It is well known that sharp increases/decreases in the solar wind dynamic pressure can result in sudden compression/decompression of the magnetosphere and subsequent magnetic positive/negative impulses (SI+/SI−) detected on the ground magnetometers. While the large-scale enhancement of aurora during an SI+ has been well established, the response of aurora to an SI− is still little known. This prompts an interesting question whether the response of the global aurora to an SI− mirrors the response to an SI+. This letter reports results from a study of auroral images, acquired from the ultraviolet imager (UVI) on board the Polar satellite, during 13 SI− events. It is found that, in most cases, the luminosity of the aurora indeed showed a clear decrease almost immediately after the decompression. In some cases, the luminosity decrease exhibits a day-to-night fading effect and is consistent with the tailward propagation of the magnetosphere decompression front. Auroral particle observations from DMSP indicate that reduction of CPS electron precipitation is the major cause of the large-scale auroral dimming. We propose that an induction electric field triggered by the sudden expansion of the magnetosphere at the expansion front along with adiabatic decompression and magnetic reconfiguration are responsible for the observed effect.
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.
 The solar wind dynamic pressure plays the major role in controlling the overall size and shape of the magnetosphere. When a fast interplanetary shock or a sudden increase in the solar wind dynamic pressure makes a contact with the Earth's magnetosphere, the magnetosphere is compressed. In response to the compression, the Chapman-Ferraro current becomes enhanced and produces a sudden commencement (SC) or positive sudden impulse (SI+) on the ground magnetometers. On the other hand, when a reverse shock or a sudden decrease in the solar wind dynamic pressure (“negative pressure pulses”) arrives at the Earth, the magnetosphere becomes decompressed and a negative sudden impulse (SI−) is detected by the ground magnetometers [Araki, 1994].
 Interplanetary shocks have been attributed to optical auroral enhancements seen on the ground [Vorobyev, 1974] and in space [Craven et al., 1986; Spann et al., 1998]. Sudden compression of the magnetosphere by shock impact results in sudden brightening of aurora in the dayside oval that propagates antisunward [Zhou and Tsurutani, 1999] and in the dayside subauroral regions moving equatorward [Liou et al., 2002; Zhang et al., 2002]. It is generally believed that enhanced pitch angle scattering of pre-existing magnetosphere particles into the loss cone is responsible for the enhanced aurora luminosity. On the other hand, studies of aurora associated with SI−s are rare. An important question concerning the solar wind pressure effect is whether or not the response of the global aurora to SI−s is just the opposite to that of SI+s. There appears to be only one report touching on the subject [Sato et al., 2001]. Based on ground-based all-sky camera images, they found enhancements of aurora in the afternoon sector of the southern oval after an SI−. The result of Sato et al.  seems to suggest that any type of disturbance creates more pitch angle scattering and more aurora. Because of the limited field of view imposed on their observations, the response of the aurora to SI−s cannot be addressed on a global scale. It is likely that depending on the location and the physical mechanism involved the effect of an SI− could be to increase or decrease aurora. The purpose of this study is to examine this global response.
 This study starts with a list of 28 negative sudden impulses (SI−s) identified by Takeuchi et al.  from the mid-latitude Sym-H index [Iyemori and Rao, 1996]. These SI− events were associated with a sharp decrease in the solar wind dynamic pressure originating from a variety of solar wind plasma sources such as corotating interaction regions, magnetic clouds, heliospheric plasma sheet, and plasma holes. After examining auroral image data from the ultraviolet imager (UVI) [Torr et al., 1995] on board the Polar satellite, we found 13 concurrent events. These events are listed in Table 1.
 Polar UVI images are then used to determine changes of auroral luminosity associated with the 13 SI− events. Specifically, we use UVI images of Lyman-Birge-Hopfield long (LBHl) band (160–180 nm) to measure the luminosity changes because auroral emissions in this band are approximately proportional to the energy flux of precipitating electrons [Strickland et al., 1983]. It is found that in most events the auroral luminosity showed clear decreases after the SI− onset. The last event, 3 December 1999, in the list will be used to demonstrate the effect.
2.1. Case Event: 3 December 1999
 This event is associated with the heliospheric plasma sheet (HPS). The sharp decrease in the solar wind dynamic pressure was observed by ACE (XGSM ∼ 222 RE) during the crossing of the HPS trailing edge at ∼1008 UT. The solar wind plasma and IMF parameters for the 1000–1300 UT period are plotted in Figures 1a–1c. The sudden pressure decrease induced a large SI− (ΔH ∼ 26 nT) at 1114 UT (see Figure 1e). The z-component of IMF was fluctuating but remained mostly negative, whereas the y-component of IMF was positive with a sharp increase at the discontinuity.
 A sequence of auroral images in the LBHl band from UVI between ∼1100 and 1128 UT with a time separation of ∼2 min are shown in Figure 1f. During the negative impulse (∼1114 UT), the field of view of the Polar UVI was off the center of the oval toward the postmidnight sector, leaving the postnoon oval outside the field of view. Auroras were active in the postmidnight sector before the SI−. One minute after the SI− onset, the luminosity of the auroras showed a significant reduction (the first image at the third row) and the reduction lasted more than 15 min. The reduction of the aurora occurred first on the dayside and then nightside. The eastern end of the aurorally intensified region in the prenoon-to-dawn section shows a clear antisunward retreat. To quantify the auroral activity in response to the SI−, we have integrated auroral power over the 21–09 MLT sector above 60° MLAT and the result is shown in Figure 1d. The auroral power showed a 60% decrease (from ∼100 gigawatts at ∼1114 UT, the SI− onset time, to ∼40 gigawatts at ∼1120 UT, the end of the SI− time), and is well correlated with the Sym-H index.
2.2. General Results
 We have carefully examined the 13 events of negative impulses and the result is summarized in Table 1. During these impulse events, the auroral display may show complex variations; new aurora may appeared in some part, mostly nightside, of the oval but the overall auroral luminosity show noticeable decreases for all events, except the one event on 1 August 1998, within a few minutes of the SI− onset. Nightside auroral enhancements, probably associated with substorms, occurred in a few events within ∼10–20 min of the SI− onset. In such cases, the nightside auroral luminosity may increase but dimming of the aurora at some places, especially in the day oval, was still discernible. Percentage changes in auroral power inferred from UVI images is listed in the last column of Table 1. No correlation between the amplitude of SI−s and auroral power was observed. Note that the auroral power was computed over viewable oval area, which varies from events to events; therefore, only the percentage change is provided.
 In some of the events studied the reduction of auroral luminosity can be seen first on the dayside and then nightside. This fading effect can be demonstrated best in magnetic local-time keogram [Meng and Liou, 2002] shown in Figure 2. In this event, the dawnside aurora started fading from 1000 MLT at ∼1116 UT, a couple of minutes after the SI− onset, to 0100 MLT at ∼1125 UT. Note that the afternoon oval was not imaged during this time. The small time delay may be due to the timing uncertainty of the SI− onset. Note that not all events showed such a day-to-night fading effect. After examining those events with the day-to-night fading feature, we found the typical time scale is ∼10 minutes, which translates to ∼8 km/s in the ionosphere (assuming 70° oval latitude). Therefore the day-to-night auroral fading is consistent with the tailward propagation of magnetosphere decompression front.
 Ultimately the quench of aurora is associated with a reduction of precipitating particle (probably mostly electron) energy fluxes. To understand possible causes of the decrease of precipitating particles, it is useful to study the characteristics of particle precipitation. We have searched DMSP particle data for the 13 events and found a few events (06 January 1999, 11 February 1999, and 16 April 1999) in which two DMSP satellites made a consecutive oval crossing along a similar orbit (within 2 hours of MLT) during the SI− onset. Here we show the 6 January 1999 event in Figure 3. In this fortuitous event, both F12 and F14 satellites flew over the southern oval in the mid-morning sector along ∼08:30 (±00:20) MLT within 20 minutes from each other. The onset of SI− occurred at ∼14:19 UT, which is ∼10 min after the F12 crossing but ∼10 min before the F14 crossing. Therefore, Figures 3a and 3b represent precipitating particle spectra before and after the impact of the negative impulse. A significant difference between the two spectrograms is the dramatic reductions in both the energy flux and average energy of the structureless precipitating electrons in the low-latitude part of the oval, perhaps associated with the dayside extension of the plasma sheet. For example, between 76° and 77° MLAT on the first pass (BPS), electron precipitation averaged about 0.5 ergs/cm2-s, whereas over the second pass, the oval in this same region averaged about 0.2 ergs/cm2-s. Indeed, it is the electron in the keV energy range and above which are reduced. Unfortunately auroral images were not available during this time period. The similar feature was observed in the other events even though the two DMSP satellites (F11 and F13) were separated by up to ∼2 hours of MLT.
 The thirteen events that we have studied provide enough evidence of “rapid” drops in the overall auroral luminosity after the arrival of negative interplanetary pressure pulses at the Earth. The effect is opposite to the positive pressure pulses, which enhance the auroral luminosity [e.g., Craven et al., 1986; Spann et al., 1998; Zhou and Tsurutani, 1999]. However, a day-to-night propagation of the effect is seen in both cases, indicating a tailward moving magnetospheric source associated with the tailward motion of the newly compressed/decompressed regions. The time scale for the reduction of the aurora is ∼10 minutes, which is comparable to the duration of the SI−. Therefore, the oval regions that show auroral luminosity variations must map to the interaction regions between the negative impulses and the magnetosphere.
 Recently, Sato et al.  reported ground observations of discrete aurora in the afternoon sector during a solar wind negative pressure impulse event. They concluded that the discrete aurora was associated with field-aligned current enhancements triggered by the pressure impulse. Our observations do not necessarily disagree with their results because such a small scale aurora may not be detectable by UVI because of its relatively low spatial resolution and sensitivity. On close examination of the UVI images, it is found that enhancements of small-scale dayside aurora did occur in a few events studied (e.g., 4 May 1998, 26 June 1998, and 11 February 1999), while the overall auroral luminosity decreased. Furthermore, unlike the discrete auroral feature observed by Sato et al. , particle spectrograms from a few DMSP crossings indicate that the gross reduction of the auroral luminosity observed by UVI was associated with the reduction of the diffuse electron precipitation originating from the plasma sheet. Therefore, different mechanisms must have acted on different regions of the magnetosphere and produce different auroral displays in the ionosphere.
 One possible explanation for the rapid quench of aurora by negative impulses is the adiabatic expansion of the magnetosphere. During the impact of a negative pressure impulse on the magnetosphere, the outer magnetosphere expands outward. Conservation of the first adiabatic would decrease the perpendicular-to-parallel ratio of particle energy and inhibit the growth of loss-cone instability [Johnstone et al., 1993]. This in turn reduces the amount of particles that may be scattered into the loss cone by wave-particle interactions. This is opposite to the adiabatic compression mechanism proposed by Zhou and Tsurutani  to explain shock-triggered aurora. However, Sato et al.  argued that such a theory would predict no auroral enhancements when applying to negative pressure impulses and is in contradiction with their findings. Such a contradiction can be resolved if one can differentiate discrete aurora from the background diffuse aurora. Based on DMSP data presented in Zhou et al.  and the present study, variations of the overall auroral luminosity associated with changes in solar wind dynamic pressure are controlled mainly by the appearance/disappearance of non-structured electron precipitation. On the other hand, the smaller scale auroral intensification reported by Sato et al.  during negative impulses was associated with inverted-V electrons, and therefore, must be associated with different mechanisms.
 Another possible explanation for the reduction of aurora associated with negative impulses is due to the reduction of the mirror ratio. It is expected that the effect of magnetospheric decompression caused by the impact of a negative impulse is strongest at the equatorial plane. A larger decrease in the magnetic field at the equatorial plane than in the mirror points would decrease the mirror ratio hence reduce the amount of pre-existing particles precipitation. However, such an effect may be compensated by the reduction of pitch angle of the plasma due to the first adiabatic invariant.
 While the adiabatic expansion mechanism and reduction of the mirror ratio may explain the reduction of particle precipitation, they cannot explain the decrease in the average energy of precipitating electrons. A possible cause is the combination of an induction electric field and the second (longitudinal) invariant or “Fermi acceleration.” Sudden expansion of the magnetosphere can produce an induction electric field due to the temporal change in the magnetic field. The induction electric field in the magnetosphere equatorial plane associated with the magnetospheric expansion is duskward. An E × B drift could move particles sunward to larger L-shells. Because the bounce time for 1-keV electrons is only a few seconds in the dayside auroral zone, which is much smaller than the magnetosphere response time (several minutes), the second adiabatic invariant must be conserved. Therefore, the parallel energy of precipitating electrons will be reduced because of longer magnetic field lines. Such an explanation may also explain the gaining of precipitating electron energy during magnetospheric compression reported by Zhou et al. . To test these ideas, one needs to first quantify the compression/decompression effect on the auroral particle energy flux and average energy with much expanded events. Such work is out of the scope of the present paper and will be performed in the future.
 We have studied the response of the aurora to negative pressure pulses (SI−s), an interesting phenomenon that explores a complementary aspect of the previously established effects of (positive) pressure pulses in the solar wind. Based on Polar UVI images acquired during 13 SI− events, it is found that a rapid reduction of the overall auroral luminosity is quite common. The reduction of aurora is found to be associated with a reduction of the diffuse type of electron precipitation primarily in the keV energy range and above. The reduction is more significant in the dayside than in the nightside part of the oval and sometimes reveals a day-to-night fading effect, with a time scale of ∼10 min. We proposed that adiabatic decompression and magnetosphere reconfiguration are likely the cause of the reduction of auroral precipitation, whereas Fermi acceleration can be the cause of the decrease in the energy of precipitating electrons.
 We acknowledge M. Torr, who built the Polar UVI instrument, and G. Parks, the current principal investigator. The Wind plasma and magnetic field data were courtesy of K. W. Ogilvie (PI of SWE) and R. P. Lepping (PI of MFI), respectively. The ASY/SYM indices are provided by the World Data Center for Geomagnetism, Kyoto, Japan. NASA grant NNG-05GB72G issued through the Polar Mission Program supported this work.