MAVEN observations

Figure 1 shows MAVEN observations from 22:17 UT (universal time) to 22:22 UT on 25 February 2017 (more detailed MAVEN observations available in Supplementary Figs. S1 and S2). These observations are made at a LT of roughly 20 h, a solar zenith angle of about 100° (near the dusk terminator), and over the strongest crustal fields (part of the orbit tracks shown in Fig. 1e). In Fig. 1c, there are two time periods of electron fluxes significantly elevated from the surrounding time periods, particularly at high energies, and the energies of peak electron fluxes are somewhat time-dispersed, suggesting that these are electrons that have experienced field-aligned acceleration similar to those creating Earth’s discrete auroras3,21.

Fig. 1: Relevant MAVEN observations of the miniature Dungey-like cycle.Fig. 1: Relevant MAVEN observations of the miniature Dungey-like cycle.

Time series of MAVEN observations on 25 February 2017: a magnetic perturbation Br in the local horizontal plane, b the derived field-aligned current density from magnetic pertubation \({B}_{EW}^{r}\) (\({j}_{//}^{b}\), blue) and superthermal electron fluxes (\({j}_{//}^{e}\), orange), c superthermal electron energy spectra (differential energy flux in units of eVcm−2sr−1s−1eV−1), and d the flow velocity of O2+ (V(O2+)) in the local horizontal plane. e The derived FAC (\({j}_{//}^{b}\)), the east-west component of O2+ flow velocity (∣VEW(O2+)∣ > 1 km/s), and electron acceleration observations with \({j}_{//}^{e} > 0.1\,\mu A/{m}^{2}\) as orange dots, overlain on a color map of the modeled radial crustal magnetic field at 250 km altitude35. The sign of \({j}_{//}^{b}\) refers to upward (\({j}_{//}^{b} > 0\)) or downward (\({j}_{//}^{b} < 0\)) FAC with respect to the local horizontal plane, regardless of whether the local magnetic field is radially upward or downward.

Using an empirical linear relation between electron energy fluxes and the emission brightness16,22, these energized/accelerated electrons could produce CO Cameron-band auroral emission (a specific set of ultraviolet emission features of the carbon monoxide molecule) in the Martian upper atmosphere with a limb brightness of about 10 kilo Reighley (kR). This is comparable to some of the brightest auroral observations by MAVEN’s Imaging UltraViolet Spectrograph (IUVS) instrument13. In fact, the IUVS instrument observed a CO Cameron-band auroral emission brightness ranging from 0.6 to 1.4 kR at 22:18:37-22:19:30 UT, 25 February 2017, similar to the interval studied, but estimated to occur at [−39.9°, −36.7°] in latitude and [182.5°, 183.0°] in longitude, different from the locations of electron observations. The estimated auroral brightness from auroral electrons at 22:18:20 (closest to the IUVS observations in time) is 4–6 kR, which is a few times the IUVS measurements. It means that the observed electrons in this case study have sufficient energy fluxes to produce observable auroral emissions by IUVS. Their differences are probably due to the time variations in the electron precipitations as well as their different spatial locations.

In the vicinity of the energized electrons, the measured magnetic fields deviate from the local crustal magnetic fields such that the magnetic perturbation Br from the crustal field (Fig. 1a) is mainly in the local east-west direction. This magnetic perturbation is consistent with a FAC sheet along the east-west direction. With this assumption and the fact that the spacecraft is moving mainly from north to south (Fig. 1e), we derive the FAC density \({j}_{//}^{b}\) from \({B}_{EW}^{r}\), shown as the blue line in Fig. 1b. Note that the sign of \({j}_{//}^{b}\) refers to upward (\({j}_{//}^{b} > 0\)) or downward (\({j}_{//}^{b} < 0\)) FAC with respect to the local horizontal plane, regardless of whether the local magnetic field is radially upward or downward. The uncertainty of \({j}_{//}^{b}\) is about 7%, shown in Supplementary Figs. S1f, S3f and S4f. The detailed explanations of the uncertainty calculations for \({j}_{//}^{b}\) and other key parameters are also provided in Supplementary Information. Comparing Fig. 1b, c, accelerated electrons are mostly associated with upward FAC, and the upward FAC is bracketed by two downward FACs for both time periods.

We can also derive the FAC carried by electrons \({j}_{//}^{e}\) based on the observed electron number fluxes, shown as the orange line in Fig. 1b. The uncertainty of \({j}_{//}^{e}\) is about 15%, also shown in Supplementary Figs. S1f, S3f and S4f. The calculated \({j}_{//}^{e}\) has a decent agreement with \({j}_{//}^{b}\) for \({j}_{//}^{b} > 0\) (22:18:15–22:18:28 UT and 22:21:06–22:21:35 UT), supporting the argument that the upward FAC is carried by accelerated downgoing electrons. For times of \({j}_{//}^{b} < 0\) (22:17:55–22:18:15 UT and 22:21:35–22:21:50 UT), \({j}_{//}^{e}\) is near 0, as the downward FAC should be supported by upwelling (denser) ionospheric electrons, and no downward electron acceleration is expected. We note for two particular measurements of \({j}_{//}^{e}\) at approximately 22:18:30 UT and 22:21:05 UT, \({j}_{//}^{e} > 0\) while \({j}_{//}^{b} < 0\), where we expect \({j}_{//}^{e}\) to be around zero. At this time, \({j}_{//}^{b}\) (Fig. 1b, as well as Supplementary Figs. S3 and S4) has rapid variations in its magnitude and signs at a time scale of <1 s, indicating rapid temporal and/or spatial dynamics that are unresolved by the 2-s cadence of the MAVEN electron measurements. In addition to the main case study presented above, we identify six additional events, with their corresponding MAVEN observations shown in Supplementary Figs. S5S10. In most cases, \({j}_{//}^{e}\) has a decent agreement with \({j}_{//}^{b}\) for \({j}_{//}^{b} > 0\), except for when the change in \({j}_{//}^{b}\) is less than 2 s and too rapid to be resolved by the electron measurement.

Lastly, as shown in Fig. 1d, the calculated bulk flow velocities of \({{{{\rm{O}}}}}_{2}^{+}\) (\({{{{\bf{V}}}}}_{{O}_{2}^{+}}\); solid) and O+ (\({{{{\bf{V}}}}}_{{O}^{+}}\); dashed) in the local plane are very similar. The uncertainties in \({{{{\bf{V}}}}}_{{O}_{2}^{+}}\) and \({{{{\bf{V}}}}}_{{O}^{+}}\) are overplotted as error bars in Fig. 1d, also shown in Supplementary Figs. S1j, S3j and S4j. Both ion flows have a prominent east-west component (VEW (\({{{{\rm{O}}}}}_{2}^{+}\)) and VEW(O+)) in the vicinity of energized electrons, suggesting the ionospheric bulk flow is in the east-west direction. In particular, there is an east-west flow reversal at 22:18-22:19 UT. More detailed observations and the derivations of Br, \({j}_{//}^{b}\), and \({j}_{//}^{e}\) are provided in the Supplementary Information.

Miniature Dungey-like cycle at Mars

Figure 1e synthesizes these key observations: the FAC density derived from Br (\({j}_{//}^{b}\)) shown as magenta (\({j}_{//}^{b} > 0\)) or cyan (\({j}_{//}^{b} < 0\)) lines, (auroral) electron observations with \({j}_{//}^{e}\) with a magnitude  > 0.1 μA/m2 shown as orange dots, and VEW(O2+) with a magnitude >1 km/s as the black arrows, all projected onto a crustal field map. Taken together, these can be explained by a miniature Dungey-like cycle of magnetic reconnection and flux and plasma circulation, as illustrated in Fig. 2.

Fig. 2: Schematics of the miniature Dungey cycle4 at Mars.Fig. 2: Schematics of the miniature Dungey cycle4 at Mars.

a The precondition for the first magnetic reconnection between the closed crustal magnetic fields and draped interplanetary magnetic field (IMF − BY) at the dusk terminator. The yellow and gray shaded regions are the dayside and nightside, separately. The blue and red shaded regions are crustal magnetic fields, blue for radially inward magnetic fields (Br < 0) and red for radially outward magnetic fields (Br > 0). b The magnetic topology change and the magnetic flux circulation of the miniature Dungey cycle. c A zoomed-in view of the corresponding ionospheric plasma circulation and current system of the miniature Dungey cycle, with the numbered circles marking the footpoints of the numbered magnetic field lines in (b). All the schematics are not-to-scale.

Figure 2a is a sketch of the precondition, where the crustal fields are located near the dusk terminator, the upstream IMF before and after the time of interest being predominantly Bx > 0 and By < 0 (the preferred IMF condition for auroral occurrence). The incident draped IMF is tilted southward in the southern hemisphere as a result of momentum exchange between the solar wind and planetary ions, creating a favorable geometry for magnetic reconnection with the local crustal fields, which have a northward horizontal component (the upper arcade)17.

Figure 2b illustrates the resulting sequence of magnetic topological reconfiguration consistent with the miniature Dungey-like cycle. It starts with the first magnetic reconnection between the draped IMF and the upper arcade. This reconnection reforms the previously closed field line (1) to two open field lines marked as 2 and 2′. As these open field lines are convected further down the tail by the solar wind flow, the second magnetic reconnection occurs between the oppositely directed open field lines (3 and 3′), generating newly closed field lines (4). These newly closed field lines relax and are then convected Mars-ward (5 and 6), completing the Dungey-like cycle.

Associated with this reconnection-driven circulation of magnetic flux, the footpoints of these field lines in the ionosphere are convected by ionospheric plasma flows, as in the case of Earth. As illustrated in Fig. 2c, the flow circulation is most likely clockwise in the upper crustal patch and counterclockwise in the lower crustal patch, so that the Mars-ward convected closed loops (4→5→6) are located towards the center, underneath which are (unopened) closed crustal loops, and open field lines are located outside of closed loops. As these ionospheric flows are driven by the E × B drift, considering the direction of the local magnetic fields, a set of converging electric fields E is needed to generate such flow circulations at both crustal patches. Note that the actual flow circulations do not necessarily extend tens of longitudinal degrees, but could be much more localized than those illustrated, depending on the crustal field pattern. This illustrated miniature Dungey-like cycle resembles half of the Dungey cycle at Earth’s dusk hemisphere, and the two crustal patches where the Mars ionospheric flow circulations take place would be equivalent to the dusk halves of Earth’s northern and southern polar regions, as illustrated in Fig. 3b–d.

Fig. 3: Schematics of three types of plasma and magnetic field cycles.Fig. 3: Schematics of three types of plasma and magnetic field cycles.

a Vasylinuas cycle30, operating at giant planets such as Jupiter and Saturn. b Dungey cycle, operating at magnetized planets such as Earth and Mercury. c Miniature Dungey-like cycle operating at Mars from this study. d The flow circulations of the Dungey cycle viewed from North and South, respectively. e The flow circulations of the miniature Dungey-like cycle from this study. All the schematics are not-to-scale. IMF is the interplanetary magnetic field.

The converging electric fields set up pairs of cross-field currents in the ionosphere that close via FACs flowing between the ionosphere and the magnetosphere. FACs flowing away from the ionosphere are supported mainly by precipitating electrons, and if these electrons have insufficient flux to carry the current, magnetic field-aligned large-scale electric fields (double layers) could develop to accelerate electrons toward the planet (resulting in nearly monoenergetic energy spectra) to carry the needed current density. These monoenergetic electrons then produce discrete auroral emissions11,12,21.

The MAVEN observations shown here contain the key elements of this process, as highlighted in Fig. 1e. The configuration of downward-upward-downward FACs in both crustal patches in Fig. 2c is consistent with the derived \({j}_{//}^{b}\) shown in Fig. 1b, e. Electron accelerations also mostly coincide with the upward FAC (\({j}_{//}^{b} > 0\)) observations (Fig. 1b, e), and the FAC carried by these accelerated electrons \({j}_{//}^{e}\) has a reasonable agreement with the upward FAC \({j}_{//}^{b}\). These relationships support the interpretation that electrons are accelerated to carry the needed FACs. Moreover, a prominent east or west (equivalent to tailward or sunward near the dusk terminator, respectively) plasma flow is observed mostly between downward-upward FACs (Fig. 1d, e), consistent with the illustrated flow circulation pattern in Fig. 2c.

In addition to the main case study presented above, Fig. 4 shows the FAC, flow velocity, and electron acceleration observations for all 7 identified case examples overlain on the geographic coordinates. These additional case studies (Events 1, 3–7) have similar signatures to the main case study (Event 2) and occurred at various geographic latitudes and longitudes. Meanwhile, Events 4 and 6 occurred at almost identical geographic locations and LT, despite being observed 11 days apart. Additionally, most of the identified events occurred at latitude [−40°, −55°] during post-dusk, while Event 3 occurred south of these events with the opposite crustal field polarity during pre-dawn (LT 04), consistent with previous findings of the dawn-dusk asymmetry in auroral observations14. The occurrence of these events at different crustal fields or at the same location but at different times suggests that this miniature Dungey-like cycle operates regularly at these miniature magnetospheres.

Fig. 4: All event examples of MAVEN observations in the geographic coordinates.Fig. 4: All event examples of MAVEN observations in the geographic coordinates.

The derived FAC (\({j}_{//}^{b}\)), the east-west component of O2+ flow velocity (∣VEW(O2+)∣ > 1 km/s), and electron acceleration observations with \({j}_{//}^{e} > 0.1\,\mu A/{m}^{2}\) as orange dots, overlain on a color map of the modeled radial crustal magnetic field at 250 km altitude35, for all identified case examples. The date and hour in UT and the local time (LT) of the MAVEN observations for these examples are listed in the lower right corner. The numbers in the brackets are the event numbers, and a and b refer to the two segments of the same event.