Examples of EPIs observed by Swarm
Figure 2b, c presents two examples of EPIs observed by Swarm during the first study period. For both events, the top panel presents the latitudinal profiles of the original electron density time series (Ne, 2 Hz resolution) with different colors for different spacecrafts. The epochs, altitudes and longitudes when the spacecraft passed the geographic equator are listed in the topside. The middle panel shows the reconstructed background Ne and the bottom panel presents ΔNe with background variations subtracted. For the first event observed on December 10, 2013, the three spacecrafts flew at an altitude of about 501 km. Swarm B was the leading spacecraft, crossing the geographic equator 1 min earlier than A and C. In this case, Swarm A was 0.1° (about 11 km) westward of B and 0.2° (about 22 km) eastward of C, respectively. As the three spacecrafts were so close in time and space, they observed similar plasma density irregularities in both hemispheres. In the northern hemisphere, the correlations of the density depletions, R
max, attained values of 0.91, 0.83 and 0.76 for the pairs Swarm B/A, A/C and B/C, respectively. As Swarm B/A were most closely spaced, R
max is largest, as expected. Furthermore, a correlation less than 1.0 between Swarm B/A also reveals that the EPI had fine structure with zonal extent less than 0.1° (about 11 km) or that the plasma density structure did change within the 27 s leading time between Swarm B/A. As described in section “Processing approach,” to calculate the correlations, for each Swarm pair, we took the ΔNe series from the eastside spacecraft as reference and time-shifted the other one. In the northern hemisphere, we found Δt is about 2.0, 3.5 and 5.0 s for the three pairs, respectively. The positive values of Δt result from the EPI inverted-C shell structure. In fact, from the latitudinal profiles of electron density, we can also see that Swarm C observed the density depletion at highest latitudes, as expected for the most westerly spacecraft.
However, the correlations are significantly reduced in the southern hemisphere with R
max of 0.56, 0.55 and 0.42 for the three pairs, compared to the northern hemisphere. The corresponding Δt with values of −10.0, −14.5 and −24.0 s are also somewhat larger in absolute value than those in the northern hemisphere. And we found the electron density latitudinal profiles show much finer structures between 6° and 12°S MLAT, which might contribute to the lower correlations in the southern hemisphere. In addition, the negative values of Δt in the southern hemisphere seem not support the westward-tilted inverted-C shell structure of EPI.
Figure 2c presents another example of EPI on December 19, 2013. The electron densities measured by Swarm also show similar depletions in both hemispheres. In this case, the three spacecrafts were practically at the same altitude of about 501 km, and Swarm B was 1 and 2 min ahead of Swarm A and C, respectively. The longitudinal separation between Swarm B/A increased to 0.2° (about 22 km), and 0.3° (about 33 km) between Swarm A/C. In the northern hemisphere, R
max was larger than 0.7 for all the three pairs and again was largest for Swarm B/A. In the southern hemisphere, R
max is somewhat smaller but still around 0.7. The time shifts (Δt) were again mainly positive in the northern and negative in the southern hemisphere.
Statistic results
With the approach described above, there were about 60 EPI events detected by all the three spacecrafts during the first period and about 1000 events detected by Swarm A and C during the second period. Figure 3 shows the global distribution of the equatorial plasma irregularities for both periods. During the first period, most of the plasma irregularities occurred in the longitude sector between −90 and 0°E, which reflects well the seasonal dependence of EPI’s occurrence almost exclusively above the south American/Atlantic sector during December solstice months (e.g., Stolle et al. 2008). During the time of the final constellation phase considered in this study, the EPIs were more evenly distributed in longitude; however, the increased concurrency between longitude −90° and 0°E is again visible.
The longitudinal separation and local time of sampling varied simultaneously during the first period. Therefore, it is difficult to separate their effects on EPIs from the data during this period. Conversely, the longitudinal separation between Swarm A and C was almost fixed (about 1.4° in longitude) during the second period, which allowed us to check for a possible local time dependence on EPI scale sizes. Figure 4 shows the local time distribution of EPI events observed by Swarm A and C during the second period. The green and red bars indicate EPIs detected by each of the spacecraft A and C, respectively. The blue bars represent the number of EPIs simultaneously observed by the two spacecrafts. The number of detections for the individual satellite is always higher than the number when both satellites detect depletions, because the individual satellite may also monitor depletions when the other satellite does not. When comparing the EPI number of detected events between the two satellites, we found EPI occurrence is slightly higher from Swarm A than from Swarm C for almost all local times. Differences in data availability have been excluded, and both satellites provided continuous data sets. Differences in the sensitivity of the instrument can be possible, but we did not explicitly investigate such an effect. Further statistical investigation into the difference in detection is interesting and should be continued with a larger database. However, the occurrence rates of EPIs observed by the two spacecrafts are similar over local time evolution, which increases dramatically between 1900 and 2000 LT and gradually decreases during post-midnight hours.
For statistical analysis of the EPI scale sizes, we calculated R
max and Δt only when EPI was detected by both spacecrafts. Figure 5a, b shows the relation between R
max and Δt in both hemispheres. The circles represent the individual EPI events marked by different colors for the three different pairs. R
max reached larger values when Δt was close to zero, and the focusing is more prominent in the northern hemisphere. Frames c and d present the dependence of R
max on the longitudinal separation (ΔGlon) between the spacecrafts. The gray triangles and vertical lines denote the mean values and the standard deviations for each longitudinal bin (0.1°). The longitudinal separation between spacecrafts varied from 0.1° to 1.2° during this period. A clear trend of the R
max mean values can be found in both hemispheres: R
max
gradually decreases with ΔGlon untill ΔGlon approaches 0.4° (about 44 km). Higher correlations are found for smaller separations, especially in the northern hemisphere. When ΔGlon > 0.4°, the mean values of R
max stay around 0.4 in both hemispheres. The second peak of the R
max mean value at ΔGlon = 0.9° is possibly caused by the small event number (only three events) within that bin.
We have also calculated the correlations between EPIs but only for those events simultaneously observed by Swarm A and C during the second period. Figure 6a, b shows the relation between R
max and Δt separately for the two hemispheres, with blue circles representing the individual EPI events. Similar to the result during the first study period, we find generally larger R
max when Δt is close to zero. Frames (c) and (d) present the dependence of R
max on local time. The gray triangles and vertical lines are again the mean values and standard deviations for each local time bin (1 h). For given spacecraft longitudinal separation of about 150 km, R
max shows a considerable spread with a mean value of around 0.4 and standard deviations of ca. 0.2, especially after 2000 LT. These results suggest that the majority of events include plasma density scale sizes less than 150 km. Figure 6e, f shows local time distribution of the occurrence ratio for those EPI events with R
max > 0.7 between Swarm A and C. We see clearly that shortly after sunset (around 1900 LT), about 18 % EPIs are found with zonal scale length larger than 150 km (R
max > 0.7), and these larger-scale-length EPIs decay fast during later local times.
The second study period comprised a data set covering one and a half years, which allowed us for a more detailed investigation of EPIs local time dependence. Since all observations have been made at zonal separation of about 150 km, little correlations are expected between the ΔNe records (2 Hz) of Swarm A and C. We further low-pass-filtered ΔNe with different window sizes (in seconds). One of such example is presented in Fig. 7, which was observed on May 14, 2014. Considering a satellites speed of 7.6 km/s, the filter window sizes of 0.5, 1.5, 2.5, 5, 10, 15 and 20 s correspond to moving-average length about 3.8, 11.4, 19, 38, 76, 114 and 152 km along-track, respectively. At larger moving-average length, the small-scale density irregularities are smoothed out and the R
max values gradually increase. When a 15-s filter was applied, the small-scale plasma irregularities were totally smoothed out, and the R
max values surpassed 0.8 in both hemispheres in this example.
The local time dependence of R
max for different filter window sizes of ΔNe during the second study period is presented in Fig. 8. With larger moving-average lengths, the R
max values gradually increase. When the filter window size increased to 20 s, better correlations (R
max around 0.7) are obtained for all local times. Additionally, relatively larger R
max values are found shortly after sunset, 1900–2000 LT (also seen Fig. 6), which we interpreted as resulting from larger zonal scale sizes during the earlier evening local time as will be discussed in section “Discussion and summary.” These scale sizes drop rapidly with local time and are insignificant after 2000 LT.