Signatures of Open Magnetic Flux in Jupiter's Dawnside Magnetotail

Jupiter's magnetosphere exhibits notable distinctions from the terrestrial magnetosphere. The structure and dynamics of Jupiter's dawnside magnetosphere can be characterized as a competition between internally driven sunward flow and solar wind‐driven tailward flow. During the prime mission, Juno acquired extensive data from dawn to midnight, sampling the magnetodisc and higher latitude regions. Numerical moments from the Jovian Auroral Distributions Experiment (JADE‐I) plasma (ion) instrument revealed a mid‐latitude region of anticorotational (−vϕ) flow. While the magnitude of the flow is subject to uncertainty due to low count rates in these rarefied regions, we demonstrate in the raw JADE‐I data that the sign of vϕ is a robust measurement. Global Grid Agnostic Magnetohydrodyamics for Extended Research Applications simulations show a similar region of strongly reduced flow in proximity to open field lines. Additionally, we use Jupiter Energetic‐particle Detector Instrument integral moments to determine the Hen+/H+ ratio (where n refers to He+ or He++) and show that a transition to solar wind‐like composition occurs in the same region as the anticorotational flow. We conclude that the global simulations are consistent with the Juno data, where the simulations show a crescent of open magnetic flux that is bounded by the magnetodisc and a closed high‐latitude polar region (nominally the polar cap), which is never observed in the terrestrial magnetosphere. The distinct distribution of open flux in Jupiter's dawnside magnetosphere suggests the significance of planetary rotation and may represent a characteristic feature of rotating giant magnetospheres for future exploration.


Introduction
The interaction of Jupiter's magnetosphere with the solar wind has been a decades long enigma.Following Voyager flybys, it was widely understood that Jupiter's magnetosphere was largely open based on wave data and energetic particle observations.Voyager 2 electric field measurements showed spectra (e.g., cutoff frequency) outside of the magnetodisc to be similar to solar wind spectra (Gurnett et al., 1980).These regions were referred to as the "tail lobe" (i.e., by terrestrial analog, the lobes are open field lines) and it was widely considered that a dawn/dusk asymmetry existed with much of the dawnside magnetosphere containing open flux (Cowley et al., 2003;Khurana et al., 2004).However, others argued instead for a largely closed magnetosphere due, in part, to the preponderance of auroral activity in the polar regions (Delamere & Bagenal, 2010;McComas & Bagenal, 2007, 2008).A viscous-like interaction (governed by, e.g., the Kelvin-Helmholtz (KH) instability) with the solar wind was proposed by Delamere and Bagenal (2010) to account for the strong dawn/dusk asymmetries that must, fundamentally, stem from the solar wind interaction.
Using the Grid Agnostic Magnetohydrodyamics for Extended Research Applications (GAMERA) global simulation for Jupiter, Zhang et al. (2021) showed that the dawnside magnetosphere is composed of a large bundle of closed flux mapping to the high latitude polar region, a corotating inner region (i.e., magnetodisc), and a crescent of open flux separating the inner and outer closed regions (Figure 1).To compare the meridional structure of the GAMERA simulations with Juno data from the dawn-side magnetosphere, a comparison of plasmadisc properties as a function of distance from the centrifugal equator plane (z cent ) was made (Schok et al., 2023).It was found that the strong dependence on z cent exists in the inner plasmadisc (e.g., <50 planetary radii, or R J ∼ 7.14 × 10 4 km).But in the outer magnetosphere (e.g., >80 R J ), variations in density with z cent diminish, implying a large scale height.The inferred variation in centrifugal confinement suggests that the inner plasmadisc corotates, while the outer regions strongly subcorotate, possibly becoming part of the non-corotating closed flux-pileup region seen in the GAMERA simulations.However, there were no obvious signatures in the Juno magnetic field, waves, or ion (JADE mean counts) data that would indicate possible regions of open flux.
In addition, the Juno mission has provided an unprecedented glimpse into the high latitude polar aurora (e.g., polarward of the main oval of emission).In particular, the Juno Ultraviolet Spectrograph (UVS) detailed the polar aurora for a wide range of local times.To simplify the geometry, the polar region has been divided into two different spectral regions determined by the "color ratio" (defined as the ratio of the radiance at 155-162 nm to the radiance at 125-130 nm (Bonfond et al., 2017)).The high-latitude "swirl" region exhibits a high color ratio and is circumscribed by a "polar collar" with significant local time asymmetry (Greathouse et al., 2021).The dusk side (11-22 LT) manifests as a bright "active region" and the dawn collar is the "dark region" (0-11 LT) (Dunn et al., 2022;Stallard et al., 2003).
The polar aurora regions can be further bounded by examining the current systems, particle, and wave data (Sulaiman et al., 2022).The main emissions have been subdivided into Zone-I and Zone-II which could carry respective upward and downward currents.Zone-I has the expected characteristics of an upward current region with energetic upward H + beams and downward energetic electrons (Allegrini, Mauk, et al., 2020;Mauk et al., 2020;Szalay et al., 2021).Zone-II does not have a clear poleward boundary and tends to have sporadic signatures that would be considered typical in Earth's downward current regions with upward energetic electrons, downward iogenic ions, and downward H + inverted Versus (Clark et al., 2020;Mauk et al., 2020).
Finally, the polar-most region has been shown to exhibit Zone-II characteristics with heavy ion composition (Szalay et al., 2022).While Juno traversed the polar region, JADE time-of-flight count rates showed similar heavy ion (e.g., O + , S ++ ) features compared with measurements made in the magnetotail at 110 R J .Thus, it was concluded that Jupiter's polar-most field lines can be topologically closed.The polar-most closed topology is consistent with GAMERA simulations, with field lines mapping to a flux-pileup region on the dawn flank (Zhang et al., 2021).However, there remains the issue of the existence of open flux and whether the polar collar can contain open flux as predicted by simulation.In this paper we present model/data comparisons in the dawn/tail meridian to determine possible signatures of open flux in Jupiter's magnetosphere.In particular, we compare the azimuthal flows from JADE numerical moments and the He n+ /H + ratio as determined by JEDI densities (integral moments).

JADE Numerical Moments
The Jovian Auroral Distributions Experiment (JADE) (McComas et al., 2017) on Juno observes the thermal plasma of the jovian magnetosphere.At launch it was intended to carry out short 11-day orbits with apojoves under 40 R J , with limited local time coverage before eclipses behind Jupiter froze the solar powered spacecraft.However, during Juno's capture orbits, it was decided to remain in a 53-day orbit configuration.This still addressed the mission's prime science goals, but in addition allowed for magnetospheric mapping since these new orbits had apojoves out to 113 R J , with the ability to skip over a mission ending "eclipse season," keeping power and allowing for an extended mission.
While the JADE instruments were designed for the aurora encounters, they were not designed to observe these rarefied, middle-to outer-magnetospheric regions, so some data summing (at the expense of time resolution) is required to build up signal in the data.JADE-I is the ion sensor of JADE, covering an energy range of 0.01-50 keV/q over 64 approximately logarithmically spaced energy steps every 2 s.There are 12 anodes that cover 270°in spacecraft elevation from 90°to +180°, such that anodes 0-3 cover the same field of view as anodes 4-7 half a spin later.In low rate telemetry modes, individual 2 s energy sweeps are summed into multiples of 30 s, and the ion species product only uses 8 of the anodes (4-11) and remaps them onboard in to one of 78 look directions to cover 4π steradians for three mass/charge ranges, known as species 3, 4, 5. Species 3 is protons only, species 4 is "lights" (approximately 2-5 amu/q), and species 5 is "heavies" (approximately 5-64 amu/q).These 78 look directions were chosen to have approximately the same solid angle coverage, but are averaged over 1, 2, 3 or 5 energy sweep views (see Figure 9 of the JADE Software Interface Specification document (Wilson, 2018)).This all-sky-coverage data set can then be used to calculate moments.
The proton moments utilized here are calculated from the JADE-I ion species 3 data set, that contains only protons.Generally, there is no significant signal from heavies in the mid to high latitudes.However, due to instrument design, some fraction of incoming protons are registered in species 4 and thus missed from species 3. Using ground calibrations, we appropriately multiply up the species 3 counts, per energy step, to account for the fraction of protons that were missed.Finally, to improve count statistics in the very rarefied region of interest, we carefully sum records together to obtain new records of 600 s accumulation before calculating moments.The low rate ion species data used is Level 3 Version 04 files in units of counts/second (Allegrini, Wilson, et al., 2020), which is converted to phase space density ( f ) using the calibrations given by equation E1 of Kim et al. (2020).
The numerical moments calculations to derive density (n), bulk flow (V) and temperature (T ) are textbook (i.e., ith moment given by M i = ∫∫∫fv i dv x dv y dv z , then n = M 0 , etc.), carried out assuming that a small spacecraft potential (i.e., ∼few V) is negligible to the analysis (i.e., plasma energy is 100s of eV), and have the bulk flow velocities converted into the Jupiter-De-Spun-Sun frame (JUNO_JSS frame in SPICE (Acton, 1996), see kernel file juno_v09.tfor later version, after accounting for Juno's own velocity), using spherical coordinates.

JEDI Numerical Moments
To complement the JADE numerical moments, energetic particle data obtained from the Jovian Energetic particle Detector Instrument (JEDI) was used to determine light ion composition (i.e., He n+ /H + ) (Mauk et al., 2017) as the JADE numerical moments used in this study are for protons only.JEDI is composed of three sensors that are mounted in various locations on the spacecraft deck to optimize angular coverage.Two of sensors, specifically JEDI-90 and JEDI-270, can determine composition using a time-of-flight (TOF) technique coupled with solid state detectors.This method requires multiple timing and positional coincidences to be satisfied which results in high-confidence of ion species determination.In this study, we rely on this JEDI measurement technique (sometimes referred as TOF × E).The species dependent energy ranges are: (a) H: ∼50 keV to 2.5 MeV; and (b) He: ∼90 keV to ∼2 MeV.JEDI also measures electrons from ∼25 keV to 1 MeV as well as heavier ions such as oxygen and sulfur from 10s of keV to 10s of MeV.The partial densities are computed from the integral moment of the ion intensity spectra using a method similar to the one developed by (Mauk et al., 2004) for Galileo observations of Jovian magnetosphere.The error associated with the integral moments are typically <15%.In particular, we are interested in the He n+ /H + (where n+ denotes either doubly or singly ionized charge states) ratio as a proxy for identifying regions of the magnetosphere that have a higher abundance of solar wind particles.We limit our analysis to regions outside of Callisto's orbit (26.3 R J ) to avoid measurement contamination due to highenergy radiation.Therefore, we present JEDI data for r > 30 R J .Finally, JEDI cannot detect charge state, so we cannot definitively determine whether the signal is from He + or He ++ .We assume that it is likely the latter, but use He n+ to acknowledge this uncertainty.Figures S5 and S6 in Supporting Information S1 show how JEDI can distinguish between H + and He n+ .

GAMERA Simulations
The GAMERA model, based on the Lyon-Fedder-Mobarry (LFM) code, is a global magnetosphere model which has been used extensively to study solar wind -magnetosphere interactions (Lyon et al., 2004;Zhang et al., 2019).The ideal magnetohydrodynamics equations are solved with a finite volume (FV) numerical method.The model uses high-order flux reconstructions with a total variation diminishing limiter to preserve steep gradients while minimizing numerical dissipation and dispersion.A full description of the simulation used in this study was given by (Schok et al., 2023).
Briefly, the spherical grid extends 100 R J in the sunward direction, 1000 R J in the anti-sunward direction, and ±300 R J in the directions perpendicular to the Sun-Jupiter axis.The grid resolution is non-uniform with the highest resolution of 0.2 R J near the magnetopause and approximately 0.15 R J near the inner magnetosphere.The inner boundary of the simulation is at 6 R J .The simulation uses only a single fluid with a mass loading rate of 1,000 kg/s (Delamere & Bagenal, 2003), introduced in the equatorial region of the inner boundary.As a single-fluid simulation, the inner "iogenic" source was the only source of mass besides the solar wind.Ionospheric outflow was not included.We note that as a fluid simulation, only the mass density determines the dynamics when finite gyroradius effects are considered unimportant for global-scale processes.The combined (bulk) density profiles from a multifluid simulation, where solar wind and Io plasma are separate fluids, were qualitatively similar to the single-fluid results.The single-fluid simulation has the benefit of being computationally less expensive than the multi-fluid simulation.The simulations were driven by constant solar wind and interplanetary magnetic field (IMF) conditions (V = 400 km/s, B y = 0.5 nT), consistent with the average solar wind dynamic pressure (∼0.05 nPa) at Jupiter (Ebert et al., 2014;Jackman & Arridge, 2011).No dipole tilt of Jupiter's magnetic field was included.The magnetosphere-ionosphere coupling is adapted from the Magnetosphere Ionosphere Coupler/Solver (MIX) model with a Pedersen conductance of 0.5 mho and zero Hall conductance (Merkin & Lyon, 2010).We solve the ionospheric current closure in the rotating frame and the electric field at the top of the ionosphere (assumed to be = 1 R J ), which includes both the corotation electric field and the electrostatic potential from current closure.
One of the most remarkable GAMERA results is the peculiar magnetic field topology associated with the polar region.Other global simulations tend to produce an open polar cap region under typical Parker-spiral IMF conditions, for example, with an open polar cap containing roughly 200 GWb of open flux (Sarkango et al., 2019), while the GAMERA simulation, produced ∼25 GWb of open flux distributed in a crescent region poleward of the main emissions (Zhang et al., 2021).These simulations are remarkably consistent with auroral observations of Jupiter's northern polar region, assuming that aurorally dark regions are plausible signatures of open flux.Also remarkable is the variability of the open flux region on a sub-rotation (<10 hr) time scale, suggesting a very dynamic auroral signature, which is also consistent with observations of Jupiter's northern polar region (Greathouse et al., 2021;Grodent, 2014).The GAMERA simulations support the hypothesis that aurorally active regions occur on closed magnetic field lines.

Results and Discussion
The plasma densities are low (e.g., <10 3 cm 3 ) at high southern latitudes (during the outbound trajectories), leading to low count rates.Figure 2 shows one such example beginning 2019-256, where an enhancement of protons is seen every Jovian rotation (10 hr).As shown these counts are in the 0.1 to 1 counts/second range when calculated over a 600 s accumulation period.These are very low when one considers the 1-count per accumulation level of a single JADE-I energy step over the native 2 s sweep (after included settling time of the onboard voltages) translates to 34.19 counts/sec.However summing the data to 600 s accumulation times greatly improves the counting statistics to reliable values.Panels 2-4 of Figure 2 show the calculated proton numerical moments over these regions.To avoid the perils of low count statistics, we present moments only when the density is 0.001 cm 3 or higher.During each enhancement, the density increases as Juno passes through the middle of the feature, then decreases again on the other side.The temperature does the opposite; it begins higher, decreases as we approach the middle and then increases again on the other side.However, the v ϕ component (panel 3) begins each enhancement with a positive value (the usual sub-corotating sense), but rotates around (mostly in the v ϕ v θ plane) to have a negative value by the end of the enhancement.We note little deflection of the magnetic field in the bottom panel.
Given the low densities, it is reasonable to ask if these negative v ϕ values are real, or a manifestation of low counts going through moments calculations.To address these questions, panel 5 of Figure 2 shows a lower level data product of the ion species data.The y-axis is now of the 78 look directions in low rate science mode, summing over all energy steps 7 to 47 (of 0-63).This subset of energy steps was chosen based on the top panel, to pull out counts from the enhancement and not noise from the higher energy steps.White lines separate the different anodes (based off Figure 9 from Wilson, 2018), with the middle 4 bands highlighting the "belly band" of the spacecraft with 15 look directions to cover the 360°azimuth.As we pass through each enhancement, the peak counts clearly changes look direction from one side to the other (the diagonal bands in the data), proving that the change in direction is real.
In Figure 3 we show the mean binned statistic of the azimuthal flows (v ϕ ) in the meridional plane (r, θ) for orbits 5 through 23.We have used values with a density uncertainty filter following Huscher et al. (2021) (i.e., σ n /n < 10, where σ is the error and n is proton density) and we only plot bins with ≥3 data points.(In the electronic supplement (Figures S1-S4 in Supporting Information S1) we show the negative and positive v ϕ cases separately, the azimuthal flow normalized by rigid corotation, and the number of records per bin.)These orbits cover much of the midnight to dawn sector where significant variations would be expected for plasma under solar wind or planetary control.Gaps in the data (e.g., inner, mid-latitude region) are due to incomplete spacecraft sampling on the inbound and outbound portions of the polar orbit.Keep in mind that multiple orbits can contribute to a given bin and thus the flows should be considered as averages.The moments have been organized by vertical distance from the magnetic equator as defined by the magnetic coordinate system (JUNO_MAG_VIP4, https://naif.jpl.nasa.gov/pub/naif/JUNO/kernels/fk/juno_v12.tf).Alternatively, one could use the centrifugal equator plane (z cent ) as defined by Phipps and Bagenal (2021).As noted by Schok et al. (2023), the outer magnetosphere does not exhibit a clear centrifugal confinement of the magnetodisc, so the specification of z cent may only be relevant for the inner magnetosphere.Furthermore, the Phipps and Bagenal (2021) model converges to the magnetic equator beyond ∼25 R J anyway.
In the equatorial plane, the flows are corotational (positive v ϕ ) with maximum values greater than 400 km/s.In the outer magnetosphere, on average, these flows are negative, with speed approaching solar wind-like values ∼ 400 km/s.We suggest the outermost region is strongly influenced by tailward flows in the closed field line region extending along the dawn flank (Zhang et al., 2021).In the inner region and at mid latitudes, there is a region of v ϕ flanked by +v ϕ at higher latitudes.Open field lines in this region would be expected to have a v ϕ component of flow, while closed, corotating field lines would have a +v ϕ component of flow.In what might be considered the "intermediate" regions (e.g., 60 to 100 R J (Schok et al., 2023)), the flows exhibit a mix of positive and negative flows.
Figure 4 shows the occurrence percentage of closed flux and the average v ϕ in the 3 LT meridian from the GAMERA simulation from 2 rotations.Zhang et al. (2021) showed that open/closed flux content varies significantly on the planetary rotation period, indicative of ongoing and/or intermittent reconnection.Therefore, we suggest that the open/closed boundary is highly variable, but, on average, maintains a crescent-shaped (i.e., green region in bottom left panel of Figure 1), mid-latitude open topology.Note that Zhang et al. (2021) used steady solar wind conditions yet the open flux content varies.The variability would be expected to be substantial during strong variations in solar wind dynamic pressure.The azimuthal flows in the midlatitude open field region is strongly reduced, but on average, does not indicate a change in sign as seen in the data.However, there are instantaneous cases with negative v ϕ (Figure 5).Qualitatively, the sharp contrast in flow between low-, mid-, and high-latitude regions is similar in both model and data, but the quantitive details will be reserved for future studies when we investigate the sensitivity of the ionospheric boundary condition to magnetosphere-ionosphere coupling.Additionally, composition could also be used to identify open field lines.Here we use the He n+ /H + density ratio as determined by integral moments from the JEDI data (Mauk et al., 2004).Figure 7 shows the He n+ /H + ratio in the meridional plane (number of records per bin are shown in Figure S4 in Supporting Information S1).It is clearly evident that the He n+ /H + ratio increases substantially in the region where v ϕ transitions to anti-corotational flow, consistent with solar wind composition (e.g., >0.1).However, in the highest latitude regions of closed flux, the ratio remains consistent with solar wind composition.While the closed field lines in the extreme high latitude have been shown to contain distributions of heavy ions (Szalay et al., 2022), the light ion composition is similar to the solar wind.The evolution of the high latitude closed flux was described as a flux pile up region by Zhang et al. (2021), and should, therefore, involve the cross-tail nightside flow, which was shown to contain significant solar wind plasma (Cheng & Krimigis, 1989).Future studies will investigate the evolution and composition of the   (1) So the acceleration of the reconnected flux tube is

Conclusions
We conclude as follows: • Global simulations show Jupiter's magnetosphere contains open flux in a limited, mid-latitude regions while the extreme high latitude polar region is closed.Future studies should compare plasma properties in regions of positive and negative v ϕ .Such a study should also involve energetic particle data from the JEDI instrument.Additionally, model sensitivity to ionospheric boundary conditions and solar wind variations will be considered.GAMERA-based test particle simulations could address the mixed solar wind/iogenic plasma on the extreme high latitude closed polar region.

Figure 1 .
Figure 1.Topological classes of Jovian magnetic field lines from Zhang et al. (2021), showing inner closed (red), polar closed (black), distant lobe (blue) and interplanetary magnetic field (IMF) open (green).The IMF open field lines map to a crescent region and the blue field lines map to a patch in the ionosphere.

Figure 2 .
Figure 2. JADE-I Species 3 proton data from 2019-256T18:07:33.869 to 2019-258T17:57:35.844 after summing into 10-min bins.First panel: Look direction averaged spectrogram of Species 3 data.Second, third, and fourth panels show the numerical moments of density, velocity and temperature respectively.The fifth panel is an energy summed spectrogram of Species 3. The sixth (bottom) panel shows the 60 s resolution magnetometer data in spherical coordinates for this interval.

Figure 5
Figure 5 shows an example snapshot of the azimuthal flow and density in the 3 LT meridian from the GAMERA simulation.Here we see that significant negative v ϕ can occur on open field lines (closed = hatched area) and the lowest density regions flanking the magnetodisc need not be on open field lines.In fact, higher density regions can exist at higher latitudes.If we consider, hypothetically, a spacecraft traversing through the simulated open flux region, the density is sufficiently high across the open/close boundaries to determine plasma moments.The example case in Figure 2 could be at the highest latitude open/closed boundary and in a region where the plasma density does not vary considerably over multiple rotations.The variability of open flux, azimuthal flow, and density is further illustrated in Figure 6.Here we take the azimuthal flow and density at 3 LT and at a radial distance of 40 R J as a function of latitude and time.The hatched areas correspond to closed flux.The open flux region does vary in latitude with time and is highly structured, showing regions of embedded closed flux within the open flux regions.In the particular case of v ϕ in the

Figure 3 .
Figure 3. Azimuthal flow from Jovian Auroral Distributions Experiment moments.Only cases with ≥3 counts per bin and with density uncertainty filter σ n /n < 10 are shown.

Figure 4 .
Figure 4. Average (over 2 rotations) open flux (left) and azimuthal flow (right) from the Grid Agnostic Magnetohydrodyamics for Extended Research Applications simulation in the 3 LT meridian.

Figure 5 .
Figure 5. Snapshot of azimuthal density (left) and flow (right) in the Grid Agnostic Magnetohydrodyamics for Extended Research Applications simulation in the 3 LT meridian.Hatched areas are of closed flux.

Figure 6 .
Figure 6.Timeseries of azimuthal flow and density as a function of latitude taken from 3 LT and a radial distance of 40 R J .The hatched regions correspond to closed flux.The vertical line indicates the time of the meridional snapshot shown in Figure 5.
dp c dt = 2ρ o Av A v c (t).
) with v c (t) = v 0 e t/τ (3) where τ = (ρ c L)/(2ρ 0 v A ), and where v 0 is the velocity of the corotating plasma relative to the open region, A is the cross section of the interchanged flux tube, ρ 0 is the mass density in the open region, and L is the field-aligned length of the originally closed flux tube section.As an example, consider the conditions in Figure 2 where the magnetic field strength is ∼25 nT, the mass densities can be considered comparable, and the Alfvén speed is roughly 3 × 10 6 m/s.If L ∼ 50 R J , then τ ∼ 10 min.This is a rough estimate because the Alfvén speed and mass density will vary significantly on the open field lines, for example, magnetospheric plasma loss on open field lines is not considered here.Nevertheless, the momentum transfer time scale would be much less than the solar wind advection time scale (hours) and the planetary rotation period.So we suggest here that intermittent reconnection between open and closed regions can quickly modify the flows from corotation to anti-corotation direction.However, due to the momentum transfer "delay," instantaneous v ϕ should not be considered as a definitive indication of open/closed flux in such a highly variable situation.
• Global simulations show that the dawn sector azimuthal flows on open field lines are, on average, strongly reduced and occasionally anti-corotational.•JADE numerical moments show, on average, a mid-latitude region of anti-corotational flow.Comparisons with the model results suggest that anti-corotational flows may occur on open field lines.• JEDI integral moments show that the He n+ /H + ratio changes abruptly at near the transition to anti-corotational flows, consistent with open field lines and solar wind composition.•The magnetometer data, in the example case presented here, do not show any distinct characteristics in the transition between open and closed field lines.Magnetic field variations should be explored in more detail in future studies.