In order to clarify the equatorial electrojet effects on ground
magnetic pulsations in central South America, we statistically analyzed the
amplitude structure of Pc3 and Pc5 pulsations recorded during days considered quiet to
moderately disturbed at multiple equatorial stations nearly aligned
along the 10
Magnetic pulsations, or ULF (ultra-low-frequency) waves, contain information that can be used to study different physical processes acting in the Sun–Earth system. They can be generated internally in the magnetosphere by various types of disturbances, such as substorms or instabilities associated with cyclotron resonances or drift-bounce resonances, or externally by disturbances in the magnetopause (McPherron, 2005). Its detection on the ground at very low latitudes shows that a significant portion of the energy of these hydromagnetic waves penetrates deep into the magnetosphere and the plasmasphere. Particularly at these latitudes, since ground-based magnetometers respond mainly to changes in E-region currents, pulsations bring records of equatorial ionosphere disturbances associated with complex electrodynamic changes during magnetospheric energy transfer processes.
It is well known that the very low-latitude region is characterized by high
zonal ionospheric conductivity alongside the dip equator. The most obvious
effect of this increased conductivity is that in a narrow latitude range,
centered on the dip equator, the eastward flow of the dynamo current is
intensified during daytime. This enhanced current is known as the equatorial
electrojet (EEJ). The effect of the EEJ appears as a significant increase of
amplitude in the horizontal component of diurnal quiet-time geomagnetic
variations, Sq (
The mechanisms involved in the generation, propagation, and amplification of equatorial pulsations have not yet been clearly identified. At very low latitudes, a significant portion of the geomagnetic field lines lies within the ionosphere, and therefore the field line resonance theory that explained numerous observations at middle and high latitudes cannot be applied (Yumoto, 1986). Generally, two models are proposed as the possible mechanisms for the generation and propagation of magnetic pulsations near the ground magnetic equator. In the first model, compressional hydromagnetic waves, which propagate from the magnetosheath across the dayside magnetosphere, arrive vertically at the equatorial ionosphere and are interconnected with magnetic disturbances observed on the ground through the ionosphere. It has been proposed that this mechanism is the main source of Pc3 pulsations at low latitudes (Yumoto and Saito, 1983). In the second model, electric fields driven by Alfvén waves in the high-latitude ionosphere propagate horizontally through the atmosphere to low-latitude regions and the Equator in a waveguide bounded by the conducting ionosphere and the ground (Kikuchi and Araki, 1979). Global observations of Pc5 have been interpreted using this equatorward transmission mechanism of the electric field which originated in the polar region (Trivedi et al., 1997). The first model would produce an attenuation of pulsation amplitude at the dip equator as a result of shielding effects, while the second model would generate equatorial enhancement due to the concentration of ionospheric currents at the magnetic equator (Hughes and Southwood, 1976; Shinohara et al., 1997; Tanaka et al., 2004).
Another remarkable feature of the magnetic field in equatorial latitudes is
related to the strong longitudinal gradient of ionospheric conductivity near
the solar terminator. This feature changes the patterns of the ionospheric
currents and consequently of the geomagnetic field reaching the ground.
Several authors have studied these effects in regular and irregular
geomagnetic pulsations, particularly for the dawn terminator (Saka et al.,
1982; Saka and Alperovich, 1993; Tanaka et al., 2007; Imajo et al., 2016).
These studies have shown that one of the most prominent effects observed
around dawn at very low latitudes is a change in pulsation polarization,
characterized by the increase in the
Most of these previous results were based on data from a single station
(Saka et al., 1982; Saka and Alperovich, 1993) or only two stations, one
located near the dip equator and the other outside that region (Sarma and
Sastry, 1995). Consequently, a detailed mapping of the latitudinal
dependence of the pulsation amplification and polarization characteristics
at very low latitudes has not yet been performed. In this paper, the
horizontal spatial structure of equatorial Pc3 and Pc5 pulsations is
investigated using data acquired by a meridional magnetometer profile in
central South America, with five stations around the dip equator (within
The influence of the equatorial ionosphere on the amplitude of Pc3 and Pc5
pulsations is derived here using geomagnetic data recorded at six stations
roughly aligned along the 10
Figure 1 shows a map of South America with the location of the six
geomagnetic stations. The location of the Huancayo station (HUA; to be discussed
in the text) is also shown. Relevant information from the geomagnetic field
for the period in which the measurements were performed is presented. These
include the dip equator and the inclination of
Geographic location of the six geomagnetic stations in South
America and contours of the geomagnetic field in 1994. The solid line shows
the dip equator (
Five stations were placed in the equatorial region under the influence of the EEJ currents, with one of them at the dip equator and the other four approximately in conjugate points to the north and south of the Equator. The latitudinal spacings in the equatorial zone are sufficient to obtain a measurable amplitude change, while maintaining a high level of signal coherence between the stations. The other station (CUI) was located immediately outside the EEJ belt and will be used here as a reference to evaluate the EEJ effects on the pulsation events.
All stations were positioned well inland, far away from any ocean or mountain range, but a recent study has shown an extensive anomaly in the underground conductivity beneath the CUI site (Padilha et al., 2017). An amplitude enhancement up to tens of a percent is observed on the ground geomagnetic variations recorded at this station. Using transfer functions between two nearby geomagnetic stations, inside and outside the anomaly, these authors established the amplification factors of the geomagnetic field. The amplification varies as a function of frequency due to the reflection of EM (electromagnetic) waves at the interface with the very good conductor and to differences in the damping of the EM wave amplitude during its propagation through the conductive medium inside the Earth. These amplification factors are used here to correct the amplitudes of the geomagnetic variations recorded in CUI.
Geographic and geomagnetic coordinates of the six stations are presented in Table 1. The geomagnetic coordinates refer to the dipole components of the geomagnetic field, whereas the inclination values are derived from the International Geomagnetic Reference Field (IGRF) model. The dip latitude values were calculated from the IGRF inclination values using the formula of Matsushita and Maeda (1965) and will be utilized as a reference for the geomagnetic station location in the final data analysis. The local noon occurs at nearly 16:00 UTC in all stations.
Geomagnetic and geographic information of the six geomagnetic stations.
Simultaneous data from the six geomagnetic stations were obtained during 60 d from 3 September to 1 November 1994. Since we are interested in the
ionospheric contribution to Pc3 and Pc5 amplitudes, we preferentially
considered the data corresponding to geomagnetically quiet or moderately
disturbed conditions. Typical solar daily variations (Sq) in the
Figure 2 shows the behavior of the geomagnetic field during the measurement
period through geomagnetic indices and the activity level of filtered Pc3
and Pc5 pulsations. The bandpass filter period ranges were 10–45 s (Pc3) and
150–600 s (Pc5). The Dst (disturbance storm time) index indicates geomagnetically active conditions
during this period with the occurrence of one intense (minimum Dst of
Geomagnetic indices and filtered ground geomagnetic pulsations
during the period from 3 September to 1 November 1994.
The time series data were digitally filtered in Pc3 and Pc5 frequency bands
using a recursive Butterworth band-pass type, with unit gain within the
chosen frequency band (Kanasewich, 1981). Since ground pulsations near the
geomagnetic equator are known to be strongly polarized to the
Band-pass-filtered geomagnetic variations (
Visual inspection indicates that the Pc3 pulsation is apparently damped at
the PRM station (under the dip equator) compared to the other equatorial
stations. On the other hand, the Pc5 pulsation is strongly amplified in all
equatorial stations in relation to CUI. In addition, the strong equatorial
polarization to the
The same as Fig. 3, for Pc5 pulsations with periods ranging from
150 to 250 s, in the afternoon on 15 October 1994. An event is observed in
the
To emphasize the effect of equatorial ionospheric currents on the pulsation
amplitude, we chose days with little magnetic activity in the equatorial
region. Only those days where the minimum Dst value was greater than
Following the procedure described in Roy and Rao (1998), the characteristic period of each event was determined from spectral analysis. In this procedure, a fast Fourier transform (FFT) is applied to intervals of the filtered time series containing the pre-chosen events. An analysis is made to verify that the same major spectral peak is observed simultaneously in all stations, taken as indicative of a coherent pulsation event. The characteristic period of each signal is then defined from this maximum spectral peak. In addition, the CUI station signals were corrected using the period-dependent amplification factors defined by Padilha et al. (2017). The correction factor in the Pc3 frequency band is given by a 2nd-degree polynomial function, whereas a 4th-degree polynomial function was used for the Pc5 band (pulsations in the CUI station shown in Figs. 3 and 4 were previously corrected for the underground amplification).
Also following Roy and Rao (1998), the EEJ effect for each event was
estimated by the ratio between the spectral amplitudes (in units of
Power spectral analysis for the Pc3 pulsation on 15 October 1994
shown in Fig. 3. Left panels
Figure 6 shows the occurrence distribution of Pc3 and Pc5 relative amplitudes for all events at each equatorial station as a function of local time (LT) and universal time (UTC). It can be seen that most of the Pc3 events are slightly amplified in the equatorial region, during both daytime and nighttime. However, there is a larger number of damped events around noon. This is particularly salient for the dip equator (PRM station), where most of the events are significantly damped. An increase in the Pc3 amplification is observed in stations POV, ARI, PRM, and VIL shortly before 06:00 LT. This amplification is associated with the sunrise effect and will be discussed later in greater detail. On the other hand, almost all Pc5 events are amplified at equatorial stations. The highest amplifications are observed during daytime (06:00 to 18:00 LT) in all stations.
Local time dependence of occurrence distributions of Pc3 and Pc5 relative amplitudes at the equatorial stations. In red are values greater than one (amplification); in blue are values less than one (damping).
The relative amplitudes of the Pc3 pulsations were also analyzed as a function of the wave period. For this analysis we used only nighttime events (18:00 to 06:00 LT) and noon events (10:00 to 14:00 LT). It can be observed in Fig. 7 that nighttime events are not dependent on the period, being preferably amplified throughout the Pc3 band. On the other hand, the noon events present significant differences as a function of the period. It should be noted that, in the latter case, no pulsations with a period shorter than 27 s were observed. Besides, there is an evident cutoff period around 35 s for amplification or damping under the dip equator. Events with periods shorter than 35 s are damped, while longer periods are amplified. This observation explains why some events around noon in Fig. 6 show amplification at the PRM station (these events have periods longer than 35 s). In summary, these results show that at the dip equator there is an amplitude damping in Pc3 pulsations with periods of less than 35 s, when EEJ currents are well developed around local noon.
Period dependence of the relative amplitudes of nighttime and noon Pc3 pulsations in equatorial stations.
We found that Pc3 amplitudes undergo a depression under the dip equator around noon, and this is a period-dependent phenomenon. On the other hand, Pc5 amplitudes are enhanced at all equatorial latitudes, regardless of the period. During nighttime, Pc3 and Pc5 amplitudes are slightly increased for all stations in the equatorial region. These results from the central South American region agree with many previous studies at similar latitudes, based on both ground observations and theoretical models of excitation and propagation of MHD (magnetohydrodynamic) waves that lead to pulsation activity on the Earth's equatorial surface (e.g., Yumoto et al., 1985; Itonaga and Kitamura, 1993; Roy and Rao, 1998; Takla et al., 2011). However, they contradict other studies that showed daytime Pc3 amplification in the equatorial region, without specifying any damping interval (Matsuoka et al., 1997; Zanandrea et al., 2004).
Our results can be explained by considering the mechanisms for the generation and propagation of the magnetic pulsations until their detection on the ground at the equatorial region. As discussed by Yumoto (1986), two models have been proposed to explain Pc3 waves observed at very low latitudes. In the first model, upstream waves generated by ion-cyclotron instabilities on the bow shock propagate in the form of compressional waves along the equatorial plane of the magnetosphere, cross the lines of the magnetic field, and arrive directly at the equatorial ionosphere. In the second model, surface waves generated by instabilities at the diurnal magnetosphere boundary (such as Kelvin–Helmholtz instabilities) propagate to the high-latitude ionosphere and generate large-scale ionospheric current oscillations at these latitudes. These high-latitude currents leak into low latitudes and can cause Pc3 pulsations near the magnetic equator. Following Yumoto (1986), the latter model cannot explain the occurrence of equatorial Pc3 because they can only be transmitted to the high-latitude ionosphere due to the high damping of these waves in their propagation in the radial direction. Therefore, equatorial Pc3 pulsations are more likely to be related to the direct transmission of the compressional fast mode towards the ground. These waves are usually controlled by the solar wind parameters and the interplanetary magnetic field (IMF), with their transmission into the Earth's magnetosphere connected to small cone angle values (Russell et al., 1983).
Theoretical considerations about variations on pulsation amplitude due to the direct incidence of a plane compressional MHD wave on the equatorial ionosphere were obtained in different studies (e.g., Itonaga and Kitamura, 1993; Itonaga et al., 1998). These models predict a depression in the amplitude of the pulsation on the ground at the dip equator, since the induced ionospheric currents act as an obstacle in the propagation of compression waves. Such depression arises from the ionospheric screening effect which becomes more marked as the ionospheric conductivity increases. Consequently, these models predict a greater damping in the pulsation amplitude at the dip equator around noon (when the ionospheric conductivity reaches its maximum), as observed in our data. On the other hand, although some of these models take into account the frequency dependence for this pulsation amplitude depression under the dip equator, there is still no clear physical explanation for the occurrence of a maximum period that limits this damping effect.
Pc5 waves are a persistent component of a disturbed magnetosphere, and their possible sources include Kelvin–Helmholtz oscillations in the magnetopause, the excitation of cavity or waveguide modes, the direct control by oscillations in the solar wind, the fluctuation of field-aligned currents in the auroral zone, and drift-bounce resonance with ring current particles (Hughes, 1994; Elkington, 2006; Kessel, 2008). They are often observed near auroral latitudes, but they can also be observed at middle and low latitudes (Ziesolleck and Chamalaun, 1993). The occurrence of these waves at equatorial latitudes (Reddy et al., 1994; Trivedi et al., 1997) led to the proposition of different penetration mechanisms. Kikuchi and Araki (1979) argued that the ULF wave energy arrives at high latitudes first from the magnetosphere and then propagates to low latitudes through an ionosphere waveguide, while Chi et al. (2001) proposed a combination of fast and shear Alfvén waves which allow ULF waves to penetrate low latitudes directly from the magnetosphere.
The Earth–ionosphere waveguide model of Kikuchi and Araki (1979) was initially developed to elucidate the simultaneous observation of the preliminary reverse impulse (PRI) of the magnetic sudden commencement (SC) at high latitudes and at the magnetic dip equator. It was later expanded to explain the equatorial enhancement of a class of short-period fluctuations, including Pc5 pulsations. In this model, surface waves generated by instabilities at the dayside high-latitude magnetosphere boundary generate large-scale ionospheric current oscillations at the polar ionosphere. These Pc5-related polar electric fields can propagate to the low-latitude ionosphere at the speed of light through the zeroth-order transverse magnetic (TM0) Earth–ionosphere waveguide mode and can generate zonal ionospheric currents at low latitudes. A numerical simulation by Tsunomura and Araki (1984) has shown that the amplitude of the polar electric field decreases as it propagates towards the Equator due to the small proportion of the size of the polar electric field in relation to the propagation distance. Nevertheless, the process is still efficient enough to allow the polar waves to reach the dip equator and be abruptly amplified by the high ionospheric Cowling conductivity in that region. Thus, this model predicts an enhancement of Pc5 amplitudes in the vicinity of the dip equator.
Using a global network of magnetometers with high time accuracy, Chi et al. (2001) observed differences in the arrival time of PRI signals at middle and low latitudes. This result raised questions about the validity of the Earth–ionosphere waveguide propagation model of Kikuchi and Araki (1979), which predicts that the PRI onset should be seen simultaneously at all locations on the Earth's surface (for a more complete review of the public discussion between these authors, the reader is referred to Kikuchi and Araki, 2002, and Chi et al., 2002). Chi et al. (2001) suggested that these differences could be explained by MHD wave propagation along the path proposed by Tamao (1964), in which a compressional wave propagates along the Equator until it is converted into a shear Alfvén wave that then propagates along the field lines to the ionosphere. This penetration mechanism has also been recently proposed for different ULF waves (e.g., Yizengaw et al., 2013). Due to the screening effect of the enhanced ionospheric conductivity at the Equator on an MHD wave signal incident from the magnetosphere, this model would lead to a damping of the wave amplitude at the dip equator and a phase lag of this signal compared to an off-equatorial region.
Both characteristics were tested in our data set. Figure 8a shows the mean
relative amplitudes of Pc5 pulsations around noon for the five stations in the
equatorial region, with standard deviation error depicted by error bars.
Although the 1
Generally, the detailed characteristics of the Pc5 amplification and phase delay in the South American equatorial region are not consistent with only one of the proposed wave transmission mechanisms. The Earth–ionosphere waveguide propagation model agrees well with the overall equatorial enhancement and phase delays of these pulsations, but it is inconsistent with the lower amplification observed at the dip equator reported in the study. On the other hand, the MHD wave propagation model explains the local damping in Pc5 amplitudes as well as the expressive phase lag at the dip equator, but it does not account for the general equatorial enhancement of the Pc5 signals. Probably, when the two mechanisms work in an interactive coupled manner, they may explain the variability of the equatorial enhancement. A more advanced theoretical model is needed, which would consider the simultaneous effects from MHD wave propagation and from the Earth–ionosphere waveguide. It must be also considered that fine-scale amplification factors in the equatorial zone should be taken cautiously since ground-based magnetometers have poor spatial resolution because the pulsation signal arises from the effect of ionospheric currents integrated over a transverse ionospheric region of dimensions comparable with the height of the ionosphere (Engebretson et al., 1995).
The local time distribution of Pc3 relative amplitudes in Fig. 6 shows
expressive amplification around sunrise at the stations closest to the dip
equator (ARI, PRM, and VIL). As an example of such events, Fig. 9 shows
band-pass filtered data (10 to 45 s) of geomagnetic
Band-pass-filtered geomagnetic variations (
Figure 6 presents a small number of events around sunrise. To clarify the
significance and effect of the sunrise terminator on Pc3 pulsations, the
data were reevaluated to obtain statistically more robust results. Only two
stations were used to detect coherent events, one under the dip equator and
affected by the dawn effect (PRM) and the other unaffected by this effect
(CUI). Using these stations, 32 events were detected between 02:00 and 10:00 LT, and
their relative amplitudes (PRM/CUI) are shown in Fig. 10. In this figure,
squares are relative amplitudes considering the spectral amplitudes of
the
Distribution of Pc3 relative amplitudes between the PRM and CUI
stations around sunrise. Squares are ratios of the
The amplification characteristics observed by our data are certainly related
to the strong longitudinal gradient of the ionospheric conductivity near the
dawn terminator. This feature changes the patterns of the ionospheric
currents and consequently of the magnetic field on the ground driven by
these currents. Considering the strong north–south polarization of the
ground pulsations, we suggest that an increase of zonal (east–west)
ionospheric currents in the sunlit region adjacent to the dawn terminator
controls the observed amplification of the
An explanation for this difference can be sought in terms of the existence of large longitudinal variations in electrodynamic processes in the South American equatorial ionosphere, as verified by various types of ground- and satellite-based observations. These variations were reviewed by Abdu et al. (2005) and are associated with the strong longitudinal variation in the geomagnetic declination angle and total field intensity due to the presence of the SAMA. The magnetic declination angle controls the development of the F-layer dynamo in dawn and dusk hours that couples the E and F regions and plays a key role in the electrodynamics of the equatorial ionosphere. The enhanced ionization by energetic particle precipitation increases the background conductivity distribution in the ionosphere over the SAMA central region and may also extend for several degrees in longitude and latitude, reaching the equatorial ionosphere, even in quiet conditions. The combination of both effects is a unique feature of the South American region and causes significant longitudinal/seasonal differences in the equatorial plasma density distribution.
The dissimilarities in the results between central South America (PRM
station) and western South America (HUA station) may therefore arise mainly
from differences in the geomagnetic parameters of the two stations. During
the experiment of Saka and Alperovich (1993), the magnetic declination angle
and the total field intensity in HUA were 1.5
Sketch illustrating the processes responsible for the enhancement
of zonal electric fields around sunrise (partially modified from Farley et
al., 1986; Kelley et al., 2014). Neutral thermospheric winds (
Following Abdu et al. (2005), the extra ionization at PRM due to the SAMA develops polarized electric fields in the vicinity of the enhanced conductivity gradient of the dawn terminator, particularly when external electromagnetic fields (e.g., geomagnetic pulsations) are imposed. A secondary longitudinal electric field structure is produced, which is superimposed, in phase, on the regular zonal electric field. This enhancement of the zonal electric field around dawn can be explained in part following the same mechanism proposed by Farley et al. (1986) for the development of pre-reversal electric field enhancement in the evening hours. In this case, F-region thermospheric winds blowing eastward through the sunrise terminator are the source of sunrise enhancement. Figure 11 illustrates the creation of the polarization electric field in the equatorial F region and its mapping to the low-latitude E region (as presented by Kelley et al., 2014). The zonal gradient of the Hall conductance generates an accumulation of positive charges across the sunrise terminator, and, as a consequence, an enhancement of zonal electric fields is developed to the east (dayside) of the terminator.
In addition to this dynamo action, it is known that compressional MHD waves propagating near the solar terminator are also expected to induce zonal ionospheric currents in the vicinity of the dip equator (Saka and Alperovich, 1993). Moreover, according to the model of fast-mode incidence on the ionosphere described by Alperovich and Fedorov (2007), these zonal currents induced by the compressional waves are frequency dependent, with higher frequencies creating strong ionospheric currents and consequently more amplified magnetic fields on the ground. These same authors also presented experimental evidence that the polarization of pulsations in the Pc3 frequency band is not sensitive to the presence of the terminator during the equinoxes. This is consistent with our results and the period of our measurements (September–October). Thus, assuming that our Pc3 pulsations occurring around sunrise are also related to compressional waves vertically incident on the ionosphere (as recognized for the noon events), we can explain their ground amplification and no change in polarization. This is probably due to the net effect of zonal currents generated mainly by the electric field carried by the fast-mode waves, substantially enhanced by the secondary longitudinal electric field structure related to the dynamo action through the F-layer thermospheric winds. These effects may also be controlled by the large longitudinal variations of electrodynamic processes in the South American equatorial ionosphere.
The enhancement of the integrated zonal electric field related to the proposed dynamo mechanism attains its largest values when the sunrise terminator is aligned with the magnetic meridian. In central South America, this condition prevails during a period close to the December solstice, when there is a seasonal maximum in the intensity of the secondary zonal electric fields associated with the negative magnetic declination (Zhang et al., 2015). This corresponds exactly to the period in which our measurements were taken. On the other hand, the proposed dynamo mechanism for the zonal electric field enhancement over PRM around sunrise would be seasonally dependent on the solar zenith angle. Due to the limited temporal coverage of our ground measurements, we cannot test this hypothesis at PRM, but Saka and Alperovich (1993) reported a seasonal modulation in the polarization parameters at HUA as a function of the solar angle. Also, despite a large amount of evidence of the longitudinal ionospheric variability in the South American equatorial sector, it is difficult to verify our proposition to explain the observed differences between PRM and HUA due to the lack of direct and simultaneous measurements of ionospheric electric fields at these places.
Using multipoint observations at equatorial ground stations in central South America, we studied the spatial variation in the amplitude of continuous pulsations (Pc3 and Pc5) observed in the zone of influence of the EEJ currents. Nighttime events in the equatorial region exhibit a systematic pattern of minor amplification, while pulsation amplitude changes during daytime as a function of the period and as the stations approach the dip equator. These results can be explained by the low intensity of the equatorial ionospheric currents at night and by the development of a high ionospheric conductivity along the dip equator during daytime.
We found attenuation in the Pc3 amplitudes around local noon. This amplitude
damping is observed exclusively at the station closest to the dip equator
and for wave periods shorter than
Another important result is the sunrise effect observed on Pc3 ground
pulsations at the stations closest to the dip equator. Its main effect on
our data is to increase the
In summary, the study presented new information about the effects of EEJ currents on the geomagnetic pulsations recorded on the ground at equatorial latitudes. The detailed characterization of the Pc5 pulsation amplification in the region around the dip equator was not previously known and was only possible by the availability of several stations operating simultaneously under the EEJ. Also noteworthy are the results on the peculiar feature of the sunrise effect in this region. A very unique combination of factors contributed to the anomalous behavior of Pc3 pulsations around the dawn terminator. On the other hand, interpretation is limited by the small amount of data with the stations operating simultaneously and the absence of simultaneous measurements of other ionospheric parameters during the study period which may support the proposed interpretation.
Data for the Dst and AE indexes are publicly available at
GBDS processed the magnetic data, performed the analysis, and wrote the paper. ALP and LRA contributed to the interpretation of the data.
The authors declare that they have no conflict of interest.
This article is part of the special issue “7th Brazilian meeting on space geophysics and aeronomy”. It is a result of the Brazilian meeting on Space Geophysics and Aeronomy, Santa Maria/RS, Brazil, 5–9 November 2018.
The study was supported by a research grant from FAPESP and fellowships from CNPq. The geomagnetic experiment was designed and carried out by Tai Kitamura (Kyushu University, Japan) and Nalin B. Trivedi (INPE). We are grateful to two anonymous reviewers for their helpful comments.
This research has been supported by the FAPESP (grant no. 92/04764-7) and the CNPq (grant nos. 304353/2013-2 and 131675/2015-0).
This paper was edited by Marcos D. Silveira and reviewed by two anonymous referees.