Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
 Analysis of a sequence of auroral images acquired by the Polar Ultraviolet Imager (UVI) during a long-duration (∼1.5 hr) solar wind dynamic pressure enhancement reveals auroral intensity variations in the Pc5 band. Power spectral analysis indicates that the “auroral pulsations” appeared in multiple frequency bands predominantly in the day sector and the dawn-dusk flank of the oval, with the maximum wave power skewing toward the dawn sector. Ground-based magnetometer recordings showed Pc5 micropulsations modulated with the auroral pulsations. The lower frequency (1–3 mHz) auroral pulsations were modulated with solar wind dynamic pressure pulses moving tailward, whereas the higher frequency (6–8 mHz) ones were not. We concluded that the lower frequency auroral pulsations were driven directly by the solar wind, whereas the higher frequency ones were associated with global compressional cavity mode. This study shows for the first time the snapshot of global Pc5 ultralow-frequency (ULF) waves and demonstrates that global far-ultraviolet (FUV) imaging such as Polar UVI provides a powerful ancillary tool for studying ULF waves on a global scale.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Ultralow-frequency (ULF) geomagnetic pulsations in the Pc5 band (∼1.67–6.67 mHz) [Jacobs et al., 1964] are considered to be the longest-wavelength (comparable to the size of the magnetosphere) hydromagnetic waves in the magnetosphere and can be observed over a large area. There are three basic types of Pc5 ULF waves: the field line resonant (FLR) toroidal mode, global compressional cavity mode, and breathing mode. It is proposed that FLRs are driven by the magnetosheath flow [Dungey, 1954; Samson et al., 1971]. Under this scenario, large-amplitude surface waves excited at the magnetopause via a Kelvin-Helmholtz instability can launch magnetohydrodynamic (MHD) fast mode waves propagating earthward across the Earth's magnetic field line to couple with the shear mode Alfvén waves through a mode conversion, whereby the energy of the surface waves can transfer to the local field line resonance [Southwood, 1974; Chen and Hasegawa, 1974].
 Global compressional ULF oscillations are also expected to arise when the magnetosphere is impacted by abrupt variations in the solar wind dynamic pressure [Warnecke et al., 1990]. Such a global compressional mode is monochromatic, and its frequency depends on the size of the cavity within which waves are trapped [Kivelson and Southwood, 1985]. It is also suggested that very low-frequency monochromatic FLRs can tap energies from the magnetospheric cavity mode [Kivelson and Southwood, 1985]. However, the breathing mode is a direct response of the magnetosphere to solar wind pressure variations that drive a forced breathing magnetopause into a global compressional ULF waves within the magnetosphere [e.g., Korotova and Sibeck, 1995; Stephenson and Walker, 2002; Kepko et al., 2002].
 A number of expected features can be used to identify the excitation mechanisms of Pc5 ULF waves. For example, FLRs are localized and their frequencies are L-shell dependent, as the eigenfrequency of the field line depends on the length and mass of the field line. Second, the cavity mode is expected to be global and its frequency is determined by the size of the cavity in which the waves are trapped. Third, the forced breathing mode is also expected to be global but its wave frequency is determined by the oscillating frequency intrinsic to the solar wind plasma density/dynamic pressure. However, studies of Pc5 ULF waves are limited to point or multi-point measurements and thus cannot provide instantaneous ULF wave information on a global scale required for mode identification. In this letter, we demonstrate, with one event study, that this problem can be resolved to some extent by satellite-based global auroral imaging such as the Ultraviolet Imager (UVI) [Torr et al., 1995] onboard the Polar satellite.
 This event is associated with a large, long lasting (∼1.5 hr) density/pressure pulse, which took place on September 26, 1999 [Liou et al., 2007]. According to Solar Wind Electron Proton Alpha Monitor (SWEPAM) onboard the ACE spacecraft [McComas et al., 1998], a square wave function-like enhancement in the proton number density (with a peak value of ∼50 cm−3) was first observed at ∼1820 UT, and the associated solar wind bulk flow was large (∼480 km/s) but relatively steady (not shown).
 The auroral display in response to the pressure pulse exhibits significant enhancements in both the dawn and the dusk sectors, forming the so-called “compression aurora” (see Liou et al.  for a detailed discussion of the auroral response to the pressure pulse). Figure 1 shows auroral keograms at eight magnetic local times (MLTs), from 03 to 24 MLT with 3-hr separations. These keograms are derived from a sequence of Polar UVI images at the Lyman-Birge-Hopfield band (∼160–180 nm) from 19:00 to 21:00 UT. The time resolution of these images is 36.8 s. The sudden brightening of the aurora in the dawn and dusk sectors is caused by the magnetospheric compression. Within the enhancement region, variations in auroral intensity exhibit no significant latitudinal variation (hereafter called auroral pulsations) and can be seen clearly from the dawn (03 MLT) through the noon to the dusk sector (18 MLT). Periodic auroral forms are most obvious at 06–09 MLT and least obvious in the midnight sector. The frequency for the intensity variations in the dawn sector appears to decrease in time.
 To gain more insights into these auroral pulsations, we integrate the intensity of the aurora along latitudes (60°–90°) for each magnetic local hour sector and perform spectral analysis of these latitudinal-integrated auroral luminosities using the Fast Fourier Transform. The result is shown in Figure 2. A noticeable result is that the auroral pulsation is revealed in discrete bands spreading over the day sector and forming a midnight gap, except for the lowest frequency (<1 mHz). Most of these discrete bands showed little change over a large spatial extent. Generally speaking, in the Pc5 band (up to the Nyquist frequency of ∼13.6 mHz), the largest power centered at ∼1.8 mHz and covered from ∼04 to 11 MLT, followed by a peak at ∼7.7 and ∼6.4 mHz that covered from 07 to 13 MLT. Between 2 and 5 mHz, there are distinctive, localized (∼2–3 MLT) power peaks centered at ∼3.6 and 4.8 mHz. These peaks were very localized and showed frequency changes in local time. Distribution of the auroral pulsation power was not symmetric about noon; the auroral pulsation power maximizes in the prenoon (06–12 MLT) sector.
 In order to understand if any relationship exists between the auroral pulsations and Pc5 geomagnetic pulsations, we examine magnetometer data in the dawn sector, where auroral pulsations exhibit greatest amplitude. Figure 3 shows the H-component of high-latitude magnetometer data (1-min resolution) from four 210° MM stations [Yumoto and the 210° MM Magnetic Observation Group, 1996]: Kotel'nyy (KTN), Tixie (TIX), Chokurdakh (CHD), Zyryanka (ZYK), and Kakioka (KAK). Large Pc5 pulsations (δB > 100 nT) were observed at TIX and CHD between ∼19:30 and 20:30 UT. A close examination of the auroral images (not shown here) one by one, we found that during the magnetospheric compression, TIX and CHD were inside the oval while KTN was in the polar cap and ZYK was in the subauroral region. Figure 3b shows TIX observations of five large-amplitude, short-time Pc5 (f ∼ 2 mHz) between ∼20:10 and 20:40 UT, with the first four wave cycles coinciding with the four large-amplitude auroral pulses (red trace), as indicated by a green horizontal bar in Figure 3b. These Pc5 oscillations were more or less seen at other stations from high to low latitudes. Similarly, large-amplitude Pc5 waves of higher frequency (f ∼ 3–4 mHz) can also be seen at CHD between ∼19:40 and 20:20 UT. There is a short period of time, ∼19:40–20:10 UT (as indicated by a green horizontal bar in Figure 3c), when both magnetic (H-component) and auroral pulsations show in-phase modulations.
 Apparently, the transient Pc5 pulsations occurring between ∼20:10 UT and 20:35 UT are widespread as they appeared on the ground from low to high latitudes. Figure 3f shows the total magnitude of the magnetic field acquired by GOES 08 (LT ∼ UT – 5) and GOES 10 (LT ∼ UT – 9). The field changes are dominated by the parallel component as magnetospheric compression is strongest at the equator. Note that a large negative spike at ∼20:26 UT has been removed from the GOES 10 data using a median filter. Wave analysis of these data indicates that most of the wave power is distributed below 1 mHz, in agreement with the auroral power spectral data shown in Figure 2, and no apparent frequency peak above 1 mHz (not shown).
Figure 3h shows the solar wind dynamic pressure from ACE SWEPAM [McComas et al., 1998] (a time shift of 40 min is assumed to match KAK magnetometer recordings). At the trailing edge of the pressure enhancement region, there were four discernible small pressure pulses. These small pressure pulses show a good match in timing with the impulse Pc5 pulsations seen on the ground as well as with the auroral pulsations (see Figure 3i).
 In the near-Earth plasma sheet, the Geotail Magnetic Field Experiment (MFE) [Kokubun et al., 1994] observed four large-amplitude pulses in the transverse component of the magnetic field at (–9.6, –5.6, 0) RE GSM between ∼20:10 and 20:40 UT (Figure 3g). The dominant frequency for the pulses is ∼1.5 mHz, which is slightly smaller than the dominant auroral pulsations at ∼1.8 mHz at Geotail's ionospheric foot point (∼65.2° MLAT and 0236 MLT based on the T96 model [Tsyganenko, 1995]). No corresponding variations in the electric field measurements (not shown) was observed. Because Geotail was in the plasma sheet [Liou et al., 2007], the large magnetic field pulses are consistent with the flapping of the magnetic tail caused probably by solar wind pressure variations.
Figure 4a shows a study of the phase relationship of low-frequency (1–3 mHz) auroral pulsations. A day-to-night propagation is clearly shown. Using a time-lagged cross-correlation technique, the phase velocity for the dominated lower-frequency pulsations is estimated to be 1.6°/s (∼40 km/s at 70° MLAT) anti-sunward at the dawn flank (04–08 MLT), corresponding to ∼400 km/s at the magnetopause equator (assuming 10 RE from the Earth's center). This further justifies the scenario that the four larger pressure pulses were directly driven by the solar wind. On the contrary, the higher frequency (6–8 mHz) component of the auroral pulsations that occurred in the first half of the pressure enhancement as seen in Figure 1 does not show noticeable MLT dependence in the phase (see Figure 4b).
3. Discussion and Conclusions
 The encounter of a long-duration (∼100 min) pressure pulse on September 26, 1999, produced one of the most striking auroral phenomena observed by the Polar UVI imager – auroral pulsations, with discrete frequencies in the Pc5 range centered around ∼1.8, 3.5, 4.8, 6.3, and 7.7 mHz and a wide local-time extent that leaves a void of pulsations in the midnight sector (∼21–03 MLT). Coincidentally, two of the lowest frequency bands are consistent with those observed repeatedly by the ground magnetometers [Samson et al., 1991]. Large-amplitude, in-phase geomagnetic pulsations were also observed at two sites underneath the oval. Because the oval maps out to a large part of the magnetosphere, this result thus provides the first observation of a “global” mode of ULF oscillations excited by solar wind pressure enhancements. Using the T96 model [Tsyganenko, 1995], the ionospheric foot points for GOES-08 and GOES-10 at 20:00 UT are (67.52° MLAT, 15.32 MLT) and (68.07° MLAT, 11.18 MLT), respectively. These foot points were located at sub-auroral region. Therefore, we conclude that such a global mode is confined within the magnetosphere but outside the geosynchronous orbits.
 The global appearance of relatively constant frequency Pc5 pulsations over a large local time rules out the possibility of the FLR mechanism for this event, although the solar wind velocity was moderately high (∼480 km/s) during this time period. However, these two features are consistent with the global compressional cavity mode, and the discrete Pc5 pulsations are indicative of the cavity mode harmonics [Kivelson and Southwood, 1985, 1986]. The appearance of the midnight gap is also a good indication of the global cavity mode as there is no outer reflection boundary there. Because cavity mode frequencies are determined by the geometry of the cavity, systematic local-time variations in the wave frequencies are expected. A closer look indicates that some frequency bands at ∼09 hr and ∼13 hr local time show small changes. It is not obvious if these small nonsystematic changes would be consistent with the geometry of the magnetospheric cavity. A detailed analysis of the auroral data in a two-dimensional space combined with a realistic magnetosphere model is planned in the future to help justify this thought.
 An apparent driver of the observed discrete Pc5 band auroral pulsations is the enhancement of the solar wind dynamic pressure. Recent research by Stephenson and Walker  and Kepko et al.  has demonstrated that some preferred discrete ULF waves reported previously [Samson et al., 1991] are observed in the solar wind dynamic pressure. During this transient magnetospheric compression event, the auroral intensity increased in response to the dynamic pressure increase. Within the intense aurora, especially toward the end of the compression, four cycles of auroral pulses (f ∼1.9–2.7 mHz) were found to be nearly in phase with the “small” pressure fluctuations (∼3 nPa peak-to-peak). During such (slow) pressure changes, the magnetosphere is expected to respond by changing its magnetic fields and currents within to achieve a balanced state. These lower frequency transient waves appeared to move antisunward with velocities comparable to the solar wind speed, further justifying the direct solar wind driving mechanism. However, the higher frequency (>6–8 mHz) Pc5 waves were not propagated and did not appear in the solar wind; therefore, we conclude that they are likely to be associated with the magnetospheric cavity mode, though they are still excited by the solar wind dynamic pressure enhancements.
 So far, we have assumed in this study that periodic auroral intensity variations are associated with ULF hydromagnetic waves that also produce geomagnetic pulsations on the ground. Precipitation pulsations are often explained by modulations of electron pitch-angle scattering with a compressional hydromagnetic mode [Coroniti and Kennel, 1970]. For this particular event, because compressions make unstable particle distributions, waves are generated and particles are scattered into the loss cone. It is also may be that field strengths increase more at the equator and the loss cone widens during compressions. Note that either process must be strong enough to overcome the increase in the pitch angles owing to the conservation of the first adiabatic invariant.
 Finally, the availability of global auroral images provides a promising technique to address the global issue of Pc5 ULF waves, because the aurora is geomagnetically mapped to the greater part of the magnetosphere. However, it is not clear as to what extent the global auroral image can contribute in this field. For example, do all ULF waves produce auroral pulsations and can we distinguish different ULF wave modes from the global auroral image? We will explore the potential of the new technique in the near future.
 G. Parks is the principal investigator for Polar UVI. The ACE SWEPAN, Geotail MFE, and GOES magnetometer data were provided by the Space Physics Data Facility (SPDF) and the National Space Science Data Center (NSSDC). This work was supported in part by NSF GEM grant ATM-0703414 and in part by NASA grant NNG05GB72G to the Johns Hopkins University Applied Physics Laboratory.