral imaging would likely cast significant light on the formation and dynamics of the arcs.

Conclusions
We have reported the observation of cusp-aligned arcs during periods of near-zero IMF clock angle.Similarly to Q.-H.Zhang et al. (2020), we have shown that these arcs are associated with flow shears in the polar convection pattern.We propose that duallobe reconnection is first responsible for a significant closure of the magnetosphere, but subsequently complicated lobe reconnection geometries involving open and closed magnetospheric field lines could produce these flow shears.This pattern would evolve with time, but highly similar patterns would be expected in the two hemispheres.We expect that high-cadence (minutes rather than 10s of minutes) imaging of the auroral emissions are required to fully resolve the formation mechanism of the arcs.
Figure 1.A schematic diagram of the reconnection geometries available for southward and northward IMF.(a) Southward IMF leading to reconnection of closed magnetic flux near the subsolar magnetopause, in turn leading to magnetotail reconnection, together driving the Dungey cycle.(b) Northward IMF and single lobe reconnection with open magnetic flux in the northern hemisphere; the open flux content of the magnetosphere is unchanged.(c) Dual lobe reconnection occurring with open lobe flux in both hemispheres; open flux is closed.(d) Single lobe reconnection in the northern hemisphere in a closed magnetosphere, opening magnetic flux.(e) Dual lobe reconnection in a closed magnetosphere; the magnetosphere remains closed.Adapted from Cowley (1981).
The interval of interest is 17 to 19 March 2013, inclusive, encompassing the St. Patrick's Day storm of 2013.The IMF conditions, derived from the OMNI dataset (King & Pa-pitashvili, 2005), which drove this storm are shown in panel (g) of Figure 2. A step in solar wind ram pressure arrived at Earth shortly before 06 UT on 17 March (not shown), accompanied by an enhancement in the IMF magnitude.The GSM B Y and B Z components of the IMF fluctuated until 15 UT; from 15 UT B Z was consistently negative until approximately 00 UT on 18 March, producing the main phase of the storm.For the majority of the next 50 hours, B Z was positive and B Y near zero, such that the clock angle, θ = atan2(B Y , B Z ), shown in panel (h), was close to 0 • .Periods of small clock angle are highlighted by vertical lines, being 01 to 12 UT and 15 to 20 UT on 18 March, and 04 to 15 UT on 19 March; we will refer to these as intervals I, II, and III, respectively.

Figure 2 .Figure 3 .
Figure 2. Observations of auroras, field-aligned currents, and interplanetary magnetic field for three days, 17 to 19 March 2013.(a) to (d) Auroral observations along the dawn-dusk meridian in the northern and southern hemispheres from the SSUSI instruments on DMSP F16 and F17.(e) and (f) AMPERE observations of field-aligned current density along the dawn-dusk meridian in the northern and southern hemispheres.Red and blue shading indicate upwards and downwards FACs, respectively.(g) IMF BY (blue) and BZ (red) components, and total IMF field strength (grey).(h) IMF clock angle.Three periods of near-zero clock angle, labelled I, II, and III, are indicated by vertical dashed lines.Tick marks at the top correspond to panels in Figure 3.
: in panel (c) the auroras have contracted to high latitudes, but a clear polar cap is still observed; in panel (d) cusp-aligned arcs appear poleward of the dawn and dusk sectors of the auroral oval; and in panel (e) the polar regions are filled with auroras, with the appearance of multiple cuspaligned arcs.Milan et al. (2020) interpreted similar observations as the closure of open flux by DLR, producing sunward ionospheric flows across the noon sector polar boundary, and the distribution of this newly-closed flux to the high-latitude dawn and dusk sectors by antisunwards flows; however, in this example that process has continued and the whole polar regions have filled with auroras.Prior to DLR the open field lines are evacuated of plasma, so the polar cap is largely devoid of auroras; however, the newlyclosed flux is expected to be laden with captured solar wind plasma, which could give rise to the auroral emission seen at high latitudes.
Figure 4. Left and middle panels show AMPERE observations of field-aligned current density in the northern and southern hemispheres at times of selected passes by DMSP.The data are presented on a geomagnetic latitude and local time grid, with noon at the top and circles representing latitudes of 60, 70, and 80 • , with the solar terminator indicated in green.Crosstrack ionospheric flow vectors from IDM are superimposed on the hemisphere in which the pass occurred; the scale is indicated in panel (d).The dial plots show the IMF BY and BZ components.Right panels show the auroral emission measured by SSUSI (panels a, c, and d) or the ionospheric flow pattern inferred from SuperDARN observations (panel b); in the latter, the green circle is the low latitude extent of the convection pattern.In the latter case, the location of radar observations and the convection speed are indicated by shading; contours of electrostatic potential are indicated in steps of 6 kV.

Figure 5 .
Figure 5. Observations from a pass of the northern hemisphere by DMSP F17 around 09:51 UT on 18 March 2013.Top panels show AMPERE observations of field-aligned current density in the northern and southern hemispheres at the time of the pass (averaged over the duration of the pass), presented on a geomagnetic latitude and local time grid with noon at the top.Circles indicate geomagnetic latitudes of 60, 70, and 80 • .Ionospheric flow vectors from IDM are superimposed on the hemisphere in which the pass occurred.Below this, on the same grid, and auroral observations from SSUSI, presented twice, once with and once without the flow vectors superimposed.Arrows indicate type 1 and 2 HiLDAs (see text for details).The five panels below show parameters along the spacecraft track, over the duration of the pass.These are: auroral emission intensity from SSUSI, precipitating electron and ion spectrograms from the SSJ/4 instrument, the speed of the cross-track flow (positive sunwards) from IDM (also shown on the electro spectrogram), and the field-aligned current density from AMPERE.

Figure 6 .
Figure 6.Similar to Figure 5, for the northern hemisphere pass of DMSP F17 around 13:02 UT on 18 March 2013.
, F16 passes antisunwards of the pole, crossing several prominent cuspaligned arcs.The flows are highly structured, with multiple narrow sunward / antisunward flow shears with amplitudes in excess of ±1 km s −1 .Negative gradients in V ⊥ are associated with inverted-V signatures and colocated cusp-aligned arcs.These signatures are highly reminiscent of the observations reported by Q.-H.Zhang et al. (2020).Ion precipitation is observed across most of the polar traversal, suggesting that the magnetic flux is closed; a relatively narrow region is devoid of ions between the two main cuspaligned arcs, which might be interpreted as the poleward edges of the dawn and dusk regions of the HCA, indicating that the central region of flux could be open.No clear FAC signatures are seen by AMPERE along the F16 track, probably as a consequence of the relatively coarse spatial resolution of the observations.

Figure 9 (
Figure 9 (corresponding to Figure 3m) shows superimposed observations from F16and F18 as they traverse the SH polar regions at almost exactly the same time, with F18 and F16 sunward and antisunward of the pole.Once again the flows are highly structured, with flow shears being colocated with inverted-V electron precipitation and cuspaligned auroral arcs.F18 observes mostly sunward flows near the noon meridian, but with large perturbations superimposed.There is consistency between the flows observed in the dawn sector where the spacecraft pass close to each other, but it is difficult to discern the how the patterns fit together at dusk.It is probable that the nature of the flows are ordered by the pattern of cusp-aligned arcs, but the convection is clearly complicated and presumably time-variable.Ion precipitation present along both spacecraft tracks, especially on the nightside, suggest that much of the polar magnetic flux is closed.

Figure 7 .
Figure 7. Similar to Figure 5, for the southern hemisphere pass of DMSP F16 around 17:49UT on 18 March 2013.Note that for this pass of the southern hemisphere, the time axis has been reversed so that left and right correspond to dusk and dawn.

Figure 8 .Figure 9 .
Figure 8. Similar to Figure 5, for the southern hemisphere pass of DMSP F16 around 09:19UT on 18 March 2013.Note that for this pass of the southern hemisphere, the time axis has been reversed so that left and right correspond to dusk and dawn.

Figure 1
Figure 1 summarises the reconnection geometries available at the magnetopause for southward and northward IMF: (a) low latitude reconnection, (b) single lobe and (c) dual-lobe reconnection in an open magnetosphere, and (d) single lobe and (e) dual-lobe reconnection in a closed magnetosphere.Scenario (a) increases the open flux (the polar cap flux, F P C ) in the magnetosphere and drives the Dungey cycle, (b) neither increases nor decreases F P C but produces lobe stirring, (c) reduces F P C and causes reverse convection within the dayside polar regions.Scenario (d) opens flux, increasing F P C , while (e) does not open flux but induces reverse convection throughout the outer magnetosphere and polar regions, as modelled by Song et al. (1999) and Siscoe et al. (2011).For NBZ with near-zero IMF B Y we expect first scenario (c) and then (e) to close the magnetosphere and then drive reverse convection within it.However, if B Y becomes non-zero then (b) will halt the closure of the magnetosphere, and (d) will even reverse it.This is consistent with the disappearance of auroral emissions near the pole in Figure 3 panels (t) and (w).The southward turning of the IMF at the end of interval III, Figure 2, leads to low latitude reconnection which reopens the magnetosphere and drives Dungey cycle flows, as shown in Figure 3x.It is interesting to note that although the reconnect in both NH and SH could be caused by small variations in IMF B Y , turbulence within the magnetosheath, or transient and/or episodic reconnection.It seems likely that during prolonged periods of near-zero clock angle, a complicated interleaving of regions of open and closed flux could be produced in the polar regions, with the magnetosphere nearly closed, but not entirely.Such bursty reconnection and the interleaving of open and closed flux could be related to the appearance of cusp-aligned arcs and fast flow channels, as discussed below.

Figure 11 .
Figure 11.A schematic diagram of the ionospheric flows which could be driven by lobe reconnection in a magnetosphere that is nearly closed but has regions of open magnetic flux.The grey semicircle is the low latitude extent of the dayside convection pattern, with noon towards the top; the red dashed line is the ionospheric projection of the lobe reconnection x-line.Green and blue arrows show flow streamlines which might be driven for different reconnection geometries (see text for details).Blue shading indicates where vorticity in the flow will be associated with upwards field-aligned currents producing auroral emission, including type 1 and 2 HiLDAs (right and top-left) and cusp-aligned arcs.