The presence of ionospheric scintillation at high latitudes is typically attributed to some form of irregular plasma structuring, and subsequent signal refraction and diffraction. In this section we present cases for the presence of both energetic particle precipitation and large-scale plasma structures as catalysts for the observed GPS phase scintillation during the period of study.
4.1. Particle Precipitation
 We focused our attention on two periods of interest. The first period was the ∼16:30 UTC time period on 5 April 2010 (see Figure 4). During this period, South Pole and site Eagle (81°S W22°) were both in the magnetic local noon sector at ∼(12:30, 13:10) MLT, respectively, while South Pole was at ∼74°S corrected geomagnetic (CGM) latitude and Eagle was at ∼67°S CGM latitude. Both the South Pole GPS receiver and the Eagle GPS receiver showed significant scintillations, but in two significantly different magnetic latitude sectors; South Pole would typically have been considered to be in the dayside cusp region while Eagle would have been in the dayside auroral zone, although magnetic storm conditions can alter the cusp and auroral boundaries. The second period of interest was the ∼07:30 UTC period on 6 April 2010 (Figure 5). Again, both stations showed significant scintillations. However, they were then in the magnetic local midnight sector, and there was very little large-scale structuring at all. Thus, it appears that the observations indicate three distinctly separate scintillation events, with differing geophysical conditions.
 The first event was the 16:30 UTC period for the South Pole scintillations. These scintillations were located near the dayside cusp. Over the 1 hour period from 16:00 to 17:00 UTC, four separate satellites observed scintillations, with three satellites observing two separate periods of scintillations over the 1 hour period. Of the total seven scintillation periods, five of them exhibited very similar properties. Each scintillation event consisted of a short burst of phase fluctuations that lasted ∼30 s. Within the 30 s burst there were several pseudoperiodic oscillations that lasted ∼5–6 s. Each of the scintillation bursts were located in local magnetic time near 13:00 MLT, and at a magnetic latitude of 73–74°S. In addition, the amplitude of each of the scintillation events varied by 3–4 dB over the 30 s period, indicating that the source was likely diffractive in nature. These similar characteristics indicate a single source of the scintillations located near 13:00 MLT, 73–74°S latitude and extending in time for at least 30 min. It seems likely this source region was due to cusp precipitation of some kind. However, without correlative data this must remain only a plausible supposition. Figure 6 presents a representative example over a 200 s period for one of the 30 s burst events for PRN 4. The phase scintillations have been filtered with a sixth-order Butterworth filter (with 0.1 Hz cutoff frequency). The short burst of scintillations is similar for all five events.
Figure 6. High data rate phase scintillations from PRN 4 recorded at South Pole during 1500–1700 s after 16:00 UTC on 5 April 2010. A 30 s burst period included several pseudoperiodic oscillations that lasted ∼5–6 s. Similar signatures from other satellites during the same period suggest a source of signal diffraction fixed in magnetic local time for ∼30 min, likely due to cusp precipitation.
Download figure to PowerPoint
 The second event was the 16:30 UTC period for the Eagle scintillations. These scintillations were located in or near an electron density depletion region at ∼70°S geographic latitude and ∼330°W longitude. This depletion region appeared at the “break-off” point of the plasma enhancement, between the solar-produced dayside plasma and the resulting patch-like body of enhancement that drifted southward. There are three interesting questions regarding this event. First, what was the cause of a density depletion region, at local noon, which was in a sunlit region? Second, what was the cause of the phase scintillations and third were the physical causes of the density depletions and scintillations linked? To help in answering these questions, Figure 7 presents a pass of DMSP 17 over Antarctica from 16:10 to 16:15 UTC. The third panel down presents the electron energy and energy flux along the pass, while the fourth panel presents the ion energy and energy flux. Unfortunately, this pass was in the ∼16:00 MLT sector rather than the 13:00 MLT sector of the observations. However, it is at least possible that the precipitation observed by DMSP in this sector was similar to precipitation events at ∼03:00 MLT away, considering the same magnetic latitude regions. If we focus on the elevated red electron precipitation from the time 16:12 UTC for ∼35 s, we notice electron precipitation with >1 KeV energies and large energy fluxes. This would imply fairly hard E region precipitation. In addition, the proton precipitation in the same time period showed 1–10 KeV energies which were also E region. During this period the magnetic latitudes were ∼70–68°S, which is very similar to the Eagle magnetic latitudes. Particle data from the POES N18 satellite also showed strong electron precipitation and some proton precipitation in the ∼1400 MLT sector (geographic longitude 344.7°, latitude 69.9°S at ∼16:38 UTC), lending further support to the suggestion of precipitation energy spread over the entire postnoon sector. Thus it seems likely that the Eagle phase scintillations manifested as a result of mixed plasma structuring; kilometer-scale (or larger) E region precipitation irregularities, and large-scale plasma density gradients associated with the enhancement structure break off. However, there was also significant <1 KeV soft electron precipitation especially at ∼16:12 UTC. This was probably F region precipitation, and was located at similar geomagnetic latitudes to the large density depletion observed in Figure 4. It is possible that the F region precipitation had elevated the electron and ion temperatures, which would lead to enhanced recombination of O+, and thus reduced electron densities. Valladares et al.  show evidence for this mechanism causing multiple TOI break-off events in the northern high latitudes.
Figure 7. A particle spectrometer record from DMSP satellite F17 during 16:10–16:15 UTC, 5 April 2010. Labeled longitudes are eastern. Electron energies of >1 KeV, proton energies of 1–10 KeV, and elevated energy flux levels indicate a period of hard E region precipitation within a few sectors of MLT from the plasma enhancement patch break-off point.
Download figure to PowerPoint
 The third event of interest was on 6 April 2010, from 06:00 to 08:00 UTC, and is represented in Figure 5. Here there was significant phase scintillation on a number of different satellites from both Eagle and South Pole. During this period, the ionosphere was unstructured in the region of scintillations and had low overall TEC values. Figure 8 presents a DMSP F18 pass from 06:30 to 06:40 UTC. The pass trajectory went right through the region of the scintillation observations, particularly the period from 06:33 to 06:37 UTC. During this period there were very high energies of electron precipitation (>10 KeV), with high-energy flux, that extended across the region from ∼350° to 293° geographic longitude, and from ∼74°S to 80.4°S latitude. This suggests hard auroral precipitation that was probably directly causing kilometer-scale irregularities, which produced the observed phase scintillations. There was also a narrowly confined region of high-energy proton precipitation located at approximately 315° longitude and 80°S latitude that appears to correlate well with some of the larger scintillations presented in Figure 5 for the 06:30 UTC and 07:30 UTC maps.
Figure 8. A particle spectrometer record from DMSP satellite F18 during 06:30–06:40 UTC, 6 April 2010. Very high electron energies (>10 KeV) with high-energy flux are present, extending from 293° to 350° geographic longitude and 74°–80°S geographic latitude, that is, in the observation area of site Eagle. This suggests hard auroral precipitation. Note also the short period of high-energy proton precipitation at ∼06:35 UTC.
Download figure to PowerPoint
 Figure 9a shows a sample single-satellite time series of detrended 50 Hz L1 phase fluctuation during 07:04 UTC on 6 April 2010 at site Eagle; the low-frequency component due to satellite movement has been removed by a sixth-order Butterworth filter with cutoff frequency 0.1 Hz. Figure 9b shows an accompanying energy density spectrum of phase fluctuations from PRN 16, obtained by Fourier transform of 9000 50 Hz samples (3 min in time). Cycle fluctuations were occurring on a time scale of several seconds, implying that the ionospheric changes were of correspondingly longer time scales than those associated with classical diffractive scintillation. This is also supported by the phase scintillation spectrum (9b) that shows significant phase fluctuation occurrence at lower frequencies. It is likely that TEC gradients, associated with local precipitation energy input, were the cause of the phase scintillation in this case.
Figure 9. (a) A sample of detrended 50 Hz L1 phase fluctuation from PRN16 recorded at site Eagle during 07:04 UTC on 6 April 2010 and (b) accompanying smoothed energy density spectrum of the phase fluctuation using a total of 9000 samples around this time. Detrending was performed using a high-pass sixth-order Butterworth filter with cutoff frequency of 0.1 Hz. Cycle fluctuations were occurring on a time scale of several seconds, implying that the ionospheric changes were of correspondingly longer time scales than those associated with classical diffractive scintillation.
Download figure to PowerPoint
 Incoherent scatter radar measurements [Haldoupis et al., 2000; Yin et al., 2008; Mitchell et al., 1998] have previously shown that during strong precipitation events, enhancements in E region electron density can reach up to 4 × 1011 electrons per cubic meter over an extended altitude range. Projecting this electron density over, for example, an arbitrary 50 km vertical distance equates to ∼2.5 TECU. Taking into account also the low-elevation nature of GPS raypath geometry at high latitudes, phase fluctuations such as those in Figure 9a could feasibly have arisen from E region irregularities.
4.2. Large-Scale Plasma Structuring
 The time sequence of large-scale (hundreds of kilometers) plasma structuring in Figure 4 shows a plasma enhancement structure “breaking off” from the lower-latitude solar-produced plasma, resulting in a large southward drifting plasma patch and apparent depletion region at the break-off point. The reconstructed enhancement patch in the 16:30 UTC image of Figure 4 is approximately 1000 km long and 500 km wide; however 500 km is approaching the limit of the reconstruction resolution. The plasma depletion region is interesting since it occurred in the sunlit sector of the polar region, and appeared to be well equatorward of the cusp, within the auroral oval. It has been suggested above that the depletion region was due to enhanced recombination of O+ due to soft electron precipitation. However, it is important to consider whether the depletion region could be an artifact of the tomographic imaging process, particularly because there is not an abundance of data in the Antarctic region. To address this issue, a completely separate imaging method, Ionospheric Data Assimilation Four-Dimensional (IDA4D) [Bust et al., 2004], based on data assimilation techniques, was run over the same time periods as MIDAS. For the 5 April 2010 analysis, IDA4D used International Reference Ionosphere (IRI) as a background initial model and ingested the same dual-frequency GPS data set from 50 sites shown in Figure 1; that is, identical raypath observations were processed by both MIDAS and IDA4D assimilation tools. In addition, IDA4D ingested TEC from the ∼55 ground Doppler Orbitography and Radio-positioning Integrated by Satellite (DORIS) stations, including four on the continent of Antarctica, LEO satellite occultation TEC from five Constellation Observing System for Meteorology, Ionosphere and Climate (COSMIC) satellites, LEO satellite topside TEC data from five COSMIC satellites and Satellite de Aplicaciones Cientifico-C (SAC-C). IDA4D uses a Gauss-Markov Kalman Filter to predict the solution forward in time.
 Figure 10 shows IDA4D specification of TEC over Antarctica at 16:45 UTC, overlaid with the data coverage ingested in its production. There is broad agreement with the MIDAS result shown in Figure 4. The two imaging methods use different algorithms and assumptions and further, IDA4D has used radio occultation data from COSMIC. Of particular note is the improved data coverage in the IDA4D specification across the region of depletion in plasma entering the polar cap. This provides some confidence in the accuracy of the TEC maps.
Figure 10. IDA4D TEC image over Antarctica at 16:45 UTC on 5 April 2010, overlaid with ingested data coverage. Solid yellow lines show COSMIC occultation intercepts (derived from the longest great circle paths between receiver and satellite, along the occultation trajectory up to 800 km altitude). Orange dots represent GPS raypath intercepts at 350 km. Dashed yellow lines show topside TEC intercepts at 1000 km from COSMIC, CHAMP, and GRACE. Red squares and lines show DORIS coverage at 350 km intercepts. This reconstruction used GPS raypath observations that were identical to the MIDAS method, with the addition of TEC input from DORIS, COSMIC, and SAC-C instruments and a Kalman filter approach.
Download figure to PowerPoint
 Further verification is provided by the GPS receivers on board the CHAMP and GRACE satellites. Figure 11a shows the upward looking vertical TEC projection above the CHAMP satellite altitude of ∼300 km. It is clear that there is an enhancement of electron density in the topside ionosphere above the satellite in the vicinity of the plasma enhancement patch from the tomographic images. Figure 11b shows a similar plot of vertical TEC for the GRACE satellite altitude of ∼480 km. It is noted that the high altitude of the patch may be indicative of the Carlson et al.  mechanism of formation. Thus, we believe the plasma enhancement region to be a real physical effect, most likely a result of antisunward plasma drift from a TOI separation being sustained by field-aligned soft electron precipitation.
Figure 11. GPS vertical TEC projections over Antarctica (a) above the CHAMP satellite between 16:15 and 16:45 UTC and (b) above the GRACE satellite between 16:00 and 16:25 UTC. Both show enhancements in TEC at the crossing of the plasma enhancement patch, with the GRACE levels slightly lower owing to its orbital altitude above the bulk of the F region.
Download figure to PowerPoint