Dayside Magnetospheric and Ionospheric Responses to a Foreshock Transient on 25 June 2008: 1. FLR Observed by Satellite and Ground-Based Magnetometers

As one type of driver of magnetospheric Alfvén waves, foreshock transients have received less attention than, for example, the Kelvin-Helmholtz instability, discrete and broadband frequency solar wind dynamic pressure oscillations, and interplanetary shocks. Previous works show that foreshock transients can induce both Alfvén mode and compressional mode Pc 3–5 ULF waves inside the magnetosphere. However, to our knowledge, none of these reported Pc 3–5 waves, induced by foreshock transients, are proved to be localized in the magnetosphere. In this paper, using in situ and ground-based observations, we report the generation of localized magnetospheric compressional waves and ﬁeld line resonances (FLRs) by a foreshock transient. Both the foreshock transient and Pc 5 ULF waves were found on the duskside; while on the morning side of the magnetosphere, no clear wave signatures were captured. Our results demonstrate that in addition to the global eﬀects of foreshock transients on the magnetosphere reported earlier, foreshock transients can also generate localized magnetospheric responses in the Pc 5 range with clear dawn-dusk asymmetry. A suite of eight dayside spacecraft plus ground magnetometer measurements make possible the determination of the foreshock transient driver and dawn-dusk asymmetry of the magnetospheric response not previously reported with such a complete data set. KeyPoints: and Alfvén are found to be generated by a foreshock transient • Magnetospheric the are localized with clear dawn-dusk asymmetry The dawn-dusk asymmetry of wave power with MLT


Introduction
Magnetospheric Pc 3-5 ULF waves are those continuous pulsations with period between 10 and 600 s (i.e., 1.7 to 100 mHz; Jacobs et al., 1964). Via wave-particle interactions, Pc 3-5 waves play an important role in the acceleration, transport, and loss processes of relativistic electrons near or within the outer radiation belt (e.g., Elkington et al., 1999;Hudson et al., 2000Hudson et al., , 2017Turner et al., 2012;Zong et al., 2009Zong et al., , 2012. For example, Zong et al. (2009) reported that electrons can be resonantly accelerated by the azimuthal electric field carried by poloidal modes. Elkington et al. (1999) and Hudson et al. (2000) studied electron acceleration that is caused by toroidal mode-driven inward radial transport and conserving the first adiabatic invariant. Additionally, outward radial transported electrons driven by toroidal modes could contribute to electron loss via magnetopause shadowing effects (e.g., Turner et al., 2012).
A few studies have reported that foreshock transients could induce magnetospheric Pc 3-5 waves (Eastwood et al., 2011;Hartinger et al., 2013;Zhao et al., 2017). Using Cluster, Rosetta, and ground observations, Eastwood et al. (2011) observed transient Pc 3 ULF waves induced by an HFA; Archer et al. (2015) studied the global effects of a foreshock bubble on the magnetosphere, including ULF wave activity; Hartinger et al. (2013) reported that Pc 5 ULF waves can be generated by HFAs, foreshock bubbles, and motion of the foreshock compressional boundary. Recently, Zhao et al. (2017) reported a global Pc 3 ULF wave event induced by an HFA, using Cluster and ground-based observations. However, none of these reported Pc 3-5 waves, induced by foreshock transients, are proved to be localized in the magnetosphere. Additionally, no Pc 5 standing Alfvén waves, which are mainly regarded as fundamental field line resonances (FLRs; e.g., Chen & Hasegawa, 1974;Claudepierre et al., 2010), have been reported to be driven by foreshock transients.
In this paper, we report on the generation of localized compressional and standing Alfvén waves by a foreshock transient. During this event, Pc 5 compressional waves were first observed near noon, then propagated to the afternoon sector based on in situ and ground observations. FLRs (standing Alfvén waves) were excited in the afternoon sector, based on in situ observations, while on the morning side of the magnetosphere, no clear ULF wave signatures were captured by satellites or ground magnetometers.
The paper is organized as follows: section 2 introduces the data sets that have been used in this work. Section 3 presents the observations of the solar wind, magnetosheath conditions, and ULF waves in the dayside magnetosphere. Section 4 discusses the wave energy source and FLRs. A summary is given in the last section. In a companion paper, Wang et al. (2018 hereinafter Paper 2) presents 2-D evolution of the magnetosphere and ionosphere deduced from optical observations.

Data Set
The solar wind, IMF parameters, and magnetospheric measurements used in this paper are obtained from the Time History of Events and Macroscale Interactions During Substorms (THEMIS) mission (Angelopoulos, 2008), which consists of five identical probes (THA, THB, THC, THD, and THE). Each probe is equipped with a fluxgate magnetometer (Auster et al., 2008), which provides magnetic field data and an electrostatic analyzer (McFadden et al., 2008), which provides thermal (5 eV to 25 keV) ion and electron data. We use 3-s resolution spin-fit data from those instruments.
We have also used supporting data from Geostationary Operational Environmental Satellite spacecraft (G10, G11, and G12) and ground-based magnetometers (KIAN, FSMI, RBAY, CDRT, ATU, STF, NAQ, SCO, and BJN stations). Figure 1 shows the locations of the satellites in (left) the GSE x-y plane and (right) x-z plane. During 1530 to 1600 UT, THB and THC satellites were located in the solar wind, monitoring solar wind plasma and IMF conditions. THA was situated most often in the magnetosphere and made intermittent passages into the magnetosheath. THD and THE were located in the duskside magnetosphere, observing wave signatures. G10 and G12 were in the dayside magnetosphere near local noon. G11 was located on the dawnside of the magnetosphere. The black curve indicates the location of the magnetopause estimated from the Shue et al. (1998) model.  Figure 2 shows the solar wind conditions observed by (left) THB and (right) THC satellites. THC (X GSE is approximately 10.0 R E ) was located closer to the Earth than THB (X GSE is approximately 22.7 R E ). Figures 2a-2e show THB observations of IMF, magnetic field strength, ion density, ion velocity, and ion energy flux, respectively. During ∼1536 UT to ∼1541 UT, depletions in magnetic field and density were identified. At the same time, an enhancement in the flux of high-energy ions, ∼1-10 keV, can be seen from the ion energy flux spectrum.

Solar Wind and Magnetosheath Conditions
Figures 2f-2l show THC observations of IMF, ion density, ion velocity x component, dynamic pressure, ion temperature, electron energy flux, and ion energy flux, respectively. During ∼1537 UT to ∼1545 UT, depletions in magnetic field, density, and velocity were identified. Meanwhile, plasma temperature was elevated. A distribution of hot plasma can be seen from the ion energy flux spectrum. Yellow-shaded areas mark similar phenomena observed by the two satellites. Comparing the THB and THC observations, one can find that these    features are similar but more clear at THC, which is closer to the bow shock region. Both of them observed depletions in magnetic field and ion density while ion temperature was elevated (not shown for THB). Energetic ions (∼1-10 keV) have also been identified by the two satellites during the shaded time period. These characteristics indicate the possibility that THC and THB observed a proto-hot flow anomaly (proto-HFA) event close to the bow shock region  or a foreshock cavity. In other words, a foreshock transient event was observed during that time by THB and THC. Figure 3 shows the THA observations near the magnetopause. Figures 3a-3f show magnetic field (GSM coordinates), ion density, ion velocity (GSM coordinates), ion temperature, electron energy flux, and ion energy flux, respectively. The region between two dashed black lines shows the crossing from the magnetosphere (characterized by low density, high magnetic field magnitude, low ion velocity, and high ion temperature) to the magnetosheath (characterized by high density, low magnetic field magnitude, high ion velocity, low ion temperature, and clear ion energy flux distribution around 1keV) at ∼1544 UT, then back to the magnetosphere at ∼1546 UT. The yellow-shaded time period shows that THA was located in the magnetosheath. Figures 3g-3i zoom in to the time period of interest from 1543 to 1548 UT indicated by the two dashed lines in Figure 3a. From top to bottom, panels show the ion density, ion velocity components in GSM coordinates, and total velocity, respectively. The total velocity in the sheath (yellow-shaded area) varies from ∼110 km/s to ∼250 km/s, suggesting that a magnetosheath high-speed flow may exist. This is consistent with Table 2 of Dmitriev and Suvorova (2015), where this time period is listed as a magnetosheath high-speed flow event. Figure 4 shows (from top to bottom) the magnetic field, ion density, ion velocity, electron energy flux, and ion energy flux from THD observations. One can find that after ∼1541 UT, disturbances were observed in the magnetic field, ion density, ion velocity, electron energy flux, and the ion energy flux. Ion velocity shows nearly monochromatic wave signatures. The 1-10 keV electron flux pitch angle distribution shows 90 ∘ flux enhancements that vary with period close to the wave period observed in the ion velocity, indicating the existence of wave-particle interactions. Ion density shows modulations with double the frequency of the ion velocity waves, which could be caused by low-energy ions (≤100 eV) accelerated locally by the electric field (no matter in which direction the acceleration is perpendicular to the background magnetic field), seen in the bottom panel of Figure 4. . Wave modes and phase difference analysis. Top four panels show the 3.1 to 3.5 mHz filtered electric and magnetic fields associated with compressional mode, compressional mode, toroidal mode, and poloidal mode, respectively. Last panel shows the phase differences between the electric and magnetic fields for these modes near 3.3 mHz. Dashed lines in the last panel mark ±90 ∘ . Figure 5 shows the (left) electric and magnetic fields from THD observations in field-aligned coordinate (FAC) system (e.g., Hartinger et al., 2011;Takahashi et al., 1990) and (right) the wavelet analysis of them. In the FAC, a points in the azimuthal direction, r points in the radial direction, and p is parallel to the field line. The amplitude of electric and magnetic fields reached 6 mV/m and 6 nT, respectively, which is comparable to that of shock-induced ULF waves (Walker et al., 1982;Zong et al., 2009). Narrow band frequency distribution, 2 to 4 mHz, was captured from wavelet analysis of the electric and magnetic fields. According to Jacobs et al.'s (1964) classification, the wave belongs to the Pc 5 ULF wave category. The wave power of E r is stronger than that of E a . Figure 6 shows the phase difference analysis between the electric and magnetic fields obtained from the THD satellite. The top four panels show the 3.1-to 3.5-mHz filtered electric and magnetic field components for the two compressional modes, the toroidal mode, and the poloidal mode. The bottom panel shows the phase difference between the electric and magnetic field components near 3.3 mHz. Time window we applied to calculate the phase differences is 768 s (256 points multiple 3-s/point time resolution) with 96 s (32 points) shifted time window. Therefore, time resolution of the calculated phase difference is 96 s. Black dots represent the phase difference for the toroidal mode; blue dots represent the phase difference for the poloidal mode; green and red dots represent the phase differences for two compressional modes. Dashed black vertical lines in the bottom panel of Figure 6 mark ±90 ∘ . The toroidal mode shows near 90 ∘ phase difference, indicating that standing Alfvén waves are formed along the magnetic field lines (Tamao, 1965). Electric field is leading magnetic field by +90 ∘ phase difference in the north of magnetic equator (about 5.2 ∘ magnetic latitude, MLAT). This is a feature of fundamental mode (or odd harmonics modes), instead of second harmonic mode, which should show −90 ∘ in the north of the magnetic equator (Takahashi et al., 2015). Figure 7 shows the THE observations of electric and magnetic fields in FAC. During 1530 to 1600, THE was closer to the Earth than THD. Even though the amplitude was weaker than that from THD observations, one could still find distinct wave signatures in the magnetic field (amplitude: less than 3 nT) from ∼1540 UT. The electric field did not show clear wave signatures, which may be due to the fact that the THE satellite was away from the FLR region. During 1540 to 1550 UT, THD was located at L = 8.4-8.6 and THE was located at L = 6.6-6.8. Figure 8 shows the magnetic field components measured by G11, G12, G10, THE, and THD (note that the sequence of these satellites is based on the MLT of these satellites from dawnside to duskside of the magnetosphere). The top panel presents the dynamic pressure changes monitored by THC upstream of the bow shock for reference. Three columns, from left to right, show the radial, azimuthal, and parallel components of the magnetic field, respectively. G11, located the most dawnward of the five spacecraft inside the magnetopause in Figure 1, did not observe clear wave signatures during the event. G12, G10, and THE were close to noon and observed strong Bp variations, which is associated with compressional waves. On the duskside of the magnetosphere, THD observed strong Br and Ba variations, which are associated with Alfvén waves, as well as Bp variations. From the stack plot, one can find that waves were first observed at G12 satellite, which is located at about 10.3 MLT, then propagating duskward and observed by G10, THE, and THD, respectively, at 11.3, 12.2, and 14.0 MLT. The speed of wave is 0.43 to 0.65 hr MLT per minute (79 to 120 km/s at L = 6.6 in the equator). Please see Figure 4 and figure caption of paper 2 for details. This speed is much lower than local magnetosonic wave speed, which should be in the order of 1,000 km/s with plasma and magnetic field parameters from THD, but close to the azimuthally propagating speed of the foreshock transient, inferred from the propagating speed of the discontinuity in the solar wind, please see Table 1 in the Paper 2. It is suggested that fast mode waves launched by the small-scale foreshock transient may damp fast and not propagate very far. The fast mode waves observed by satellites should be launched by the foreshock transient at the magnetopause at different MLT. Figure 9 shows the magnetic fields from ground-based magnetometers. From top to bottom, panels show magnetic fields observed from different ground stations ranging from ∼3.04 to 17.39 MLT (MLT was calculated at 1540 UT) with magnetic latitude (MLAT) close to 70 ∘ . Magnetic field vectors are shown in HDZ coordinates, where the H direction points to magnetic north, D is along magnetic east, and Z points vertically downward. Clear wave oscillations can be seen in the ATU, STF, NAQ, and SCO stations (from noon to afternoon sector). On the morning side, there is no obvious wave signature compared to the afternoon sector.

Ground Observations of ULF Wave
The wave was first observed at the ATU station (MLT = 12.75) near ∼1535 UT, then propagated to the afternoon sector. It should be noted that there is a time delay of the observation of wave signatures at STF station. For this event, scale of the upstream disturbance is small (comparing to a shock), usually less than several Earth radii and is propagating duskward with a speed of 0.5-hr local time per minute. The fast mode waves observed at different L and MLT may be lauched at the magnetopause with different MLT at different time by the foreshock transient. This could be a possible reason of causing the delay at STF station. It took about ∼13 min elapsed for the wave to propagate from the ATU station (MLT = 12.75) to the BJN station (MLT = 17.39). Considering the fact that MLT of two stations changed 0.22 hr during 13 min, the wave is eastward (duskward) propagating with a speed of about 0.56-hr local time per minute, which is consistent with satellite observations. As mentioned above, this speed is suggested to be associated with the azimuthal propagation of the foreshock transient.

Discussion
On 25 June 2008, a foreshock transient was identified by THB and THC satellites in the duskside outside the bow shock at ∼1536 and ∼1537 UT, respectively. THA was located in the magnetosphere near the magnetopause at that time and observed inward and outward motion of the magnetopause during ∼1544 to ∼1546 UT. Around ∼1535 to ∼1600 UT, distinct wave signatures were captured by THD, THE, G10, and G12 satellites in the noon to afternoon sector of the magnetosphere. Furthermore, standing toroidal mode waves at ∼3.3 mHz were observed by THD.

Wave Energy Source
During this event, geomagnetic activity was low (AE ≤ 50 nT, Kp ≤ 3, Dst ∼ −2 nT). The solar wind speed is less than 400 km/s, which is a normal solar wind speed (Hapgood et al., 1991) and of low possibility to induce K-H waves at the dayside flank magnetopause (Ogilvie & Fitzenreiter, 1989).
From THB and THC observations (see Figure 2), a foreshock transient occurred from ∼1536 UT to 1537 UT, accompanied with solar wind dynamic pressure perturbations. THA observed inward and outward motion of the magnetopause at ∼1544 UT and ∼1546 UT, respectively; please see Figure 3. The magnetopause motions may be modulated by the solar wind dynamic pressure perturbations imbedded in the foreshock transient (see Figure 2). While THA traveled into the magnetosheath, a high-speed sheath jet was observed. The jet traveled mostly along the magnetopause (inferred from comparable V X and V Y , see panel two in the right plot of Figure 3). We suggest that the magnetopause motions were induced by the foreshock transient in order to maintain pressure balance. The observation of magnetopause motions is later than expected (several minutes). Paper 2 shows that the foreshock transient measured by THB and THC is associated with IMF discontinuities. Based on the timing, the discontinuities are tilted dawnward and swept through the magnetosheath duskward. The discontinuity propagation explains the timing and propagation of ULF signals from dawn to dusk by the foreshock transient.

Wave Distributions
ULF oscillations were observed at G10 and G12 near noon MLT at ∼1535 UT (see Figure 8). THD and THE in the afternoon sector observed a corresponding wave signature around 1541 UT (Figures 4 and 7). Ground magnetometer observations show coincidental observations at station ATU at 12.75 MLT around ∼1535 UT. Stations STF and NAQ at ∼ 13 MLT observed the waves at ∼1540 UT. Station BJN in the duskside observed the waves at ∼1548 UT.
The compressional mode wave frequency observed by THD was 2-4 mHz. The magnetic field fluctuation observed by THE, G10, and G12 deeper near the noon MLT was ∼2 mHz. Additionally, ground magnetometer station observations of the magnetic fields show that the frequency is ∼2-4 mHz from ∼65 ∘ to ∼74 ∘ magnetic latitude.
The satellites and ground observations show that compressional mode waves were launched near noon by the ion foreshock disturbance via pressure balance at the magnetopause. Then these waves spread eastward to the afternoon sector, while in the morning sector, there are no obvious wave signatures captured by satellite or ground magnetometer. These observations lead to a suggestion that compressional waves were excited most strongly when the ion foreshock disturbance moved into the afternoon sector, as indicated by THB/THC observations. Additionally, a ∼3.3 mHz toroidal mode standing Alfvén wave was driven at the location of THD. With assumptions that the magnetic field is dipolar (Degeling et al., 2010) and density is a power law distribution (Denton et al., 2006), with power law index equals 4.0, the eigenfrequency of the field line near THD was calculated to be 4.3 mHz (poloidal mode) and 5.1 mHz (toroidal mode); please see Rankin et al. (2005) for calculation method. The calculated eigenfrequency of the field line is close to the observed 3.3 mHz field line resonance, and ratios of observed to calculated frequencies of approximately 0.7 are reasonable for distorted, nondipolar field geometries (e.g., Berube et al., 2006). As mentioned in section 3.2, the wave mode should not be a second harmonic mode due to the +90 ∘ phase difference between electric and magnetic fields in the north of magnetic equator. Thus, it is suggested to be a fundamental mode toroidal wave.
Wave mode number m is estimated using ground magnetometer measurements. The wave number m = ∕ (Sarris et al., 2009), where is the phase difference of the H magnetic field component and represents the longitudinal difference between the two stations. Even a small separation in geomagnetic latitude between two stations may result in considerable error (Sarris et al., 2013). To minimize error, we choose stations SCO (MLT: 14.99, MLAT: 71.39) and BJN (MLT: 17.39, MLAT: 71.68) with small difference in geomagnetic latitudes. The parameter is calculated to be ∼291 ∘ and is 36 ∘ for the two stations. Thus, the wave mode number m is ∼8. However, if these waves are propagated from different locations of the magnetopause at different times, the calculated wave mode number may not be trustable. Figure 10 shows the schematic diagram of a foreshock transient accompanied with dynamic pressure changes impacting the magnetosphere, making the magnetopause move inward and outward. Then compressional mode waves are launched at the magnetopause. These waves are propagated from noonside to the afternoon side of the magnetosphere. At the location where the compressional mode frequency is harmonically matched to the FLR frequency, FLRs will be excited (Hughes, 2013;Lee & Lysak, 1989). The magnetic field direction was obtained based on the 5-min average of the magnetic field before the THC observation of foreshock transients, please see Figure 2f.

Summary
In this paper, we studied a FLR event on 25 June 2017 from satellite and ground magnetometer observations. In the upstream region, a foreshock transient was observed by THB and THC. THA, near the magnetopause, monitored an inward and outward motion of the magnetopause. In the dayside magnetosphere, G10, G11, THD, and THE observed Pc 5 ULF wave activity. The amplitude of electric and magnetic fields observed by THD reached 6 mV/m and 6 nT, respectively. On the morning side, no clear wave signatures were captured based on space and ground observations.