Strong Thermal Emission Velocity Enhancement (STEVE) is a purple/mauve-colored arc that forms in the upper atmosphere equatorward of the auroral oval (MacDonald et al., 2018). STEVE can be associated with a green ray structure called the picket fence. Figure 1 shows two photographs of STEVE on 3 September 2016. The purple/mauve arc is STEVE. As in these photographs, STEVE sometimes consists of two emission layers (Liang et al., 2019), where the top part is more purplish and the bottom part is more whitish. The green rays underneath STEVE in the image at 5:22 UT are the picket fence. The two photographs were taken only 6 min apart, but the picket fence was not present earlier at 5:16 UT. Typically, STEVE appears first, and then the picket fence, if it exists, emerges during the course of STEVE (Yadav et al., 2021; Anderson and Anderson, 2021; Martinis et al., 2022). The picket fence can remain after STEVE disappears. Figure 2 shows a photograph from another event when the picket fence is present but STEVE is not visible. As described further in Section 2, STEVE was present earlier in this event, and the picket fence without STEVE appeared near the end of the lifetime of STEVE.

STEVE generally appears before magnetic midnight during the recovery phase of large substorms and lasts ∼1 h (Gallardo-Lacourt et al., 2018; Yadav et al., 2022). STEVE has a strong seasonal dependence, where its occurrence is the highest in equinoxes. STEVE is a narrow (∼20 km width) arc but can extend over ∼2000 km azimuthally. Unlike typical auroral emissions that are bright at particular spectral lines, the purple/mauve color is characterized by continuum emissions across 400–700 nm wavelengths (Gillies et al., 2019). STEVE is associated with intense subauroral ion drifts (SAID), ionospheric upflow, plasma density depletion (trough), temperature enhancements, and electromagnetic waves in the downward field-aligned currents (MacDonald et al., 2018; Nishimura et al., 2019; Nishimura et al., 2020a; Archer et al., 2019). These features are generally seen for enhanced subauroral flows (subauroral polarization streams or SAPS), but they are more pronounced and latitudinally narrow when STEVE appears. STEVE lacks energetic particle precipitation from space.

While STEVE gained attention through recent citizen scientist photographs, a similar phenomenon was reported in historical observations of the night sky. One of the most well-documented STEVE-like events was seen on the evening of 11 September 1891 across Ireland and the United Kingdom. It was reported as “a rare phenomenon” in a series of letters in Nature as summarized in Table 1. Wilson (1891) described the emission as “a luminous band stretching from the eastern horizon to the west … like a straight tail of a large comet.” A number of readers responded to his letter and reported their observations. This emission band was characterized by its white color (Rix, 1891) and southward drift (Dreyer, 1891), and it occurred during a magnetic disturbance (Tuckwell, 1891). It also involved short-lasting (10–15 s) rays (Marshall, 1891). These emission characters are distinctly different from regular auroral emissions and strikingly similar to STEVE. The observations were spread across ∼500 km in clear skies and did not appear on the next day, excluding the possibility of other types of lights. The interaction among the journal readers is analogous to the lively communication about STEVE sightings on social media in the modern era. The review by Bailey et al., (2018) documented an extensive list of STEVE-like emissions that were found in historical records dating back to the 18th century or even earlier. Hunnekuhl and MacDonald (2020) described the work by Stormer (1935), who observed STEVE-like emissions and recognized them as feeble homogeneous arcs.

The following sections review more properties of STEVE and the picket fence and discuss outstanding unsolved issues about these phenomena. Readers are also encouraged to refer to the review papers by Gallardo-Lacourt et al., (2021) and Lyons et al., (2021) about the current understanding of STEVE.

As mentioned above, STEVE is most frequently observed in equinoxes (Gallardo-Lacourt et al., 2018). Although the occurrence statistics have not been normalized by total observation times, their results can be perceived as an equinox preference of STEVE, considering that observation times during equinoxes and winter are comparable. Many citizen scientists and general public watch the night sky in the winter, but reports of STEVE during the winter are much more limited. This seasonal dependence raises a question: why does STEVE occur preferentially in spring and fall?

Substorms are generally stronger during equinoxes (Tanskanen et al., 2011), due to Earth’s dipole tilt being closer to the Parker spiral of the interplanetary magnetic field (IMF) (Russell and McPherron, 1973). However, the difference in the substorm strength between equinoxes and winter is small (∼10%), and substorms occur more frequently during winter. The substorm occurrence characteristics do not explain the large contrast in the occurrence of STEVE between equinoxes and winter.

STEVE is closely associated with intense SAID and SAID tend to occur more often during equinoxes (He et al., 2014). The dusk side occurrence of SAID in the region of the downward field-aligned currents to the subauroral ionosphere could explain the dusk side preference of STEVE (see Section 5 on the mechanism). Thus, the equinox preference of SAID would be linked to the seasonal dependence of STEVE. However, the question remains as to why SAID are more common during equinoxes. Since SAID are closely associated with low conductance in the subauroral ionosphere, ionospheric structures may play a role in the seasonal dependence of SAID and STEVE.

In addition to the seasonal dependence, it is not understood whether STEVE has any longitudinal or hemispheric asymmetry. Observer locations and sky conditions pose a difficulty in comparing occurrence characteristics between regions. Space-based observations are necessary to answer these questions.

As Wilson (1891) recognized, another important characteristic of STEVE is that it is a rare phenomenon. STEVE tends to occur during moderate-to-large substorms under a weakly southward IMF (Gallardo-Lacourt et al., 2018). However, although SAID also favor a southward IMF, SAID strength and occurrence do not correlate with IMF B _z, B _y, storm strength, or substorm strength (Nishimura et al., 2022a). Many substorms can occur under a weakly southward IMF and without STEVE. It is puzzling why STEVE occurs only for a small number of substorms and why STEVE does not appear to require unique solar wind or geomagnetic conditions. The imaging of the auroral oval suggests that the substorm auroral surge during STEVE expands far duskward, while the substorm surge without STEVE stays till pre-midnight (Nishimura et al., 2020b). These works showed that substorms with STEVE inject more electrons to the dusk side inner magnetosphere and create stronger subauroral flows.

During the STEVE event in Figure 2, Time History of Events and Macroscale Interactions during Substorms (THEMIS) satellites were located in the dusk side magnetosphere. Figures 3A–D show selected mosaics of THEMIS all-sky imagers (ASIs). An auroral intensification started around 23 h MLT and 2:46 UT (Figure 3A). It rapidly moved duskward, crossed the footprint of THEMIS-D (Figure 3B) and even reached <19 h MLT (Figure 3C). STEVE appeared soon after the auroral surge that passed that region (Figure 3C). Figure 3D shows the snapshot near the end of STEVE at 3:16 UT, when the photograph in Figure 2 was taken near the region of the STEVE observation in the THEMIS ASI. The IMF B _z was mostly negative and AL and SYM-H were moderately elevated for many hours around this STEVE (Figures 3E–G). The auroral intensification at 2:46 UT was associated with a dipolarization (Figure 3H), flow burst (Figure 3I), and injection (Figures 3J, K) at THEMIS-D and an enhancement in |AL| (Figure 3F), marked by the vertical line. Interestingly, the strengths of those parameters during this auroral intensification were comparable to those for the other injections that are seen in this figure (at 0:20, 1:50, and 4:20 UT). The IMF and SYM-H also did not show any unique features for the 2:46 UT event. The only exception is the electron injection strength. The electron flux above a few tens of keV for the 2:46 UT event was substantially higher than that for the other injections (Figure 3K). This event supports the finding by Nishimura et al., (2020b) that the strong electron injection to the dusk side magnetosphere could be the driver for intense subauroral flows and STEVE. There may be special yet unknown conditions why the substorm energy release in the magnetotail occurs far away from midnight and transports energetic electrons more efficiently to the pre-midnight inner magnetosphere. It is important to understand what coupling processes in the global solar wind−magnetosphere−ionosphere system control the azimuthal progression of substorms, injection locations, and the development of subauroral flows.

STEVE, in some images, appears as a homogeneous arc with little structure along the arc [e.g., Archer et al., 2019b; Martinis et al., 2022; Parnikov et al., 2022]. Wilson (1891) and Stormer (1935) perceived STEVE-like emissions as a luminous band such as a comet tail and homogeneous arc. The high-altitude portion of STEVE in the right panel of Figure 1 is another example of a homogeneous arc. However, citizen scientist photographs have also revealed that STEVE has rich fine-scale structures within the arc. The photograph in Figure 4 shows that STEVE has ray structures, and the rays of STEVE are connected to individual rays of the picket fence. The low-altitude portion of the STEVE in Figure 1 also shows a similar connection to the picket fence. A number of STEVE photographs also show similar fine-scale structures along STEVE [e.g., MacDonald et al., 2018; Nishimura et al., 2019; Mende et al., 2019]. The fine-scale structures of STEVE drift westward along the arc, and the speed seems to correspond to the intense SAID flow speed (Gillies et al., 2020). The exposure time may affect the detectability of fine-scale structures in photographs, but the photographs by Martinis et al. (2022) and Figures 1 and 4, have similar exposure times (10, 13, and 10 s). There may also be conditions in the magnetosphere−ionosphere system that create fine-scale structures of STEVE. The presence of fine-scale structures raises questions about what the elementary structures of STEVE are and what mechanisms create the fine-scale structures of STEVE. There may also even be finer-scale structures that may be missed in photographs with exposure times longer than several seconds. High-speed imaging observations are necessary to identify what these elementary structures of STEVE are.

Figure 5 shows the evolution of the STEVE event of Figure 4 using the Transition Region Explorer (TREx) true color imager and spectrograph observations at Lucky Lake, Canada. STEVE appeared at around 6:10 UT and was illuminated at all wavelengths shown in Figures 5G–J, including the airglow continuum. It was initially a homogeneous arc in the purple/mauve color without the picket fence (Figure 5A). After ∼6:20 UT, the arc moved equatorward and the bottom part of the arc showed wave-like fine-scale structures as highlighted by the white arrows (Figure 5B). In keograms, fine-scale structures appear as the periodic modulation of the STEVE intensity (Figures 5H, J). This temporal variation at a fixed longitude occurs because fine-scale structures of STEVE drift westward. The fine-scale structures are tilted westward with an increasing altitude, which apparently starts first at higher latitudes in the keogram format. The azimuthal wavelength of fine-scale structures became shorter as time progressed (Figure 5C). After 6:43 UT, the picket fence appeared adjacent to (at a lower altitude) the bright parts of STEVE (Figure 5D). The picket fence can be seen as a nearly pure 557.7 nm intensification (Figure 5H). The picket fence became brighter and extended to lower altitudes, while it maintained the connection to the bright parts of STEVE, and a group of green rays developed under each bright part of STEVE (Figure 5E). The western (right) portion of Figure 5F corresponds to the citizen scientist photograph taken at Strathmore, about 500 km to the west of Lucky Lake. While STEVE that is highlighted by the right three arrows in this image was nearly aligned with the picket fence, STEVE that is highlighted by the left two arrows was shifted to the west of the picket fence. In the sequence of images, the trailing edge of STEVE was connected to the picket fence but the leading edge drifted westward faster than the picket fence. STEVE and the picket fence disappeared soon after 7:00 UT. This event showed interesting developments of the fine-scale structures of STEVE, the initiation of the picket fence, and the connection between the fine-scale structures of STEVE and the picket fence. It raises questions about how wave-like fine-scale structures of STEVE form and how they give rise to the picket fence. More observations about the evolution of STEVE and the picket fence are needed to understand how those emissions evolve over time and how they are related to each other.

Knowledge about the picket fence is more limited than that for STEVE. The picket fence typically appears with STEVE, but STEVE can appear without the picket fence. As shown in Figure 2, there are moments when the picket fence can be present without STEVE. Both STEVE and the picket fence drift westward, but the picket fence is substantially slower (Gillies et al., 2020). What are the processes that determine the occurrence, drift speed, and duration of the picket fence? It is also unclear what conditions set the wavelength of the picket fence.

Another peculiar appearance of the picket fence is that the picket fence can be localized to a small portion of STEVE. While the picket fence typically forms an arc with many rays, the photograph in Figure 6 shows that the picket fence can consist only of a few rays, despite that STEVE is much more extended. It is a challenge to explain what mechanisms can confine the picket fence to a small portion of STEVE. STEVE in this photograph has a small kink near the picket fence. A distortion of the SAID flow channel or trough density gradients may play a role in the formation of the picket fence. It is important to determine flow and density structures along STEVE in order to understand how the picket fence forms.

Although the picket fence was originally recognized as periodic ray structures (in an analogy to the pickets of a fence), high-speed imaging showed that the picket fence also has fine-scale structures (Semeter et al., 2020). The picket fence can have wave-like modulation, and the picket fence can have green streaks appearing below it. While the picket fence drifts westward along STEVE, the green streaks appear poleward of STEVE and drift toward STEVE. The picket fence in Figure 4 shows such streaks underneath and poleward of the picket fence. The picket fence may be a time-aliased form of more dynamic emission structures, and a simple notion of ray structures (i.e., pickets) may not correctly represent the actual form of the green emission. High-speed imaging is important to resolve the true structure and evolution of the picket fence. The 3D structure of each green emission, including the wavelength and alignment to the magnetic field, should also be determined.

How the magnetosphere plays a role in the generation of the picket fence is another important question. Although the green ray structure resembles rayed auroral arcs that are created by energetic electrons precipitating from the magnetosphere, so far there is no consensus about whether electron precipitation is involved in the picket fence. Several works support the precipitation hypothesis. Low-altitude satellite observations show energetic electrons when the picket fence is present (Nishimura et al., 2018). The altitude profile of the green emission can be explained by the energetic electron precipitation (Bennett and Bourassa, 2021). A simulation of the feedback instability indicates that electrons are accelerated along magnetic field lines and create a wave structure along STEVE (Mishin and Streltsov, 2019; 2022). On the other hand, the picket fence has a minimal N 2 + 1N emission but intense N 2 1P emission, and thus only <18.75 eV of the excitation energy is required to illuminate the picket fence (Mende et al., 2019). The fine-scale structures of the picket fence mentioned above (Semeter et al., 2020) suggest that waves or turbulence in the ionosphere are responsible for the picket fence.

The characteristic features of STEVE in the ionosphere are the extreme flow velocity and heating in the subauroral region that generally lacks precipitation over STEVE. This unique state of the ionosphere seems to be the condition for the formation of STEVE. The continuum emission of STEVE led to a hypothesis that the STEVE emission arises from the reaction between N O and O ( N O + O   →   N O 2 + h ν ) (Gillies et al., 2019; Mende et al., 2019; Harding et al., 2020; Mishin and Streltsov, 2022). This reaction requires vibrationally excited N 2 that interacts with O and produces N O ( N 2 + O   →   N O + N ). The key question is whether there is a sufficient amount of the excited N 2 and resulting N O in the upper atmosphere to create the brightness of STEVE. An investigation is required as to whether the extreme velocity and temperature in the ionosphere sufficiently heat the thermosphere and provide N 2 at STEVE emission altitudes [∼130–270 km (Archer et al., 2019b)]. In the model calculation by Harding et al. (2020), the emission through these chemical reactions is stronger at lower altitudes (>130 km), while the emission is minimal at >200 km. Whether the mechanism that involves molecular ions can explain the emission well above the 200 km altitude is an interesting question. It is important to conduct neutral composition observations and simulations to address whether chemical reactions can explain the STEVE brightness and its altitude profile or whether other reactions need to be considered.

Extreme temperature enhancements are a characteristic feature of STEVE. The extreme flow velocity could be related to the heating, but a simulation by Liang et al. (2021) showed that frictional heating alone does not explain the extreme electron temperature. Another heat source such as heat fluxes from the magnetosphere is required, but because of the difficulty of measuring heat fluxes, it is not known how large the heat fluxes over STEVE are. Enhanced heat fluxes are also thought to be the source of stable auroral red (SAR) arcs, another type of subauroral emission that is dominated by the red color, but what mechanisms distinguish the nature of heating between STEVE and SAR arcs are not understood (see also the next section).

The extreme velocity also brings a challenge to simulating ionosphere−thermosphere dynamics. An intense SAID velocity of ∼5 km/s corresponds to ∼1 eV (or ∼11,600 K) energy of O + ions. Since the measured ion temperature in the topside ionosphere is only ∼2000–3000 K (Nishimura et al., 2020a), the ion kinetic energy is larger than the thermal energy. This supersonic condition is not generally expected in simulation schemes of the ionosphere, and it may bring a challenge about how to treat dynamics and chemistry properly under such fast flow conditions.

The extreme velocity and heating of the ionosphere during STEVE have substantial impacts on ionospheric dynamics. The F-region plasma density decreases with the flow speed due to enhanced collisions between O + and neutrals and their subsequent recombination (Schunk et al., 1975; Liang et al., 2021a). The major species also change from O + to N O + due to the O + + N 2 → N O + + N reaction. The plasma loss creates the density trough, and the steep density gradient can be a source of the plasma instability. For weaker and latitudinally broader SAPS flows, the trough density gradient has been shown to create density irregularities through the gradient drift instability and Kelvin−Helmholtz instability (Rathod et al., 2021). Steeper density and velocity gradients for intense SAID may create an even stronger instability. Figure 7 shows an event where STEVE and the picket fence show a large undulation. The undulation also has a large luminosity gradient along STEVE. This feature suggests that the SAID flow channel and trough could become unstable and deform the velocity or temperature structure in the ionosphere through instabilities. Simulations should be conducted for understanding what plasma instability develops during STEVE. The knowledge from such simulations is also important to interpret the fine-scale structures of STEVE and the picket fence.

Steep density gradients also impact radio wave propagation. High-frequency (HF) radio-wave backscatter echoes have been shown to increase (Nishimura et al., 2020b) or disappear (Perry et al., 2022) in association with STEVE. The two different outcomes suggest the complex radio wave propagation at the density trough. Radio waves at higher frequencies such as GHz waves for the global navigation satellite system (GNSS) may also be affected by density irregularities that develop at steep trough density gradients. Simulations of the density irregularity formation and radio wave propagations are required to understand the impact of steep trough density gradients on radio communication.

STEVE could also impact thermosphere dynamics. A neutral wind shear has been observed to develop around STEVE (Liang et al., 2021b). Heating in SAPS has been shown to be the source of the traveling atmospheric disturbance (TAD) (Guo et al., 2018). The extreme heating in STEVE may drastically change the thermospheric pressure and wind pattern. Thermospheric simulations under the presence of extreme flow velocity and heating should be conducted to understand how the thermospheric dynamics change under such conditions.

In addition to STEVE, SAR arcs and the subauroral proton aurora are major optical emissions in the pre-midnight subauroral ionosphere. Traditionally, the three types of subauroral emissions are often studied separately, and their relations are not well known. Limited investigations have shown that STEVE can occur after the proton aurora (Nishimura et al., 2020b) and that a SAR arc can transition to STEVE (Martinis et al., 2022). Subauroral emissions are not exclusive to each other, but STEVE and a SAR arc can be adjacent to each other simultaneously (Liang et al., 2019; Chu et al., 2019; Martinis et al., 2021). The proton aurora and red arc that transitions to a SAR arc later have also been reported to co-exist during substorms (Nishimura et al., 2022b). However, STEVE was not detected in these events.

Similar to STEVE, SAR arcs and the proton aurora occur during storms and substorms, and they are related to particle injections to the inner magnetosphere and fast plasma flows in the subauroral ionosphere. Despite the similarities, the three types of emissions generally occur separately, and it is not understood how the system controls the occurrence of the three types of emissions under similar magnetosphere−ionosphere conditions. The transition between the two types of emissions indicates that driving conditions for the two types of emissions can switch from one to the other during a substorm, e.g., the SAPS flow speed increases and becomes narrower during the transition from the proton aurora or a SAR arc to STEVE. The simultaneous existence of STEVE and a SAR arc suggests that an intense SAID and weaker SAPS flow channels are located adjacent to each other. Observations of the time evolution of flow and particle structures are required to address what magnetosphere−ionosphere conditions change during the emission transition.

This paper reviewed the key properties of STEVE and the picket fence and discussed major open questions that need to be addressed to advance the understanding of these phenomena. Although the understanding of STEVE and the picket fence has rapidly progressed over the past few years, a number of important issues have not been addressed. STEVE should be placed in the context of the subauroral magnetosphere−ionosphere−thermosphere coupled system. The system-level behavior needs to be understood in order to address questions about STEVE occurrence conditions and substorm development, particularly the behavior of the electron injection during STEVE and the relation to SAR arcs and the proton aurora. Ionosphere−thermosphere modelings that consider extreme velocity and heating should be conducted to answer what chemical and dynamical processes occur and how much the STEVE luminosity can be explained. The modeling of instabilities is also required to understand how fine-scale structures of STEVE and the picket fence form. High-resolution observations are required to resolve fine-scale structures of STEVE and the picket fence, and such observations are important to understand the underlying processes in the ionosphere and thermosphere.