As the optical manifestation of atmospheric ionization through precipitating electrons from the magnetosphere, auroral arcs are commonly observed as narrow bands in the nightside auroral oval. They can, however, reshape and distort into numerous small- and meso-scale structures, among those auroral spirals (Paschmann et al., 2003). An auroral spiral can occur as an individual structure or in a vortex street with multiple spirals (Hallinan, 1976). On average they have diameters of 25–75 km (Partamies et al., 2001b), but can reach sizes up to 1,300 km and thereby they are the largest vortical forms found in an auroral arc (Davis and Hallinan, 1976). As stated by Partamies et al. (2001b), spirals develop primarily during quiet magnetospheric conditions. In the course of a substorm, they occur mainly during the expansion and beginning of the recovery phase (Hallinan, 1976).

Partamies et al. (2001b) observed that the drift motion of auroral spirals coincides with the direction of the ionospheric convection. As viewed from above in the northern hemisphere, auroral spirals wind in a counterclockwise direction around upward field-aligned currents (FACs). Increased auroral precipitation in the FACs connected to brighter aurora is correlated with the winding of the spiral, while its brightness decreases with unwinding (Partamies et al., 2001b). Instead of unwinding, spirals can also decay into patchy auroral structures (Davis and Hallinan, 1976). While the auroral spiral was subject to several ground observations (Davis and Hallinan, 1976; Partamies et al., 2001b) and theoretical approaches (Hallinan, 1976; Partamies et al., 2001a), correlated space and ground measurements as performed by Keiling et al. (2009b) have rarely been done. This is due to a lack of events during which satellites were positioned in a magnetospheric region conjunct by magnetic field lines with the region of a spiral observation. Therefore, the exact generation mechanism and location, which may also vary for different magnetospheric states, is not yet established. In this paper we study an eastward moving spiral event with space measurements and propose a spiral generation mechanism following from the theory by Keiling et al. (2009b).

On 18 September 2013 at 22:58 UT, an auroral spiral was observed in Tromsø during the expansion phase of a substorm. Simultaneously, measurements in the magnetotail by three Cluster and the THEMIS-A spacecraft were obtained.

The three Cluster spacecraft measured strong variations in the magnetic field components, the plasma flow velocity and the electron flux approximately 1–2 min prior to the spiral development in Tromsø (see Figure 4). The plasma flow velocity from Cl4 over a time interval of 8 min showed quasi-periodic peaks in velocity up to ∼500 km s⁻¹ in X and Y directions. These measurements correspond to the signatures of bursty bulk flows with intermittent flow bursts (Angelopoulos et al., 1992). As they commonly are (Paschmann et al., 2003), these bursty bulk flows were accompanied by dipolarization signatures seen in the B _Z magnetic field components. The vector plots of the Cluster magnetic field projected onto the L-M plane of the local LMN coordinate system, which varies for each spacecraft, also show vortical signatures (see Figure 7). Cl2 encountered a magnetic field vortex street approximately oriented in the X-Y plane (GSE) with opposed rotating vortices on either side of its trajectory, as shown in Figure 9. Cl1 and Cl4 only measured a pair of opposed rotating vortices located roughly in the X-Z plane (GSE). Cl1 and Cl4 were positioned close together, so it is likely that they observed the same pair of magnetic field vortices.

If the frozen-in condition is valid, it is expected that the magnetic field vortices are connected to plasma flow vortices. The plasma flow velocity observed by Cl4 exhibited a change of direction at the occurrence of the magnetic field vortices, which suggests a vortical structure. Clear signatures of plasma vortices were not detected by the Cl4 spacecraft, likely because the CIS instrument used to measure the plasma velocity has a time resolution of ∼4 s. The magnetic field vortices however occurred at a temporal scale of ∼30 s.

These magnetic field vortices were probably associated with the bursty bulk flows that were also measured by Cl4. Bursty bulk flows are known to cause vortical structures in plasma flow in the X-Y (GSM) plane, when they encounter the dipolarized region of the inner magnetosphere (e.g., Panov et al., 2010; Birn et al., 2011). It was also observed by e.g., Volwerk et al. (2007) that Kelvin-Helmholtz instabilities appear together with bursty bulk flows. Hasegawa et al. (2004) detected plasma flow and magnetic field signatures caused by a Kelvin-Helmholtz instability at the magnetopause. Their magnetic field perturbations are different than shown in Figure 7, because the perturbations in our study are probably caused by pairs of oppositely rotating vortices, not by a single vortex street. TH-A was located closer to Earth and did not measure any vortical structures (see Figure 8). As seen in Figures 4J, K, Cl1 observed signatures in the electron temperature and density that are typical for Kelvin-Helmholtz instability, similar to Cluster observations described by Hasegawa et al. (2009). Therefore, the Cluster satellites likely observed a Kelvin-Helmholtz-like vortex street in the magnetic field induced locally by the measured bursty bulk flows. It has however to be taken into account that part of these results are based on the MVA, which produced results that were not ideally reliable.

Keiling et al. (2009b) observed short-lived equivalent ionospheric current vortices of 600–800 km size associated with the appearance of auroral spirals. In this study, such short-lived vortical J _eq structures could not be resolved. This can be due to the size of the auroral spiral (189 ± 45 km) and the comparative spatial resolution of the IMAGE magnetometer network. The stations in the vicinity of Tromsø have an average calculated distance of ∼ 140 km to Tromsø, but the coverage in this area is limited by the ocean to the north and west. Therefore, the only station in the spiral region (marked by the green solid ellipse in the middle panel of Figure 2) is Andenes (AND). According to the estimates of the spiral location and size at the time of the photograph in Figure 1 the spiral was not located in the zenith over Tromsø, and only marginally over Andenes. This can potentially lead to the missing deflection of the ionospheric currents during the optical spiral. To better resolve such structures, more mesoscale observations as described by, e.g., Gabrielse et al. (2023) would be helpful. Another possibility for the absence of vortical J _eq structures is that the particle precipitation during this observed spiral event can have been too weak to deflect E-region plasma flows.

To be able to correlate the ground observations and space measurements, the ionospheric footprints of the satellites were derived with the T96 external magnetic field model. They were located in the pre-midnight sector of the auroral oval (see Figure 5). The footprints lay at a distance from Tromsø of ∼ 68 ° and ∼ 34 ° in longitude for Cluster and TH-A, respectively. Since the spiral was developing up to ∼ 5 ° in longitude west of Tromsø, as inferred from the field of view from Tromsø (Figure 1), the longitudinal distance of the footprints to the visible spiral may have been smaller than their distance to Tromsø. In the magnetosphere, the possible source region of the auroral spiral was located ∼10.1 R _E in Y (GSE) direction from the Cluster satellites. It has to be taken into account that topological magnetic mapping is not very reliable during substorm conditions in particular (Paschmann et al., 2003). The T96 model may describe the overall conditions of this event better than the newer models, but it does not specifically represent substorms (Tsyganenko, 1996). For a strongly disturbed magnetosphere, it is known not to display the dawn-dusk asymmetry of the inner magnetosphere and to overstretch the field lines (Tsyganenko, 2001). Also, the discrepancy between the results of the newer models T02 and T05 and the T96 model is very high, since the newer models generate an entirely different field line topology in placing the spacecraft on open field lines. This shows the high uncertainty in the mapping and therefore also in the interpretations. With these aspects considered, the footprints calculated with the T96 magnetospheric model are regarded only as an approximation for the real ionospheric footprints of the spacecraft. The derivation of the magnetospheric source region of the auroral spiral is treated equally. Due to the uncertainty of magnetic mapping during substorm conditions, the Cluster satellites may have been located closer to or further from the source region of the spiral than the computations suggest. Also, the plasma velocity measurements discussed above exhibited significant flow bursts and the electron temperature and density showed signatures typical for KHI. Approximately 2 min before the spiral development, the plasma velocity started to increase to an absolute value of v ≃ 960 km s⁻¹. The flow bursts could have produced the vortices in the magnetic field and plasma. With a distance d = 12.1 R _E between the spacecraft and the possible source region and t = d/v, a propagation time t ≃ 80 s of the vortex street observed by Cl2 to the generator region can be estimated. A schematic illustration of this movement is given in Figure 10. It is assumed that the propagation of the plasma flow in X direction is slowed due to the dipolarized magnetic field near the Earth. In Figure 5, the ionospheric footprints of this motion are depicted.

Following from the above calculation it can be possible that the magnetic field vortices, associated bursty bulk flows, KHI and dipolarizations detected by the spacecraft and described above were correlated with the spiral formation. What must be emphasized again, however, is the high uncertainty in the field line mapping during substorm conditions. This makes these interpretations mostly suggestions for a generation mechanism of the auroral spiral, based more on the temporal correlation of the vortices in the magnetotail and the optical spiral than on the exact spatial positions.

This suggestion proposes the spiral formation taking place already in the magnetosphere, not only in the ionosphere. Therefore, taking into account the available data, the spiral generation theory presented by Hallinan (1976) is not applicable to this event, since it states that the spiral only fully develops in the ionosphere.

According to a theory by Keiling et al. (2009b), plasma flow vortices in the magnetosphere are the initiator of a spiral, which then maps to ionospheric altitudes. These vortex structures can in turn be caused by either ballooning instabilities (e.g., Voronkov et al. (1997)) or bursty bulk flows (Keiling et al., 2009a; Panov et al., 2010). On encountering the dipolarized magnetic field in the inner magnetosphere, the high-speed plasma flows are decelerated (Panov et al., 2010; Birn et al., 2011). By their deflection around the dipolar magnetic field, vortices with opposed sense of rotation form on both sides of the flows (Panov et al., 2010).

The observations of dipolarizations and bursty bulk flows in this study coincide with the theory by Keiling et al. (2009b). In contrast to the detected structures of this event, Keiling et al. (2009b) also observed opposed rotating J _eq vortices that developed and decayed concomitant with auroral spirals. They associated these vortices with upward and downward field-aligned currents, which were driven in the magnetosphere by a pair of vortical plasma flows. According to Keiling et al. (2009b), the generation of the auroral spiral was likely initiated in the region of strong plasma flows between the plasma flow vortices in the magnetosphere. Here, such conclusions can not be drawn in the same way, because there were no equivalent ionospheric current vortices detected. The evidence of a vortex pair of J _eq in the ionosphere connected to plasma flow vortices in the magnetosphere was also not given. This can be due to the lack of mesoscale observations that can resolve these structures. However, with ∼500 A km⁻¹ the westward electrojet current observed was stronger than an average substorm expansion current (Juusola et al., 2015), which may have helped establish an the instability for the spiral formation. Also, the magnetic field and possibly plasma vortices observed in the magnetotail caused by the eastward bursty bulk flows can have propagated toward the spiral source region and then mapped to the ionosphere as proposed similarly by Keiling et al. (2009b).

For future research on the interaction of magnetosphere and ionosphere in the generation of auroral spirals extended mesoscale observations in the ionosphere are needed. Also, for optical detection of auroral spiral events complementary to all sky cameras, citizen science projects like Aurorasaurus (MacDonald et al., 2015) are helpful.