During the VLF-CHAIN campaign, several interesting features of VLF/ELF emissions were observed, such as QP emissions, falling-tone and rising-tone chorus, hiss, as well as ‘bursty-patch’ emissions, first reported by Shiokawa et al. (2014). We visually inspected the VLF spectra from ATH from 17 to 25 February, looking for clear chorus, QP, and other emissions, then applied the FFT to obtain spectral and polarization parameters. We then chose the cases with relatively high spectral intensity and coherence values, retaining five instances of rising-tone chorus, one falling-tone chorus, four QP emissions, two intense chorus mixed with hiss, and three instances of bursty-patches. The time span of these emissions varied from 1 to 2 min for the bursty-patches to several hours for the other emissions. Using the polarization parameters, we found several interesting characteristics for different types of waves. We first concentrate on the variation of the polarization angle with time and frequency using data from ATH. Then, we pinpoint the source location of three events observed simultaneously at ATH and FVM.
Frequency dependence
We observed several instances in which the polarization angles of the VLF/ELF emissions were frequency-dependent, which we illustrate with the following two examples that include chorus emissions mixed with a QP emission or with hiss:
(1) On 17 February, there was a long and intense emission that lasted for approximately 4 h, starting around 09:50 UT as a QP emission with rising tones which, after 20 to 30 min, had a chorus-like structure mixed with an intense hiss band ending at around 13:30 UT. This corresponds to post-midnight to early morning, local time (02:50 to 06:30 LT). Figure 2 shows the spectral and polarization parameters of a 30-min time frame at the beginning of this emission. The rising-tone chorus can be seen in the PSD panel from frequencies of approximately 1.5 to 3.5 kHz (Figure 2d), accompanied by some QP structures that can be more clearly observed in the first 5 min. In Figure 2a, high coherence values (≥0.6) between N-S and E-W components can be seen for most of the emission around 2 kHz, as well as at 3.5 kHz for the first 8 min. Similar results are observed for the polarization degree in Figure 2c. In the polarization angle panel (Figure 2b), we observe that for lower frequencies (approximately 2 kHz), the angle is negative, approximately −35° (blue). However, for higher frequencies (approximately 3 kHz), the angle is positive, approximately 20° (light yellow). We also note that in the three polarization panels shown in Figure 2a,b,c, there is a gap at around 2.8 kHz that is not visible in the PSD panel (Figure 2d).
(2) On 18 February, another long emission, lasting for almost 6 h, was observed. It also started as a QP emission around 14:00 UT, lasting for about 2 h, then became a hiss at around 16:00 UT, mixed with a chorus emission at about 18:20 UT, and finally ended around 20:30 UT. This corresponds to morning to early afternoon, local time (09:00 to 13:20 LT). Figure 3, in the same format as the previous figure, shows a 30-min interval in the middle of the chorus emission. We can clearly identify a hiss band mixed with chorus from 2.5 to 4.0 kHz, composed of some rising-tone elements visible around the 8- to 10-min mark in the PSD panel (Figure 3d). As in the previous example, high coherence and degree of polarization values are observed (Figure 3a,c). In Figure 3b, we can see that the polarization angles at higher frequencies, above 3 kHz, have positive values close to approximately 15° (light yellow), whereas for frequencies below 3 kHz, the angle is negative, closer to approximately −20° (light blue).
This frequency dependence of the polarization angle was observed a total of six times during the campaign. It was usually seen once per event, per day (except on 19 February, when it happened twice in 1 day during a particularly long event showing chorus-like structures mixed with QP), lasting from several minutes to several tens of minutes (up to 40 min).
Time dependence
In addition to the frequency dependence of the polarization angle, we noticed several occasions in which the polarization angle of the VLF/ELF emission was time-dependent, as illustrated by the next two examples:
(1) On 23 February, we observed an emission suddenly start during a quiet time, lasting for almost 5 h. It started at 16:25 UT with very clear QP elements, quickly showing chorus-like structures, to then mix with hiss and finally disappearing around 21:10 UT, corresponding to morning and early afternoon, local time (09:25 to 14:10 LT). Figure 4d clearly shows the starting point and subsequent 25 min of the emission for frequencies of 2.5 to 4.0 kHz. Figure 4a,c shows high coherence between the N-S and E-W components, as well as high values of degree of polarization (both ≥0.8). These panels can help us identify a wave structure with a period of several minutes within the emission, which could be an overlapping QP emission but that only lasts for about 20 min. Figure 4b shows that during this 25-min interval, the polarization angle values change from negative (approximately −30°, blue) to positive (approximately 25°, light yellow), then back to negative (approximately −40°, dark blue) for another few minutes, only to return to positive values (approximately 20°, light yellow) in a similar time frame. Additionally, we can see the appearance of a gap at around 3.2 kHz, which is not detected in the PSD in Figure 4d but visible only in the last 10 to 15 min of the polarization panels (Figure 4a,b,c).
(2) On February 19, there was an extremely long QP emission lasting from 14:30 to 23:00 UT, showing rising-tone elements between frequencies of 1 and 2.5 kHz. In this case, the emission did not turn into a chorus-like band as in previous examples, but it did show chorus elements during some intervals. Figure 5 shows a 30-min span towards the end of the QP emission: in Figure 5d, we can clearly see the rising tone centered on 1.5 kHz, as well as the QP structures with an uneven periodicity of several tens of seconds. Figure 5a,c shows high coherence between N-S and E-W magnetic field components and polarization degree values (both ≥0.7). In Figure 5b, we can see that for the first 6 to 7 min of this chosen time frame, the polarization angle is negative (approximately −10°, light blue) or close to 0° (light green), then changes to positive (approximately 15°, yellow) for about 9 min, only to return to negative (approximately −40°, dark blue) for a few minutes before finally stabilizing to positive values. Unfortunately, at this point, the intensity of the emissions lessens to a degree that the polarization analysis cannot be properly visualized.
Time-dependent emissions such as the ones shown in these examples, were observed eight times during the 9-day period, six of which were found during the very long event on 19 February. In all of the occurrences, intervals of polarization angle sign changes lasted for several tens of minutes, within which sign changes took place every few minutes (from 2 to 10 min).
Polarization analysis of bursty-patches
Shiokawa et al. (2014) reported short time-scale bursty emissions during geomagnetically disturbed periods, observed during the VLF-CHAIN campaign, giving them the name ‘bursty-patches’. Figure 6 shows a 10-min interval of the polarization and spectral parameters of the very long event observed on 19 February (also shown as an example in Figure 5), accompanied by a bursty-patch. Figure 6d shows the bursty-patch starting at the 8-min mark falling from 5 to 4 kHz for about 1 min, then rising in frequency to slightly above 5 kHz for the next minute or so. In this panel, we can also see a QP emission mixed with chorus with a period of about 20 s, accompanied by rising tones with frequencies of 1.0 to 2.5 kHz. Compared to the other examples, Figure 6a shows relatively low coherence (between 0.5 and 0.7), but Figure 6c shows polarization degree values greater than 0.6. In Figure 6b, the main chorus emission centered at 2 kHz starts with slightly positive polarization angle values, then, in a few minutes, becomes highly negative (up to −50°, dark blue). On the other hand, the polarization angle of the bursty-patch remains positive, varying from approximately 5° (green-yellow) to approximately 20° (yellow) for its whole duration.
Chorus structures
With the help of the very high sampling rate of data (100 kHz), we can create high temporal resolution plots of the spectral and polarization parameters that allow us to see the chorus’ structure. To improve the temporal resolution from 0.16384 to 0.04096 s, we lowered the frequency resolution from 97 to 388 Hz, averaging every 64 frequency data points. Figure 7 shows a 1-min interval of a chorus emission observed at ATH on 19 February at 18:15 UT, just before noon, local time. In Figure 7d, we can see structures with a periodicity of approximately 10 s, as well as chorus elements with a time scale of a few seconds. Figure 7a,c shows high coherence between N-S and E-W components and high polarization degrees, both with values greater than 0.7. The polarization angle is shown in Figure 7b. Due to the lowered frequency resolution, however, the detailed view of the variation of the polarization angle for each chorus element is not conclusive. Nonetheless, we can say that the polarization angle for the approximately 10-s structures seems to be fairly stable, between 0° (light green) and approximately 20° (yellow).
Locating the ionospheric exit point
From 17 to 25 February 2012, we found several chorus events that were simultaneously observed at ATH and FVM. Only three of them were intense enough to yield significant results from the polarization analysis at both stations. The three cases were observed on successive days from 18 to 20 February and at similar times in the early afternoon, local time. Figure 8 shows, in order, from top to bottom, the resulting polarization angle values as a function of time for 18, 19, and 20 February. The raw angle values, each for 30 s of data per 10 min, are shown in light blue for ATH and orange for FVM. Their calculated mean angle and their corresponding error bars are shown in dark blue for ATH and red for FVM. Among the examples previously shown in the sections ‘Time dependence’, ‘Polarization analysis of bursty-patches’, and ‘Chorus structures’, only the one corresponding to the very long event on 19 February gave satisfying results for both stations (Figure 8b). We observe that all three cases show similar angle variations for both ATH and FVM. This trend can be seen more clearly in Figure 8c, as the angle values for both stations drop until 19:30 UT, then rise at around 20:10 UT, only to drop again until 20:30 UT.
We used the calculated mean value of the polarization angle, combined with triangulation, to follow the movement of the ionospheric exit point during these three events (e.g., Tsuruda and Hayashi 1975, Ozaki et al. 2008). In an effort to make the movement clearer for the reader, we chose only 4 to 5 relevant points per case, which are plotted in Figure 9. The geographic locations of the two stations and the directions of the incoming chorus waves as seen from ATH are in blue, and those from FVM are in red, for all three events. The numbers inside each dot indicate the progression with time of the location of the ionospheric exit point of the chorus emissions, from 1 to 4 for Figure 9a,c. In the case of Figure 9b, the exit point moved in order, from locations 1 to 4, then went back to location 1. The stations are separated by 450 km, so we can estimate that the exit point hovered for a maximum of approximately 90 km in case (a), approximately 125 km in case (b), and approximately 275 km in case (c). We also note that in Figure 9a,b, the ionospheric source of the chorus seems to move in an overall straight line, moving away from ATH or FVM with forward and backward movements. On the other hand, in the case of Figure 9c, the source seems to move in a more disorderly manner, hovering between the two stations, but also in a forward/backward motion. We note that in all three cases, the source location always seems to be between the two stations and closer to FVM than to ATH.