The response of high latitude ionosphere to the 2015 June 22 storm

This work investigates physical mechanisms triggering phase scintillations on L-band signals under strong stormy conditions. Thanks to selected ground-based Global Navigation Satellite Systems (GNSS) receivers, located both in Antarctica and in the Arctic, an interhemispheric comparison between high latitude ionospheric observations in response to the peculiar solar wind conditions occurred on June 22, 2015 is here shown. To trace back the observed phase scintillations to the physical mechanisms driving it, we combine measurements from GNSS receivers with in-situ and ground-based observations. Our study highlights the ionospheric scenario in which irregularities causing scintillation form and move, leveraging on a multi-observation approach. Such approach allows deducing that scintillations are caused by the presence of fast-moving electron density gradients originated by particle precipitation induced by solar wind variations. In addition, we show how the numerous and fast oscillations of the north-south component of the interplanetary magnetic field (Bz,IMF) result to be less effective in producing moderate/intense scintillation events than during period of long lasting negative values. Finally, we also demonstrate how the in-situ electron density data can be used to reconstruct the evolution of the ionospheric dynamics, both locally and globally.


INTRODUCTION
The geomagnetic storm occurred from 21 to 24 June 2015 is one of the largest geomagnetic storms of the 24 th solar cycle. Beside its intensity, the peculiarity of the storm lies in its occurrence during the summer solstice and in resulting from the superposition of several solar events. In fact, it was caused by a series of three inter− planetary shocks hitting the Earth's magnetosphere at 16:45 UT on June 21, at 05:45 UT on June 22 and at 18:30 UT on June 22, 2015, respectively. Unlike the 2015 St. Patrick's storm that was caused by multiple Inter− planetary Coronal Mass Ejections (ICMEs), all the June shocks resulted from single ejecta (Liu et al., 2015).  (B z, IMF ), the solar wind (SW) density, the SW proton temperature, the SW velocity, the SW dynamic pressure P, the SYM-H index and the AU (black) and AL (red) indices from 21 to 24 June 2015. The shaded region indicate the ICME interval and the dashed vertical lines mark the corre− sponding shocks arrivals at WIND.
As reported by several authors [see e.g. Astafyeva et al., 2017;Piersanti et al., 2017;Cherniak and Za− kharenkova, 2017] and shown in Figure 1, the first D'ANGELO ET AL. shock (black dashed line), accompanied by sharp changes in the SW density (panel e), temperature (panel f) and speed (panel g), compressed the Earth's magne− tosphere and caused a sudden increase of the SYM-H of ~40 nT (panel i). Anyway, since the North−South com− ponent (B z, IMF ) of the IMF remained at most positive at the shock arrival (panel d), no substorm activity follows (panel l).
The second shock (red dashed line) was accompa− nied by a small solar wind density increase (panel e), causing an enhancement in the SYM-H of ~20 nT (panel i). At the shock's arrival, the B z, IMF turned nega− tive (panel d) causing a sharp enhancement of the au− roral activity, as visible in panel l. Then, B z, IMF fluctuated around zero until the arrival of the third shock (panel d) and a smaller decrease in the SYM-H index was ob− served (~ −40 nT, panel i). The last shock (green dashed line), accompanied by a large and sudden increase in solar wind and IMF components, caused a large and sudden increase in the SYM-H index up to ~88 nT at 18:37 UT (e.g., storm sudden commencement, panel i). The ejecta following this shock was characterized by a large negative B z, IMF (−39 nT, panel d), that caused a drop of the SYM-H index to −208 nT (on June 23, panel i) and, consequently, a strong auroral activity (panel l). Such conditions led to unusual responses of the iono− sphere−thermosphere system labelled by both inter− hemispheric asymmetries and latitudinal differences.
The multi−instrumental works by Prikryl and co− authors [2011Prikryl and co− authors [ , 2013Prikryl and co− authors [ and 2015 on the interhemispheric response to geospace forcing highlighted how the main asymmetries are due to the IMF dawn-dusk component, being responsible of the cusp location, and of the main orientation of the plasma convection within the polar cap. The combination of both effects results into a dif− ferent occurrence of plasma patches between the two hemispheres, so leading to causing significant differ− ences in the scintillation patterns over high−Arctic re− gions and over Antarctica. However, the solstice conditions, under which the June 2015 storm occurred, make the explanation of the inter−hemispheric asym− metry and latitudinal development very challenging, as reported in the recent literature [Astafyeva et al., 2016;Mansilla, 2017;Cherniak and Zakharenkova, 2017]. In fact, Mansilla [2017], which performed a global study of the ionospheric Total Electron Content (TEC) from high to low latitudes, observed asymmetries of the TEC re− sponse in both hemispheres. More in detail, he observed a TEC increase in the Southern hemisphere well corre− lating with an increase of the O/N 2 ratio. Correspond− ingly, he observed a decrease in the Northern hemisphere, not associated with a decrease in O/N 2 ratio. In addition, Astafyeva et al. [2016], who analysed variations of the ionospheric vertical TEC and electron density in the topside ionosphere during the initial and the main phases of the storm, observed a pronounced hemispheric asymmetry in the nigh time topside iono− sphere. Specifically, in the Northern Hemisphere (sum− mer), they observed an extreme enhancement in the investigated parameters attributed to the combination among the prompt penetration electric fields, the dis− turbance dynamo and the storm−time thermospheric circulation. Cherniak and Zakharenkova [2017], inves− tigating the ionospheric irregularities through a chain of Global Navigation Satellite System (GNSS) located in both hemispheres from middle to high latitudes, found a good correlation between the occurrence of ionos− pheric irregularities and the variations of the AE and the SYM-H indices.
To provide further insights about the physical mech− anisms leading to the irregularities formation and their effect on GNSS satellites, our study focuses on the scin− tillation events occurred in the high latitude regions of both hemispheres on June 22, when the bulk of the storm−driven ionospheric disturbances are observed. In particular, we analyse scintillation events recorded by five stations distributed both in the Arctic and in Antarctica. To trace back the observed scintillations to the physical background triggering the irregular iono− sphere, we combine the information coming from TEC and from scintillation indices with the observations ob− tained from in-situ data (Swarm and Polar−orbiting Operational Environmental Satellites -POES -constel− lations) and from ground−based acquisitions (Super Dual Auroral Radar Network -SuperDARN). Moreover, since this day was characterized by largely varying solar wind and IMF conditions, we discuss about the role played by different solar drivers in the scintillation production.
This paper is organized as follows: section 2 illus− trates the method adopted in the study; section 3 dis− cusses the results; section 4 provides discussion of the results; section 5 draws the conclusions.

DATA AND METHODS
This paper reports the analysis of GNSS data acquired by GPS Ionospheric Scintillation and TEC Monitor [GISTM, Van Dierendonk et al. 1993] receivers on June 22, 2015. Specifically, we use data from Eureka (EURC), Resolute Bay (RESC) and Ny−Ålesund (NYA0) stations located in the Arctic, and from Concordia (DMC0) and Zhongshan (ZSGN) stations located in Antarctica.
The geographic and corrected geomagnetic coordinates of the receivers are reported in Table 1. Under quiet ge− omagnetic conditions, EURC, RESC and DMC0 look for most of the day at the polar cap, while NYA0 and ZSGN look at polar cusp/auroral region. This allows both to investigate the ionospheric response in different geo− magnetic sector and to perform an interhemispheric comparison.
Each station in Table 1 is equipped with a Novatel GSV4004, able to sample GPS signal phase and ampli− tude at 50 Hz for each satellite being tracked on L1 (1575.42 MHz). The receiver's firmware provides am− plitude and phase scintillation by computing the S 4 [Yeh and Liu, 1982] index every 60 seconds and the σ Φ [Van Dierendonck et al., 1993] index by considering the standard deviation of detrended carrier phase averaged over intervals of 1, 3, 10, 30 and 60 seconds. It is also able to provide TEC and the rate of TEC change (ROT) values every 15 seconds from combined L1 and L2 (1227.6 MHz) pseudorange and carrier phase measure− ments.
In order to detect and investigate the ionospheric ir− regularities induced by the interplanetary medium con− ditions on June 22, we use σ φ values from 60 seconds average and ROT over 1 minute interval. We avoid re− porting here amplitude scintillation, as no meaningful amplitude events were found during the day. The impact on measurements of longer paths through the iono− sphere, due to signals from low−elevation satellites, is accounted by using σ φ index verticalization approach proposed by Spogli et al. [2009Spogli et al. [ , 2013. According to such approach, we project the σ φ index according to the following equation: (1) where σ φ is the index directly provided by the receiver at a given elevation angle along the slant path, while F(α elev ) is the obliquity factor that is defined as (Man− nucci et al. 1993): (2) In equation (2), R e is the Earth's radius and H IPP is the height of the Ionospheric Piercing Point (IPP), which in the present investigation is assumed to be located at 350 km of altitude. According to Rino [1979aRino [ , 1979b and as described by Spogli et al. [2009], the exponent a is equal to 0.5. Although the angular dependence of the σ φ index in equation (1) is valid only under weak scattering condition, we decide to apply the formula for weak scattering to characterize a moderate/strong scin− tillation scenario as well. In fact, according to Rino [1979a] and as critically discussed by Spogli et al. [2013], the projection to the vertical leads to a σ φ value smaller than the original slant index, so leading to un− derestimate the corresponding scintillation. As the pro− jection to vertical could underestimate the scintillation level, our method ensures that the detection of high values of σ φ are associated to actual ionospheric effects.
In order to reduce the impact of non−scintillation related tracking errors (such as multipath) a mask of 20° on the elevation angle of the satellites is applied on scintillation data used in this work. In fact, although the 50 Hz sampling frequency adopted by GNSS re− ceivers is useful to investigate transient ionospheric ef− fects, it cannot distinguish the scintillations caused by ionospheric irregularities from multipath due to phys− ical obstacles (buildings, trees, etc.) that may be present in the environment surrounding the receiver antenna.
As shown by D'Angelo et al. [2015], the choice to apply a mask of 20° on the elevation angle of the satellites is not always effective in eliminating all the multipath ef− fects. Nevertheless, it is difficult to perform a site char− acterization like the one proposed by D'Angelo et al. [2015] in remote sites such as those selected for the in− vestigation proposed in this work. On the other hand, the polar regions are usually characterized by very few environmental constrains making the choice of 20° el− evation mask a good compromise between loss of scin− tillation data and multipath effects filtering. In addition, to minimize possible mismeasurements of the scintil− lation indices and ROT following a loss−of−lock event, we take into account only data characterized by lock time on L1 larger than 240 s [Smith et al., 2008]. Lastly, to discriminate between scintillating and not scintillat− ing signals, we choose a threshold of 0.25 radians, which, according to Spogli et al. [2009], identifies scin− tillations from moderate to strong levels.
In order to infer information about the electron den− sity gradients leading to the observed scintillations and the scale sizes of the ionospheric irregularities involved, we also study the ROT behaviour [Wernik et al., 2004;Zou and Wang, 2009;Alfonsi et al., 2011].
To support the reconstruction of the ionospheric fea− tures detected by ROT and scintillation parameters, we combine information coming from GNSS measurements with concurrent supporting information provided by ground−based and space−borne observations. Namely: • For the local characterization of the ionospheric electron density distribution, we analyse in-situ measurements of plasma density in the topside ionosphere provided by Langmuir probes on board  Chisham and Freeman, 2004]. To provide an overall representation of the high−lat− itude ionospheric irregularities causing the observed phase scintillations, we adopted the same method used in D'Angelo et al., [2018], in which the Altitude Ad− justed Corrected GeoMagnetic [AACGM; Baker and Wing, 1989] coordinates (MLat, MLon) and Magnetic Local Time (MLT) are used as the reference frame. For each hemisphere, we provide maps in which Super− DARN observations, Swarm electron density measure− ments and the projection of the GNSS tracks experiencing scintillation are given simultaneously. It is worth noting that all the scintillation events were recorded after the arrival of both the interplanetary shocks and appear longer lasting in the Southern Hemi− sphere. In addition, the EURC receiver did not record significant scintillation events during the day, while the RESC receiver, although located at few magnetic lati− tude degrees southward than EURC (Table 1), recorded three scintillation events at ~20:45 UT, ~22:00 UT and ~23:30 UT with mean intensities of ~0.50 radians, of ~0.60 radians and of ~0.42 radians, respectively. The DMC0 receiver recorded two scintillation series between ~08:00 UT and 10:00 UT and between ~19:00 UT and 23:00 UT, with mean intensities of ~0.45 radians and of ~0.57 radians, respectively. The NYA0 and ZSGN σ φ time profiles show similar behaviours during the entire day, even if the receiver at ZSGN recorded more intense scin− tillation events than NYA0. An exception occurred be− tween ~20:00 UT and 21:00 UT, when NYA0 recorded a scintillation of ~0.99 radians. Figure 3 shows the ROT time profiles for the same stations and in the same time interval as above. Also in this case, the dashed vertical lines mark the arrival at the magnetopause of the first (red) and second (green) interplanetary shock and different colours refer to dif− ferent satellites in view. The most intense and sudden ROT excursions were recorded after the arrival of the two interplanetary shocks, as visible in all panels of Figure 3. Such excursions appear longer lasting and more intense in the Southern Hemisphere and after the arrival of the second interplanetary shock. They also occur mainly in correspondence with the scintillation events in Figure 2. Figure 4 shows polar view maps, covering |50°|-|90°| MLat and 00:00-24:00 MLT for the Northern (top) and the Southern (bottom) Hemispheres. Each map displays an overview of the ionospheric convection patterns (ionospheric electrostatic potential reconstructed by Su− perDARN, red/blue isocontours mean positive/negative values of the potential) in the time interval 07:00−10:00 UT, corresponding to the first significant scintillation peak recorded on June 22 (Figure 2). Furthermore, each map reports the intensity profile of electron density (black line, whose thickness expresses the electron den− sity variation) recorded by Swarm A (SWA) or B (SWB). In addition, each map reports the projection of iono− spheric scintillation (colour dots), simultan eously recorded by Eureka, Resolute Bay and Ny−Ålesund sta− tions in the Northern Hemisphere (top panels) and by Concordia and Zhongshan stations in the Southern Hemisphere (bottom panels). In each map, we also report all SuperDARN observations of spectral widths: we highlight spectral widths values greater than 200 m/s by means of blue squares, while values lower than 200 m/s are represented by black squares. The SuperDARN data were collected along the two minutes after the time shown at the top of each map. Figure 4 confirms that scintillation originated in the cusp region of both hemispheres. Moreover, high spec− tral width values (blue squares) occur, in the Northern Hemisphere, mainly in the early pre−dawn sector of the auroral oval (around 3 MLT, panels a and e), while, in the Southern Hemisphere, they occur mainly in the cusp and between the two convection cells (panels b and d). The overall convection patterns, both in the Northern and in the Southern Hemispheres, are coherent with a scenario dominated by B z, IMF predominantly negative, with con− vection cells tilted towards dawn by the effect of a mainly positive B y, IMF (panels c and d in Figure 1). Figure 5 shows the polar maps for the Northern (top) and the Southern (bottom) Hemispheres in the time in− terval 19:00 to 23:00 UT (i.e. when the second signifi− cant scintillation peak occurs, see Figure 2). It is worth noticing that scintillations mainly occur in the Southern Hemisphere and appear between the two convection cells. Several authors [see, e.g., De Franceschi et al 2008;Mitchell et al 2005;Moen et al. 2013] identified the re− gion between the two convection cells as the best can− didate to host the ionospheric irregularities causing scintillations. In addition, scintillations occur close to the region in which Swarm observes clear increases of electron density (evidenced by an enhanced thickness of the Swarm track, Figure 5 bottom panels). Concerning the spectral width, in the Northern Hemisphere, the higher values are mainly concentrated in the cusp and in the early morning sector of the auroral oval. In the Southern Hemisphere, they are almost missing. The comparison between Northern and Southern maps in Figure 5 shows that the low/null number of spectral width observations above threshold (blue squares) recorded in the Southern Hemisphere could be associated with ionospheric absorption rather than with quiet ionospheric conditions. In fact, in the Southern Hemi− sphere also the number of spectral width observation < 200 m/s is low (black squares). The overall convection symmetry is very similar with respect to the earlier in− terval considered above, with B z, IMF fluctuating around zero between 19:30 and 21:00 UT, supported by a strongly positive B y, IMF (see Figure 1, panels c, d). After about 21 UT, B z, IMF reaches a more stable trend, keeping positive values: this is consistent with the shrinking of the polar cap, particularly evident in the Northern H − emisphere ( Figure 5, panels p and r). Southern Hemi− sphere seems to keep the usual symmetry of the convection patterns, even after 21:00 UT, but this could be an artefact of the convection model, being the actual SuperDARN measurements very sparse in the Southern Hemisphere during this interval.     Figure 6 shows the time profile of the rms of the ionospheric in-situ electron density measured by Swarm A (SWA, circles) and Swarm B (SWB, stars) during June 22, in both polar regions within 50° of magnetic latitude. Dashed vertical lines mark the arrival at the magne− topause of the first (red) and second (green) interplane− tary shock. Until ~18:30 UT, the rms in the Southern Hemisphere (red) is three times smaller than in the Northern Hemisphere (blue). At the same time, in the Southern Hemisphere the SWA rms (red circles) is roughly comparable with SWB rms (red stars). Con− versely, in the Northern Hemisphere, rms appears greater (blue circles) for SWA than for SWB (blue stars). After the shocks arrivals at the Earth's magnetopause (red and green dashed lines), the two satellites observed a higher electron density rms in both hemispheres. After the ar− rival of the second shock (green line), the rms in South− ern Hemisphere (red) becomes greater than in the Northern Hemisphere (blue). In addition, the rms of SWA (blue circles) becomes comparable with those of SWB (blue stars), in the Northern Hemisphere.

DISCUSSION
Although the June 2015 geomagnetic storm was one of the most intense storms of the 24 th solar cycle [Astafyeva et al., 2017;Piersanti et al., 2017], the ob− served scintillation was not as intense ( Figure 2) as during the 2015 St. Patrick's day storm [D'Angelo et al., 2018]. This unexpected behaviour might be associated with two possible reasons. The first is the peculiar con− ditions of the B z, IMF recorded on June 22, 2015. In fact, B z, IMF first showed fluctuations around zero until the ar− rival of the second shock, and then a large−brief neg− ative excursion (see panel d in Figure 1). Such condi− tion is less effective in producing moderate/intense scintillation events than a negative, long lasting B z, IMF condition. The second reason is the insufficient geo− magnetic sectors coverage of the GNSS receivers used in this analysis. In fact, according to Cherniak and Za− kharenkova, [2017], a large number of plasma density fluctuations just occurred in both auroral regions that are not in the field of view of our GNSS network.
In order to verify how the observed scintillation is related to particle precipitation, in Figure 7 we show the electron and proton total atmospheric integral energy flux as a function of the magnetic latitude from POES [Evans and Greer, 2004] constellation observations, in− tegrated at 120 km on June 22, 2015. In particular, starting from the electron and proton total atmospheric integral energy fluxes we calculate the total particle− precipitation fluxes in two time intervals: the first, be− tween the two shocks arrivals (Figure 7a), the second between the arrival of the second shock and the end of the day (Figure 7b). After the arrival of the first shock (panel a), the most intense particle precipitation occurs in the Southern Hemisphere. Correspondingly, DMC0 and ZSGN show longer lasting phase scintillations  with respect to the Northern receivers. Figure 7b shows that at the arrival of the second shock a quite symmetrical particle precipitation occurs in both hemi− spheres. At the same time, Figure 2 shows scintillations in all the investigated ionospheric regions of both hemispheres, with the exception of the Northern polar cap. It is important to note that the absence of ampli− tude scintillations allows correlating the phase scintil− lation with refractive effects [Yeh and Liu, 1982;Kintner et al., 2007] that may be associated with the plasma dy− namics caused by the particles precipitation. In fact, the latter determines an increase in the ionospheric elec− trojets that may inhibit the formation of irregularities near the first Fresnel radius.
Furthermore, Figure 2 shows that our network recorded during the day phase scintillations character− ized by almost the same intensities in both hemispheres. Such behaviour can be associated with the geometry of the shocks hitting the Earth's magnetopause. In fact, the estimated shocks normal orientations, evaluated by ap− plying the Rankine Hugoniot's conditions [see e.g. Tid− man and Krall, 1971] on solar wind data, (Table 2), lie almost in the ecliptic plane and hit the magnetopause almost at noon [Villante et. al, 2008;Alberti et al. 2016;Piersanti et al. 2016]. Actually, the rms variations (Fig−  ure 6) show a similar behaviour in both hemispheres, despite the offset between Northern and Southern rms. Such offset can be associated with seasonal effects. In fact, the different ionospheric background, in terms of ionization, produces, in this case event, a larger rms in the summer hemisphere (the Northern one). The conse− quence of a more Northern hitting of the first shock onto the magnetopause (Table 2a) coupled with the B z, IMF oscillations around zero, can be responsible of the larger rms of SWA's electron density in the Northern Hemi− sphere (blue circles in Figure 6). In addition, the conse− quence of a more Southern hitting of the second shock onto the magnetopause (Table 2b) coupled with the huge negative B z, IMF values, can be responsible of the enhancement of the Southern rms, which becomes comparable to the Northern one.
Lastly, following the analysis proposed by Alfonsi et al., [2011] and Wernik et al., [2004], we compare the ROT time profiles (Figure 3) to the σ φ variations ( Figure  2), recorded by all GNSS receivers. Such comparison with the absence of amplitude scintillation events al− lows deducing that irregularities of largely varying scale−sizes occupied the investigated regions although their dimensions were far from the first Fresnel radius. This result agrees either with Cherniak and Za− kharenkova [2017], who showed the presence of a large number of plasma density fluctuations over both hemi− spheres during the day, and with Swarm electron den− sity observations, which showed the presence of electron density enhancements in the both polar regions (Figure 4, Figure 5). Moreover, the electron density vari− ations induced by the storm drivers, confirmed also by the Swarm rms electron density variations (Figure 6), can be related to the concurring manifestation of scin− tillations (Figure 2), confirming the direct link between scintillation and abrupt changes in plasma distribution.

CONCLUSIONS
This paper investigates how ionospheric irregulari− ties, triggered by the June 2015 storm, lead to phase scintillation events in the high−latitude ionosphere of both hemispheres. To investigate the origin and the evolution of the ionospheric irregularities causing scin− tillations, we combine information from scintillation parameters and ROT (derived from GNSS receivers) with measurements acquired by SuperDARN, Swarm and POES constellations.
This study highlights how a detailed reconstruction of the ionospheric scenario in which irregularities caus− ing scintillation form and move can be achieved through a multi−observation approach. Such approach, indeed, allows deducing that scintillations were due to the presence of fast−moving electron density gradients, of several scale−sizes that are far from the first Fresnel radius, originated by particle precipitation associated with the arrival, at the Earth's magnetopause, of two interplanetary shocks. Nevertheless, although the June storm was the second largest storm of the 24 th solar cycle, it did not produce severe scintillation events in the investigated regions. We attributed such effect to the peculiar conditions of the B z, IMF that showed nu− merous and fast oscillations for most of the June 22, which are less effective in producing moderate/intense scintillation events than during period characterized by long lasting negative values of B z, IMF .
Additionally, such approach provides insights about the link between the GPS phase scintillation and the ionospheric plasma distribution. In fact, the comparison between σ φ and ROT variations and the rms of Swarm electron density measurements suggests that the ionos− pheric regions, characterized by strongly variables elec− tron density, are most likely to give rise to phase scintillation.
Our results show that the combined use of data from in-situ and ground−based sensors allows a detailed characterization of ionospheric dynamics during a ge− omagnetic storm. Being the ionospheric scintillation a complex effect and hard to predict, the study of the ionosphere at different heights and with different sam− pling frequencies can provide information useful to better characterize the scintillation triggers in the high latitude ionosphere. In the Space Weather context, this can open the door to new approaches to realize scintil− lations prediction tools, especially in terms of identifi− cation of the geomagnetic sectors most likely to be affected by scintillations.
The authors thank PNRA (Programma Nazionale di Ricerche in Antartide) for supporting the upper atmosphere observations at Concordia Station (Antarctica), and CNR (Consiglio Nazionale delle Ricerche) for supporting the upper atmosphere observations at Dirigibile Italia Station at Ny Alesund (Svalbard). The authors thank Dr. Giorgiana De Franceschi, Dr. Vincenzo Romano and Dr. Ingrid Hunstad for management of the INGV (Istituto Nazionale di Geofisica e Vulcanologia) stations. The authors acknowledge the use of SuperDARN data. SuperDARN is a collection of radars funded by national scientific funding agencies of Australia, Canada, China, France, Italy, Japan, Norway, South Africa, United Kingdom and the United States of America. Eureka and Resolute Bay GNSS data are provided by Canadian High Arctic Ionospheric Network (CHAIN, http://chain.physics.unb.ca). Infrastructure funding for CHAIN was provided by the Canada Foundation for Innovation and the New Brunswick Innovation Foundation. CHAIN and CGSM operation is conducted in collaboration with the Canadian Space Agency (CSA). Zhongshan GNSS data are provided by Beijing National Observatory of Space Environment IGGCAS through the Data Center for Geophysics, National Earth System Science Data Sharing Infrastructure. The solar wind plasma and magnetic field data of WIND, as well as the POES data, were obtained from the NASA's cdaweb site (http://cdaweb.gsfc.nasa.gov/ istp_public/). Swarm data are provided by the European Space Agency upon registration (https://earth.esa.int/web/guest/ swarm/data-access/). We kindly thank Dr. Stephan Buchert from IRFU, Sweden, for providing the Swarm Langmuir Probe Extended Dataset. This research work is supported by the Italian MIUR-PRIN grant 2012P2HRCR on The active Sun and its effects on Space and Earth climate.