Magnetospheric conditions
The event which we focused on had a 3-h Kp index of less than 1+ all day before August 19, 2010, which indicates that the magnetosphere remained in a quiet and steady state. Figure 2a shows the provisional auroral electrojet (AL) index from 09:00 to 10:00 UT on August 19, 2010. The decrease from −70 to −250 nT in the AL index indicates that the substorm onset occurred at 09:12 UT, as represented by the vertical dashed line. To show that Pi2 pulsations occurred globally at low latitudes, Fig. 2b illustrates the time series plots of the H component at Honolulu (HON) and Memanbetsu (MMB) at low latitudes during the interval between 09:00 and 10:00 UT. The substorm onset can also be identified by the positive bay in the geomagnetic field data at HON, which occurred at midnight. Clear Pi2 pulsations appeared simultaneously at the substorm onset at HON and MMB.
The electric field experiments on the THEMIS satellites allowed us to estimate the electron number densities and to determine the location of the plasmapause. Figure 3 shows the electron number density derived from the THEMIS spacecraft potential data plotted as a function of an AACGM latitude in the ionosphere at a height of 100 km, which was determined by use of the Tsyganenko 96 model. The electron number density increased steeply around 65.5° AACGM latitude, which was the location of the plasmapause. During the time intervals of Pi2 pulsations observed by the satellites, which are indicated by the horizontal lines in Fig. 3, all three satellites were located in the plasmasphere.
Satellite observations of Pi2 pulsations
Figure 4c shows the perturbations of the electric and magnetic fields from the satellites, which were observed simultaneously at the low-latitude MMB station when THA (THD and THE) was (were) located in the Northern (Southern) Hemisphere pre-midnight sector, as shown in Fig. 4a, b at 09:12 UT. The THD and THE were very close to each other during the Pi2 pulsations. For all three satellites, the E
R, E
A, and B
// components exhibited oscillations in the Pi2 frequency band with slightly different waveforms from the Pi2 pulsations in the H component at MMB. The Pi2 pulsation did not appear clearly in the B
R and B
A components. Note that regular oscillations with periods of 1 min in the B
A component were artificial. The E
R, E
A, and B
// components oscillated with periods of 40 and 70 s throughout the event, whereas Pi2 pulsations at MMB oscillated with dominant periods of 70 s. The amplitude of the 40-s oscillations was much smaller than that of the 70-s oscillations. Oscillations in the B
// and E
R components were in-phase relative to those on the ground, whereas the Pi2 pulsations in the E
A component had an out-of-phase relation with those on the ground. These phase characteristics indicate that the Pi2 pulsations from satellites were not standing waves but rather propagating waves.
Figure 5 shows the spectral properties of the event. The electric and magnetic field data from satellites were linearly interpolated and resampled at the same 1-s time rate as the geomagnetic field data. After subtracting 300-s boxcar data from the original data to remove long-term variations, which were defined as dH, dB, and dE, a fast Fourier transform (FFT) was applied to the residual dH, dB, and dE in the time interval from 09:11 to 09:19 UT (512 data points), and three-point smoothing was conducted in the frequency domain, thus providing a Nyquist frequency of 500 mHz and a frequency resolution of 1.9 mHz. The upper left, middle, and right panels of Fig. 5 indicate the spectral properties of the Pi2 pulsations in the electric and magnetic fields observed by the THA, THD, and THE satellites, respectively. The power spectral densities for the dE
A, dE
R, and dB
// variations from the satellites and the dH variations at MMB, which represent S
(dEA) (f), S
(dER) (f), S
(dB//) (f), and S
(dH) (f), exhibited clear peaks around 12 to 14 (f
1) mHz. At 23 to 25 (f
2) mHz, S
(dEA), S
(dER), S
(dB//), and S
(dH) had their second highest peaks. The total S(f
2)/S(f
1) power ratio for the dE
A, dE
R, dB
//, and dH components was lower than 0.7. At both f
1 and f
2, the dE
A–dH, dE
R–dH, and dB
//–dH coherences were nearly perfect (approximately 1.0), and the cross-phase spectra showed that dH was in-phase with dB
// and dE
R was out of phase with dE
A.
In order to identify the direction of the wave energy propagation, we calculated the Poynting flux P by using the following formula:
$$ \mathbf{P}=\frac{\mathrm{d}\mathbf{E}\times \mathrm{d}\mathbf{B}}{\mu_0} $$
Figure 6 shows the Poynting flux in the radial, azimuthal, and parallel components (positive radially outward, duskward, and northward, respectively). The transverse components exhibited non-zero mean values during the Pi2 interval. The negative P
A and P
R components imply that wave energy propagated westward and earthward. These features were inconsistent with the cavity mode model, in which the Poynting flux in the radial component had a mean value close to zero because the cavity mode consisted of standing fast-mode waves. The amplitude of the Poynting flux in the parallel component was smaller than that in the transverse components. The sign of P
// from THA was positive, whereas the signs of P
// from THE and THD were negative. This means that the electromagnetic energy flux was directed away from the magnetic equator and toward both hemispheres.
SuperDARN observations
Figure 7 shows the time series plots of the line-of-sight Doppler velocities obtained with the UNW radar (beam 14, ranges of 3–5 and 14–22 (Fig. 7a)); with the TIG radar (beam 4, range of 13–27 (Fig. 7b)); and with the HOK radar (beam 4, ranges of 21–29 and 53–57 (Fig. 7c)) on August 19, 2010, at 09:10–09:20 UT. The black arrows indicate the plasmapause locations, which were estimated by the electron number density from the THEMIS satellites, as shown in Fig. 3. The plasmapause was located between the range of 21 and 22 for the UNW, and it occurred at latitudes higher than the observation points of the HOK and UNW radars. Furthermore, we present a time series plot of the geomagnetic field variations in the dH component at MCQ in the bottom panels of the UNW and TIG observations and those in the dH component at St. Palatunka (PTK) in the bottom panels of the HOK observations. These geomagnetic observations were made at almost the same latitudes as the observation points in the field of view of each radar. At lower latitudes, the amplitudes of Doppler velocity variations from the HOK radar at lower latitudes were smaller than those at higher latitudes. The periods of Doppler velocity variations from the HOK radar at ranges 53–56 were almost identical to those of the Pi2 pulsations in the dH component at PTK with a 180° phase difference. Data from the UNW radar showed that the periods of Doppler variations were much shorter than those of the Pi2 pulsation in the dH component at MCQ and the largest amplitude was recorded around a range of 16. At higher ranges (14–22), the waveforms of Doppler variations from the UNW radar were almost identical. In the Doppler velocities from the TIG radar, the periods of the variations were identical at all ranges between 09:13 and 09:15 UT, whereas the periods at higher latitudes (range of 19–27), which were located near or outside the plasmapause, were longer than those at lower latitudes (range of 13–18) after 09:15 UT. In the Doppler velocities from both the UNW and TIG radars, Pi2 pulsations followed those at MCQ. Waveforms of Doppler variations from the UNW and TIG radars were quite different from those at MCQ.
Figure 8a–c shows the corresponding power spectral densities S(f) for Doppler velocity variations (dV), as shown in Fig. 7a–c. The plasmapause locations are identified by black arrows. The spectra of dV from the HOK radar, which exhibited echoes at a lower latitude than the UNW and TIG radars, showed a dominant peak at approximately 14 mHz with a maximum power density near the range of 53. In comparison, two dominant spectral peaks at f
1 and f
2 appeared in the S(f) of dV from the UNW radar, as shown in Fig. 8a. The S(f) of dV from beam 14 of the UNW radar exhibited a dominant peak at 24 mHz, which reached a maximum power at a range of 16, and a smaller peak at 11 to 14 mHz, which archived maximum power at lower latitudes in the range of 5. The amplitudes of the 24-mHz signal in the range of 14–22 for the UNW radar were much larger than those of the 11- to 14-mHz and 24-mHz signals for the HOK and TIG radars. The 11- to 14-mHz signal exhibited maximum power in the range of 15 for the TIG radar, whereas the 23- to 24-mHz signal had a smaller or comparable peak to those of other frequencies. The S(f) of dV from the TIG radar at ranges of 23–27 also showed a single peak in the frequency range from 13 to 18 mHz, which was centered at approximately 15 mHz. The S(f) shapes of dV from the TIG radar at ranges of 23–27 were different from those of TIG at ranges of 13–18, which had double peaks at f
1 and f
2. These features indicate that Pi2 pulsations observed at ranges of 23–27 for the TIG radars, which were located outside the plasmasphere, were excited by different mechanisms from those at lower latitudes in the plasmasphere.
Figure 9 shows the latitudinal characteristics of the f
1 and f
2 peaks that appeared in the Doppler velocities of the HOK, UNW, and TIG radars. To identify the location of the latitudinal power peak of the f
2 signal that appeared in the radars, the latitudinal profile for the f
2 peak normalized by the f
1 peak S(f
2)/S(f
1) was plotted, and the results are shown in Fig. 9a. The vertical dashed lines indicate the plasmapause locations. The latitudinal S(f
2)/S(f
1) peaks greater than two in the Doppler data were localized at 61° to 65° AACGM latitude and were located near the plasmapause. However, the S(f
2)/S(f
1) of dV from the TIG radar at 61° to 65° AACGM latitude was much smaller than that from the UNW radar. This could be attributed to the longitudinal difference between the UNW and TIG radars. The HOK radar, which was located at a lower latitude than the other radars, observed S(f
2)/S(f
1) to be less than two. Figure 9b, c shows the latitudinal profile for the cross-phase between dV from radars and dH at MMB with high coherence (>0.7) between dV and dH at f
1 and f
2, respectively. The latitudinal phase profiles at both f
1 and f
2 exhibited changes at 55° AACGM latitude. At lower latitudes, the cross-phase was approximately 180° at both f
1 and f
2, whereas at higher latitudes, the cross-phases were between 0° and 90° (around 90°) at f
1 (f
2).
Ground observations
Figure 10a, b shows the geomagnetic field data in the H and D components on the ground from 09:08 to 09:23 UT. The bottom two panels show the geomagnetic field data in the Southern Hemisphere. All ground stations except for Dumont d’Urville (DRV) were located in the plasmasphere. The waveforms of Pi2 pulsations at MQR and DRV were different from those in both the H and D components at MMB, PTK, and Magadan (STC), which were located in the plasmasphere, whereas Pi2 pulsations in the H component at MMB, PTK, and STC oscillated with the same periods and with a slight phase difference.
Figure 11 shows the spectrum densities of the geomagnetic field in the dH and dD components. The power spectra at MMB, PTK, and STC had clear peaks in the frequency range from 9 to 18 mHz, and peaks were centered at approximately 14 mHz. However, the frequency of peaks in the dH component at MCQ was slightly different from those at lower latitudes. Moreover, the frequency of Pi2 pulsations at DRV was much lower than those at lower latitudes. These spatial characteristics of frequencies indicate that the Pi2 pulsations inside the plasmapause were excited by the same source, whereas those outside or close to the plasmapause were excited by different sources.