Fabry-Perot interferometer observations of sustained strong vertical wind activity in the auroral thermosphere

The vertical component of the neutral wind velocity is critical for understanding the thermospheric dynamics at high latitudes. The neutral vertical wind circulation, driven by auroral activity, redistributes heat and momentum within the lower and upper thermosphere between 100 and 500 km. This flow affects the neutral atmospheric composition, thus contributing to the auroral thermospheric density variability, which is important to the monitoring of neutral density in general and to measuring the orbital satellite drag specifically. We used ground-based vertical wind and temperature observations from Eagle and Fort Yukon narrow field Fabry-Perot interferometer observatories in Alaska along with nearby and collocated Scanning Doppler Imagers, spectrographs, and incoherent scatter radar observations to further understand the physics underlying the production of unexpectedly strong sustained vertical wind (SSVW) events. Our observations found one example of vertical wind velocities of 100 ${\text{ms}}^{-1}$ together with a temperature increase of 450–500 K. Our analysis suggests that while Joule heating within the lower thermosphere played a significant role during these events, the Alfven F-region heating contribution may also be significant and comparable to the energy associated with soft electrons. This view is based on the work that found that significant auroral heating can result from the thermospheric absorption of Alfvén waves populating “broadband” auroral displays. Overlapping with our FPI observations of strong sustained vertical wind (SSVW) events was detected thermospheric wave activity with maximum vertical wind speeds of 20–30 ${\text{ms}}^{-1}$ and varying periods from 20 to 40 min, which may be part of the Alfven wave heating process taking place. Thus, both Joule and Alfven wave heating sources taken together may account for the routine incidence of SSVW events as seen in many ground-based narrow field FPI measurements reported over the years.

1 Introduction

The neutral vertical velocity is a critical parameter for understanding the dynamics of the high-latitude thermosphere. The neutral vertical circulation, driven by auroral activity, redistributes heat and momentum within the lower and upper thermospheric region between 100 and 500 km (Larsen and Meriwether, 2012; Dhadly et al., 2023; Sarris, 2019; Lotko and Zhang, 2018; Hogan et al., 2020). Also important in this context is how this flow might affect the neutral composition, thus contributing to the auroral thermospheric density variability, which is important to the monitoring of atmospheric density in general and orbital satellite drag specifically, i.e., Qian and Solomon (2012); Emmert (2015); Nwankwo et al. (2021), Richmond (2021) for example. The electrodynamic coupling in the F-region extends both upward to the magnetosphere and downward to the E-region, contributing to the development of plasma instabilities that are also partially governed by this vertical flow.

Measurements of the thermospheric vertical velocities are challenging, but the available measurements, both from ground-based and in-situ instruments, have formed a consistent picture over the last few decades, showing that unexpectedly large and sustained vertical velocities are a common feature of the region (Rees et al., 1984; Aruliah and Rees, 1995). Thus, understanding the physics underlying the production of sustained strong vertical wind (SSVW) events in the auroral thermosphere has been a long-standing challenge in studying the neutral dynamics of this region, especially during geomagnetically active conditions. This problem extends further to understanding the polar cusp dynamics where density observations show a significant density increase of nearly a factor of 2 at 400 km (Lühr et al., 2004). The concern about the forcing source of SSVW events also extends to mid-latitudes where it is difficult to reconcile the detection of vertical winds in excess of 10–20 ${\text{ms}}^{-1}$ with theoretical predictions (Schmidt et al., 2025; Hernandez, 1982).

Larsen and Meriwether (2012) presented examples illustrating the SSVW events for periods of auroral activity which were unusual not only because of the large magnitude of the vertical velocities, reaching speeds as large as 100 ${\text{ms}}^{-1}$, but also because the upwelling is often seen to be sustained over periods of 60 min or longer. Thus, the upwelling speed in SSVW events tends to be quite large and often greater than 20 ${\text{ms}}^{-1}$. Of particular interest are the high upward speeds found for the winds described in observations obtained with three FPI instruments located in central Alaska during the intense geomagnetic storm that occurred during the night of 17 March 2015 (Lu et al., 2023). In the example shown in Figure 1, wind speeds exceeding 50 to 100 ${\text{ms}}^{-1}$ for a few minutes were seen for each of the three instruments showing that for this storm event the upwelling flow was widespread, extending over an area at least a third the size of Alaska. The plasma convective flow was quite high corresponding to an electric field, 80 mV/m. The auroral electrojet index (AE), generally used as a measure of geomagnetic activity, reached a level of 2000 nTs for this event, indicating that this geomagnetic storm was an extremely active period.

Figure 1

Figure 1. (a) An example showing SSVW events taking place simultaneously in central Alaska during the St. Patrick storm on 17 March 2015; (b) a map of central Alaska illustrating the locations of the three FPI observatories that made thermospheric wind and temperature measurements at Toolik, Kaktovik, and Eagle (from Lu et al., 2023).

A similar finding was reported by Conde et al. (2018), who used a multistatic ground-based FPI instrument configuration, combining the data from the Scanning Doppler Imager (SDI) network and the Poker Flat Incoherent Scatter Radar (PFISR) instruments to study the behavior of thermospheric wind fields during strong variations in the ion convection patterns for different geomagnetic storm cases. Their estimates found large localized vertical winds greater than 80 ${\text{ms}}^{-1}$ over a significant portion of central Alaska even during periods of small horizontal vorticity and divergence. An example of these observations is presented in Figure 2 for SDI data obtained by Conde et al. (2018) on 31 January 2016 at 0920 UT. In this case the AE index reached a magnitude of $\sim 700$ nT, which might be regarded as representative of active but moderate geomagnetic activity. Comparison of the results obtained by the narrow-field FPI instruments that were used in this study with the results obtained by the operation of the SDI imaging systems (Dhadly et al., 2015) showed the wind and temperature results for the two FPI instruments of quite different designs to be in very good agreement.

Figure 2

Figure 2. Left, 630 nm multi-static SDI measurements obtained on the 21 January, 2016 at 0920 UT applied to determine the thermospheric wind vector field (white) combined with the PFISR ion drift vector field (orange). The SDI intensities (green) are shown in the background. Right, top panel, color shade map showing the spatial distribution of vertical winds derived from multi-static SDI observations. The other two panels show color shade maps of the divergence and vorticity. Figure selected from Conde et al. (2018).

Results generally consistent with the ground-based measurements have been obtained from the analysis of accelerometer data collected during multiple orbital passes by the GOCE satellite (Visser et al., 2019). Vertical wind velocities were calculated from GOCE satellite accelerometer data. Examination of the vertical wind spread showed significant increases for AE greater than 800 nT for high-latitude regions. The approach used in that paper is similar to that used by Innis and Conde (2001) and is based upon assessing the standard deviation of the vertical wind rather than its magnitude. Evidence for large vertical wind structures was found, especially for the geomagnetic midnight sector.

Since the summary of results on large and long-lived vertical velocities in the thermosphere presented by Larsen and Meriwether (2012), progress in advancing our understanding of the underlying physics of SSVW’s has been relatively slow. More recent FPI vertical wind observations have shown additional SSVW events (Meriwether et al., 2019; Conde et al., 2018). Moreover, recent modeling studies to investigate such events have been undertaken, but with limited success. Lu et al. (2023) compared the FPI observations presented in Figure 1 with modified Thermosphere Ionosphere Electrodynamics General Circulation Model (TIEGCM) predictions to show the neutral dynamics over the area in central Alaska in the St. Patrick Day 2015 storm, including the neutral upwelling. Their study quantitatively assessed the magnitude of the drivers that contributed to the development of the observed SSVW events. Their model improvements centered on the inclusion of both the DC and AC components of the electric field from AMISR measurements to the overall Joule heating rate, the assimilation of particle energy data to calculate the extent of both hard and soft particle heating, and increased horizontal and vertical spatial model resolution within the region of central Alaska. These improvements were significant but achieved for the region of central Alaska a maximum upwelling speed of only $\sim 50$ ${\text{ms}}^{-1}$, lasting for 10–30 min near 250 km during the period of strongest geomagnetic activity. Whether other more sophisticated models with better spatial resolution can produce results more consistent with the observations is not clear.

The mystery extends to the polar cusp region where it is known that the atmospheric density inferred from the Champ satellite accelerometers has been seen to increase by as much as a factor of 2 Lühr et al., (2004), and on average, more typically to be 33% greater (Kervalishvili and Luc, 2013). Joule heating has been cited as a driver of SSVW events, especially within the cusp region. However, Carlson et al. (2012) has pointed out that such E-region heating normally would lack the energy needed to produce the magnitude of upwelling observed at heights perhaps 100 km above the E-region heights where the maximum Joule heating associated with plasma convection takes place. The atmosphere’s total weight above 110 km is nearly 300 times the weight of the air above 190 km, for example. Thus, consideration of the energy required to account for the mass of neutral air moved upward by this upwelling phenomenon into the thermosphere shows that the energy required is two orders of magnitude more than what might be available from nominal Joule heating in the lower E region. However, heating near 190 km and above would make thermal expansion to affect the neutral density at 400 km easier to accept.

Other modeling studies have attempted to explain the mass density enhancements (e.g., Deng et al., 2013) with results that are typically of order $\sim 10\%$. In contrast, upwelling speeds as large as 200 ${\text{ms}}^{-1}$ have been observed by FPI instruments at the Kjell Hendriksen Observatory located in Svalbard (private communication, A. Aruliah).

Thus, we suggest that while enhanced Joule heating within the E-region by DC E-fields may contribute to the thermospheric upwelling that produces density anomalies for the upper thermosphere region, we also believe there is a need for a broadened view of the physics associated with auroral heating. Clearly, this puzzle is an important problem in space physics and represents a challenge to our current understanding of the complex interactions of the dynamical and electrodynamical processes active in the auroral thermosphere.

Larsen and Meriwether (2012) suggested that thermospheric wave activity as a possible contributing factor in the vertical circulation within SSVW events. Ford et al. (2007) reported on the climatology of wave events based upon the analysis of FPI data for three sites in the auroral region. Numerous detections of events were found with a broad range of periods between 60 min and many hours. These waves were taken to be gravity waves (GW) that they suggest were likely generated within the altitude range of the 630 nm volume emission profiles. This inference was based on the fact that the thermosphere would dissipate high frequency GW modes more quickly than low frequency modes. The observed range of periods, however, was only seen to extend to intervals of 20–30 min, near the thermospheric Brunt-Vaisala period of 20 min. A second finding was that the inferred wave propagation direction tended to be perpendicular to rather than along the orientation of the auroral oval. Moreover, their data did not show any strong indication of a correlation of the occurrence frequency of the wave events with geomagnetic activity.

Itani and Conde (2023) reported on the detection of oscillatory perturbations in scanning Doppler imager (SDI) thermospheric line-of-sight (LOS) wind data that were observed over a broad range of look directions simultaneously on many nights. Oscillations were also seen in E-region LOS wind and F-region Doppler temperature data. They found that the wave structure amplitudes were closely correlated with auroral and geomagnetic activity. Phase relationships between perturbations observed in different look directions were used to identify time intervals when the oscillations were likely to be due to traveling GW events. However, a number of instances were noted in which the oscillations had characteristics suggesting production within the auroral ionosphere that were not identified with GW events.

In this paper we present several case studies featuring examples of SSVW events observed with the Fabry-Perot instrument installed at the Eagle Observatory ($64.79°$N, $147.2°$W) during moderate and strong geomagnetic activity. We will also show that blended within these examples are episodes with active wave activity with periods of 30 min or longer, suggesting that these waves may play a role in contributing to the dynamics of these events.

2 Instrumentation

The Eagle and Fort Yukon results reported in this paper are based upon the FPI vertical wind and temperature measurements that were acquired by the Eagle (64$°$ 47′ 10″ N, 141$°$ 12′ W) and Fort-Yukon (66$°$ 34′ 5″ N, 145$°$ 15′ 36″ W) imaging FPI instruments. These instruments have a narrow field of view of $\sim$1.5 $°$. The etalon aperture diameters and the etalon spacer gaps for these two instruments were 4.2 cm and 10.0 cm, and 1.5 cm and 2.049 cm, respectively. An objective lens of 299.2 mm and 393.5 mm focal lengths produced 9 and 8 complete rings for the 1D interferograms for the Eagle and Fort-Yukon FPIs, respectively. The bandpass for the 630 nm filter mounted ahead of the etalon was $\sim$0.8 nm. The CCD camera used for both instruments has a 1024 × 1024 pixel array with a dark noise of 0.0001 counts/sec or less for the cooler temperature of $\sim -80°C$. A binning of 2 × 2 was chosen for these imaging observations to reduce the readout noise impact upon these observations.

The low noise CCD camera has a 90% imaging quantum efficiency at the 630 nm wavelength. Furthermore, the field of view of each of the nine rings in the FPI interferogram seen at the 630 nm wavelength is equivalent to the field of view of a single-order scanning FPI such as that discussed by Hernandez (1982). The overall FPI field of view for the Eagle FPI is $\pm 1.5°$ Unlike a scanning instrument, which passes only $\sim 10\%$ of a full order at any given time, the FPI imaging optics collect light continuously over each full order.

Considering these three instrumental factors affecting sensitivity, despite the reduced aperture diameter, these FPIs achieve an overall sensitivity gain of $\sim 100$ compared to a gap-scanning FPI equipped with a 15 cm etalon and a photomultiplier. This enhanced sensitivity allows measurements of zenith wind speeds and temperatures with typical standard errors of 3–5 m ${\text{s}}^{-1}$ and 10–15 K, respectively, for 2-min exposures under 630 nm nightglow intensities of 20–50 R. These errors decrease significantly for auroral intensities of 100 R or greater.

The analysis of FPI images to extract wind and temperature measurements from FPI data is based upon the implementation of the profile fitting analysis as described by Harding et al. (2014). This approach is based upon first determining the ring center position followed by performing an annular summing of the 512 × 512 superpixel signals about the ring center as described by Coakley (1995) to determine a one-dimensional interferogram. Typically, 500 points are chosen along the radius with 55 points per order at 630 nm. The FPI instrument function at the 630 nm emission wavelength is calibrated using a frequency stabilized HeNe 632.8 nm laser source. The position of the ring center is determined by analyzing cross-sections of the observed interferogram image at 15° intervals. For each cross-section cut, the two fringe peaks of the innermost order relative to the ring center were estimated assuming a Gaussian function for the spectral profile. The ring center pixel coordinates were averaged using these 24 cross-section estimates of the ring center position.

Once the position of the image ring center is calculated, a 1-D interferogram is generated by the application of annular summing of the pixel signals for each annular ring as described byCoakley (1995). Then the FPI temperature parameter is derived from the determination of the Doppler broadening detected relative to the instrumental width seen in the FPI instrument function. The relative intensity is calculated using the area of the 1D spectral profile. These three parameters are determined using a non-linear least square fitting procedure applied to the 9 order 1D interferogram ((Innis et al., 1999; Meriwether et al., 2011; Shiokawa et al., 2012; Nicolls et al., 2012; Harding et al., 2014)).

The FPI analysis algorithm follows the methods outlined in Harding et al. (2014). The Harding et al. (2014) blurring model is used to fit the interferogram peaks under the assumption that some vignetting is causing the peak height to decrease slightly and the instrumental width to broaden somewhat. The usage of a non-linear Airy function with optical distortion parameters as the instrument function allows for the best estimate of accurate wind velocities.

The calculation of the Doppler zero assumes that the averaged amplitude of the zenith wind speed over long periods of time (5–10 h) is small relative to the horizontal speeds. Data associated with twilight periods are removed from the set of zenith measurements used for this determination. The value of the Doppler zero reference may be in error by a few ${\text{ms}}^{-1}$ if there is a substantive period of vertical wind activity within the period of time chosen for this baseline determination. Data associated with cloudy periods were also excluded from our analysis.

One concern regarding the analysis of 630-nm observations for this instrument is whether the 0.8-nm passband of the 630-nm filter is sufficient to remove or attenuate the weak OH contamination present at 627.9 nm. Recent work by Kerr et al. (2023) has demonstrated that the existence of this OH emission will introduce an error in the Doppler shift and Doppler broadening retrieval of the LOS wind velocities and the temperature derived from the comparison of the forward model to the data. While the OH emission does represent a possible source of error, we suggest that it is not serious, as the 630.0 nm emission rate for periods of auroral activity is much greater than the attenuated intensity contributed by the OH emission. Filter fabrication technology has improved such that more narrow bandwidth filters may be used to help ameliorate this concern in future measurements with new FPI instrumentation.

We note that it is easier for the narrow field FPIs, such as the one at Eagle to detect these waves, as opposed to the all-sky scanning Doppler imagers operated in Alaska. The FPIs have narrow fields-of-view less than 2$°$ in contrast to the much larger size of the SDI “superpixels” field of view. The narrow field FPIs would thus be more sensitive to the passage of Alfven wave events as compared with the broadening associated with the SDI observations made by these pixels.

3 Observations

Before presenting specific examples of the upwelling phenomenon observed with the Eagle FPI, we show in Figure 3 monthly climatology of vertical winds for clear nights using data selected for quiet and active geomagnetic conditions. The vertical bars indicate the range of vertical wind data for each hour in the monthly climatology selected. The quiet time averaged vertical winds are seen to be weak, within a range of a few ${\text{ms}}^{-1}$, and the long term variation is consistent with a semi-diurnal tidal variation as found by Mikkelsen and Larsen (1991) to apply in the polar region. In contrast, the active night monthly climatology indicates a substantive perturbation of $\sim 50$ ${\text{ms}}^{-1}$ near the midnight magnetic local times when typical auroral substorm events are more frequent, i.e., between 10 and 14 UT.

Figure 3

Figure 3. Eagle nightly-averaged vertical wind data over a 35 day period in January and February of 2015 for quiet (Kp < 3) and active nights (Kp > 3). The vertical bars illustrate the range of vertical wind speeds observed in each UT time interval.

We now show specific examples of vertical wind activity for a variety of different geomagnetic disturbances, namely 1) quiet, 2) a night associated with extended geomagnetic activity featuring AE levels generally between 500 and 1,000 nTs, 3) a night for which the AE sequence started with typical values of less than 100 nT but then over several hours the AE index increased substantively, exceeding 700 nTs toward dawn, and 4) a night showing results obtained at the Fort Yukon Observatory starting as disturbed with AE in excess of 600 nT, but then with weakening geomagnetic activity diminishing near the end of the night to values near 50–100 nT. These examples are representative of the results generally seen over the 4 years of EAA and FYU data collected between 2013 and 2016.

Figure 4 (left and right) illustrate the results for two nights showing vertical wind activity being quiescent and the geomagnetic AE index and Kp indices being quite weak with values within the ranges of 50–100 nTs and 0 to 2, respectively. The vertical wind data for both nights show the appearance of occasional wave trains during the night with amplitudes of 5–10 ${\text{ms}}^{-1}$. It is interesting that the waveform for the oscillations observed tends to be spikey rather than sinusoidal in profile as seen in the observations reported by Schmidt et al. (2025). The red line intensities, even for the weak level of geomagnetic activity, are still slightly enhanced as expected for the appearance of weak auroras.

Figure 4

Figure 4. Plots showing the temporal behavior of Kp and AE FPI indices for two nights (top panel). Eagle FPI vertical wind, temperature, and relative intensity data are plotted, respectively, in the bottom three panels. Typically, for these two quiet nights, with an exposure time of 2 min, the wind and temperature standard deviations are 3–5 ${\text{ms}}^{-1}$ and 10–15 K. Left, 17 February 2014. Right, 30 March 2015.

In contrast Figure 5 shows the vertical winds being quite active, with numerous events near 0430 UT, 0500 UT, 0600 UT, 0800 UT and 1145 UT showing upwelling amplitudes greater than 50 ${\text{ms}}^{-1}$ and durations typically 20–30 min. In this case the geomagnetic activity ranged from the maximum value of 500–600 nT seen at the beginning of this night to a low of $\sim 100$ nT toward the end of the night. A significant degree of activity in temperature, decreasing from the initial value of $\sim 1000$K to the morning twilight value of $\sim 750$K, can be seen coincident with the vigorous vertical wind activity. Also seen are several temperature oscillations that are weakly correlated with the SSVW events at 06 UT, 0745 UT, but there are other temperature perturbations that show no match to a corresponding counterpart in the SSVW behavior. The 630-nm intensities are elevated relative to the nightglow almost entirely throughout the night, indicating a general level of enhanced soft particle precipitation lasting toward dawn.

Figure 5

Figure 5. Geomagnetic activity indices AE and Kp, FPI vertical wind, temperature, and relative intensity plots for 21 January 2016. Typically, the standard deviations (plotted as vertical bars) are 2–5 ${\text{ms}}^{-1}$ and 5–15 K for wind and temperature errors, respectively.

The vertical wind data for 4 January 2015 presented in Figure 6 illustrate an example of a SSVW event showing vertical wind upwelling occurring between 11 UT and 15 UT with speeds increasing linearly from nearly zero at 1100 UT to a maximum of 100 ${\text{ms}}^{-1}$ at 1220 UT and then reducing slowly to zero by 1,445–1500 UT. The duration of this SSVW event is thus $\sim 3.5$ hrs, running from 1130 UT to 1500 UT. Superimposed upon this SSVW structure are numerous fluctuations with peak to valley magnitudes of 50–60 ${\text{ms}}^{-1}$ seen to occur throughout the night with a period typically 25–30 min. This oscillating structure is similar to that seen for the two quiet nights except that the amplitudes are about twice as large and the wave structure occurs continuously throughout the night. The spectral profile of these oscillations is narrow with a duration of $\sim 10$ min, about one fifth to one third of the wave period. Preceding the large SSVW event between 3 and 6 UT the vertical wind data tend to be dominated by an overall downwelling of $\sim 20-25$ ${\text{ms}}^{-1}$. Coordinated with this vertical wind increase during the 1,130 to 1500 UT period the temperature data show a substantive increase from 650–700 K–1150 K, indicating that a major heating event was taking place over several hours. The red line relative intensity was at the nightglow level following the evening twilight decay until $\sim 0530$ UT when the intensity grew steadily to a maximum at 07 UT showing a sustained period of soft particle precipitation causing the red line intensity enhancement.

Figure 6

Figure 6. Geomagnetic activity indices (AE, Kp) and Eagle vertical wind, temperature, and relative intensity plots for 4 January 2015. Typically, the standard deviations are 2–4 ${\text{ms}}^{-1}$ and 6–12 K for wind and temperature errors, respectively.

Figure 7 (left) presents Poker spectrograph data showing the 15 h keograms for the four wavelengths (630 nm, 486.1 nm $H_{\beta }$, 427.8 nm, and 557.7 nm) between 04 and 18 UT illustrating the spatial distribution of spectral emissions occurring along the N-S geomagnetic meridian. A keogram is generated by assembling the spectrograph images in sequence. Details regarding the design and brightness calibration of this imaging instrument are described by English et al. (2024).

Figure 7

Figure 7. Optical keograms (left) and plasma density profiles (right) for 4 January 2015 (from the UAF Geophysical Institute archive and from the PFISR MIT madrigal database, respectively).

This instrument is quite useful for characterizing the type of aurora activity occurring within the meridional range of coverage. Mono-energetic aurora activity would be associated with emissions at these wavelengths because the characteristic energy ${\text{E}}_0$ is typically 10–12 keV and thus sufficient to excite these emissions at 110–120 km altitude (Strickland et al., 1989; Newell et al., 2009; Gabrielse et al., 2021). Broadband aurora is associated with incoming soft energy particles with energies within the range of 200–2000 eV.

Figure 7 (right) presents the plasma density profiles collected by the Poker Flat incoherent incoherent radar for the radar beam directed toward the geographic zenith. In examining both sets of profiles obtained at Poker and comparing with the Eagle FPI data, it must be remembered that the Eagle instrument is located $\sim 300$ km to the east of Poker Flat. It is assumed that the spatial structures seen in the optical keograms and ${\text{N}}_e$ plasma profiles observed at Poker Flat would be similar to what was taking place over Eagle.

What is shown in Figure 7 (left) in the 630 nm keogram is that the intensity data in the direction toward the south near 150$°$ elevation is dominated by an arc with an intense red line aurora that moved quickly northward at 13 UT, extending all the way toward the northern limit of the keogram. Coincident with this arc enhancement and northward movement are major increases in both the 427.8 nm and 557.7 nm emissions suggesting a period of hard electron influx that is associated with mono-energetic auroral activity. Coincident with this strong increase in auroral brightness was also a significant increase at 13 UT in the ${\text{H}}_{\beta }$ emission indicating proton precipitation extending across the whole of the auroral oval. Multiple periods of auroral brightness enhancements were seen during this night with significant brightness maxima seen at 1230 UT, 1300 UT, and 1345 UT.

What is also seen in Figure 7 (right) is that a great deal of ${\text{N}}_e$ structure between 100 and 300 km is evident in the time history of the plasma densities from 0700 UT to 1715 UT. The ${\text{N}}_e$ structure between 130 km and 250 km beginning at 11 UT and concluding at 17 UT is consistent with auroral activity that is broadband, with characteristic energies in the range of 200 eV to 2 keV accounting for the major enhancement of the red line intensity seen toward south. It is likely that superimposed upon this soft electron precipitation beginning at 12–13 UT is a period of hard particle precipitation accounting for the major brightness enhancements seen in the 557.7 nm, ${\text{H}}_{\beta }$, and 427.8 nm emissions.

Figure 8 illustrates the results of the Doppler shift analysis of the Poker Flats Incoherent Scatter Radar (PFISR) plasma echo returns for this night. PFISR is a phased radar array located in Alaska (65.2 N, 143.3 W). It is boresighted at 74 and 15°. This geometry allows for pulse-to-pulse steering over multiple directions. The data from the line of sight (LOS) radar returns is processed for the multiple directions selected to analyze frequency shifts with reference to a calibrated local oscillator. The LOS plasma velocity corresponds to the Doppler shift of the spectrum, and a spectral moment method is generally used to estimate the frequency phase shift.

Figure 8

Figure 8. Latitudinal variation of PFISR plasma flow vectors plotted against Universal Time for 4 January 2015. Graphic selected from the MIT madrigal data base.

The experimental configuration employed 10 beam positions, including three pairs spread in magnetic latitude, three pointed along the magnetic meridian, and one pointed up the magnetic field line. This experimental configuration is useful for electric field and ionospheric-parameter measurements close to zenith, although more latitudinal coverage is possible using beams with lower elevation angles. The dip angle varies from beam to beam, but is in the range of 75°–85° in the F region. Further detail describing the analysis processing is given by Nicolls and Heinselman (2007).

What is particularly interesting in Figure 8 is the reversal of the plasma flow direction at 945 UT from westward to eastward with the speed increasing from $\sim 100-200$ to a maximum of $\sim 1500$ ${\text{ms}}^{-1}$ at 1,400–1430 UT that lasted for 60–90 min with a variability of typically 20–30° in the eastward direction.

Figure 9 shows the time sequence of the vertical wind data obtained at Fort Yukon for 3 February 2016, and again, numerous oscillations can be seen that are similar to those seen for Eagle (Figure 6) with respect to the periods and amplitudes, typically 25–30 min and 10 to 25 ${\text{ms}}^{-1}$, respectively. A strong SSVW event highlights the beginning of this night with two maxima at 0300 UT and 0345 UT having amplitudes with peak values reaching 50 ${\text{ms}}^{-1}$ and 75 ${\text{ms}}^{-1}$, respectively. The AE index was active starting at a level of 750 nT at 0500 UT but decreasing toward nearly zero by 1000 UT. The keogram data and PFISR ${\text{N}}_e$ profile data are presented as Figure 10 (left) and 10 (right). Here in contrast to the night of 4 January 2015 the aurora appears near the northern edge of the keogram profiles. Moreover, toward the southern edge, the brightness of ${\text{H}}_{\beta }$ emission shows the presence of a strong proton aurora seen to be persistent throughout the night. These results show that the magnetospheric configuration for this night was quite different than for 4 January 2015 in the sense that the aurora is seen developing toward the north relative to Ft. Yukon. Particle precipitation as represented by the 630 nm brightness data was generally quite soft with little luminosity seen for the 557.7 nm and 427.8 nm emissions throughout the night.

Figure 9

Figure 9. Geomagnetic activity indices (AE, Kp) and Fort Yukon FPI vertical wind, temperature, and relative intensity plots for 3 February, 2016. Typically, the standard deviations are 3–5 ${\text{ms}}^{-1}$ and 10–15 K for wind and temperature errors, respectively.

Figure 10

Figure 10. Poker keograms (left, four wavelengths) and PFISR Ne profiles (right) obtained on 3 February, 2016 (from the UAF Geophysical Institute archive and from the PFISR MIT madrigal database, respectively).

This SSVW example exhibits how the vertical wind oscillation structure persists throughout much of the night. This example shows how the amplitudes were first large at 20–25 ${\text{ms}}^{-1}$ but became weaker with time disappearing completely by 1100 UT as the geomagnetic activity also became weak. The temperatures show signs of oscillations as well with amplitudes of $\sim 20$K that faded by 1200 UT.

4 Discussion

The aim of this paper has been to identify what might be the major drivers for producing the SSVW events observed. As illustrated by the results presented in Figures 6–8, it is quite likely that DC Joule heating played a significant role for this event in producing the SSVW winds with speeds as much as 100 ${\text{ms}}^{-1}$ and the 630 nm temperature increase of $\sim 450-500$K. Figure 8 illustrates how the F-region plasma convective flow reversed from westward to eastward during this midnight period between 09 and 14 UT with speeds as much as 1 to 1.5 ${\text{kms}}^{-1}$. The convective flow of plasma would then be in a direction opposite to that of the thermospheric neutral wind which was being driven geomagnetically westward by the ion-neutral coupling. Such a reversal is well known (Richmond, 2021) to create significant Joule heating as a result of momentum transfer from ions to the neutral atmosphere that contributed at least partly toward the development of this large heating pulse.

Hogan et al. (2020) noted Alfven waves reaching the thermosphere from the magnetosphere in the midnight sector is associated with the phenomenon known as “broad band aurora” with auroral particle energies within the range of 200 eV to 2 keV in contrast to the energy spectrum of mono-energetic auroras that tend to have characteristic energies within the range of 10–12 keV (Strickland et al., 1989). These waves reach the upper atmosphere from the magnetosphere within the geomagnetic midnight sector and may contribute to the overall auroral heating. Such auroral activity would feature enhanced upper E-region plasma densities caused by the extended vertical distribution of electrons absorbed within the lower thermosphere. These electrons are associated with the low energy particle precipitation with energies within the range of 200 ev to 2 kev.

Figure 11a selected from the Hogan paper illustrates the enhancement of thermospheric densities in the polar distribution of upper thermospheric density anomalies observed by the Champ satellite at 400 km in the polar region. Also presented in Figures 11b,c are the measured global distributions of estimated “broadband” electron precipitation and the DC E field variability that are seen to be enhanced in the midnight sector. Figure 11d show measured rates of Alfven wave deposition together with the superposition of the Feldstein oval. These results clearly show the association of broadband electron precipitation with the enhanced frequency of Alfven wave detection in the midnight sector.

Figure 11

Figure 11. Figure taken from Figure 1, Hogan et al. (2020): (a) 1-year average difference between air density recorded by CHAMP in the Northern Hemisphere and estimated from MSIS (Liu et al., 2005); (b) statistical number flux of broadband electron precipitation from DMSP. The Feldstein statistical auroral oval is superposed as dotted lines (Newell et al., 2009); (c) statistical root-mean-square electric field variability from DE-2, with same illustrative convection lines (Matsuo and Richmond, 2008). (d) occurrence of Alfvén waves recorded by FAST, also with Feldstein oval superposed (Chaston et al., 2007). All figures are in MLT-MLAT coordinates. The nominal altitude of measurements is indicated at the bottom of each panel.

Lotko and Zhang (2018) calculated the Joule heating for a spectrum of Alfven waves and concluded that the Joule heating can reach the upper heights of the E-region near 200–220 km. Their figure showing the vertical distribution of the total Joule heating divided between ${\text{W}}_{\mathit{Alf}}$ and ${\text{W}}_{DC}$ is reproduced in Figure 12. Hogan et al. (2020) carried out a general circulation model simulation of the thermospheric upwelling that might be induced by the spectrum of Alfven waves that was used to produce Figure 12 and found that vertical wind upwelling would be sufficient to cause thermospheric density enhancement within the polar cusp by 20%–30%. Thus it is reasonable to infer that Alfven heating may have contributed to the development of the upwelling speed observed.

Figure 12

Figure 12. Height profiles of volumetric Joule heating rates calculated from assumed Alfven wave spectra and a 10-mV/m static, uniform dc electric field (${\text{W}}_{dc}$, Figure 6 of Lotko and Zhang (2018)). This reference provides discussion regarding the details on the calculation of the DC electric field heating rate and the 1-s average of the Alfvenic Joule heating rate ${\text{W}}_{Al}$ added to each altitude. ${\text{W}}_{\mathit{A26}}$ refers to the Alfvenic Joule heating averaged over the time period of 26 s ${\text{W}}_{QS}$ refers to the computed Joule heating rate for a pseudo quasistatic electric field at 400 km that is electrostatic rather than inductive. The dotted line refers to the total Joule heating rate W = ${\text{W}}_{DC}$ + ${\text{W}}_{\mathit{quasistatic}}$, which has a peak Joule heating rate of 58 nW/m³ at 132 km altitude.

The ISR ${\text{N}}_e$ profiles in Figure 7 (right) showing enhanced plasma densities between 150 km and 225 km relative to quiet time behavior do suggest that the type of aurora manifested for this period of 11–15 UT was indeed a broadband auroral event. Then the question becomes to what extent the Alfven wave Joule heating associated with such aurora events may have contributed to the development of the vertical wind upwelling observed. A more conclusive statement regarding the importance of Alfven wave heating will require a measurement that can quantify the extent of Alfven wave activity contributing to the overall Joule heating. Analysis of magnetometer records such as described in the work by Ishii (2005) and Frissell et al. (2024) may be useful in assessing the extent of Alfven wave heating contribution to the overall thermosphere heating budget. A preliminary analysis of the Eagle magnetogram data found the recording exhibited complex behavior that may be related to Alfven wave energy deposition but no conclusions can be drawn at this point due to the aurora complex current structure caused by multiple arcs. Further study is required.

5 Summary

Our paper suggests that the Alfven F-region Joule heating contribution described by the work of Lotko and Zhang (2018) and Hogan et al. (2020) may be significant and comparable to the particle energy input associated with soft electrons. One question is whether the individual “spikey” wave events observed represent individual Alfven wave Joule heating interactions giving rise to separate vertical wind fluctuations. Thus, a SSVW event may be a consequence of a multitude of Alfven waves continually arriving in which the Joule heating provided by the deposition of each Alfven wave instigates a thermospheric wave event contributing in part to the development of the SSVW event. Observations with exposure times of a minute or less by a more sensitive narrow field FPI instrument will provide data that can be used to answer this question. The pulse shape of these thermospheric wave events would be more clearly resolved as a result of the higher temporal resolution, and a detailed frequency analysis of the spectral profile of these waves might provide further insight into the wave origin.

In examining the extensive collection of narrow field FPI and also SDI data, it is a surprise to see how frequent the vertical wind upwelling phenomenon is evident in these results, especially for a weak or moderate level of geomagnetic activity. Thermospheric wave activity with amplitudes in excess of 20 ${\text{ms}}^{-1}$ are seen in almost all of the nights, even for nights featuring quiet geomagnetic activity. SSVW events such as the one on 4 January 2015 look to be superimposed upon a background of enhanced wave activity. These waves generally have 25 to 40 ${\text{ms}}^{-1}$ amplitudes and periods varying from 20 min to 40 min. In agreement with the work published by Ford et al. (2007), our results suggest that these waves associated with SSVW events tend to be generally high frequency, i.e., with periods near the 20 min Brunt Vaisala limit for the thermosphere, but on occasion may also be lower in frequency. Coupled with this finding is the fact that the SSVW events are usually not seen when the geomagnetic activity is weak as represented by the examples presented in Figures 3A,B showing the vertical wind activity for two nights with quiet geomagnetic activity. However, wave events with large amplitudes of 15–25 ${\text{ms}}^{-1}$ do appear.

It should be recognized that the deposition altitudes for most Alfven waves would be below the 240–250 km centroid altitude of the auroral 630-nm volume emission profile. This would be a result of the height distribution of the deposition of soft electron energies associated with broadband aurora (200 ev–2000 ev) that would typically be within the altitude range from 150 to 220 km (Strickland et al., 1989). Thus, the wave events seen in the vertical wind data may not be entirely representative of Alfven wave activity as the bulk of the Alfven wave activity is expected to be taking place at lower altitudes.

We also note that the spatial horizontal extent of these SSVW events as characterized by what is shown in Figures 1, 2 suggests the driver for these SSVW events is wide-spread. A possible reason for this may lie in the broad spatial extent of Alfven wave deposition for waves originating in the magnetosphere region.

Thus, summarizing, the SSVW results and the literature regarding Alfven wave phenomenology that went into the development of Figure 11 support the idea that Alfven wave Joule heating may occur simultaneously with broad-band electron particle precipitation that would also contribute to the elevated 630 nm intensities. Thus, we suggest the role of Alfven wave energy deposition in the auroral thermosphere is important though it is difficult to quantify the extent of its contribution to the overall heating of the auroral thermosphere. The implication is that an accurate modeling of the production of SSVW events and the related thermospheric density anomalies in our view will require a prescription for the calculation of Alfven wave Joule heating that is based upon either an assumed spectrum of Alfven waves with a reasonable frequency range or upon actual measurements. Moreover, the variability of the Alfven wave spectral distribution would likely be a significant factor contributing to the temporal changes in the total auroral Joule heating.

This question of how the partition between the direct DC Joule heating and the Alfven wave Joule heating can be measured is certainly challenging. We have considered two ways that this partition analysis might be done. One is based on the analysis of ground-based magnetometer recordings. Preliminary analysis of the Eagle magnetometer readings for the 4 January 2015 example discussed in this paper did find evidence for a complex oscillatory behavior. However, this approach is complicated by the possibility of the existence overhead of multiple auroral arcs implying multiple currents that would all contribute to the overall modulation of the magnetometer readings. Possibly this approach might be extended by the use of a network of magnetometers that would allow sophisticated means of tomographic inversions of the Alfven wave source function. This approach is certainly a possible pathway but further research into the intricacies of the analysis would be necessary.

The other way (suggested by Dr. Dong Lin, private communication) is described in a recent paper (Lin et al., 2026) that has been just recently made possible by the launch of the twin TRACER satellites designed to study magnetic reconnection phenomena. Overflights of the ground-based FPI stations by the two TRACER satellites carrying magnetometers designed to measure Alfven wave oscillations in both low and high frequency ranges would possibly enable the quantification of the extent of Alfven wave activity and the related Joule heating contribution to the development of vertical wind upwelling. This approach is promising.

Data availability statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Author contributions

JM: Software, Writing – review and editing, Methodology, Supervision, Writing – original draft, Investigation, Conceptualization, Data curation, Funding acquisition. ML: Investigation, Writing – review and editing, Validation, Funding acquisition, Conceptualization, Project administration. LN: Visualization, Data curation, Software, Methodology, Writing – review and editing, Resources, Investigation. DH: Resources, Data curation, Project administration, Writing – review and editing, Methodology, Investigation, Funding acquisition.

Funding

The author(s) declared that financial support was received for this work and/or its publication. We would like to acknowledge funding support from the New Jersey Institute of Technology Center for Solar-Terrestrial Research. PFISR/AMISR facilities are operated and maintained for the US National Science Foundation (NSF) by SRI International under a cooperative agreement AGS-1840962. AMISR data are produced and made publicly available for scientific and academic research purposes under the same award. The Poker Flat ISR data and the optical keogram data are archived at the MIT madrigal database and the optics.alaska.gi.edu database, respectively. Financial support for JWM, DH, and ML was provided by a grant from the National Science Foundation, AGS-1243077. Support for ML was also provided by the NSF grant AGS-2012994 to Clemson University, and for JWM a NSF grant provided by the NSF grant AGS-2431687 to the New Jersey Institute of Technology.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer AA is currently organizing a Research Topic with the author(s) JWM, MFL.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

References

Aruliah, A., and Rees, D. (1995). The trouble with thermospheric vertical winds: geomagnetic, seasonal and solar cycle dependence at high latitudes. J. Atmos. Terr. Phys. 57, 597–609. doi:10.1016/0021-9169(94)00100-3

CrossRef Full Text | Google Scholar

Carlson, H. C., Spain, T., Anasuya, A., Asmund Skjaeveland, A., and Joran, M. (2012). First-principles physics of cusp/polar cap thermospheric disturbances. Geophys. Res. Lett. 39. doi:10.1029/2012GL053034

CrossRef Full Text | Google Scholar

Chaston, C., Carlson, C., McFadden, J., Ergun, R., and Strangeway, R. (2007). How important are dispersive alfvén waves for auroral particle acceleration? Geophys. Res. Lett. 34. doi:10.1029/2006GL029144

CrossRef Full Text | Google Scholar

Coakley, M. M. (1995). “Application of the CCD fabry-perot annular summing technique to thermospheric O(1)D, ph.D. thesis,”. The University of Wisconsin - Madison. Available online at: https://www.proquest.com/dissertations-theses/application-ccd-fabry-perot-annular-summing/docview/304256876/se-2 (Accessed February 24, 2023).

Google Scholar

Conde, M. G., Bristow, W. A., Hampton, D. L., and Elliott, J. (2018). Multiinstrument studies of thermospheric weather above Alaska. J. Geophys. Res. Space Phys. 123, 9836–9861. doi:10.1029/2018JA025806

CrossRef Full Text | Google Scholar

Deng, Y., Fuller-Rowell, T. J., Ridley, A. J., Knipp, D., and Lopez, R. E. (2013). Theoretical study: influence of different energy sources on the cusp neutral density enhancement. J. Geophys. Res. Space Phys. 118. doi:10.1002/jgra.50197

CrossRef Full Text | Google Scholar

Dhadly, M. S., Meriwether, J., Conde, M., and Hampton, D. (2015). First ever cross comparison of thermospheric wind measured by narrow- and wide-field optical doppler spectroscopy, J. Geophys. Res. Space Phys., 120, 9683, 9705. doi:10.1002/2015JA021316

CrossRef Full Text | Google Scholar

Dhadly, M., Sassi, F., Emmert, J., Drob, D., Conde, M., Wu, Q., et al. (2023). Neutral winds from mesosphere to Thermosphere—past, present, and future outlook. Front. Astron. Space Sci. 9, 1050586. doi:10.3389/fspas.2022.1050586

CrossRef Full Text | Google Scholar

Emmert, J. T. (2015). Altitude and solar activity dependence of 1967–2005 thermospheric density trends derived from orbital drag. J. Geophys. Res. Space Phys. 120, 2940–2950. doi:10.1002/2015JA021047

CrossRef Full Text | Google Scholar

English, A., Stuart, D. J., Hampton, D. L., and Datta-Barua, S. (2024). Automated nighttime cloud detection using keograms when Aurora is present. Earth Space Sci. 11, e2022EA002808. doi:10.1029/2022EA002808

CrossRef Full Text | Google Scholar

Ford, E. A. K., Aruliah, A. L., Griffin, E. M., and McWhirter, I. (2007). High time resolution measurements of the thermosphere from fabry-perot interferometer measurements of atomic oxygen. Ann. Geophys. 25, 1269–1278. doi:10.5194/angeo-25-1269-2007

CrossRef Full Text | Google Scholar

Frissell, R., Kim, H., Gerrard, A., and Frissell, N. (2024). Climatology of the open–closed boundary using ULF wave observations from south pole, McMurdo, and distributed antarctic AGOs, frontier of astronomy and space. Science 11, 1396527. doi:10.3389/fspas.2024.1396527

CrossRef Full Text | Google Scholar

Gabrielse, C., Nishimura, T., Chen, M., Hecht, J. H., Kaeppler, S. R., Gillies, D. M., et al. (2021). Estimating precipitating energy flux, average energy, and hall auroral conductance from THEMIS all-sky-imagers with focus on mesoscales, frontier of astronomy and space. Science 9, 744298. doi:10.3389/fphy.2021.744298

CrossRef Full Text | Google Scholar

Harding, B. J., Gehrels, T. W., and Makela, J. J. (2014). Nonlinear regression method for estimating neutral wind and temperature from fabry-perot interferometer data. Appl. Opt. 53, 666–673. doi:10.1364/AO.53.000666

PubMed Abstract | CrossRef Full Text | Google Scholar

Hernandez, G. (1982). Vertical motions of the neutral thermosphere at midlatitude. Geophys. Res. Lett. 9, 555–557. doi:10.1029/GL009i005p00555

CrossRef Full Text | Google Scholar

Hogan, B., Lotko, W., and Pham, K. (2020). Alfvénic thermospheric upwelling in a global geospace model. J. Geophys. Research:Space Phys. 125, e2020JA028059. doi:10.1029/2020JA028059

CrossRef Full Text | Google Scholar

Innis, J. L., and Conde, M. (2001). Thermospheric vertical wind activity maps derived from dynamics Explorer-2 WATS observations. Geophys. Res. Lett. 28, 3847–3850. doi:10.1029/2001GL013704

CrossRef Full Text | Google Scholar

Innis, J., Greet, P., Murphy, D., Conde, M., and Dyson, P. (1999). A large vertical wind in the thermosphere at the auroral oval/polar cap boundary seen simultaneously from mawson and davis, Antarctica. J. Atmos. Solar-Terrestrial Phys. 61, 1047–1058. doi:10.1016/S1364-6826(99)00060-7

CrossRef Full Text | Google Scholar

Ishii, M. (2005). Relationship between thermospheric vertical wind and the location of ionospheric current in the polar region. Adv. Polar Up. Atmos. Res. 19, 63–70.

Google Scholar

Itani, R., and Conde, M. (2023). Wavelike oscillations in high latitude thermospheric doppler temperature and line-of-sight wind observed using all-sky imaging fabry-perot spectrometers. J. Geophys. Res. Space Phys. 128, e2022JA031069. doi:10.1029/2022JA031069

CrossRef Full Text | Google Scholar

Kerr, R. B., Kapali, S., Harding, B. J., Riccobono, J., Migliozzi, M. A., Souza, J. R., et al. (2023). Spectral contamination of the 6300Å emission in single-etalon fabry-perot interferometers. J. Geophys. Res. Space Phys. 128, e2023JA031 601. doi:10.1029/2023JA031601

CrossRef Full Text | Google Scholar

Kervalishvili, G. N., and Lu, hr, H. (2013). The relationship of thermospheric density anomaly with electron temperature, small-scale FAC, and ion up-flow in the cusp region, observed by CHAMP and DMSP satellites. Ann. Geophys. 31, 541–554. doi:10.5194/angeo-31-541-2013

CrossRef Full Text | Google Scholar

Larsen, M. F., and Meriwether, J. W. (2012). Vertical winds in the thermosphere. J. Geophys. Res. Space Phys. 117. doi:10.1029/2012JA017843

CrossRef Full Text | Google Scholar

Lin, D., Hartinger, M., Lotko, W., Wang, W., Shi, X., Sorathia, K. e. a., et al. (2026). Efficiency of electromagnetic energy transfer from solar wind to ionosphere through magnetospheric ultra-low frequency waves. Geophys. Res. Lett. 53, e2025GL118532. doi:10.1029/2025GL118532

CrossRef Full Text | Google Scholar

Liu, H., Luhr, H., Henize, V., and Kohler, W. (2005). Global distribution of the thermospheric total mass density derived from CHAMP. J. Geophys. Res. Space Phys. 110. doi:10.1029/2004JA010741

CrossRef Full Text | Google Scholar

Lotko, W., and Zhang, B. (2018). Alfvénic heating in the cusp ionosphere-thermosphere. J. Geophys. Research:Space Phys. 123 (10), 368–383. doi:10.1029/2018JA025990

CrossRef Full Text | Google Scholar

Lu, X., Wu, H., Kaeppler, S., Meriwether, J., Nishimura, Y., Wang, W., et al. (2023). Understanding strong neutral vertical winds and ionospheric responses to the 2015 st. Patrick’s day storm using TIEGCM driven by data-assimilated Aurora and electric fields. Space weather. 21, e2022SW003308. doi:10.1029/2022sw003308

CrossRef Full Text | Google Scholar

Lühr, H., Rother, M., Köhler, W., Ritter, P., and Grunwaldt, L. (2004). Thermospheric up-welling in the cusp region: Evidence from champ observations, Geophys. Res. Lett., 31 (6). doi:10.1029/2003GL019314

CrossRef Full Text | Google Scholar

Matsuo, T., and Richmond, A. (2008). Effects of high-latitude ionospheric electric field variability on global thermospheric joule heating and mechanical energy transfer rate. J. Geophys. Res. Space Phys. 113. doi:10.1029/2007JA012993

CrossRef Full Text | Google Scholar

Meriwether, J. W., Makela, J. J., Huang, Y., Fisher, D. J., Buriti, R. A., Medeiros, A. F., et al. (2011). Climatology of the nighttime equatorial thermospheric winds and temperatures over Brazil near solar minimum. J. Geophys. Res. Space Phys. 116. doi:10.1029/2011JA016477

CrossRef Full Text | Google Scholar

Meriwether, J., Hampton, D., Conde, M., and Westerhout, C. (2019). “Observations of vigorous vertical thermospheric neutral winds at high latitudes,” in Paper SA24A-06, presented at 2019 fall meeting. San Francisco, CA: AGU.

Google Scholar

Mikkelsen, I. S., and Larsen, M. F. (1991). A numerical modeling study of the interaction between the tides and the circulation forced by high-latitude plasma convection. J. Geophys. Res. Space Phys. 96, 1203–1213. doi:10.1029/90JA01869

CrossRef Full Text | Google Scholar

Newell, P., Sotirelis, T., and Wing, S. (2009). Diffuse, monoenergetic, and broadband Aurora: the global precipitation budget. J. Geophys. Res. Space Phys. 114. doi:10.1029/10.1029/2009JA014326

CrossRef Full Text | Google Scholar

Nicolls, M., and Heinselman, C. (2007). Three-dimensional measurements of traveling ionospheric disturbances with the poker flat incoherent scatter radar. Geophys. Res. Lett. 34. doi:10.1029/2007gl031506

CrossRef Full Text | Google Scholar

Nicolls, M. J., Vadas, S. L., Meriwether, J. W., Conde, M. G., and Hampton, D. (2012). The phases and amplitudes of gravity waves propagating and dissipating in the thermosphere: application to measurements over Alaska. J. Geophys. Res. Space Phys. 117. doi:10.1029/2012JA017542

CrossRef Full Text | Google Scholar

Nwankwo, V., Denig, S., Chakrabarti, S. K., Ajakaiye, M. P., Fatokun, J., Akanni, A. W., et al. (2021). Atmospheric drag effects on modelled low earth orbit (LEO) satellites during the July 2000 bastille day event in contrast to an interval of geomagnetically quiet conditions. Ann. Geophys. 39, 397–412. doi:10.5194/angeo-39-397-2021

CrossRef Full Text | Google Scholar

Qian, L., and Solomon, S. (2012). Thermospheric density: an overview of temporal and spatial variations. Space Sci. Rev. 168, 147–173. doi:10.1007/s11214-011-9810-z

CrossRef Full Text | Google Scholar

Rees, D., Smith, R., Charleton, P., McCormac, F., Lloyd, N., and Steen, A. (1984). The generation of vertical thermospheric winds and gravity waves at auroral latitudes I. Observations of vertical winds, planet. Space Sci. 32, 61–84.

Google Scholar

Richmond, A. (2021). Joule heating in the thermosphere, Geophys. Monogr. Ser., 285, 1, 18. doi:10.1002/9781119815631.ch1

CrossRef Full Text | Google Scholar

Sarris, T. (2019). Understanding the ionosphere thermosphere response to solar and magnetospheric drivers: status, challenges and open issues. Philosophical Trans. R. Soc. 337, 20180101. doi:10.1098/rsta.2018.0101

PubMed Abstract | CrossRef Full Text | Google Scholar

Schmidt, A. L., Meriwether, J. W., Gerrard, A., Goodwin, L. V., Zhang, S., and Ying, Y. (2025). Detection of wave activity in measurements of thermospheric vertical winds and temperatures at subauroral latitudes, Front. Astronomy Space Phys., 12, doi:10.3389/fspas.2025.1613164

CrossRef Full Text | Google Scholar

Shiokawa, K., Otsuka, Y., Oyama, S., Nozawa, S., Satoh, M., Katoh, Y., et al. (2012). Development of low-cost sky-scanning fabry-perot interferometers for airglow and auroral studies. Earth, Planets Space 64, 1033–1046. doi:10.5047/eps.2012.05.004

CrossRef Full Text | Google Scholar

Strickland, D., Meier, R., Hecht, J., and Christensen, A. (1989). Deducing composition and incident electron spectra from groundbased auroral optical measurements: theory and model results, J. Geophys. Res., 94, 13527, 13539. doi:10.1029/2009JA014326

CrossRef Full Text | Google Scholar

Visser, T., March, G., Doornbos, E., de Visser, C., and Visser, P. (2019). Characterization of thermospheric vertical wind activity at 225- to 295-km altitude using GOCE data and validation against explorer missions, J. Geophys. Res. Space Phys., 124, 4852, 4869. doi:10.1029/2019JA026568

CrossRef Full Text | Google Scholar