Simultaneous FPI and TMA Measurements of the LowerThermospheric Wind in the Vicinity of the PolewardExpanding Aurora After Substorm OnsetShin-ichiro Oyama1,2 , Ken Kubota1,3, Takatoshi Morinaga4,5, Takuo T. Tsuda6 ,Junichi Kurihara7 , Miguel F. Larsen8 , Masayuki Yamamoto4, and Lei Cai9

1Institute for Space-Earth Environmental Research, Nagoya University, Nagoya, Japan, 2Space and Upper AtmosphericSciences Group, National Institute of Polar Research, Tachikawa, Tokyo, Japan, 3Now at KUBOTA, Inc., Osaka, Japan,4Electronic and Photonic Systems Engineering Course, Kochi University of Technology, Kami, Japan, 5Now at NishimuElectronics Industries Co., Ltd, Fukuoka, Japan, 6Department of Computer and Network Engineering, Universityof Electro-Communications, Chofu, Japan, 7Department of Cosmosciences, Hokkaido University, Sapporo, Japan,8Department of Physics and Astronomy, Clemson University, Clemson, SC, USA, 9Ionospheric Physics Unit, University ofOulu, Oulu, Finland

Abstract Lower thermospheric wind fluctuations in the vicinity of an auroral arc immediately before andafter a substorm onset were examined by analyzing data from a ground-based green line Fabry-Perotinterferometer (FPI; optical wavelength of 557.7 nm) at Tromsø, Norway, and in situ measurements from atrimethyl aluminum (TMA) trail released from a sounding rocket launched during the Dynamics andEnergetics of the Lower Thermosphere in Aurora 2 (DELTA-2) campaign on 26 January 2009. Soon after therocket launch but before disappearance of the TMA trail, a substorm onset occurred. The DELTA-2 TMAexperiment appears to be the first case in which the substorm onset occurred during the TMA windmeasurement. It is known that energy dissipation induced by the ionospheric closure current is compacted atthe poleward side of the discrete arc in the ionospheric morning cell. Both FPI and TMA measurementswere made at the poleward side, but the FPI measured winds nearer to the poleward edge of the arc than theTMA by 110–130 km. The FPI winds at distance of 53–74 km relative to the arc edge showed clear fluctuationsimmediately after the substorm onset, but there was no obvious similar fluctuation in the TMA-measuredwinds. The difference in the response at the two locations suggests that energy dissipation sufficient to bedetected as the FPI/TMA wind perturbations was confined to the area from the poleward edge of the arc to arelative distance shorter than 163–203 km but longer than 53–74 km in this event.

Plain Language Summary Lower thermospheric winds at 100–130 km altitude can be significantlymodulated by external energy inputs from the magnetosphere in association with auroral activity. Substormonset, which coincides with sudden poleward expanding aurora, dissipates kinetic energies of the plasmainto the lower thermosphere. We measured the lower thermospheric winds with two methods duringDynamics and Energetics of the Lower Thermosphere in Aurora 2 campaign on 26 January 2009 in Norway.One is ground-based optical technique applying the Doppler shift of the emitting neutral particles, that is,Fabry-Perot interferometer (FPI). Two is a time-derivative measurement of the trimethyl aluminum (TMA)trail released from a sounding rocket. This experiment was the first success of the TMA measurementconducted at a substorm onset. The FPI measurement was taken place nearer to the poleward expandingauroral arc than the TMA measurement by about 100 km. Lower thermospheric winds from the FPI variedclearly in the magnitude but those from the TMA did not. It is considered that the wind acceleration or theenergy dissipation occurred at least within an area of 53–72 km away from the arc edge but not reaching163–203 km away from the arc edge.

1. Introduction

Understanding the flow of energy and mass throughout the magnetosphere-ionosphere-thermospherecoupled system is a fundamental goal of solar-terrestrial physics. Since substantial energy accumulated inthe substorm growth phase in the magnetospheric tail flows into the polar ionosphere immediately afterthe substorm onset, investigating the energy dissipation process at high latitudes around the time of sub-storm onset can contribute significantly to achieving that objective. Sudden commencement of the

OYAMA ET AL. FPI/TMA WINDS NEAR SUBSTORM ONSET ARC 10,864

PUBLICATIONSJournal of Geophysical Research: Space Physics

RESEARCH ARTICLE10.1002/2017JA024613

Key Points:• Lower thermospheric winds adjacentto a substorm onset arc weremeasured

• Winds at horizontal distance of53–74 km varied clearly, but those at163–203 km distance did not

• Energy dissipation to cause the windvariations was confined to the areanearer to the arc

Correspondence to:S.-i. Oyama,soyama@isee.nagoya-u.ac.jp

Citation:Oyama, S.-i., Kubota, K., Morinaga, T.,Tsuda, T. T., Kurihara, J., Larsen, M. F., …Cai, L. (2017). Simultaneous FPI andTMA measurements of the lowerthermospheric wind in the vicinity ofthe poleward expanding aurora aftersubstorm onset. Journal of GeophysicalResearch: Space Physics, 122,10,864–10,875. https://doi.org/10.1002/2017JA024613

Received 21 JUL 2017Accepted 10 OCT 2017Accepted article online 13 OCT 2017Published online 30 OCT 2017

©2017. American Geophysical Union.All Rights Reserved.

poleward expansion of the auroral arc is one of the typical features in the first stage of the substorm expan-sion phase. Upward field-aligned current (FAC) induced by the inverted-V type potential pattern in the accel-eration region of the magnetosphere is spatially encompassed by the converging electric field, E⊥,perpendicular to the local magnetic field, B, or the arc at ionospheric heights (Aikio et al., 1993; de laBeaujardiere & Vondrak 1982; de la Beaujardiere et al., 1977, 1981; Marklund, 1984, 2009; Marklund et al.,1982; Opgenoorth et al., 1990; Oyama et al., 2009). The perpendicular electric field accelerates ionosphericions in the E⊥ × B direction, which in turn transfers the ion momentum to the neutral particles via collisions.That is the fundamental mechanism of the Lorentz force. In the case of no horizontal gradient in the conduc-tivity or a gradient parallel to E⊥, the Pedersen current, which is parallel to the electric field, is the curl-freecurrent or the principal ionospheric closure current near but outside of the arc (Wilkinson, Nielsen, & Luhr,1986). In more general cases of nonzero horizontal gradient in the conductivity, both Pedersen and Hall cur-rents can contribute to the curl-free current. The closure current flowing in the thermosphere, which acts asthe resistive medium for the current, results in an exchange of energy from the kinetic energy of the iono-spheric ions to the thermal energy of the thermospheric particles, according to Ohm’s Law. This is theJoule heating process. While direct heating due to electron precipitation through the recombination processcan also dissipate a small amount of thermal energy, Lorentz force and Joule heating are the two major pro-cesses capable of modulating the dynamics of the thermosphere.

At F region heights, the pressure gradient between the sunlit and dark hemispheres is a predominant forcefor controlling the thermospheric wind dynamics (Dickinson et al., 1984; Kohl & King, 1967), resulting in diur-nal tides (Kawamura et al., 2000; Witasse et al., 1998). At E region heights, in particular below 110 km, the diur-nal tide is weakened due to less in situ solar EUV absorption by the atmosphere, making the semidiurnalcomponent induced by solar UV absorption by ozone in the stratosphere dominant. Thus, a dominant tidalcomponent shifts from semidiurnal to diurnal with increasing height in the lower thermosphere at high lati-tudes (Oyama et al., 2003, and references therein). During geomagnetically quiet periods, tidal winds aredominant at all thermospheric heights, and the vertical wind is considerably smaller in magnitude thanthe horizontal wind because the wind tends to flow on the isobar, which is almost parallel to the terrestrialsurface. Note that “vertical wind” in the context refers to a geographically vertical wind throughout thetext. Histograms of the vertical component of the ion velocity measured with the European IncoherentScatter (EISCAT) radar has a single peak at 0 m/s at 99–111 km heights for periods of geomagnetically quietcondition (Oyama et al., 2005). Since ion motions are well controlled by the collisional process with neutralparticles at these heights for the quiet periods, it is considered that the vertical neutral wind also tends tohave such a small magnitude on average for the quiet periods, although it can frequently fluctuate in asso-ciation with auroral evolution as mentioned below. Thus, we can use the vertical wind as a proxy of theenergy/momentum inputs from the external environment.

While tidal motions resulting from lower atmospheric forcing and the pressure gradient in the global scaleat F region heights are predominant in the polar thermospheric dynamics as mentioned above, both theLorentz force and Joule heating during periods of active aurora can change polar thermospheric dynamicsover various spatiotemporal scales. Many of the earlier FPI and other instrumental observations are sum-marized by Larsen and Meriwether (2012), and enhancements of the vertical wind may be evidence ofsuch effects (Ishii et al., 2001, 2004, 1999; Peteherych et al., 1985; Price & Jacka, 1991; Price et al., 1995;Smith & Hernandez, 1995). Price et al. (1995) found that wind variations tend to appear adjacent to thearc on the poleward side but inside the arc. The peak neutral upwelling measured with a green line(557.7 nm) Fabry-Perot interferometer (FPI) was 42 m/s. Since the emission has a peak in the lower Eregion, such a large vertical velocity could not be attributed to tidal motions. The authors concluded thatthermal expansion induced by Joule heating was the major force responsible for uplifting the air parcelsrather than the Lorentz force. The conclusion is reasonable because large vertical winds cannot be gener-ated by the Lorentz force that acts in the plane perpendicular to the geomagnetic field, which is nearlyvertical at auroral latitudes. A simulation study suggests that the time derivative of the Joule heating rateis more important than the amount of dissipated energy for generation of the wind fluctuations at Eregion heights (Shinagawa & Oyama, 2006). This suggests that the lower thermospheric wind may onlyrespond weakly to a gradual increase in the Joule heating rate even if the total energy accumulatedthrough the event is large. However, the wind can respond quickly when the Joule energy dissipation rateincreases suddenly and reaches its peak value within a relatively short time (Aruliah et al., 2005; Kosch

Journal of Geophysical Research: Space Physics 10.1002/2017JA024613

OYAMA ET AL. FPI/TMA WINDS NEAR SUBSTORM ONSET ARC 10,865

et al., 2010; Kurihara et al., 2009). The response time may be as short as several minutes based on ourobservational experience. This situation can occur in the vicinity of fast-drifting arcs, in particular immedi-ately after a substorm onset. As far as we know, there is no comprehensive theory to help interpret quan-titatively such a short response time. The Lorentz force may be able to establish horizontal patterns in thethermospheric wind field resembling the ionospheric plasma convection in a statistical manner, even at Eregion heights (Richmond, Lathuillére, & Vennerstoem, 2003). This is consistent with other statistical ana-lyses made of EISCAT radar data, which imply that the effects of the Lorentz force are detectable down to105 km, but dependent on the geomagnetic activity (Fujii et al., 1998; Oyama et al., 2005). It would requirea time of the order of an hour to establish the horizontal pattern in the thermospheric wind field at Eregion heights resembling the convection pattern, based on the e-folding time for momentum transferfrom ions to neutrals (Deng et al., 2009). The e-folding time can be shortened inversely proportional tothe electron density, but any response on minute time scales of the type mentioned above cannot bereproduced well by the Lorentz force alone.

The earlier work cited above supports the idea that the lower thermospheric wind at high latitudes can devi-ate significantly from the tidal flows during active geomagnetic conditions, and that both Joule heating andLorentz forcing contribute to the deviations. However, many details related to the phenomena still remain asopen questions, in particular, a quantitative understanding of the horizontal distributions of the wind varia-tions in the vicinity of an auroral arc. The study described here focuses on developing a better understandingof the horizontal patterns of the lower thermospheric wind deviations around the arc observed in the event.There are a number of earlier publications that describe the horizontal variations of ionospheric parametersin and near arcs because the essential ionospheric parameters, such as the electron density and the electricfield, can be measured remotely with incoherent scatter radars and HF radars from the ground. Also, somesatellites, such as Cluster and FAST, directly measure the FAC and the perpendicular electric field while cross-ing the auroral-particle acceleration region in the magnetosphere (Marklund, 2009; Oyama et al., 2009, andreferences therein). Ground-based imaging of the thermospheric wind at F region heights have been alsoreported using an all-sky type of the Fabry-Perot interferometer (Anderson et al., 2012a, 2012b, 2012c;Conde et al., 2001; Conde & Smith, 1998). The lower thermospheric signatures, by contrast, have been mea-sured rarely due to the difficulty of making such measurements. Ishii et al. (2004) reported thermosphericwind measurements made with two spatially separated FPIs deployed at almost same geomagnetic latitude(65.6°N and 66.2°N) in Alaska. Green line temporal wind variations from the two FPIs had a high degree ofcorrelation, despite being about 300 km apart zonally. This suggests that the wind acceleration process orkinetic energy conversion from ions to neutrals tends to be stretched along the L shell, at least, at E regionheights. This pattern is acceptable particularly for the system nearby the zonally elongated arc. However,we have not yet known the perpendicular pattern to the arc. In some occasions, the arc has a sharp edgeof the emission intensity along with E⊥ enhancements outside of the arc and depressions in the arc (Aikioet al., 2002; Oyama et al., 2009). While this pattern must be attributed to the inverted-V potential pattern,we should also take into account the relative distance to the arc with regard to the kinetic energy conversionin the lower thermosphere.

This study analyzes simultaneous wind measurements at two locations separated by a distance of~100 km. One measurement is from an FPI (557.7 nm) located at the Tromsø European IncoherentScatter (EISCAT) radar site in Norway (69.6°N, 19.2°E in geographic coordinates and 66.7°N, 102.2°E in geo-magnetic coordinates). The other is from a trimethyl aluminum (TMA) trail released from a sounding rocketlaunched during the Dynamics and Energetics of the Lower Thermosphere in Aurora 2 (DELTA-2) cam-paign. The rocket was launched from the Andøya Rocket Range in Norway at 00:15 UT on 26 January2009. The areas covered by the two wind measurements were inside of the field of view of the all-skycamera at Tromsø. Furthermore, a substorm onset took place during the TMA wind measurement. TheFPI and TMA measurements were made at about 100 and 200 km, respectively, from the poleward edgeof the auroral arc associated with the substorm onset. A couple of distant measurements might be toosparse to resolve the horizontal pattern of the wind, but this study will show a difference between thewind perturbations at the two locations that are not due to differences in the measurement method.Section 2 describes the instruments operated for the DELTA-2 campaign. Section 3 presents observationresults comparing the two wind measurements together with discussion. The summary and conclusionswill be presented in section 4.

Journal of Geophysical Research: Space Physics 10.1002/2017JA024613

OYAMA ET AL. FPI/TMA WINDS NEAR SUBSTORM ONSET ARC 10,866

2. Instruments

The DELTA-2 campaign was conducted in January 2009 in the northernNorway. Various optical instruments and radars were operated to obtainsimultaneous measurements of ionospheric and thermospheric para-meters. The scientific objective was to improve our understanding ofthe energy dissipation process associated with the substorm onset.One of the unique aspects of this campaign was the height-resolvedin situ measurement of the thermospheric wind with the TMA technique(Larsen, 2013). The rocket launch window was set for the period from 14to 28 January 2009, and a Japanese S-310-39 rocket was launched fromthe Andøya Rocket Range at 00:15 UT on 26 January 2009 at azimuth

and elevation angles of 0° and 77.5°, respectively. Soon after the launch but before the TMA trail dissipated,a substorm onset took place at 00:22 UT. The DELTA-2 TMA experiment might be the first case in which thesubstorm onset occurred during the TMA wind measurement. A puffed TMA trail was released starting at190 s after the launch on the downleg porting of the trajectory with an alternating 2 s on/2 s off for the dura-tion of the release. Images of the trail obtained with ground-based cameras at Tromsø (Norway) and Abisco(Sweden) were to determine the positions of points along the trail as a function of time and altitudes.Cameras were also deployed at the launch site, but clouds at the launch location prevented the trail frombeing imaged. The change in the trail position with time provided the height-resolved wind vector profilein the altitude range from 86 to 132 km during the interval from 00:19 to 00:27 UT. The intermittent releaseof TMA was designed for deriving the vertical wind. However, the interval was too short to identify the inter-stice to be used for deriving the vertical wind.

The FPI at the Tromsø EISCAT radar site measured the line-of-sight thermospheric wind speeds along fivedirections, including the four cardinal directions with a zenith angle of 15° and along the zenith. The557.7 nm optical filter was selected during the DELTA-2 campaign for measuring the lower thermosphericwind. More than 10 fringes obtained through a 116 mm etalon can be captured simultaneously in oneCCD image (1,024 × 1,024 pixels) due to the relatively wide field of view of the instrument (the full widthat half maximum is about 5°). The Doppler shift or the line-of-sight velocity is calculated from individualfringes then averaged for the overall image. The statistical standard deviation derived from the averagingprocess is taken as the measurement uncertainty and is displayed as error in the subsequent plots. Eachsequence, including the laser fringe calibration measurement, required 2 min 45 s. The combination of datafrom the four cardinal positions provides the horizontal components of the neutral wind velocity (see themethodology developed by Shiokawa et al., 2003).

An all-sky camera (ASC) collocated with the FPI is capable of measuring at six optical wavelengths, namely,557.7, 630.0, OH band (720–1000 nm), 589.3, 572.5, and 732.0 nm. For the night of 26 January 2009, theASC was operated with the 557.7 nm optical filter. The resulting images and meridional keograms were usedto analyze the auroral pattern.

The EISCAT UHF radar was operated with the antenna position sequence summarized in Table 1. The timeinterval between successive antenna positions was about 90 s, and the duration of one complete sequencewas 6 min. The pulse scheme used was “beata,” which is the 64 subcycle alternating codes employed by theEISCAT radars with 32 bits and 20 μs baud length or about 3 km range resolution. The EISCAT radar measuresthe electron density, the ion and electron temperatures, and the line-of-sight ion drift.

3. Observation Results and Discussion

Figure 1 shows temporal variations of the optical and radar measurements from 21 UT on 25 January to 04 UTon 26 January 2009. The five panels include themeridional keogrammade from the all-sky camera collocatedwith the FPI and the EISCAT radar (Figure 1a), the emission intensity measured with the photometer fixed tolook along the geomagnetic field line through an optical filter of 557.7 nm (Figure 1b), the northward com-ponent of the geomagnetic field (Bx) from five magnetometers in northern Scandinavia from 66.90° to70.54°N (Figure 1c), EISCAT-measured electron density from 90 to 400 km altitude (Figure 1d, and the FPI-measured winds (positive northward, eastward, and upward in black, blue, and red, respectively) with eachmeasurement uncertainty (±1σ) (Figure 1e). Horizontal dashed lines in the keogram (Figure 1a) are drawn

Table 1Antenna Positions of the EISCAT UHF Radar for the Experiment on 26January 2009

Azimuth Elevation

Southwest (SW) 226.7° 61.6°South (S) 193.5° 64.0°Field aligned (B//) 185.8° 77.4°Vertical (Z) NA 90.0°

Note. The azimuth angle is measured clockwise from north (e.g., 90° iseast). NA, not applicable.

Journal of Geophysical Research: Space Physics 10.1002/2017JA024613

OYAMA ET AL. FPI/TMA WINDS NEAR SUBSTORM ONSET ARC 10,867

at the latitudes of the individual magnetometer sites in Figure 1c. Because of the spatiotemporal variations inthe ionosphere, some vertical stripe patterns in the time/altitude diagram are evident in Figure 1d due to theEISCAT UHF radar antenna dwell time in the four directions listed in Table 1. There are typical signaturesattributable to the substorm activity in Figures 1a to 1d. The intermittent equatorward drift of arcs seen atthe poleward edge of the keogram from 21:30 to 23:30 UT is a typical feature of the aurora during thesubstorm growth phase. The Bx values (Figure 1c) at SOR, TRO, and KIL (may be also at KIR and PEL) showcontinuous negative deviations from the individual offset levels due to the development of the westward

Figure 1. Ionospheric and thermospheric measurements of the event from 21 to 04 UT in 25–26 January 2009. (a) Keogrammade of the all-sky camera (557.7 nm), (b) 557.7 nm emission intensity from the photometer looking at the geomagneticzenith, (c) northward component of the geomagnetic field at SOR, TRO, KIL, KIR, and PEL, (d) electron density measuredwith the EISCAT UHF radar at 90–400 km, and (e) FPI-measuredwind with a 557.7 nm optical filter (eastward, northward, andupward components are drawn in blue, black, and red, respectively). Error bars of each wind component are drawn incorresponding shaded colors. Horizontal dashed lines in the keogram (Figure 1a) are drawn at latitudes of individualmagnetometer sites adopted in Figure 1c.

Journal of Geophysical Research: Space Physics 10.1002/2017JA024613

OYAMA ET AL. FPI/TMA WINDS NEAR SUBSTORM ONSET ARC 10,868

ionospheric current. While increases of the E region electron density are hardly seen in the measurementbefore 21:30 UT (Figure 1d), weak but clear enhancements can be seen from 21:30 to 22:30 UT, coincidingwith the appearance of the arcs at 70.5°N and further north. At 22:30 UT, the equatorward side of the faintaurora expands equatorward reaching the Tromsø zenith in connection with an obvious increase of the emis-sion intensity along the magnetic field line (Figure 1b) and the electron density in the both E and F regions(Figure 1d). These increases are evidence of the auroral electron precipitation at a few tens of keV energy andlower. For about 30min after 22:30 UT, a portion of the faint arc, in particular themost equatorward part of theauroral structure (intensity less than 10 kR around 69–70°N), is gradually displaced equatorward in connectionwith several sporadic enhancements of the emission intensity and other auroral activity near the northernboundary of the observing region. The equatorward drift of the arc stops at 00 UT, and the arc stays around68.5°N by 00:22 UT but seems to fluctuate in location as one can see a relatively broad distribution in the keo-gram between 67.8° and 68.5°N with intermittent decreases in the emission intensity. From 00:00 to 00:22 UT,the electron density at all the heights shown here is depressed to a relatively low level less than 2 × 1011 m�3.

At 00:22 UT, the auroral and ionospheric conditions are drastically shifted in the substorm phase from thegrowth to the expansion. The auroral emission intensity, magnitude of the negative Bx deviation, and theEISCAT-measured electron density are characterized by sudden increases in association with the substormonset. The response of the ionosphere is a manifestation of the spontaneous development of themagnetosphere-ionosphere current circuit. The ion temperature in the E and F regions is enhanced immedi-ately before the electron density increase (not shown here). Since the Tromsø site is located at the polewardside of the poleward expanding aurora, as shown in the keogram at the time of the substorm onset, the timedelay suggests that the Joule heating process develops outside the poleward edge of the onset aurora due tothe ionospheric closure current (Oyama et al., 2014). The maximum emission intensity from the photometer(Figure 1b) at 00:22 UT seems to be larger than that from the all-sky camera (Figure 1a), although both mea-surements were made at the same optical wavelength (557.7 nm). This is because the field of view and thetime resolution of the photometer are considerably narrower and shorter, respectively, than those of theall-sky camera.

Auroral emission enhancements last for about 30 min and are predominantly at the equatorward side of theTromsø site by about 01 UT (see Figure 1a). The electron density during this interval obviously increases atheights from 90 to 300 km without any notable decreases between these altitudes, although one may implya weak drop around 170 km. Some features frequently seen at the substorm recovery phase appear after 01UT; for example, faint patterns representing the diffuse aurora with decreases in the emission intensity, gra-dual decreases in the negative bay Bxmagnitude, and split of the ionized layer into F and E regions, i.e., C form(Oyama et al., 2014). In summary, the ionospheric/magnetospheric phenomena presented in Figures 1a–1dcan be recognized as typical responses expected for the substorm activity.

Lower thermospheric winds shown in Figure 1e are characterized by two types of temporal variation. One isthe relatively long-term trend, which is clearly seen in the horizontal components. For the time intervalshown in Figure 1, the flow is in a northwestward direction at first, then gradually rotates anticlockwise tosoutheastward by about 2 UT. The wind direction rotates by approximately 150–180° over 5 h. The windspeed is moderately small on average from 3 to 4 UT probably because of measurements of airglow fromthe mesosphere. The trend from 21 to 03 UT can be regarded as a part of the semidiurnal tidal wind, whichis typically seen in the lower thermosphere at all latitudes. The other is the shorter timescale variation overlaidon the tidal trend. These fluctuations seem to take place coincident with the auroral activity. As mentioned insection 1, fluctuations in the vertical wind have been used as an indicator of the polar lower thermosphericresponse to the auroral-related energy inputs. For this event, major vertical-wind fluctuations can be seen atabout 23:30, 00:30, 01:30, and 02:30 UT. In this study, we focus on the event at 00:30 UT because it is concur-rent with the TMA measurement made with the DELTA-2 rocket. The latter two events had pulsating auroralpatches (Oyama et al., 2010, 2016).

As shown in Figure 1, a sudden increase in the emission intensity took place at 00:22 UT in association withthe poleward expansion of brighter parts of the aurora. Figure 2 was made to provide a closer look at themeasurements near that time. Figure 2 (first and second panels) shows temporal variations of the EISCAT-measured E region electron density at 110 km altitude as representative values and the FPI-measured verticalwind from 00 to 01 UT (positive upward). The electron densities for the four antenna directions are plotted

Journal of Geophysical Research: Space Physics 10.1002/2017JA024613

OYAMA ET AL. FPI/TMA WINDS NEAR SUBSTORM ONSET ARC 10,869

with different markers for each beam direction, which are identical to those shown in the inset graphpresented above Figure 2 (third and fourth panels). The measurement uncertainties (±1σ) are overlaid ateach measurement of the electron density and the vertical wind. Figure 2 (third and fourth panels) showsthe horizontal patterns of the aurora green line all-sky camera images from 00:22 to 00:30 UT. Thedirections toward the top and right are north and east, respectively, in the figure. The time interval duringwhich the auroral images are captured is marked by thick horizontal bars in Figure 2 (first and second

Figure 2. (first and second panels) The EISCAT-measured electron density measured at 110 km height and the FPI-measured upward wind speed (557.7 nm) from 00 to 01 UT in 26 January 2009. The EISCAT radar was sequentiallydirected at four positions as presented in the interpolated graph. (third and fourth panels) The horizontal pattern of theemission intensity measured with an all-sky camera (557.7 nm) from 00:22 to 00:30 UT. The time interval of the auroralimages is marked in Figure 2 (first and second panels) with horizontal yellow thick bars. Red and black crosses in Figure 2(third and fourth panels) are marked at positions measured with the EISCAT radar and the FPI, respectively.

Journal of Geophysical Research: Space Physics 10.1002/2017JA024613

OYAMA ET AL. FPI/TMA WINDS NEAR SUBSTORM ONSET ARC 10,870

panels). Near the center of each color panel, measurement directions ofthe FPI and the EISCAT radar are marked by black and red crosses,respectively. Since these marks almost overlap each other in this hori-zontal scale, we can assume that the two instrument measurements cor-respond to almost the same area. An emission height of 110 km hasbeen adopted for mapping the images in geographical coordinates,but there is no significant horizontal displacement if the height is shiftedby as much as 20 km higher or lower. A white spot seen in Figure 2 (firstand second panels) (00:22 and 00:23 UT, respectively) is the emissionfrom the TMA trail.

While a sudden brightening in the field of view of the all-sky camerabegins at 00:22 UT, the EISCAT radar and the FPI are still measuringoutside on the poleward side of the main part of the aurora as late as00:27 UT. However, as indicated by the image at 00:27 UT, the EISCATsouth (S) and southwest (SW) beams may have touched the polewardedge of the brighter zone of the aurora, resulting in earlier enhance-ments of the electron density at these two positions thanmeasurementsin the other two directions, as found in Figure 2 (first panel). From00:28 to 01:00 UT, the electron density at 110 km height remains at anelevated value of 3–5 × 1011 m�3. We conclude that from 00:22 to00:27 UT, the FPI and the EISCAT radar measured mainly the region out-side of the arc on the poleward side. After 00:28 UT the measurementswere within the arc.

The vertical wind (Figure 2, second panel) has a notable positive(upward) 15 m/s peak at 00:24 UT and a negative (downward)�10 m/s peak at 00:31 UT. Of particular interest is a time discrepancyof enhancements between the electron density and the vertical-windmagnitude. The wind perturbations appear a few minutes before theelectron density increase, i.e., beyond the poleward edge of the auroralarc in this case. Similar configurations have been reported in several ear-lier papers in the literatures (Ishii et al., 2001; Price et al., 1995). The causeis related to the ionospheric closure current in the vicinity of the arc. The

convergence electric field, which can be linked to the inverted-V type potential structure in the accelerationregion for generating the auroral electrons, tends to increase on the poleward side of the arc in the morningcell of the ionospheric convection pattern. We checked the ionospheric potential plot map made from SuperDual Auroral Radar Network radar data to verify this geometry, although the map is not shown here. The elec-tric field drives a Pedersen current, which contributes directly to the Joule energy dissipation, and the asso-ciated perpendicular electric field also enhances the Lorentz force. These processes thus result in the windperturbations poleward of the poleward boundary of the auroral arc. For generating vertical-wind fluctua-tions, thermal energy deposition due to the Joule heating process is more important than the Lorentz force,which acts in the plane perpendicular to the magnetic field.

Furthermore, the data provide a new insight into the location of the energy dissipation region by taking intoaccount differences in the relative distance to the arc, as presented in Figure 2 (third and fourth panels). Thedistance from the area sampled by the FPI to the poleward edge of the arc is 53–74 km during the intervalfrom 00:22 to 00:24 UT. The TMA trail can be identified in the images at 00:22 and 00:23 UT, and perhaps alsoat 00:24 UT. The distance from the trail to the poleward edge of the arc is 163–203 km. The FPI measurementis closer to the poleward edge of the arc than the TMA measurement by 110–130 km. Individual features ofthe wind from each instrument will be compared by reference to the relative distance between each mea-surement and between the measurements and the arc.

Figure 3 shows the height profile of the wind derived from the TMA measurement in the interval from 00:19to 00:27 UT between 86 and 132 km altitude (positive northward and eastward). Horizontal bars at individualaltitudes are the measurement uncertainty (±1σ). With increasing altitude up to 106 km, the wind rotates

Figure 3. Mean height profile of the TMA-measured wind for 00:19 to 00:27UT from 88 to 132 km ((top) northward and (bottom) eastward). The mea-surement error at each height (±1σ) is overlaid by horizontal solid line. Amean value of the FPI-measured wind is calculated for the same time intervalas made by the TMA, and marked on the panels by dashed blue lines.

Journal of Geophysical Research: Space Physics 10.1002/2017JA024613

OYAMA ET AL. FPI/TMA WINDS NEAR SUBSTORM ONSET ARC 10,871

clockwise from northward to southeastward, then remains southward from 106 to 118 km. Above 118 km, thewind rotation pattern is opposite, from southward to southeastward. A large vertical shear occurs in thewinds between 98 and 106 km and is particularly evident in the meridional component. We attributethe shift of the wind rotational pattern between 106 and 118 km to the mode change in the tidal windgradually from semidiurnal to diurnal one (Oyama et al., 2003). The large shear in the horizontal wind isa typical signature, which can be seen at all latitudes (Larsen, 2002). The height profile is thereforeconsidered to be characterized by typical features of the tidal wind.

The FPI wind direction is in the southeastward direction throughout the time interval of the TMA measure-ment (see Figure 1e). In general, there is uncertainty about the height of the peak emission intensitymeasured with the FPI, especially in active auroral conditions. In this case the FPI wind measurement canbe compared with the TMA wind profile to estimate the height region for the FPI measurement. Since theheight profile shown in Figure 3 corresponds to data obtained from 00:19 to 00:27 UT, as mentioned above,the FPI-measured winds are also averaged for the same time interval. Comparison with the mean valuemarked in Figure 3 suggests that the FPI measured the winds around 106 km or at 120 km altitude. Thereis thus still uncertainty about the height range for the FPI wind estimate, but an important point is that thelarge height gradient seen at 98–106 km does not appear to cause a significant instrumental artifact dueto emission height changes.

Figures 4a and 4b show the height profile of the wind derived from the TMAmeasurement every minute from00:19 to 00:27 UT at 86–132 km altitude (positive northward or eastward, respectively). Each profile is shiftedhorizontally by 100 m/s to avoid overlap and are marked sequentially by two different kinds of circles. Themean height profiles (same as Figure 3) are shown as thin curves and overlaid in Figures 4a and 4b. As seenin the auroral images of Figure 2, before 00:24 UT the TMA trail is outside any region with aurora, so the trailcan be clearly seen in the images. Note that the all-sky camera images presented in Figure 2 were not used forestimating the wind velocity. After 00:26 UT, a bright auroral spot appears nearby the TMA trail, and thebright emission makes it difficult to identify the TMA trail in the auroral images used for deriving the wind

Figure 4. (a and b) Height profiles of the TMA-measured wind (northward and eastward, respectively) at 00:19–00:27 UTfrom 86 to 132 km. Measurements every minute are marked by solid/open circles, and the mean profile is drawn by athin line. (c and d) Height profiles of the difference between two consecutive measurements of northward and eastwardcomponents, respectively.

Journal of Geophysical Research: Space Physics 10.1002/2017JA024613

OYAMA ET AL. FPI/TMA WINDS NEAR SUBSTORM ONSET ARC 10,872

velocity. This is the reason why the wind profiles are not plotted above about 120 km after 00:25 UT.Conversely, it shows the fact about bottom of the emission height at 120 km. This may be evidence that alti-tude of the peak auroral emission intensity is closer to 120 km than 106 km, at least, after the bright auroralspot appears, i.e., after 00:26 UT. The emission-weighted center altitude of the FPI measurement is thereforeprobably also around 120 km after that time. This supports the conclusion that the vertical gradient in thethermospheric wind produced minimal instrumental artifacts in this case.

There is a close resemblance between individual wind profiles and the mean profile overall. However, someprofiles show dissimilar structure, for example, at 120 km at 00:21 UT. The deviations from the mean profileare evident in Figures 4a and 4b, but a different way of characterizing the deviations is to calculate differencesin each wind component between consecutive measurements, i.e., to plot the component acceleration as afunction of altitude. The results are shown in panels Figures 4c and 4d. Positive deviations can be regarded asnorthward and eastward acceleration, respectively.

As mentioned above, the FPI appears to have measured winds around 120 km altitude. Taking into accountthemeasurement uncertainty (2σ = 40 and 20m/s for the northward and eastward components, respectively,at 120 km; see Figure 3), there are two events to be noted in the third and fifth height profiles from the left ofFigures 4c and 4d. The third height profile, which represents measurements at 00:21 and 00:22 UT, indicates asoutheastward or southward acceleration at 120 km. A wavelike trend can be seen in the original height pro-files at 00:21 UT above 116 km in the both components. The vertical wavelength is 10 km or longer, and thisvalue can be consistent with an atmospheric gravity wave (AGW). However, similar trends cannot be found inthe height profiles taken at 1 min before and after 00:21 UT. The oscillation period of the AGW should belonger than the Brunt-Väisälä period (5–6 min at these altitudes). It is thus unlikely that AGWs suddenlyappear and disappear within a minute. While a sudden deceleration of the wind might be able to highlightthe AGW hidden within the background flow that is faster than the AGW phase speed (refer to the wind fil-tering effect), such a deceleration cannot be recognized in the TMA-measured wind. Since there is no notableauroral activity at 00:21 UT around the TMA trail, it is unrealistic to think that the wind acceleration is due toauroral energy inputs.

Another notable event seen in the fifth height profile of Figures 4c and 4d coincides with the poleward drift ofthe brightening arc seen at 00:23 and 00:24 UT (see Figure 2, third and fourth panels). The wind deviationmayimply northwestward acceleration, but only at 122 km. The acceleration is not found below or above thisheight. Since it is unlikely that the impact of the Joule energy dissipations and the Lorentz force will be lim-ited to a 2 km thin layer, it is difficult to assess the reliability of this deviation. However, we can conclude thatthe sudden appearance of the auroral arc at 00:22 UT (see Figure 2, third and fourth panels) does not produceimmediate impacts on the winds in the TMA measurement region or at distance of 203 km of the arc sincethere are no clear deviations in the winds between 110 and 126 km altitudes, except as discussed above.

4. Summary and Conclusions

We have presented observations of the lower thermospheric wind response to a substorm onset that tookplace in northern Scandinavia at 00:22 UT on 26 January 2009. The wind was simultaneously measured fromthe ground with the green line FPI located at the Tromsø, Norway, and in situ with the TMA technique onboard a rocket launched from Andøya, Norway. Both measurements were made outside the arc on thepoleward side. The distances from each measured area to the poleward edge of the arc were 53–72 kmand 163–203 km for the FPI and the TMA, respectively. The event occurred in the morning cell of the iono-spheric convection, in which the closure current tends to develop on the poleward side of the arc. This eventprovided a unique opportunity to study wind variations at two spatially separated locations located near butoutside the arc. Comparison with the height-resolved TMA wind has allowed us to estimate the centroidheight of the emission-weighted green line emission, which was determined to be at 120 km height forthe FPI wind measurement.

The FPI vertical wind suddenly increased the magnitude between 00:22 and 00:25 UT (see Figure 1e andFigure 2, first panel). It is thus concluded that fluctuations of the lower thermospheric dynamics in the FPI-measured area (or 53–72 km away from the arc edge) at 120 km altitude began immediately after the sub-storm onset in association with the arc activation at 00:22 UT or within 3 min from the onset. However, thosein the TMA measurements conducted at 163–203 km away from the arc edge at 120 km altitude had no

Journal of Geophysical Research: Space Physics 10.1002/2017JA024613

OYAMA ET AL. FPI/TMA WINDS NEAR SUBSTORM ONSET ARC 10,873

obvious fluctuations for the time interval of 3 min. The discrepancy between the FPI and TMA winds must beattributed to the distance from the source region to generate the wind acceleration. For this event, it isconsidered that the initial acceleration occurred at least within an area of 53–72 km away from the arc edgebut not reaching 163–203 km away from the arc edge. This 110–130 km discrepancy resulted in a critical dif-ference of the lower thermospheric response. The wind fluctuations might have reached the TMA-measuredarea later, but it cannot be clearly mentioned because of no wind data above 120 km altitude after 00:25 UTas shown in Figures 4a and 4b.

ReferencesAikio, A., Opgenoorth, H. J., Persson, M. A. L., & Kaila, K. U. (1993). Ground-based measurements of an arc-associated electric field. Journal of

Atmospheric and Terrestrial Physics, 55(4–5), 797–808. https://doi.org/10.1016/0021-9169(93)90021-PAikio, A. T., Lakkala, T., Kozlovsky, A., & Williams, P. J. S. (2002). Electric fields and currents of stable drifting auroral arcs in the evening sector,

Journal of Geophysical Research, 107(A12), 1424. https://doi.org/10.1029/2001JA009172Anderson, C., Conde, M., & McHarg, M. G. (2012a). Neutral thermospheric dynamics observed with two scanning Doppler imagers: 1.

Monostatic and bistatic winds. Journal of Geophysical Research, 117, A03304. https://doi.org/10.1029/2011JA017041Anderson, C., Conde, M., & McHarg, M. G. (2012b). Neutral thermospheric dynamics observed with two scanning Doppler imagers: 2. Vertical

winds. Journal of Geophysical Research, 117, A03305. https://doi.org/10.1029/2011JA017157Anderson, C., Conde, M., & McHarg, M. G. (2012c). Neutral thermospheric dynamics observed with two scanning Doppler imagers: 3.

Horizontal wind gradients. Journal of Geophysical Research, 117, A05311. https://doi.org/10.1029/2011JA017471Aruliah, A. L., Griffin, E. M., Aylward, A. D., Ford, E. A. K., Kosch, M. J., Davis, C. J., … Jussila, J. (2005). First direct evidence of meso-scale

variability on ion-neutral dynamics using co-located tristatic FPIs and EISCAT radar in northern Scandinavia. Annales de Geophysique, 23(1),147–162. https://doi.org/10.5194/angeo-23-147-2005

Conde, M., Craven, J. D., Immel, T., Hoch, E., Stenbaek-Nielsen, H., Hallinan, T., … Sigwarth, J. (2001). Assimilated observations ofthermospheric winds, the aurora, and ionospheric currents over Alaska. Journal of Geophysical Research, 106(A6), 10,493–10,508.https://doi.org/10.1029/2000JA000135

Conde, M., & Smith, R. W. (1998). Spatial structure in the thermospheric horizontal wind above poker flat, Alaska, during solar minimum.Journal of Geophysical Research, 103(A5), 9449–9471. https://doi.org/10.1029/97JA03331

de la Beaujardiere, O., & Vondrak, R. (1982). Chatanika radar observations of the electrostatic potential distribution of an auroral arc.Journal of Geophysical Research, 87(A2), 797–809. https://doi.org/10.1029/JA087iA02p00797

de la Beaujardiere, O., Vondrak, R., & Baron, M. (1977). Radar observations of electric fields and currents associated with auroral arcs.Journal of Geophysical Research, 82(32), 5051–5062. https://doi.org/10.1029/JA082i032p05051

de la Beaujardiere, O., Vondrak, R., Heelis, R., Hanson, W., & Hoffman, R. (1981). Auroral arc electrodynamic parameters measured by AE-C andthe Chatanika radar. Journal of Geophysical Research, 86(A6), 4671–4685. https://doi.org/10.1029/JA086iA06p04671

Deng, Y., Lu, G., Kwak, Y.-S., Sutton, E., Forbes, J., & Solomon, S. (2009). Reversed ionospheric convections during the November 2004 storm:Impact on the upper atmosphere. Journal of Geophysical Research, 114, A07313. https://doi.org/10.1029/2008JA013793

Dickinson, R. E., Ridley, E. C., & Roble, R. G. (1984). Thermospheric general circulation with coupled dynamics and composition. Journal of theAtmospheric Sciences, 41(2), 205–219. https://doi.org/10.1175/1520-0469(1984)041%3C0205:TGCWCD%3E2.0.CO;2

Fujii, R., Nozawa, S., Matuura, N., & Brekke, A. (1998). Study on neutral wind contribution to the electrodynamics in the polar ionosphere usingEISCAT CP-1 data. Journal of Geophysical Research, 103(A7), 14,731–14,739. https://doi.org/10.1029/97JA03687

Ishii, M., Conde, M., Smith, R. W., Krynicki, M., Sagawa, E., & Watari, S. (2001). Vertical wind observations with two Fabry-Perot interferometersat Poker Flat, Alaska. Journal of Geophysical Research, 106(A6), 10,537–10,551. https://doi.org/10.1029/2000JA900148

Ishii, M., Kubota, M., Conde, M., Smith, R. W., & Krynicki, M. (2004). Vertical wind distribution in the polar thermosphere during Horizontal ERegion Experiment (HEX) campaign. Journal of Geophysical Research, 109, A12311. https://doi.org/10.1029/2004JA010657

Ishii, M., Oyama, S., Nozawa, S., Fujii, R., Sagawa, E., Watari, S., & Shinagawa, H. (1999). Dynamics of neutral wind in the polar region observedwith two Fabry-Perot interferometers. Earth, Planets and Space, 51(7-8), 833–844. https://doi.org/10.1186/BF03353242

Kawamura, S., Otsuka, Y., Zhang, S.-R., Fukao, S., & Oliver, W. L. (2000). A climatology of middle and upper atmosphere radar observations ofthermospheric winds. Journal of Geophysical Research, 105(A6), 12,777–12,788. https://doi.org/10.1029/2000JA900013

Kohl, H., & King, J. W. (1967). Atmospheric winds between 100 and 700 km and their effects on the ionosphere. Journal of Atmospheric andTerrestrial Physics, 29(9), 1045–1062. https://doi.org/10.1016/0021-9169(67)90139-0

Kosch, M. J., Anderson, C., Makarevich, R. A., Carter, B. A., Fiori, R. A. D., Conde, M.,… Davies, T. (2010). First E region observations of mesoscaleneutral wind interaction with auroral arcs. Journal of Geophysical Research, 115, A02303. https://doi.org/10.1029/2009JA014697

Kurihara, J., Oyama, S., Nozawa, S., Tsuda, T. T., Fujii, R., Ogawa, Y.,… Kauristie, K. (2009). Temperature enhancements and vertical winds in thelower thermosphere associated with auroral heating during the DELTA campaign. Journal of Geophysical Research, 114, A12306. https://doi.org/10.1029/2009JA014392

Larsen, M. F. (2002). Winds and shears in the mesosphere and lower thermosphere: Results from four decades of chemical release windmeasurements. Journal of Geophysical Research, 107(A8), 1215. https://doi.org/10.1029/2001JA000218

Larsen, M. F. (2013). Wind measurements: Trimethyl aluminum (TMA) chemical release technique. In K. Oyama & C. Z. Cheng (Eds.), AnIntroduction to Space Instrumentation (pp. 47–51). Tokyo: TERRAPUB. https://doi.org/10.5047/aisi.008

Larsen, M. F., & Meriwether, J. W. (2012). Vertical winds in the thermosphere. Journal of Geophysical Research, 117, A09319. https://doi.org/10.1029/2012JA017843

Marklund, G., Sandahl, I., & Opgenoorth, H. (1982). A study of the dynamics of a discrete auroral arc. Planetary and Space Science, 30(2),179–197. https://doi.org/10.1016/0032-0633(82)90088-5

Marklund, G. T. (2009). Electric fields and plasma processes in the auroral downward current region, below, within, and above theacceleration region. Space Science Reviews, 142(1-4), 1–21. https://doi.org/10.1007/s11214-008-9373-9

Marklund, G. W. (1984). Auroral arc classification scheme based on the observed arc-associated electric field pattern. Planetary and SpaceScience, 32(2), 193–211. https://doi.org/10.1016/0032-0633(84)90154-5

Opgenoorth, H. J., Hägström, I., Williams, P. J. S., & Jones, G. O. L. (1990). Regions of strongly enhanced perpendicular electric fields adjacentto auroral arcs. Journal of Atmospheric and Terrestrial Physics, 52(6-8), 449–458. https://doi.org/10.1016/0021-9169(90)90044-N

Journal of Geophysical Research: Space Physics 10.1002/2017JA024613

OYAMA ET AL. FPI/TMA WINDS NEAR SUBSTORM ONSET ARC 10,874

AcknowledgmentsWe are indebted to the director andstaff of EISCAT for operating the facilityand supplying the data. EISCAT is aninternational association supported byresearch organizations in China (CRIPR),Finland (SA), Japan (ISEE and NIPR),Norway (NFR), Sweden (VR), and theUnited Kingdom (STFC). EISCAT radaroperation during the DELTA-2 cam-paign was supported by Japan, Sweden,Norway, Finland, Germany, and France.Quick-looks of the optical measure-ments analyzed in this study can befound at http://www.soyama.org/data.For the TMA wind data make a contactto Miguel Larsen (mlarsen@clemson.edu), Junichi Kurihara (kurihara@sci.hokudai.ac.jp), and MasayukiYamamoto (yamamoto.masa-yuki@ko-chi-tech.ac.jp). The authors thank all thestaff of the Institute of Space andAstronautical Science and the AndøyaRocket Range for conducting the suc-cessful rocket experiment. This researchhas been supported by a Grant-in-Aidfor Scientific Research (15H05747,16H06286, 16K05569, and 16H02230)and Special Funds for Education andResearch (Energy Transport Processes inGeospace) from MEXT, Japan.

Oyama, S., Miyoshi, Y., Shiokawa, K., Kurihara, J., Tsuda, T. T., & Watkins, B. J. (2014). Height-dependent ionospheric variations in the vicinity ofnightside poleward expanding aurora after substorm onset. Journal of Geophysical Research: Space Physics, 119(5), 4146–4156. https://doi.org/10.1002/2013JA019704

Oyama, S., Nozawa, S., Maeda, S., Murayama, Y., Fujii, R., & Shinagawa, H. (2003). Field-aligned ion motions in the polar E-F transition region:Mean characteristics, Journal of Geophysical Research, 108(A8), 1334. https://doi.org/doi:10.1029/2003JA009830

Oyama, S., Shiokawa, K., Kurihara, J., Tsuda, T. T., Nozawa, S., Ogawa, Y., … Watkins, B. J. (2010). Lower-thermospheric wind fluctuationsmeasured with an FPI during pulsating aurora at Tromsø, Norway. Annales Geophysicae, 28(10), 1847–1857. https://doi.org/10.5194/angeo-28-1847-2010

Oyama, S., Shiokawa, K., Miyoshi, Y., Hosokawa, K., Watkins, B. J., Kurihara, J., … Fallen, C. T. (2016). Lower thermospheric wind variations inauroral patches during the substorm recovery phase. Journal of Geophysical Research: Space Physics, 121(4), 3564–3577. https://doi.org/10.1002/2015JA022129

Oyama, S., Tsuda, T. T., Sakanoi, T., Obuchi, Y., Asamura, K., Hirahara, M.,…Watkins, B. J. (2009). Spatial evolution of frictional heating and thepredicted thermospheric wind effects in the vicinity of an auroral arc measured with the Sondrestrom incoherent-scatter radar and theReimei satellite. Journal of Geophysical Research, 114, A07311. https://doi.org/10.1029/2009JA014091

Oyama, S., Watkins, B. J., Nozawa, S., Maeda, S., & Conde, M. (2005). Vertical ion motion observed with incoherent scatter radars in the polarlower ionosphere. Journal of Geophysical Research, 110, A04302. https://doi.org/10.1029/2004JA010705

Peteherych, S., Shepherd, G. G., & Walker, J. K. (1985). Observation of vertical E-region neutral winds in two intense auroral arcs. Planetary andSpace Science, 33(8), 869–873. https://doi.org/10.1016/0032-0633(85)90101-1

Price, G. D., & Jacka, F. (1991). The influence of geomagnetic activity on the upper mesosphere/lower thermosphere in the auroral zone,I. Vertical winds. Journal of Atmospheric and Terrestrial Physics, 53(10), 909–922. https://doi.org/10.1016/0021-9169(91)90004-Q

Price, G. D., Smith, R. W., & Hernandez, G. (1995). Simultaneous measurements of large vertical winds in the upper and lower thermosphere.Journal of Atmospheric and Terrestrial Physics, 57(6), 631–643. https://doi.org/10.1016/0021-9169(94)00103-U

Richmond, A. D., Lathuillére, C., Vennerstoem, S. (2003). Winds in the high-latitude lower thermosphere: Dependence on the interplanetarymagnetic field. Journal of Geophysical Research, 108(A2), 1066. https://doi.org/10.1029/2002JA009493

Shinagawa, H., & Oyama, S. (2006). A two-dimensional simulation of thermospheric vertical winds in the vicinity of an auroral arc.Earth, Planets and Space, 58(9), 1173–1181. https://doi.org/10.1186/BF03352007

Shiokawa, K., Kadota, T., Otsuka, Y., Ogawa, T., Nakamura, T., & Fukao, S. (2003). A two-channel Fabry-Perot interferometer withthermoelectric-cooled CCD detectors for neutral wind measurement in the upper atmosphere. Earth, Planets and Space, 55(5), 271–275.https://doi.org/10.1186/BF03351759

Smith, R. W., & Hernandez, G. (1995). Vertical winds in the thermosphere within the polar cap. Journal of Atmospheric and Terrestrial Physics,57(6), 611–620. https://doi.org/10.1016/0021-9169(94)00101-S

Wilkinson, P. J., Nielsen, E., & Luhr, H. (1986). Ionospheric conductances: Associations between ionospheric E region electric fields and groundmagnetometer variations. Journal of Geophysical Research, 91(A5), 5839–5849. https://doi.org/10.1029/JA091iA05p05839

Witasse, O., Lilensten, J., Lathuillere, C., & Pibaret, B. (1998). Meridional thermospheric neutral wind at high latitude over a full solar cycle.Annales Geophysicae, 16(10), 1400–1409. https://doi.org/10.1007/s00585-998-1400-3

Journal of Geophysical Research: Space Physics 10.1002/2017JA024613

OYAMA ET AL. FPI/TMA WINDS NEAR SUBSTORM ONSET ARC 10,875