Periodic oscillations in the high-latitude ionosphere driven by ultralow frequency waves: simultaneous measurements using a SuperDARN radar and GNSS-TEC technique

To elucidate the relationship between electric field and electron-density variations in the high-latitude ionosphere associated with Pc5 ultralow-frequency (ULF) waves from subauroral to high latitudes, we analyzed the global navigation satellite system (GNSS)-total electron content (TEC), ionospheric plasma flow observed by the Super Dual Auroral Radar Network (SuperDARN), and electron density in the inner magnetosphere measured by the Arase satellite. On 23 November 2022, the SuperDARN Prince George (PGR) radar in the dusk sector detected meridional plasma flow oscillations with periods and amplitudes of 5 min and 10–60 m/s, respectively. The plasma flow oscillations began at approximately 01:10 UT and persisted until 03:30 UT over a magnetic latitude range of 65°–72°. The amplitude increased as the magnetic latitude increased. The electron density profile observed by the Arase satellite did not exhibit a sharp gradient during the inner magnetosphere. This indicates that the plasmasphere extended beyond the apogee of the Arase satellite (6.1 Re, where Re is Earth’s radius) under quiet geomagnetic conditions. A detailed comparison between SuperDARN radar and GNSS-TEC data revealed that meridional plasma flow oscillations occurred in the mid-latitude trough and the auroral oval (increased TEC region). Additionally, the equatorward boundary of the auroral oval was located between magnetic latitudes of 72° and 74°. The 15-min detrended TEC measured over the Fort Simpson radar, inside the field-of-view of the PGR radar, showed oscillations similar to the ionospheric plasma flow variations. Through a spectral analysis of the detrended TEC and meridional plasma flow oscillations, we identified a phase difference of ∼135° (∼1.9 min) between them. This phase difference indicates that the upward and downward motion of the ionosphere, driven by an external electric field resulting from Alfvén waves propagating from the dusk-side magnetosphere, was responsible for the observed GNSS-TEC perturbation.

Examining the propagation features of ultralow frequency (ULF: ∼1–100 mHz) magnetohydrodynamic (MHD) waves using ground- and satellite-based magnetometer data can help us elucidate the plasma environment in the magnetosphere, plasmasphere, and ionosphere. A network of ground magnetometers is a powerful tool for diagnosing the macroscopic properties of ULF waves that reach the ionosphere; however, the ground magnetometer observations have some limitations, for example, that small-scale structures (<100 km) are smeared out through the spatial integration of ionospheric current effects (Ponomarenko et al., 2001). In this regards, ionospheric radars offer better resolutions for examining the spatial evolution of the ionospheric electric field resulting from the propagation of ULF waves (Fenrich et al., 2006). Among those radards, the Super Dual Auroral Radar Network (SuperDARN) (Greenwald et al., 1995; Chisham et al., 2007; Nishitani et al., 2019) can measure the two-dimensional distribution of ionospheric plasma flow (electric field) over a horizontal area in the ionosphere of a few thousand-kilometer scale. Previous studies combining data from SuperDARN radars and ground magnetometers have clarified detailed characteristics of the latitudinal structures and longitudinal propagations of Pc3–5 ULF waves (Ponomarenko et al., 2005; Ziesolleck et al., 1998). In addition to the electric field variation, recent total electron content (TEC) measurements made through the global navigation satellite system (GNSS) have provided a new perspective of ionospheric plasma density modulation associated with ULF waves (Kozyreva et al., 2020; Pilipenko et al., 2014; Watson et al., 2015).

Pc5 geomagnetic pulsations, with periods of 150–600 s, have the largest amplitudes among all geomagnetic pulsations and are frequently accompanied by variations in the ionospheric electric field and plasma flow (e.g., Lester et al., 2000; Norouzi-Sedeh et al., 2015; Sakaguchi et al., 2012), field-aligned currents (Fenrich et al., 2019), electron density and conductance in the E-region of the ionosphere (Lathuillere et al., 1986; Pitout et al., 2003), and TEC (Davies and Hartman, 1976; Hamada et al., 2015; Kozyreva et al., 2020; Okuzawa and Davies, 1981; Pilipenko et al., 2014; Watson et al., 2015). Compared to ground magnetometer data, Ponomarenko et al. (2003) successfully visualized ionospheric electric field variations associated with Pc5 pulsations using SuperDARN TIGER radar data. They found that half of the ULF wave events observed by the SuperDARN radar exhibited high coherence with magnetic field oscillations detected by a ground magnetometer, indicating that both radars and magnetometers see the same ionospheric disturbances brought to the ionosphere by magnetospheric ULF waves. Later on, Sakaguchi et al. (2012) statistically analyzed the line-of-sight Doppler velocities obtained along a specific beam of the SuperDARN King Salmon radar in Alaska from 2002 to 2007. Their somewhat puzzling results are that the occurrence frequency of Doppler velocity oscillations in the Pc5 frequency reached a maximum around midnight, whereas almost no such oscillations are found on the dayside. This result suggests that the local time distribution of the Doppler velocity oscillations fairly differs from the well-known occurrence distribution of Pc5 ULF wave events observed in ground-based magnetometer data (e.g., Vennerstrøm, 1999).

Several recent studies reported that TEC variations can also correlate with geomagnetic pulsations. Pilipenko et al. (2014) reported TEC variations at mid and high latitudes (∼58°–75°) associated with magnetic field oscillations in the Pc5 range with an amplitude of approximately 200 nT during the recovery phase of a strong geomagnetic storm on 31 October 2003. They found that the TEC modulations exhibited coherent oscillations in synchronization with magnetic field oscillations and proposed several possible mechanisms of the TEC fluctuations related to Pc5 ULF wave field oscillations, periodic particle precipitation, plasma compression by reflected fast-mode waves, vertical motion of the ionosphere, ion heating, and field-aligned plasma transport. Using data from the European Incoherent Scatter (EISCAT) radar, Pitout et al. (2003) discovered periodic increases in electron temperatures in the F-region ionosphere associated with Pc5 pulsations. They interpreted that the temperature variation was controlled by field-aligned currents associated with a shear Alfvén wave. Additionally, they noted that ion temperature oscillations are a good indicator of temporal variations in the ionospheric convection electric field. Kozyreva et al. (2020) uncovered periodic fluctuations in the rate of TEC (ROT) at a frequency of 2.6 mHz corresponding to the variations of ionospheric plasma flow and magnetic field by analyzing SuperDARN radar and magnetometer data. By examining the ratio between the ROT and plasma flow amplitudes, they proposed that a local steep gradient of the topside ionospheric plasma at auroral latitudes could explain the TEC variations. However, they did not examine the detailed relationship between the TEC and plasma flow oscillations in the Pc5 range, owing to issues in their analysis method for deriving a TEC value from GPS data; therefore, they were unable to identify the physical mechanisms responsible for the TEC variations.

To elucidate the generation mechanism of electron-density variations in the Pc5 range in the high-latitude ionosphere, Pilipenko et al. (2014) investigated the phase offset between TEC and ULF magnetic field measured on the ground. However, the ground magnetometer data are subject to polarization and amplitude distortions due to the ULF energy transition through the ionosphere-neutral atmosphere boundary (Hughes and Southwood, 1976). In this study, to avoid these distortions, we used ionospheric plasma-flow/electric field data obtained by SuperDARN with focus on the phase relationship with GNSS-TEC variations in the Pc5 range during the 2022 SuperDARN and Arase satellite conjugate observations (Nishitani et al., 2026). To obtain information on the background magnetospheric conditions, we analyzed electron density data from the inner magnetosphere measured by the Arase satellite. Finally, we discuss the possible generation mechanisms of TEC variations in the ionosphere.

Figure 1 shows the locations of the GNSS receivers, the field of view (FOV) of the SuperDARN Prince George (PGR) radar, and the footprint of the Arase satellite orbit in geomagnetic coordinates from 00:00 to 04:00 UT on 23 November 2022. The blue curve and red circle indicate the Arase footprint at an ionospheric altitude (100 km) calculated with the Tsyganenko and Sitnov, 2005 magnetic field model (TS05: Tsyganenko and Sitnov, 2005) and the instantaneous position of the satellite at 01:30 UT, respectively. The purple curve denotes the ionospheric pierce point (IPP) of the line-of-sight between the GPS 14 satellite and the GNSS Fort Simpson (FSI) site at an altitude of 300 km. The location of IPP in the FOV of beam 11 was indicated by the short orange curve on the purple one. Here, we selected the GPS 14 satellite data because the IPP of the GPS 14 satellite and the footprint of the Arase satellite passed through the FOV of beam 11 of the PGR radar during the interval of clear TEC variations. This geometry allowed a direct comparison among the magnetospheric plasma environment observed by the Arase satellite, TEC variations and SuperDARN line-of-sight (LOS) velocity. Further, in a time interval from 01:50 to 02:30 UT, the magnetic local time of the magnetic field line at an altitude of 300 km over the GNSS FSI station and the FOV of beam 11 was in a range of 17:01–18:05. In the same period, the magnetic local time of the Arase satellite location in the inner magnetosphere was in a range of 16:48–17:18. Therefore, they were located in the dusk sector when the TEC and plasma flow oscillations were observed on the ground and a difference of the magnetic local times between the ground and satellite was less than 1 h.

GNSS-TEC data

To analyze the ionospheric electron-density variations associated with Pc5 ULF waves, we used GNSS-TEC data with a time resolution of 30 s obtained from receiver-independent exchange format (RINEX) files provided by more than fifty institutes and universities. A list of these data providers can be found in the GNSS-TEC database (https://stdb2.isee.nagoya-u.ac.jp/GPS/GPS-TEC/gnss_provider_list.html). In this study, we selected the GPS satellites from the GNSS satellite constellation for derivation of TEC data. The GPS satellites transmit two coherent signals in the L-band with different frequencies of 1,575.42 MHz (L1) and 1,227.60 MHz (L2), respectively. Considering that the refractive index of ionized plasmas varies depending on the frequency of electromagnetic waves, we calculated the TEC value from the carrier phase difference between L1 and L2. Although we can also derive the TEC value from the difference in two pseudoranges (P1 and P2), the data accuracy is two orders of magnitude lower than that obtained with L1 and L2 signals (Jakowski et al., 1996). The TEC value derived from the pseudorange is expressed by Equation 1: where A = 1 40.3 1 f 2 2 − 1 f 1 2 − 1, λ 1 and λ 2 are the carrier wavelengths at frequencies f 1 and f 2, n 1 and n 2 are the initial values at frequencies f 1 and f 2, b r, b s, b r ′ and b s ′ are receiver and satellite biases of the pseudorange and carrier phase, respectively. Because the TEC value derived from the carrier phase is a relative value, the absolute TEC value is unknown due to the integer cycle ambiguities, λ 1 n 1 and λ 2 n 2 in Equation 2. Therefore, to compensate for the relative TEC value calculated with the carrier phases, we adjusted it to the level of the TEC value calculated with the two pseudoranges (e.g., Jakowski et al., 1996; Otsuka et al., 2002). However, because the TEC value includes the instrumental biases, we need to estimate the instrumental biases to determine an absolute TEC value.

To estimate the instrumental bias, we used the method proposed by Otsuka et al. (2002), which consists of two steps: (1) estimation of hourly TEC averages and inter-frequency biases by least squares fitting of the relative TEC value at each GNSS station, and (2) removal of the biases from the relative TEC values. A more detailed description of this procedure can be found in Otsuka et al. (2002). Finally, we converted the slant TEC values into vertical TEC values and mapped the thin-shell ionosphere at an altitude of 300 km. In this study, we set the maximum satellite zenith angle to 75° and created grid data of the absolute TEC with temporal and spatial resolutions of 30 s and 0.5° × 0.5° in longitude and latitude, respectively (Shinbori et al., 2020; 2023). The grid data in the network common data format (netCDF) and quick-look plots are available on the GNSS-TEC database website (https://stdb2.isee.nagoya-u.ac.jp/GPS/GPS-TEC/). Further, to examine the perturbation component of the TEC value related to electric field/plasma flow variations in the ionosphere with periods of 150–600 s, we subtracted the 15-min running averages from the TEC data obtained with the above TEC processing. Here, we explain the main reason why we choose the 15-min running averages. The electron-density variations in the ionosphere are generated by solar activity and atmospheric gravity waves propagating from the lower atmosphere. Among the electron-density variations, traveling ionospheric disturbances (TIDs) are frequently observed during both geomagnetically quiet and disturbed periods. The period of TID is in a range from 15 min to 3 h (Hunsucker, 1982). To remove such a perturbation component related to TID, we set the running average period of 15 min, which is sufficiently longer than the periods of 150–600 s.

SuperDARN radar data

To investigate the plasma flow variations in the ionosphere associated with Pc5 ULF waves, we analyzed the Doppler velocity data obtained by the SuperDARN PGR radar, which normally scans its FOV clockwise at 16 beams per minute, with a range resolution of 45 km for each beam. Details of the radar network, measurement techniques, and data processing have been described in several review papers (Greenwald et al., 1995; Chisham et al., 2007; Nishitani et al., 2019). This study analyzed FitACF data from SuperDARN radars, which were derived from the FITACF 3.0 library of the Radar Software Toolkit 4.7 (SuperDARN Data Analysis Working Group, 2022) and archived in the common data format (CDF). These data are available from the exploration of energization and radiation in geospace (ERG) science center (ERG-SC) website (Hori et al., 2013; Miyoshi et al., 2018a). In this study, because we focus on the zonal electric-field variation that causes the ExB drift with the meridional and vertical velocity components in the high-latitude ionosphere, where a geomagnetic field line is slightly inclined to the vertical direction, we analyzed LOS velocity data of beam 11, whose FOV includes the IPP of the GPS 14 satellite at the GNSS FSI station and the footprint of the Arase satellite (Figure 1). The magnetic longitude of the FOV of beam 11 is in a range of 296.5°–295.5° at the magnetic latitudes (65.6°–71.0°) around the IPP of the GPS 14 satellite. Therefore, the LOS velocity can be regarded approximately as the horizontal northward component because SuperDARN measures the Doppler shift along the line of sight (Booker, 1956).

Arase satellite data

To investigate whether the magnetic field line of the conjugate point of the IPP of the GPS 14 satellite and FOV of the beam 11 of the SuperDARN PGR radar is located inside the plasmasphere in the inner magnetosphere or not, we analyzed the electron-density data derived from in-situ plasma wave measurements (the upper frequency limit of upper hybrid resonance (UHR) waves) obtained by the high-frequency analyzer (HFA) (Kumamoto et al., 2018) of the plasma wave experiment (PWE) instrument (Kasahara et al., 2018a) onboard the Arase satellite (Miyoshi et al., 2018b; 2022). Additionally, electron-density data were derived from the magnetic field data obtained by the magnetic field instrument (MGF) (Matsuoka et al., 2018a). The detailed method for estimating the electron density is found in several papers (e.g., Kumamoto et al., 2018; Shinbori et al., 2021).

So far, the two methods to identify the plasmapause or plasmapause boundary layer have been proposed by previous papers (e.g., Carpenter, 1966; Carpenter and Anderson, 1992; Moldwin et al., 2002; O’Brien and Moldwin, 2003; Ripoll et al., 2022; Shinbori et al., 2025). The first method computes a steep electron-density gradient as a well-defined plasmapause (Lpp) (e.g., Carpenter and Anderson, 1992; Moldwin et al., 2002; O’Brien and Moldwin, 2003). The second method considers a dense outer edge of the plasmasphere close to the 100 cm −3 density level (L 100) (Carpenter, 1966). Recently, Ripoll et al. (2024) showed that the mean 100 cm −3 level line is located above L = 5 and mean 10 cm −3 level is located beyond the Van Allen Probes apogee for Kp < 1 based on a statistical analysis of the electron density data obtained by the Van Allen Probes. They also confirmed that the plasmasphere expands beyond L = 6.1 during geomagnetically quiet periods with both the methods. In this study, to determine the location of a well-defined plasmapause for a period of the TEC perturbation event, we applied the method proposed by Carpenter and Anderson (1992) to the Arase electron-density profile. This method determines a well-defined plasmapause as a depression in the electron density by a factor of five or more within a short radial distance (dL < 0.5). For this analysis, we first calculated the average electron density in each bin of L = 0.1 for the inbound and outbound passes and derived an L-value profile of a factor of the electron-density drop within dL < 0.5, sliding the window every L = 0.1. Next, we selected the L-value where the maximum factor of the electron-density drop in the L-value profile became 5.0 or more and determined the L-value as a plasmapause location.

Comparisons of SuperDARN radar data and GNSS-TEC

Figure 2 shows the magnetic latitude (MLAT)-time plot of the LOS plasma flow velocity and the ratio of the TEC difference (rTEC) along the beam 11 direction of the PGR radar from 00:00–04:00 UT on 23 November 2022. The rTEC is defined as the normalized difference between instantaneous TEC values and their averages over ten geomagnetically quiet days divided by quiet-day averages (Shinbori et al., 2020; 2025). In Figure 2a, clear plasma flow oscillations are evident over an MLAT range of 65°–72° after 01:10 UT, with an amplitude and period of 20–60 m/s and 5 min, respectively. The amplitude tends to increase above 68° (MLAT). As shown in Figure 2b, rTEC enhancement began at MLATs > 74° after 01:30 UT, and it further increased abruptly after 01:50 UT. This rTEC variation occurred because of enhanced electron precipitation from the magnetosphere associated with auroral activity. Below the equatorward edge of the enhanced rTEC, we observed rTEC depletion with a value of less than −0.3, corresponding to a mid-latitude trough. The trough location is correlated with the plasmapause projected along the magnetic field lines onto the ionosphere (Shinbori et al., 2021). In this case, most of the mid-latitude trough was located at a higher latitude (>68° (MLAT)) than the Arase footprint (pink curve in Figure 2b). By comparing the LOS velocity and rTEC variations, it is evident that the periodic plasma flow oscillations with a period of 5 min occurred from the mid-latitude trough to the equatorward boundary of the enhanced rTEC region, and the amplitude increased significantly inside the mid-latitude trough from 01:10–02:00 UT.

To investigate the TEC response to the plasma flow oscillations in the ionosphere with a period of 5 min, we analyzed detrended TEC data derived by subtracting 15-min running averages of the TEC measured between the GPS 14 satellite and the GNSS FSI receiver, whose ionospheric pierce point was located inside the FOV of the PGR radar. Figure 3 shows the LOS velocity observed by the PGR radar and the detrended TEC at the GNSS FSI station over 40 min, beginning at 01:50 UT on 23 November 2022. As described in Section 2.2, because the LOS velocity is dominated by the meridional plasma flow in geomagnetic coordinates, the plasma flow oscillations are considered to be caused by zonal electric field perturbations. By comparing Figures 3b,c, we can observe a period during which LOS plasma flow oscillations occur simultaneously with oscillations in the detrended TEC along the IPP between the GPS 14 satellite and the GNSS FSI station. Furthermore, a clear phase difference of approximately 2 min can be observed between LOS velocity and detrended TEC perturbations.

To quantitatively examine the phase difference between the LOS velocity and detrended TEC perturbations shown in Figure 3, we conducted a cross-spectrum analysis of the two datasets for the time interval from 01:50–02:30 UT. The analysis tool is included in the Inter-University Upper Atmosphere Observation Network (IUGONET) data analysis software (Tanaka et al., 2013) or the Space Physics Environment Data Analysis System (SPEDAS; Angelopoulos et al., 2019). The results of the cross-spectral analysis are shown in Figure 4. The power spectra of both the LOS velocity and detrended TEC exhibited peaks near 3.3 mHz (∼5 min), which was attributed to the LOS velocity and detrended TEC oscillations, as shown in Figure 3. The coherence also exhibited a maximum value (∼0.8) at approximately 3.3 mHz, indicating that both the LOS plasma flow and the detrended TEC were strongly correlated near this frequency. Further, the cross phase corrected with a time offset (5.5 s: 6.5 degrees in 3.3 mHz) between the beam 11 LOS velocity and TEC observations (Figure 4d) shows a significant phase difference (approximately 135° (1.9 min)) between the LOS plasma flow and detrended TEC near 3.3 mHz. The existence of this phase difference between the LOS plasma flow and the detrended TEC is the key to understanding the generation mechanism of the TEC oscillations associated with the zonal electric field oscillations in the Pc5 range, as further discussed in Section 4.

Solar wind and geomagnetic conditions during a TEC perturbation event

Figure 5 shows time-series plots of the By and Bz components of the interplanetary magnetic field (IMF) in the geocentric solar magnetic coordinates, solar wind proton density and flow speed, and geomagnetic indices (ASY-D and SYM-H) for a one-day period starting at 12:00 UT on 22 November 2022. The IMF and solar wind data are publicly available as part of the OMNI dataset from the NASA CDAWeb (https://cdaweb.gsfc.nasa.gov/). The ASY-D and SYM-H indices (Iyemori, 1990; Iyemori and Rao, 1996; World Data Center for Geomagnetism et al., 2022) were provided by the World Data Center for Geomagnetism (Kyoto, Japan). During the period when the TEC and plasma flow oscillations were observed at high latitude in the dusk sector, the IMF By and Bz components have the negative and positive values with no large fluctuations (Figures 5a,b). Their mean amplitudes were −3.0 nT and 2.5 nT, respectively. The solar wind proton density and speed also showed no large fluctuations during 00:00–04:00 UT on 23 November 2022 (Figures 5c,d). Their mean values were ∼8/cc (/cm -3) and ∼330 km/s, respectively. The SYM-H index (Figure 5f) showed no significant signature of ring current development in the inner magnetosphere because of low solar wind and northward IMF conditions. Therefore, the TEC and plasma flow oscillations were observed during a geomagnetically quiet period. Additionally, a gradual increase of 14 nT was observed in the ASY-D index after 02:00 UT on 23 November 2022, indicating a weak development of a high-latitude current system (Iyemori, 1990).

Spatial structure of electron density in the inner magnetosphere measured by the Arase satellite during a TEC perturbation event

To investigate the plasma environment in the magnetosphere during the Pc5 event, we analyzed the dynamic spectra of the wave electric field obtained by the PWE/HFA onboard the Arase satellite. During the current interval, as shown in Figure 6a, a low plasma wave activity was observed in the frequency range of 3 kHz–10 MHz, and only whistler mode waves with power spectral densities less than 10 −4 mV 2 /m 2 /Hz were visible below the electron cyclotron frequency. This indicates that a significant amount of free energy was not available to generate plasma waves because the solar and geomagnetic activities were quiet, as shown in Figure 5. Further, as shown in Figure 6b, the electron density dropped smoothly from 10 5 − 10 1 / c m 3 with small fluctuations as the satellite moved to an outbound pass from ∼22:00 UT on November 22 to ∼02:00 UT on 23 November 2022. It seems that such electron-density variation does not show any plasmapause signature in the inner magnetosphere of the dusk sector. Here, we examined the plasmapause signature with the Carpenter and Anderson (1992) method as described in Section 2.3. Figure 7 shows an L-value profile of electron density and factor of an electron-density drop within dL = 0.5 for the two inbound and outbound passes including the period when the TEC and plasma flow oscillations were observed at high latitudes in the dusk sector by SuperDARN and GNSS-TEC. During 00:00–04:00 UT on 23 November 2022, In panels (a) – (c) in Figure 7, a factor of an electron-density drop does not exceed the value of 5 except at the low L-value corresponding to the topside ionosphere. Especially, Figure 7b corresponds to the period when the TEC and plasma flow oscillations appeared at high latitudes. This result suggests that the Arase satellite did not cross the plasmapause and traversed inside the plasmasphere during this period. Thus, the plasmasphere appeared to have extended beyond the apogee of the Arase satellite (6.1 Re) under geomagnetically quiet conditions. Considering the footprint of the Arase satellite at the altitude of the ionosphere, we can infer that the low-latitude part of the PGR radar FOV belongs to the plasmasphere. On the next outbound pass (Figure 7d), the electron density exhibited a steep gradient at L = 4.9, where a factor of an electron-density drop exceeds 5, as denoted by the red circle in Figure 7b. This result suggests that the plasmasphere shrunk to L = 4.9 due to a small enhancement of the magnetospheric convection associated with a weak southward IMF as shown in Figure 5b.

Uniqueness of the present study through a comparison of previous studies

So far, several generation mechanisms for plasma density perturbations in the ionosphere associated with Pc5 ULF wave fields have been proposed through modeling and observational studies. Because torsional Alfvén mode waves in cold plasma do not directly produce plasma density perturbations, Sciffer and Waters (2002) conducted a simulation study and proposed that the interaction of an Alfvén mode with the ionosphere-atmosphere-ground system could cause mode conversion, resulting in plasma density variations. Pilipenko et al. (2014) showed that the TEC oscillation amplitude normalized by the background TEC value becomes larger than 0.2% for incident Alfvén waves coming into the ionosphere with large wavelengths exceeding 1,000 km. The temporal and spatial variations in energetic electron precipitation, which cause the ionization of the lower ionosphere (bottom E and D regions) as recorded by riometers, can generate plasma density disturbances (TEC perturbations) with spatially irregular structures (Spanswick et al., 2005). Furthermore, a polarized electric field caused by spatial variations in ionospheric conductivity may be applied from the E to the F-region, causing plasma density perturbations in the upper ionosphere (Shalimov and Kozlovsky, 2015). Pilipenko et al. (2014) proposed several other generation mechanisms for TEC perturbations associated with ULF waves, including periodic drifts across a lateral gradient of the ionospheric plasma, periodic shifts of the plasma vertical profile, ion heating, and field-aligned plasma transport. For the global Pc5 event on 31 October 2003, Pilipenko et al. (2014) attempted to identify the dominant generation process of TEC perturbations; however, they did not obtain any definitive conclusions because the ground magnetometer data were subjected to polarization and amplitude distortions due to transition through the ionosphere-neutral atmosphere boundaries. To solve the underlying problems of ground magnetometer observations, the present study made a direct comparison between the electron density (TEC) and zonal electric field (meridional plasma flow) perturbations in the ionosphere through an integrated data analysis of GNSS-TEC and SuperDARN radar observations. Further, combining simultaneous observations of the Arase satellite traveling in the inner magnetosphere of the dusk sector, we can identify the source region of the zonal electric field perturbations in the ionosphere at high latitudes. This is a unique point of the present study. Through such an integrated data analysis, the present study showed that there is a phase difference between the TEC and plasma flow oscillations in the Pc5 range as shown in Figure 4. This is an important result for understanding the possible generation mechanism of the TEC oscillations related to the zonal electric field at high latitudes. In the following subsections, we will explain the source of the zonal electric field oscillations and the physical meaning of the phase difference between the TEC and zonal electric field oscillations.

Source region of the zonal electric field oscillations in the high-latitude ionosphere of the dusk sector

The zonal electric field (meridional plasma flow) oscillations with a frequency of ∼3 mHz corresponding to the Pc5 range shown in the present study is the same event as that studied by Nishitani et al. (2026). Analyzing simultaneous observation data obtained by ground magnetometers, two SuperDARN radars, and the Arase satellite, Nishitani et al. (2026) discussed the source and propagation mechanisms of the zonal electric field oscillations in the Pc5 range at an altitude of the dusk-side ionosphere in the high-latitude region. The electric and magnetic field observations by the Arase satellite showed a phase offset of ∼45° between the electric and magnetic field variations, which is different from a pure standing mode. Nishitani et al. (2026) interpreted the phase difference as mixed standing and propagating characteristics from the magnetosphere to the ionosphere along the field line. Based on the observational results shown by the previous study, it can be considered that the source of the Pc5 ULF wave fields exists in the equatorial inner magnetosphere of the dusk sector. Nishitani et al. (2026) did not explicitly show whether the Arase satellite travelled inside or outside the plasmapause during the Pc5 event, but the present study revealed that the Arase satellite did not cross the plasmapause based on detailed analysis of Arase electron-density data. This result suggests that the plasmasphere expanded beyond the apogee (6.1 Re) of the Arase satellite.

Furthermore, Nishitani et al. (2026) showed that the observed zonal electric field oscillations have an eastward propagation signature from the afternoon to evening sectors based on the two SuperDARN radars. The eastward propagation suggests that the Pc5 ULF waves are likely to be triggered by some externally driven mechanisms: Kelvin-Helmholtz instability (KHI) along the flank of the magnetosphere (e.g., Rostoker et al., 1998) and solar wind dynamic pressure changes (e.g., Kim et al., 2002). Considering the solar wind and IMF conditions during the Pc5 ULF event, Nishitani et al. (2026) concluded that the KHI is a more probable generation mechanism for the ULF waves in the inner magnetosphere and ionosphere of the dusk sector. Based on the above discussion, it can be interpreted that the observed zonal electric field oscillation is brought to the ionosphere by the poloidal ULF waves propagating along the magnetic field line from the plasmasphere.

Interpretation of the phase difference between the TEC and zonal electric field oscillations

Due to the uniqueness of simultaneous observations of the TEC and zonal electric field (meridional plasma flow) oscillations by a GNSS-TEC technique and SuperDARN radar, the present study showed that the zonal electric field oscillation coincided with the TEC oscillation with a phase difference of 135° (Figure 4d). Nishitani et al. (2026) discussed the propagation direction, source region, and generation mechanisms of the Pc5 ULF waves, but they did not investigate the phase difference of the TEC and zonal electric field oscillations in the Pc5 range. To discuss the physical meaning of the phase difference between the TEC and zonal electric field oscillations shown in this study advances the findings of Nishitani et al. (2026).

So far, it has been well-known that an electric field oscillation at an altitude of the ionosphere produces nighttime medium-scale traveling ionospheric disturbances (MSTIDs) at mid-latitudes (Saito et al., 1998; Shiokawa et al., 2003). An increase (or decrease) in electron density at the bottom of the F-region associated with nighttime MSTIDs coincides with a westward (or eastward) electric field with no phase difference (Shiokawa et al., 2003; Otsuka et al., 2007). Such a relationship between the density variation and the electric field implies that the electric field that causes nighttime MSTIDs is generated such that it satisfies the current continuity in the F-region. In contrast, Shinbori et al. (2022) found that the TEC perturbations observed over Japan after the explosive eruption of the Tonga undersea volcano on 15 January 2022, coincided with the northward plasma flow with a phase difference of 100°–114° based on the GNSS-TEC and meridional component of LOS velocity data in geographic coordinates obtained by the SuperDARN Hokkaido pair of radars. Based on a simple model calculation, Shinbori et al. (2022) verified that the phase difference between the TEC and electric field oscillations is caused by an external electric field originating from the low-altitude ionosphere (E-region). These facts coincide with the results shown by the present study in terms of the existence of the phase difference between the TEC and zonal electric field oscillations. The major difference between the Tonga volcanic eruption and present events is the origin of the zonal electric field oscillation. The former is an E-region dynamo field generated in the sunlit region. The dynamo field propagates along the magnetic field line from the southern hemisphere to northern hemisphere and the propagated electric field drives the TEC and meridional plasma flow in the northern hemisphere. The latter is a poloidal ULF wave generated in and propagated from the equatorial plasmasphere. The poloidal ULF wave propagating along the magnetic field line brings a zonal electric field oscillation to the high-latitude ionosphere. In short, an externally driven electric field is imposed on the ionosphere and the upward/downward motion of the ionosphere due to the externally oscillating zonal electric field is a main cause of generation of the phase difference between the TEC and zonal electric field oscillations.

To test whether the phase difference is caused by an externally oscillating zonal electric field at high latitudes, we conducted a simple model calculation of electron-density variation in the high-latitude ionosphere. In this model, we assumed that there is no production or loss of electron density and that the distribution of electron density in the ionosphere is uniform in the horizontal direction. The electron-density perturbation n e 1 (subscript “1” means the oscillating component) from the continuity equation for electrons can be expressed as follows: where the oscillating electron velocity v 1 is meridional to the magnetic field line, and i, ω, n e, z, and I denote an imaginary unit (i = − 1), angular frequency, background electron density, altitude, and dip angle of the magnetic field, respectively. For calculation of the dip angle at an altitude of the ionosphere, we referred to the International Geomagnetic Reference Field-13 model (Alken et al., 2021). To solve Equation 3, we assumed that the perturbed quantities n e 1 and v 1 are described by a plane wave as follows: n e 1, v 1 ∝ exp ⁡ i ω t − k x, where t, k, and x denote the time, wave number, and horizontal direction, respectively. Therefore, the amplitude of n e 1 is proportional to the product of the vertical density gradient of n e, or ∂ n e ∂ z and amplitude of vertical component E × B drift, i.e., v 1 c o s I.

Figure 8 shows the height profile of the calculated electron-density fluctuations as a function of time for the meridional plasma flow oscillations with a period of 5 min and an amplitude of 30 m/s. These period and amplitude values were determined based on the observations of meridional velocity oscillations by the PGR radar, as shown in Figure 3b. The background electron density was obtained from the International Reference Ionosphere (IRI) −2016 model (Bilitza et al., 2017; Bilitza, 2018) to reproduce the ionosphere at IPP (61.2°N, 239.1°E) over the GNSS FSI station on 23 November 2022, with F10.7 (Tapping, 2013) being 115.7. The dip angle of the magnetic field at this point was 78.5°. In the upper panel in Figure 8, a large-amplitude electron density perturbation appears at the bottomside of the F-region of the ionosphere, where the vertical gradient of the electron density, ∂ n e ∂ z, is large. The bottom panel in Figure 8 shows that the phase of the meridional plasma flow oscillations v 1 is 90° (1.25 min) earlier than that of the TEC oscillations. The phase difference is slightly different from the observed one (135° (∼1.9 min)), but the difference is within the time resolution (1 min) of the SuperDARN radar data. This result verifies that an externally driven oscillating electric field leads to a phase difference between the TEC and plasma flow velocity oscillations.

In a future study, we will need to confirm the interpretations and conclusions by elucidating how the zonal electric field oscillations in the Pc5 range affect the TEC variations through a statistical analysis of multi-point SuperDARN radars, dense GNSS-TEC networks, and ground magnetometers. In this future analysis, we will focus on magnetic local time and latitude dependence of the occurrence of the TEC oscillations and the variation in the phase difference between the TEC and zonal electric field oscillations. Further, combining magnetospheric satellite observations, we will clarify what is an energy source of the Pc5 ULF waves observed in such a geomagnetically quiet period. Recently, an imaging technique of SuperDARN radar observation has been developed at several radar sites (Rohel et al., 2024). This technique enables us to obtain plasma flow velocity with high time resolution and to detect a more accurate period of the Pc5 ULF signals. Therefore, such a comprehensive observation from the magnetosphere to the ionosphere can offer more direct evidence that TEC oscillations are driven by external electric field oscillations.

To elucidate the relationship between electric field and electron-density variations in the Pc5 range observed in the dusk-side high-latitude ionosphere during a geomagnetically quiet period, we analyzed GNSS-TEC data at the GNSS FSI station, the ionospheric plasma flow observed by the SuperDARN PGR radar, and electron density in the inner magnetosphere measured by the Arase satellite. The main results of the present study are shown below. The PGR radar detected significant oscillations in the meridional plasma flow with a period and amplitude of 5 min and 10–60 m/s, respectively, after 01:10 UT on 23 November 2022.

The meridional plasma flow oscillations were accompanied by TEC oscillations with an amplitude of 0.1 TECU. The frequency of the TEC oscillations is almost consistent with that of the meridional plasma flow oscillations.

A cross-spectral analysis revealed significant coherence and a phase difference of approximately 135° (∼1.9 min) between the TEC and meridional plasma flow oscillations.

The plasmasphere expanded beyond the apogee (6.1 Re) of the Arase satellite under a geomagnetically quiet condition.

The TEC oscillations are likely to be generated by an externally driven zonal electric field oscillation carried by the poloidal ULF wave propagating along the magnetic field line from the plasmasphere to the high-latitude ionosphere.