Variable morphology of Saturn's southern ultraviolet aurora

Authors


Abstract

[1] The Space Telescope Imaging Spectrograph camera on board Hubble Space Telescope obtained 68 FUV images of Saturn's southern auroral emission between 8 and 30 January 2004, during Cassini's approach to Saturn's magnetosphere. The HST observations took place in four different solar wind regimes with a low-field rarefaction region from 8 to 16 January, a minor compression event on 17 January, a rarefaction region with intermediate field strengths from 19 to 25 January, and a major compression region from 26 to 30 January. The images have been projected onto polar maps in order to characterize and compare the general morphology of the auroral emission. The first 20 images were obtained during a period covering about 70% of one full rotation of Saturn. They show that the bright ring of auroral emission actually consists of several arcs of different width and brightness and forming along different parallels. Overall, the auroral region is shown to rotate at ∼65% of the full planetary rotation, although the angular velocity of some isolated auroral structures constantly decrease with time, down to 20%. The strongest auroral precipitations are observed in the morning sector. The polar projections of the 48 remaining images confirm dramatic changes in morphology, characterized by different average zones. During the campaign, short intervals during which the auroral region is significantly contracted and clearly forms a spiral were followed by intervals of reinflation of the auroral region. It is suggested that the two major auroral contraction events corresponded to the arrival of the solar wind shocks observed by Cassini on 17 January and 25 January. The present analysis indicates that Saturn's auroral morphology responds to the solar wind conditions at Saturn.

1. Introduction

[2] Until recently, most of our knowledge concerning the morphology of Saturn's far ultraviolet (FUV) aurora was based on spectroscopic data collected with the Pioneer 11 spacecraft [Judge et al., 1980], the Voyager UV spectrometer during their flybys of Saturn in 1980 [Broadfoot et al., 1981; Sandel and Broadfoot, 1981; Sandel et al., 1982; Shemansky and Ajello, 1983], and on data from the IUE spacecraft [Clarke et al., 1981; McGrath and Clarke, 1992]. These observations showed that the Saturnian aurora appears as a narrow circumpolar region with no apparent emission present in the polar cap and probably originates from the distant magnetosphere. Radio wave observations during the Voyager encounters [Desch, 1982; Kaiser et al., 1984] indicated that variations of the Saturn Kilometric Radio (SKR) emission in the auroral zones are strongly correlated with solar wind changes. Later in the 1990s, the Faint Object Camera on board the Hubble Space Telescope (HST) was used to obtain the first image of the north Saturnian aurora, with all images coadded to obtain a reasonable S/N ratio [Gérard et al., 1995]. A set of WFPC2 FUV images [Trauger et al., 1998] with a higher limiting sensitivity showed a northern auroral arc that appeared fixed in local time near the dawn limb. The brightness of the emission was quite variable (from a few kR to 90 kR), but the morning sector was consistently enhanced in comparison with the afternoon sector. Modeling of the auroral curtain indicated that the latitude of the auroral oval was 78° ± 2°, consistent with the earlier Voyager results and supporting a mapping into the distant magnetosphere. An image of the southern auroral region obtained in 2000 with the Space Telescope Imaging Spectrograph (STIS) was described by Cowley et al. [2004a]. They found that the emission is located at ∼72° at dawn, ∼75° in the 0700–0900 LT brightest region where the incident energy flux reached between 4 and 8 mW m−2, and ∼78° in the weaker afternoon aurora. In a recent study, Gérard et al. [2004] analyzed the morphology of Saturn's FUV aurora based on HST-STIS observations made between 1997 and 2001. The observations were obtained during a period of good viewing of the planet's south pole, although emission was occasionally observed around the north pole. The variability is shown to be important both in morphology and brightness of the aurora, which is often found to form a spiral shape around the pole. The total electron precipitated power appears to vary by a factor of 5, probably as a result of changes in the solar wind characteristics. Infrared emissions from the ionospheric H3+ ion have also been detected using ground-based telescopes [Geballe et al., 1993] and have been shown to be mainly auroral in nature [Stallard et al., 1999]. Most recently, Stallard et al. [2004] have used the Doppler shift of these emissions to show that the polar ionosphere significantly subcorotates relative to the planet, with angular velocities of ∼20–40% of the planetary angular velocity. All these observations were obtained in the absence of upstream solar wind data. As a consequence, it has not been possible to relate these to concurrent interplanetary conditions. Recently, however, Prangé et al. [2004] have discussed one HST image showing a polar auroral disturbance that they infer was triggered by the passage of an interplanetary shock associated with a coronal mass ejection, tracked from the Sun via Earth and Jupiter.

[3] In January 2004, the Cassini spacecraft was approaching Saturn prior to its insertion into orbit on 1 July 2004. During the interval 8–30 January 2004, a sequence of HST-STIS images of Saturn's southern aurora was obtained in conjunction with upstream Cassini measurements of the interplanetary medium and SKR emissions. Initial results have demonstrated that both UV aurora and SKR emissions respond strongly to the shock compressions that are associated with interplanetary corotating interaction regions (CIRs) [Clarke et al., 2005; Crary et al., 2005; Kurth et al., 2005]. The initial account of these HST-STIS images [Clarke et al., 2005] shows that Saturn's aurora exhibit morphological properties clearly different from those at both the Earth and Jupiter. In particular in response to a large increase in solar wind dynamic pressure, Saturn's aurora formed a pronounced brightening, a movement of the brightest auroral emissions to higher latitudes, and an initial filling in of the dawnside polar regions. These observations demonstrate that rather than being intermediate between the Earth and Jupiter, the different conditions at Saturn lead to auroral emissions whose behavior is fundamentally different from those at other planets. Comparison with measurements of the approaching solar wind from Cassini indicates that the auroral brightness and power correlate best with the solar wind dynamic pressure, rather than any specific orientation of the IMF. In a recent theoretical study, interpreting a limited sample of the images obtained during the January 2004 campaign, Cowley et al. [2004b] suggested that the morphology of Saturn's aurora is to a large extent controlled by the balance between the magnetic field reconnection rate at the dayside magnetopause and the reconnection rate in the nightside tail. One of the important consequences of the Cowley et al. [2004b] model is that an auroral spiral structure may result from the precipitation of hot plasma created during intervals of steady but unbalanced tail and dayside reconnection. Jackman et al. [2004] argued that magnetospheric compressions could trigger intervals of rapid reconnection in the tail, as sometimes observed at Earth. Cowley et al. [2004b] proposed that such periods of reconnection commonly occur at Saturn in response to CIR or interplanetary shock induced compression events. The immediate consequence of these theoretical views is that the auroral morphology should strongly respond to variations in the solar wind.

[4] In the present study we derive the general characteristics of Saturn's southern aurora. More specifically, we discuss the morphology, the precipitation zones, and their local time or true longitude variations. These auroral characteristics are then discussed in the context of the general solar wind conditions that prevailed during the HST campaign and of the recent models of solar wind-magnetosphere-ionosphere coupling.

2. Observations and Reduction

[5] During January 2004, the STIS camera on board the Hubble Space Telescope obtained 68 FUV images of the auroral emission surrounding Saturn's south pole. These observations were made during Cassini's approach to Saturn's magnetosphere, prior to orbit insertion on 1 July 2004. They span a period of 4 weeks extending from 8 January to 30 January 2004. The images were organized in 17 HST orbits, forming 13 visits. Each orbit lasts approximately 96 min during which Saturn is visible for about 50 min. Each orbit is made of four consecutive images. The first visit, hereafter the “full rotation visit,” consists of five consecutive orbits (V1,1 to V1,5). The other visits (V2 to V13) all consist of one orbit taken approximately 2 days apart. Table 1 provides the observation time at the beginning of the second image of each orbit, as well as the longitude of the central meridian (CML). Every longitude given in this work is based on the “Set III” prime meridian angle (S3), referred to Saturn's magnetic field rotating at a rate of 360° in 10 hours 39 min 22.4 s (IAU value). During this 1 month campaign, the sub-Earth planetocentric latitude changed from −25.65° to −25.98°, and the subsolar longitude remained ∼1° from the sub-Earth longitude.

Table 1. Timing of the Observations
VisitDateTime, UTaCML, dega
  • a

    UT time (hhmm) and CML at the beginning of the second image of the orbit. CMLs are given in S3 longitudes.

V1,18 January 2004044249.2
V1,28 January 20040617102.2
V1,38 January 20040753156.2
V1,48 January 20040929210.3
V1,58 January 20041105264.3
V210 January 20040441230.1
V312 January 20040929213.7
V414 January 20040441233.5
V516 January 20040129306.9
V618 January 20040440236.4
V720 January 20040128310.1
V821 January 20042040329.5
V923 January 20040439330.2
V1024 January 20042351349.5
V1126 January 200419038.7
V1228 January 20040127316.3
V1330 January 2004190111.3

[6] The images were obtained with the MAMA array (Multi-Anode Microchannel Array), which consists of 1024 × 1024 pixels, providing a 24.7″ × 24.7″ field of view with a ∼0.08″ full width at half maximum point spread function (PSF). The transverse distance subtended by one 0.024″ pixel at Saturn is ∼150 km. Half of the images were taken during “dark time,” that is the 25-min period during which HST is in the shadow of the Earth and therefore when minimal contamination is expected from the geocoronal Lyman-α emission. For these images, the “clear” (no filter) mode was used for which the bandpass of the solar-blind detector ranges from 115 to 170 nm and is sensitive to the H2 Lyman and Werner bands as well as the strong H Lyman-α line. For the other images, taken immediately before and after the “dark time,” the F25-SrF2 filter was added in order to reject the emission shortward of 125 nm, including the (contaminated) H Lyman-α line. After reduction following the method detailed by Grodent et al. [2003], the STIS/FUV–MAMA counts were converted to flux using a factor of 1.5 × 10−3 counts per pixel per second for 1 kR of H2 plus Lyman-α for the “clear” mode and 0.5 × 10−3 counts per pixel per second for the filtered images. These conversion factors were calculated based on a synthetic UV spectrum of H2 [Gérard et al., 2002], including the Lyman-α line (contributing ∼15% of the total H2 spectrum). The “clear” images were accumulated for 270 s and the filtered images were obtained (in time-tag mode) with exposure times varying from 640 s to 740 s. It should be stated that the reduced images are consistent with those derived by Clarke et al. [2005], even though different software packages were used.

2.1. Image Projection and Accuracy

[7] Even though the target tracking stability of HST is about one hundreth of an arcsec over the exposure time, the actual pointing precision is limited by the accuracy of the onboard guide star catalog which is on the order of one arcsec. Consequently, the center position deduced from the observation log is only known within ∼160 pixels. This imprecision is far too large to allow a reliable mapping of the auroral emission in Saturnian coordinates. As a result, a limb fitting procedure has been developed to aid in the determination of the center position of the planet. The 25 × 25 arcsec2 field of view of STIS captures the full Saturnian disk and a substantial portion of the rings. The limb fitting procedure uses ephemeris information calculated with the SPICE NAIF package provided by JPL to determine the locus of a theoretical planetary limb in the detector's field of view. These calculations assume that Saturn is a tilted oblate spheroid with equatorial radius of 60,268 km and polar radius of 54,364 km. The contour of the planet in the field of view is thus defined by the intersection of this planetary ellipsoid with the viewing cone which, to a very good approximation, reduces to a cylinder parallel to the observer's line of sight. Once this elliptical contour is found, it is fitted to the illuminated portion of the planetary disk by means of a least squares method. This fit is further refined by matching the sharp boundary of the visible portion of the A, B, and C rings as well as the sharp Cassini division. This method provides a center position with an accuracy comparable to the size of a resolution element, that is less than ∼3 pixels. Subsequently, each pixel of the initial image is assigned a planetocentric latitude and a S3 longitude. This makes it possible to project the image onto any kind of projection, such as a local time (LT) map on which features fixed relative to the Sun always appear at the same location. Figures 1 and 2 illustrates the limitations of the method for mapping the auroral emission onto a polar orthographic projection. In this case, the image was replaced by a grid of 10° spaced parallels (planetocentric) and meridians viewed under the same conditions as the actual images. This grid image was then projected on a polar view where the south pole of Saturn is seen by a virtual observer above the north pole, looking through the planet. In these viewing conditions, the direction of the Sun (1200 LT) is conveniently oriented toward the bottom, dawn (0600 LT) to the left, and dusk (1800 LT) to the right. It should be noted that near opposition the noon direction is nearly coincident with the Earth direction. The most striking feature is the stretching of the pixels in the midnight sector. This distortion stems from the proximity of these pixels to the limb (with a sub-Earth latitude of −25°) where the grazing line of sight leads to inaccuracy in the latitude of the emission source increasing with the decreasing distance from the limb. At this point it should be noted that these large projected pixels were ascribed a brightness equal to the brightness of the corresponding unprojected pixel. The effect of limb brightening of the auroral emission will be discussed later. Figure 2 shows that the accuracy of the projection is a function of both latitude and longitude. Therefore it shows that the inaccuracy is generally close to 1°, while in the nightside semicircle the uncertainty, mainly in latitude, is as large as 5°.

Figure 1.

Sample of the raw images obtained with the Space Telescope Imaging Spectrograph (STIS) camera in the FUV from 8 January to 30 January 2004. The parts show a zoom of the southern auroral region (north is up) where the south pole is located near the center of the auroral ring. A reversed color table is used for which the darker the pixel, the brighter the emission. The outermost “A” ring appears downward of the planetary limb, and the Cassini division is well apparent on both sides (“horizontal” white stripe). The annotations refer to the orbit numbers (see Table 1 for the detailed timing). Each part consists of the sum of the two “clear” images obtained during each orbit. Visit 1 consists of five consecutive orbits, two of them (V1,3, V1,5) are displayed.

Figure 2.

Projected grid illustrating the accuracy of the polar projection method. The precision is a function of both latitude and longitude. It is generally close to 1° at low latitude, along the meridian facing the observer and increases to 5°, mainly in latitude, near the nightside limb (top). A grid of 10° spaced planetocentric latitudes and longitudes is overlaid.

[8] To quantify how the uncertainty of the center position of the planet affects the mapping of the auroral emission, the same projection as above was applied to an image grid shifted by three pixels in the directions perpendicular and parallel to the projected spin axis, respectively. As expected, the perpendicular shift gives rise to a distortion of the projected grid on the order of 1°, while the parallel shift mainly affects the nightside sector by up to 5°, poleward if the shift is applied from north to south. These numbers are comparable to the inherent uncertainty of the projection method given above.

[9] A further uncertainty in identifying locations on the planet arises from the assumptions made on the vertical extent of the emission. In the filtered images, the substantial contribution from auroral H Lyman-α emission is removed. This emission is characterized by a scale height about twice as large as the Lyman and Werner band emission from the twice heavier H2 molecules. Accordingly, it is expected that the characteristic emission altitude, which may be identified with the emission peak altitude, is about twice larger in the “clear” than in the filtered images. A comparison of scans across the auroral emission perpendicular to the limb shows that the emission peaks near ∼500 km above the ultraviolet limb in the filtered images and near ∼1000 km in the “clear” images. Polar projections of the “clear” and filtered images, taken a few minutes apart, reveal that this difference in altitude is in agreement with the location of the emission region in both bandpasses. In order to estimate the effect of the emission altitude on the polar projection, the image grid described above was set up to an altitude of 1000 km and projected assuming an altitude of 500 km. As for the parallel shift of the center position, the main effect occurs in the nightside region where the pixels are shifted equatorward by almost 5°.

[10] The 68 “clear” and filtered images obtained in January 2004 were projected following the method described before. In the following section we discuss these polar projections and derive the general characteristics of the morphology of Saturn's southern aurora. We then discuss these characteristics in relation to the solar wind conditions that prevailed during the HST campaign.

3. Data Analysis

3.1. Full Rotation Visit

[11] The first five HST orbits of the present campaign were devoted to the observation of one (quasi) full rotation of Saturn. During this period, on 8 January 2004, Saturn's southern aurora was continuously observed for almost 7 hours. During this period, which represents approximately 70% of one full Saturn rotation, 20 HST-STIS images were taken at intervals ranging from a few minutes to several tens of minutes. Anticipating the discussion section, we note that these observations took place in the presence of a solar wind rarefaction region when the interplanetary magnetic field (IMF) strength was continuously very low [Crary et al., 2005; E. J. Bunce et al., Cassini magnetometer observations of the solar wind upstream of Saturn and their relation to Hubble Space Telescope aurora data, submitted to Advances in Space Research, 2005, hereinafter referred to as Bunce et al., submitted manuscript, 2005], though following a major compression during 1–5 January. This full rotation may thus be considered as characteristic of a quiet period, following an interval of major disturbance.

3.1.1. Morphology

[12] Figure 3 shows five polar orthographic projections corresponding to the second image (out of the four) taken during each of the five orbits forming the full rotation visit. These images were obtained in the “clear” mode with an exposure time of 270 s. These polar views illustrate the complexity of the auroral emission. Figure 3 shows that the bright ring of emission consists of several ∼60°-long arcs, of variable widths ranging from 1° to 10° of latitude (or ∼1000 to 10,000 km), spreading along different parallels ranging from −68° near dawn, to ∼−80° near dusk, for the fainter emission arcs, and from −72° near dawn, to −71° near dusk, for the brighter emission arcs. Bright isolated spots also appear near midnight, where they stand out from the nonauroral emission reflected by the outer ring. The latter feature forms a diffuse patch of emission equatorward of −70° in the premidnight sector where it overlaps the auroral emission. The 20 images show a quite similar morphology, and it is actually possible to characterize it with two average distributions. These can be parameterized by local time zones consisting of a set of equatorward and poleward boundaries, as a function of longitude, within which the bulk of the emission is statistically likely to occur. They are considered as functions of local time because they are directly determined from the LT projections, at various central meridian longitudes (CML). The major advantage of these zones is that they do not account for the details of the auroral structures but directly point to the global auroral morphology. The comparison of simplified zones rather than complex images allows one to pinpoint major morphological changes rather than isolated or temporal variations (six zones were derived to characterize the changing morphology of the auroral emission during the January 2004 campaign). A second advantage of the zones is that they permit one to extract light curves along the zone, as a function of longitude giving rise to two-dimensional representations of the auroral brightness distribution.

Figure 3.

Polar orthographic projections of Saturn's southern auroral region. Each part shows the projection of the second image (out of the four) taken during each of the five orbits forming the full rotation visit (V1,1 to V1,5). The time and the central meridian longitude at the beginning of each projection are given in Table 1. The 180° (S3) meridian is marked with a dashed line. The lower right part shows a local time clock precising the orientation of the projections, with noon down, midnight up, dawn left, and dusk right. In this reference frame, fixed relative to the Sun, the planet surface is rotating counterclockwise (the south pole of Saturn is seen from above the north pole, through the planet). A grid of 10° spaced planetocentric latitudes and longitudes is overlaid. The dashed ellipse, marked “K,” highlights a rotating auroral feature which was used to estimate the corotation velocity lag of the auroral emission. The “J” and “L” letters mark two other distinct auroral features. The “L” feature is shown to move poleward as it rotates from the prenoon to dusk sector. This poleward motion is highlighted with a white arrow in part V1,5.

[13] The first zone (top right of Figure 4, hereafter “zone QE”) contains the bulk of the bright emission appearing poleward of −70°. “QE” stands for “quiet-expanded,” as it will be shown in the discussion that this zone is the largest of the campaign and corresponds to quiet solar wind conditions. It forms a quasi-circular strip of emission around latitude −72°. The circle near 0400 LT marks a transition, actually an inflection point, between a relatively narrow and continuous arc, and more diffuse and unstructured emission in the nightside to predawn sector appearing near latitude −75°. The faint poleward dusk arc along ∼−82° (parts 2 and 3 of Figure 3) is conspicuous in three out of the 20 images (actually out of the 68 images) and was not considered in the determination of the average morphology.

Figure 4.

Auroral zones. The local time zones represent sets of equatorward and poleward boundaries within which the bulk of the emission is statistically likely to occur. Zone O (“O” for “outer”), contains the faint emission appearing along the equatorward boundary of the auroral region and is common to the entire data set. The circle appearing in the premidnight sector limits an artifact stemming from solar light reflected by Saturn's outer ring. Zone QE (“QE” for “quiet-expanded”), applies to the full rotation visit (V1, see Table 1). Zone QM (“QM” for “quiet-moderate”) applies to visits V2-V5 and V7-V10. Zones S1 (“Shock 1”), S2 (“Shock 2”), and R (“Relaxation”), characterize the contracted cases corresponding to visits V6, V12, and V13, respectively. There is no average zone for the special event of 26 January (V11). The circle centered near 0400 LT on zones QE to R marks a separation between relatively structured auroral emission near dawn and fainter patchy emission in the midnight sector. A grid of 10° spaced planetocentric latitudes and longitudes is overlaid on each part. Noon is directed toward the bottom and dawn is to the left of the grids.

[14] The second zone (top left of Figure 4, hereafter “zone O,” “O” is for “outer”) contains the faint emission appearing equatorward of zone QE, mainly in the predawn sector, of each polar projection displayed in Figure 3. The distinction between the two zones in the postdawn sector is rather artificial for this visit as they merge together. It will be shown below that zone O applies to all the images of the campaign.

3.1.2. Corotation

[15] The main purpose of the full rotation visit was to investigate any motion of the emission during the planetary rotation. Animated views of the polar projections clearly demonstrate that individual arc structures rotate with the planet at a substantial fraction of the full corotation velocity, in contrast to a structure truly fixed in local time, as in the case of the Earth. These animations indicate that the velocity is globally ∼70% of the full corotation with rather large variations of this fraction, depending on the reference points taken to determine this velocity [Clarke et al., 2005]. The rotational motion of the auroral emission may seem to contradict the local time nature of zone QE; however, it should be remembered that this zone sets the poleward and equatorward boundaries of the polar region where the auroral emission is likely to occur, regardless of the identity or the previous location of the auroral features. Accordingly, zone QE may be seen as a frame, fixed in local time, but inside which emission is allowed to (sub-)corotate with the planet's surface around the pole.

[16] The isolated emission structure highlighted with a dashed ellipse and marked “K” in Figure 3 may be seen as a marker allowing one to follow the counterclockwise rotation of the bulk of the auroral emission as time passes. Figure 5 shows auroral light curves for the five orbits of the full rotation visit. The different curves represent the brightness distribution integrated over zone QE, in arbitrary units, as a function of shifted longitude for each orbit. In this presentation, the partial corotation of the emission was compensated by shifting westward the curves obtained for individual images by the same fraction of corotation. For clarity, only the sum of the four shifted light curves from each orbit was displayed. In this case, the largest overlap of the isolated structure appearing between 135° and 180°, corresponding to the highlighted “K” region in Figure 3, is obtained by imposing a longitudinal shift corresponding to 65% ± 10% of the full corotation. This 65% factor also applies directly to the four individual light curves of each orbit, though with larger inaccuracy owing to the smaller rotation during one orbit. Region “J” is too weak to contribute to Figure 5; however, animated views of the polar projections indicate that its angular velocity is also about 65% of the full corotation. The overlap is not as clear for the brighter structure between 50° and 135°, corresponding to the “L” region in Figure 3. The angular velocity of this structure constantly decreases with time as it moves across the noon to postnoon sector. This slower motion is clearly established for the two last orbits where the corotation factor decreases from 55% to 20%. Interestingly, a rapid apparent poleward motion of this latter bright structure is observed (it is conspicuous in parts V1,4 and V1,5 of Figure 3 and indicated by a white arrow in part V1,5) as it moves from noon to postnoon. The other structures, near dawn and dusk, are not affected by this transverse motion. The latitude increase is significant (∼10°). It starts at the beginning of the third orbit (part V1,3 of Figure 3) and is still observed at the end of the last orbit. This localized poleward motion was not accounted for in the determination of zone QE.

Figure 5.

Light curves used to estimate the angular velocity of the auroral emission during the full rotation visit (V1). The five curves represent the brightness distribution integrated over zone QE, in arbitrary units, as a function of shifted longitude for each orbit. They correspond to the five orbits (four images each) forming visit V1. The “K” and “L” letters mark distinct auroral features that were used as auroral markers (see previous figure). In this presentation, the partial corotation of the emission was compensated by shifting westward the curves obtained for individual images by the same fraction of corotation. For clarity, only the sum of the four light curves from each orbit was displayed. In this case, the largest overlap of the isolated structure marked “K” is obtained by imposing a longitudinal shift corresponding to 65% ± 10% of the full corotation. The overlap is not as clear for the brighter “L” structure. The angular velocity of this structure constantly decreases with time as it moves across the noon to postnoon sector. This slower motion is clearly established for the last two orbits, where the corotation factor decreases from 55% to 20%.

3.1.3. Brightness

[17] Figure 6 displays the 20 emission light curves derived from the “clear” and filtered images of the full rotation visit in local time coordinates instead of shifted longitudes. The average brightnesses estimated over the inner average QE zone are given in units of kR of H2 + H-Lyman-α emission. The brightness varies from 7 kR to 35 kR. The largest emission rate is reached during the second visit, at 110° LT or postdawn (corresponding to the “L” region of part V1,2 in Figure 3). Afterward, the brightness of this moving structure decreases as it rotates toward larger local time angles (toward the night). The dashed black curve represents the expected limb brightening factor (the light curves are not corrected). It is estimated to be the inverse of the cosine of the view angle. This approximation is valid for an emission uniformly filling the planet and is restricted to angles measured from the limb smaller than ∼30° (note that the angle from the limb is different from the colatitude angle). Although the actual auroral emission is neither uniform nor very far from the limb, we consider it as a reasonable approximation. In any case, the limb brightening effect is minimum near the CML (180° LT longitude), where the emission is at the greatest distance from the limb. Accordingly, the brightenings observed for V1,2 and V1,3 in the dawn to noon region may be seen as intrinsic variations which cannot be explained by geometrical factors.

Figure 6.

Emission light curves derived from the 20 “clear” and filtered images of the full rotation visit (V1) in local time coordinates (dawn at 90°, noon at 180°). The average brightnesses estimated over zone QE are given in units of kR of H2 + H-Lyman-α emission. The dashed black curve represents the expected limb brightening factor (the light curves are not corrected). It is estimated to be the inverse of the cosine of the view angle (right vertical axis).

[18] A systematic dependence of auroral intensity on Saturn's longitude was suggested by Voyager observations, with peaks near SLS longitude 135° and brightening occurring when the Sun was near 100° [Sandel et al., 1982]. Figure 7 tests this Voyager view, even though Saturn's rotation period is uncertain and longitudes estimated in the present database have no correspondence to those at the time of the Voyager encounters in 1980–1981. Three LT boxes (1.2 arcsec by 1.7 arcsec, across and along the spin axis, respectively) were defined in the dusk, noon, and dawn sectors, as sketched out in Figure 7. The auroral power emitted over these boxes is presented in Figure 7 as a function of the CML during the full rotation visit. As expected from the above description and from Figure 6, the partial corotation of the bright “L” and “K” sectors appearing in Figure 3 gives rise to different curves. In the dawn box, the emitted power is the largest and peaks around CML = 60°. As the bright emission sector rotates away from the box, the emitted power rapidly decays to the background level. The noon box catches the continuation of this rotation motion and peaks near CML = 200°, at a lower value. Finally, the dusk box confirms the continuous decrease of the brightness as the initially bright auroral feature rotates toward midnight. It is interesting to note that the brightening observed in the noon box occurs when the CML is close to 60°, that is in a configuration such that the box includes the sector around 135°. These results do not indicate the presence of any favored planetary longitude sector. Instead, the brightest emission of the main oval appears in the morning sector, even though these auroral features are carried eastward as a result of partial corotation.

Figure 7.

Auroral power emitted over local time boxes as a function of the central meridian (CML) during the full rotation visit (V1). The three LT boxes are sketched out in the top right part. This part shows a zoom of the southern auroral region in a raw STIS image. The aurora appears as a dark string of arcs. The boxes are 1.2 arcsec by 1.7 arcsec, across and along the spin axis, respectively. They are defined in the dawn, noon, and dusk sectors. In the dawn box (plus symbol curve), the emitted power is the largest and peaks around CML = 60°. As the bright emission sector rotates away from the box, the emitted power rapidly decays to the background level. The noon box (asterisk curve) catches the continuation of this rotation motion and peaks near CML = 200°, at a lower value. Finally, the dusk box (diamond curve) confirms the continuous decrease of the brightness as the initially bright auroral feature rotates toward midnight.

3.2. Other Visits

[19] The 48 remaining HST-STIS images were obtained during 12 visits spanning a time period from 10 January 2004 to 30 January 2004 (see Table 1). During this period, Saturn's southern aurora was imaged for ∼40 min approximately every 2 days. Figure 8 shows the second polar projection extracted from each of the 12 visits. In each case, the displayed image is obtained in the “clear” mode and is directly comparable to the projections displayed in Figure 3. The dashed line represents the 180° meridian in S3 coordinates. During this 1 month campaign, the morphology dramatically changed from the quiet distribution of the full rotation visit described above (Figure 3) to the extreme case of 26 January 2004 (part V11 of Figure 8 and Figure 1), where most of the emission concentrated into a small region of the dawn sector. This latter morphology is rather untypical of the global morphology and was already discussed by Cowley et al. [2004b]. A more detailed study of V11 will be presented elsewhere.

Figure 8.

Polar orthographic projections of Saturn's southern auroral region. Each part shows the projection of the second image (out of the four) taken during visits V2 to V13. The projection have the same orientation as for the full rotation visit, with noon directed toward the bottom and dawn to the left of the grids. Each part corresponds to one visit, the letter following the visit number (“QM,” “S1,” “S2,” “R”) refers to the average zone applying to this visit (see text and Figure 4). Note that for visit V11 no such average zone could be defined. The 180° (S3) meridian is marked with a dashed line. The central meridian longitude at the beginning of each projection are given in Table 1 and the UT time is repeated in the lower left corner of each part. A grid of 10° spaced planetocentric latitudes and longitudes is overlaid on each part.

3.2.1. Overall Morphology

[20] The analysis of the individual projections of the full rotation visit already suggested that a common zone of emission appears at the equatorward boundary of the nightside auroral region. This emission is well apparent in the midnight to dawn sector of each polar projection displayed in Figure 8. In this sector, faint emission appears clearly equatorward of −70° and detaches from the brighter emission poleward of −70°. In the dusk sector, this outer emission appears to blend with the rest of the emission, though in part V13 of Figure 8 the outer emission appears equatorward of the bright dusk emission arc. Its distribution in latitude and longitude is also characterized by the equatorward zone O, which was initially derived for visit V1. This emission is usually tenuous and merely detaches from the background emission by a few kR. It is actually revealed by the limb brightening of the emission. The limb brightening curve shown on Figure 6 indicates that in the nightside sector, between 2100 LT and 0300 LT, the brightening is larger than 350% and is on the order of 200% near the dawn and dusk ansae. Its morphology and mean brightness remain almost constant over the whole data set, even during the extreme event of 26 January 2004 (part V11 of Figure 8 and Figure 1). This latter observation suggests that the outer emission is produced by a different mechanism from the rest of the emission, presumably related to inwardly diffusing energetic particles deep into the corotating magnetosphere.

[21] The bulk of the auroral emission appears poleward of zone O. Figure 8 illustrates the large variability of the size and location of this emission over the 1 month period of the HST campaign. The morphology evolved from a relatively expanded quasi-circular distribution (V2) to a contracted spiral shape, such as V12, also showing intermediate cases, like V7, and the extreme case of V11 where most of the emission concentrated in the morningside. At the same time, Figure 8 reveals more stable emission distributions such as in parts V2 to V4 where the polar projections are rather similar, even though these images span 6 days of observation. Figure 8 thus suggests that the global auroral morphology can remain unchanged for almost 1 week but can also undergo dramatic changes from one day to the other. As was already the case for the full rotation visit, the bright emission does not form a continuous ring but rather a string of arcs, several tens of degrees long, with variable width, ranging from a few to several thousands of km, and variable brightness. The global auroral morphology may be characterized by emission zones, similar to zone QE defined for the full rotation visit. However, in the case of visits V2 to V13 it cannot be asserted that these zones are fixed in local time since the time coverage during one visit is limited to ∼40 min every 2 days.

[22] The average zone shown in the middle left of Figure 4 (hereafter “zone QM”) applies to the four visits from 10 January to 16 January 2004 (parts V2 to V5 of Figure 8) and to the four visits from 20 January to 24 January (parts V7 to V10), that is to ∼50% of the images, covering ∼60% of the observing days. “QM” stands for “quiet-moderate,” as it will be shown later in the discussion that visits V2–V5 and V7–V10 correspond to quiet and moderate solar wind activities, respectively. Zone QM presents two major differences with zone QE. First, it is smaller, with mean latitudes ranging from −72.5° (near dusk) to −77° (near midnight); second, it forms a spiral shape. The spiral is well apparent in all parts of Figure 8, with the notable exception of part V11. In most cases, the spiral starts at high latitude (near −75°) in the midnight to dawn sector and then drifts equatorward with increasing local time. Near dusk it connects with zone O and starts a new, partial, revolution around the pole in the midnight to dawn sector (part V10 of Figure 8 best illustrates this behavior).

[23] As for zone QE, the circle centered near 0400 LT on zone QM of Figure 4 marks a separation between relatively structured auroral emission near dawn and fainter patchy emission in the midnight sector. This circle also points to a difference between images taken before and after 18 January 2004. In the four visits from 10 to 16 January (parts V2 to V5), the nightside portion of the zone is almost never filled, as the narrow and bright dawn arc structure abruptly starts near 0400 LT, apart from 16 January (V5) where it starts near 0700 LT. In the four visits from 20 to 24 January (parts V7 to V10) the postmidnight, poleward portion of the QM zone is filled with auroral emission. A second major difference stems from the fainter and less structured (more diffuse) emission in the dusk sector of V7 to V10 compared with V2 to V5. In addition to the emission associated with zone QM, an isolated patch of emission intermittently appears in the premidnight sector. This emission is conspicuous near 2300 LT in part V9 of Figure 8. It is located near the nightside limb and blends with the weak equatorward emission from zone O. Analysis of the four-image sequence of visit V9 reveals that this spot is actually rotating at roughly 70% of the full corotation velocity, though the positional inaccuracy near the limb makes it difficult to derive a precise velocity.

[24] Zones S1, S2, and R of Figure 4 apply to the three sets of four images taken on 18, 28, and 30 January 2004, respectively (visits V6, V12, V13). On these 3 days, the brightness was significantly larger (a factor of two) than during the rest of the campaign. It will be shown below that V6 and V12 may be associated with the arrival of solar wind shocks, “Shock 1” (S1) and “Shock 2” (S2), respectively, while V13 corresponds to the magnetopause “Relaxation” (R) period that followed S2. The S1, S2, and R zones are significantly smaller in radius than the QE and QM zones. The S2 zone is the smallest of all and extends from −76° to −82° latitude. Like the QE and QM zones, they all exhibit a kink in the emission, marked in Figure 4 with a circle centered near 0400 LT. In the dusk sector, the poleward bright emission arc does not connect with the outer zone O. The 1900–2100 LT sector in part V13 of Figure 8 displays, at different latitudes, bright emission from the R zone and fainter emission from zone O. This difference suggests that zone O may not be connected with the poleward QE to R zones. The apparent merging of the QE and QC zones with zone O near 1800 LT may therefore result from a fortuitous overlap, again suggesting that these emissions originate from different magnetospheric regions and implying different precipitation mechanisms.

3.2.2. Brightness

[25] Characteristic brightness values (not corrected for the limb brightening), as a function of local time, are displayed in Figure 9 for the five parts of Figure 3 and the 12 parts of Figure 8, with the exception of V11. It should be stressed that these parts correspond to the second image out of the four images obtained during one orbit. For the sake of clarity, Figure 9 is divided into four parts. The top part shows the maximum brightness along the QE zone for the full rotation visit (V1,1 to V1,5). The second part shows the maximum brightness along the QM zone for the V2 to V5 visits (before 18 January 2004), and the third part shows the V7 to V10 visits (after 18 January). Finally, the bottom plot shows the V6 (18 January), V12 (28 January), and V13 (30 January) visits, corresponding to the smaller S1, S2, and R zones, respectively. All parts confirm the trend that was already highlighted for the full rotation visit, where the dawn to noon sector is generally brighter than the noon to dusk sector.

Figure 9.

Characteristic brightness values (not corrected for the limb brightening), as a function of local time, for the five parts of Figure 3 and the 12 parts of Figure 8, with the exception of V11. These parts correspond to the second image out of the four images obtained during one orbit. The top part shows the maximum brightness (in kR of H2 + H-Lyman-α) along zone QE for the full rotation visit (V1,1 to V1,5). The second part shows the maximum brightness along zone QM for the V2 to V5 visits (before 18 January 2004], the third part shows the V7 to V10 visits (after 18 January). Finally, the bottom plot shows the V6 (18 January), V12 (28 January), and V13 (30 January) visits, corresponding to the contracted S1, S2, and R zones, respectively. The letters in parentheses relate to the average zone that was used to derive the brightnesses.

[26] In each of the first three parts, the maximum brightness ranges from ∼10 kR to ∼50 kR. However, these parts also show that the location and the importance of the brightenings strongly vary from one visit to the other. More specifically, the second and third parts relate to images taken before (V2–V5) and after 18 January (V7–V10), respectively. Even though these two groups of four visits are characterized by the same zone (QM), that is the same global morphology (this is particularly true for V2, V3, and V4), their brightness distribution is distinctly different, as the bright emission is truncated near noon in the third part, and the V3 light curve (second part) shows a strong enhancement near noon. It appears that the V2 to V5 visits have more in common with the full rotation visit (V1) than with visits V7 to V10. This observation somewhat reflects the different solar wind conditions that were prevailing during these visits, with very low interplanetary magnetic field strength before 18 January and a more perturbed period extending from 18 January to the end of the campaign, as will be shown below. Finally, the brightness scale of the bottom part is more than doubled compared with the other ones. This is because the V6, V12, and V13 light curves are systematically larger than in the other parts. Remarkably, the level of the V12 curve peaks at ∼120 kR. It also corresponds to the smallest (S2) zone, thus strongly suggesting that the auroral emission gets brighter at every longitude when the size of the emission zone (QE to R) gets smaller. Again, it will be shown in the discussion section that these brighter and smaller aurorae are related to strong solar wind perturbations that reached Saturn.

3.2.3. Corotation

[27] To determine the level of corotation of the emission features, the method previously used for the full rotation was applied to the other visits. Longitude shifts were applied to the light curves derived from the QM to R zones in order to maximize the overlap of bright localized auroral structures for each individual visit. As an example, Figure 10 shows four normalized brightness curves obtained by shifting by 70% (±15%) of corotation the light curves extracted from the four images of visit V2, through the QM zone. It appears that while the overlap in the two peaks near 110° (dawn) and 150° (prenoon) is maximum, the overlap near the 200° (postnoon) peak is only approximate. A better overlap is obtained for the latter peak with a corotation fraction of 40% (±15%). Such a difference was already encountered during the full rotation visit, where the rotation velocity of a structure near 225° was found to decrease as it approaches dusk. Similar results were obtained for all other visits (except for visit V11). In most cases it appears that the bulk of the auroral emission rotates at a velocity close to 70% of the corotation, though isolated broader structures near noon are found to rotate slower, down to 20%, and to shift poleward.

Figure 10.

Normalized brightness curves obtained by shifting by 70% (±15%) of corotation the light curves extracted from the four images of visit V2, through the zone QM. The overlap of the curves in the two peaks near 110° and 150° is maximum, while near the 200° peak is the overlap is only approximate. A better overlap is obtained for the latter peak with a corotation fraction of 40% (±15%) (see also caption of Figure 5).

4. Discussion

[28] During the January HST campaign the structure of the interplanetary medium was measured by the Cassini spacecraft while en route to Saturn [Crary et al., 2005; Bunce et al., submitted manuscript, 2005]. The top part of Figure 11 shows the total magnetic field measured by Cassini in units of nT. The second part shows an estimate of the dayside reconnection voltage (in units of kV) across Saturn's magnetopause associated with the production of open flux. In the estimate of Jackman et al. [2004], large dayside reconnection voltages are favored for northward pointing IMF at Saturn, opposite to the case for Earth due to the opposite sense of the planetary field. Finally, the bottom part shows the solar wind dynamic pressure in units of nPa, derived from the solar wind density and velocity measured by the Cassini Plasma Science (CAPS) investigation [Crary et al., 2005]. The highly structured nature of the IMF described by Jackman et al. [2004] and discussed in detail for this interval by Bunce et al. (submitted manuscript, 2005) appears clearly in Figure 11, assuming a radial propagation delay from Cassini to Saturn of ∼17 hours. The 31 days of January constitute more than one solar rotation characterized by two major compression regions, on 1–5 January and 25–31 January, separated by one minor compression region on 16–17 January. Each compression region has elevated field strength, dayside reconnection voltage, and solar wind dynamic pressure, more so for the major than the minor. Two crossings of the heliospheric current sheet (HCS) occurred on days 17 and 26, embedded within the minor and second major compression regions, respectively. These HCS crossings occurred just after forward shocks on days 15 and 25, signalling the commencement of the minor and major compression regions, respectively. Following the major compression during 1–5 January, an extended rarefaction interval took place on 7–15 January. This rarefaction region is characterized by very low field strength and density (when available), while in the rarefaction region following the minor compression, the field strength and density decrease somewhat but remain at intermediate values. The dayside driving voltage modulation echoes that of the CIR structure seen in the magnetic field data, though it shows much substructure owing to north-south IMF fluctuations.

Figure 11.

Solar wind parameters during the Hubble Space Telescope (HST) campaign. The top part shows the total magnetic field ∣B∣ measured by Cassini in units of nT. The middle part shows an estimate of the dayside reconnection voltage (in units of kV) across Saturn's magnetopause associated with the production of open flux [Jackman et al., 2004]. The bottom part shows the solar wind dynamic pressure in units of nPa, derived from the solar wind density and velocity measured by CAPS [Crary et al., 2004]. The HST images timings (days of year 2004) are indicated on by the vertical dashed lines assuming a radial propagation delay from Cassini to Saturn of ∼17 hours.

[29] The major morphological characteristics derived from the polar projections displayed in Figures 3 and 8 are summarized in synoptic Table 2. For each visit/orbit, Table 2 lists (1) the zone corresponding to the projected image; (2) the size of the polar region limited by this zone (categorized as small, medium, or large); (3) the overall brightness (also categorized as small, medium, large, or very large); (4) the presence of emission in the postmidnight sector of the spiral zone (it does not apply to V1,1 to V1,5); (5) the presence of emission in the premidnight sector of the spiral pattern, possibly related to previous pulses of tail reconnection; (6) the presence of emission near noon, poleward of zones QE to S1; (7) the relative importance of the postdawn emission. The last three columns of Table 2 give the absolute strength of the IMF, the solar wind dynamic pressure, and the estimated value of the dayside reconnection voltage. These three parameters were averaged over the shifted HST times ±5 hours, in order to account for the error in propagation time. Important information may be derived by interpreting and cross-comparing the columns of Table 2. The size of the auroral region (column 3), that is the area subtended by zones QE to R, shows an apparent decreasing trend from 8 January to 30 January. For V1 to V5 the magnetosphere was particularly quiet with hardly any significant dayside driving due to the very low IMF strength, typical of the solar wind rarefaction region then prevailing. According to Jackman et al. [2004] and Cowley et al. [2004b], almost no dayside reconnection takes place in such conditions, though some internally driven tail reconnection pulses may occur, particularly if the tail is significantly inflated with open flux from prior activity. This is the case during V1, when the oval was the largest seen during the campaign, presumably due to open flux production during the CIR compression event on 1–5 January. The observations during V1 itself will be discussed in terms of episodic tail reconnection in more detail below. In the following interval, during V2 to V5, the oval fits in a smaller, less circular zone (QM), indicating that tail reconnection did indeed occur over the interval and that the flux closed was not balanced by the flux opened at the magnetopause, due to the very low IMF strengths then prevailing.

Table 2. Synoptic View of the Major Morphological and Solar Wind Characteristics
VisitDate, Day/TimeaZoneSizebOverall Brightnessb0000 LT→0400 LT Emission2200 LT→2400 LT EmissionNoon EmissionPostdawn Emission∣B∣, nTp, picoPaΦ, kV
  • a

    Day of January 2004/UT time (hhmm) at the beginning of the second image of the orbit.

  • b

    Here “0” means negligible, one asterisk is small, two asterisks is medium, three asterisks is large, and four asterisks is very large. NA means not available/applicable. The ∣B∣, p, and Φc values were estimated by averaging the Cassini data (plotted in Figure 11) over ±5 h. “QE” = “quiet-expanded,” “QM” = “quiet-moderate,” “S1” = “Shock 1,” “S2” = “Shock 2,” “R” = “Relaxation.” Note that no average zone was defined for V12.

V1,18/0442QE*****NA**0**0.05NA3
V1,28/0617QE*****NA**0***0.05NA3
V1,38/0753QE****NA******0.05NA3
V1,48/0929QE****NA*****0.05NA3
V1,58/1105QE****NA*****0.05NA3
V210/0441QM***0*****0.04NA5
V312/0929QM****0******0.0638
V414/0441QM***0*****0.0739
V516/0129QM***0*****0.0735
V618/0440S1******00***0.501211
V720/0128QM*******0****0.301611
V821/2040QM**********0.281129
V923/0439QM***********0.25714
V1024/2351QM**************0.2894
V1126/1903 **** ?*0****1.203556
V1228/0127S2********00****1.5549142
V1330/1901R******0*****0.7716173

[30] The apparent decreasing trend of the size of the emission region from V1 to V13 is the result of short intervals during which the auroral region is more contracted and clearly forms a spiral shape, followed by intervals of reinflation of the auroral region. Specifically, contraction is clearly seen in V6, V11, and V12 (though V11 appears to be an extreme case). In these cases it is reasonable to suppose that the auroral contraction responded to the arrival of the solar wind shocks on 17 January (V6, corresponding to zone S1) and 25 January (V11–V12, corresponding to zone S2) observed by Cassini (the V11–V13 cases are thoroughly discussed by Cowley et al. [2004b]). For V6 and V11–V12, it is possible that the resulting magnetospheric compression induced an interval of rapid reconnection in the nightside tail which closed a substantial fraction of the preexisting open flux. This led to a contraction of the aurora, forming a spiral around the pole, starting on the nightside, with intensity decreasing with increasing local time (and decreasing latitude). Both V6 and V12 are followed by visits during which the auroral size increased back to the size before the shock (the size of V7 is similar to V5, and the size of V13 is close to that of V10). In that regard, V7 has much in common with V13, with a less pronounced spiral shape (though still conspicuous), bright postnoon emission, and the presence of subcorotating noon (poleward) emission. V7 and V13 may be considered as periods of relaxation, following the perturbed period initiated by the arrival of a shock. During these relaxation periods, magnetopause reconnection is no longer dominated by tail reconnection, and new open flux gives rise to aurora forming around the expanding open/closed field line boundary, enhanced at dawn compared with dusk. In addition, hot plasma precipitation from closed field lines (if tail reconnection is also present) may produce emission which starts postmidnight and tails off past noon and is thus also brightest in the dawn sector, together with cusp-related aurora near noon produced by magnetopause reconnection. At this point it should be noted that this latter interpretation explicitly suggests that the subcorotating emission (∼20% of full corotation) appearing near noon and poleward of the average zones is to some extent related to the cusp emission suggested by Bunce et al. [2005] and in this interpretation would require some intervals (at least a couple of hours) of northward IMF. The amount of transpolar reconnection derived by Jackman et al. [2004] (Φ in the last column of Table 2) is a more appropriate parameter for the level of dayside reconnection. It shows that for V7 the level of magnetopause reconnection was moderate, while that for V13 it was very strong. However, it should be stated that the level of noon emission is not clearly correlated with Φ so that the association of the noon emission with cusp emission is not straightforward. A thorough study of this emission and its possible relation with the cusp emission is presented by J. C. Gérard et al. (Signature of Saturn's auroral cusp: simultaneous HST FUV observations and upstream solar wind monitoring, submitted to Journal of Geophysical Research, 2005). The overall brightness listed in the fifth column of Table 2 is clearly anticorrelated with the size of the aurora. This result is in agreement with Clarke et al. [2005], who showed that the emitted auroral power is anticorrelated with the mean colatitude of the emission. Cross comparison with column nine indicates that the overall brightness is correlated with the level of emission in the postdawn sector. This second result was already highlighted in Figure 9, where it was shown that the emission occurring in the dawn to noon sector plays a major role in the overall brightness. The theoretical views of Cowley et al. [2004b] suggest that if the brightest auroral emissions are indeed induced by solar wind activity, through plasma heating during magnetic field reconnection, then it is not surprising that the brightness is correlated with the solar wind parameters. As an example, although the emissions seen during V7–V10 have much in common with V2–V5 (they follow the same “QM” zone), there are also substantial differences too, really revealed in Figure 9. This figure shows that the distribution of emission during V7–V10 is strongly biased to the dawn sector, and the bright emission is truncated close to noon. This is quite unlike V2–V5 (and V1 taken together). The main difference from an interplanetary viewpoint is that V7–V10 occur during an intermediate rarefaction region in which the IMF is not so weak as before, such that a situation prevails in which dayside processes should be active at moderate levels, and it seems possible that some modest nightside activity too was excited toward the end (i.e., low-level tail reconnection could be excited), that results in a later near-steady condition.

[31] The longer duration of the full rotation visit, illustrated by the five polar projections in Figure 3, makes it possible to suggest a more detailed history of the solar wind-magnetosphere-ionosphere coupling. This interpretation rests on the suggestion made by Cowley et al. [2004b] and Jackman et al. [2004] that dayside reconnection is usually strongly modulated by fluctuations in the IMF direction, with timescales of tens of minutes to a few hours. As a consequence, the width of the spiral of newly opened field lines will be correspondingly strongly modulated, with a wide region being created when the dayside reconnection rate is high during intervals of northward IMF and a narrow region when the reconnection rate is low during intervals of southward IMF. After their creation, these wide and narrow bands subcorotate around in the polar cap until the open field lines are removed by tail reconnection. Similarly, Cowley et al. [2004b] suggest the occurrence of intervals of impulsive flux closure in the tail, producing patches of newly closed flux which subcorotate around the outer magnetosphere until removed by dayside reconnection. These fluctuations may therefore explain the broken appearance of the auroral emission, forming strings of arcs of various latitudinal widths and spanning about 60° in longitude. The polar projections show three to four arcs along the QE zone (they are best seen in part V1,3 of Figure 3 and three of them are marked “J,” “K,” “L”). Their length varies from 40° to 70° and their width ranges from a few to several degrees of latitude. At 70% of the rigid corotation, these lengths would then correspond to magnetospheric events lasting about 1 to 2 hours, if they are formed in a narrow region. This substantial angular velocity also suggests that the field lines mapping to these auroral arcs are closed and might therefore result from three consecutive but distinct pulses of tail reconnection. In this scenario arc, J would be the oldest structure. It might have been created at least one rotation before the start of the image sequence and rotated toward midnight during all this time. In parts V1,1 to V1,5 it is embedded in a new revolution around the pole. As it travels toward noon, its width decreases as a result of the lower stretching of the polar projection (that is a geometric effect). At the same time its brightness decreases, partly because of a decreasing limb brightening effect and partly because of the declining amount of precipitating hot plasma. Arcs K and L are probably more recent. Arc K may correspond to a short reconnection pulse (on the order of 1 hour) that took place several hours before part V1,1, and arc L may correspond to a longer reconnection pulse (possibly still active in part V1,1). Its relatively large width may stem from an increased reconnection rate. While arcs J and K keep on rotating toward larger local times, at almost constant velocity inside zone QE, the motion of arc L gets perturbed as it starts to cross the noon sector (parts V1,2 to V1,5). Figures 5 and 6 show that the brightness of arc L rapidly increases from V1,1 to V1,2 by ∼20% and then continuously decreases from V1,2 to V1,5. This brightness decrease is correlated with a deceleration of the angular velocity (down to 20% of rigid corotation) and with a substantial poleward motion (up to 10° of latitude). The origins of this latter behavior are not at present understood.

[32] The overall interpretation of the V1 data in terms of pulsed tail reconnection, though largely hypothetical, is not contradicted by Cassini's observations of the upstream solar wind. Figure 11 shows that the 8 January images were taken shortly after a major interplanetary compression region on 1–5 January, during which a large amount of open flux was produced, some 40–50 GWb according to the Jackman et al. [2004] algorithm. This appears to have left the auroral emission region substantially expanded during V1, when the bright emission forms the largest (QE) oval. It is then reasonable to assume that pulses of tail reconnection could have persisted after the compression region passed, associated with subcorotating auroral patches and a related contraction of the auroral oval. A direct consequence of this interpretation is that we should recognize that the auroral morphology characterizing visit V13 is that which is going to evolve, after the compression has fully passed in a day or two, into that appearing in the images of visit V1 if the field strength falls very low or into V7 if it is not. So one may look from V13 back to either V1 or V7 to predict what happens to the morphology afterward.

5. Summary

[33] From 8 to 30 January 2004, the STIS camera on board HST obtained 68 FUV images of Saturn's Southern auroral emission. These observations were conducted during Cassini's approach to Saturn. The images were projected on orthographic polar maps in order to characterize and compare the general morphology of the auroral emission and to pinpoint local time or true longitude variations. The first 20 images of the campaign were obtained consecutively over a period of 7 hours on 8 January, covering about 70% of one full rotation of Saturn's surface. It is shown that in this time frame the morphology remained globally unchanged, allowing us to define an average auroral distribution, fixed in local time, and defining the equatorward and poleward boundaries of the region within which the bulk of the auroral emission is most likely to occur. The images show that the bright ring of emission actually consists of several auroral arcs having different latitudinal widths and brightnesses and forming along different parallels. Animated sequences of these 20 images and individual light curves show that the arc features are rotating inside the average zone. Tracking of auroral spots shows that the overall auroral region is rotating at 65% ± 10% of the full planetary rotation. However, the angular velocity of one isolated bright auroral structure constantly decreases with time as it moves across the noon to postnoon sector, where the corotation factor decreases from 55% to 20%. This deceleration is concomitant with a rapid poleward shift of the feature. The brightness of the auroral features varies from 7 to 35 kR of H2 emission. It is usually decreasing as the auroral structure is rotating from dawn to dusk through noon. Simulation of the viewing geometry of the Voyager spacecraft does not indicate the presence of any favored planetary longitude sector. Instead, the brightest emission of the main oval appears in the morning sector, even though these auroral features are carried eastward as a result of partial corotation.

[34] The 48 remaining images were obtained on 12 days from 10 to 30 January. Four images were obtained during each HST orbit, lasting ∼40 min, and repeated approximately every 2 days. Their polar projections confirm dramatic changes in morphology, ranging from the extended auroral distribution of the full rotation to the extreme case of 26 January when most of the emission came from a small region of the dawn sector. As for the full rotation images, average zones have been defined to locate the bulk of the auroral emission. Some 32 images are shown to fit in one single zone, revealing some apparent morphological stability. This zone forms a spiral rather than a closed ring and is smaller than the zone defined for the full rotation images. The spiral shape is likely a consequence of the imbalance between the processes of magnetic field line reconnection occurring at the dayside magnetosphere and in the nightside tail. Individual light curves show that the brightness distributions along the zone are systematically different before and after 18 January, with the bright emission (ranging from 10 to 50 kR) strongly biased to the dawn sector after 18 January. The other images, with the exception of those taken on 26 January, form a smaller and brighter spiral (up to 120 kR), presumably in response to large solar wind disturbances.

[35] The observations made by HST in January 2004 took place in four strikingly different solar wind regimes clearly identified in the Cassini data. The images taken between 8 and 16 January took place during a solar wind rarefaction region (but interestingly, just after the first major compression region of the interval). Here the total magnetic field strength is typically smaller than 1 nT, the dayside reconnection voltage is smaller than 10 kV, almost negligible, and the solar wind dynamic pressure (when available) is on the order of 0.003 nPa. The images obtained from 18 to 24 January also took place in a rarefaction region following a minor compression event but where the field strength remained at intermediate values of 0.2–0.4 nT. The estimated dayside reconnection voltage was then approximately 35 kV, and the solar wind dynamic pressure was close to 0.01 nPa. Finally, the images obtained between 26 and 30 January took place just following the start of a major compression region when the field strength was increased to 0.5–2.0 nT, the dayside reconnection voltage was typically 150 kV, and the solar wind dynamic pressure was on average increased to 0.03 nPa.

[36] The apparent decreasing trend of the size of the emission region during the campaign is the result of short intervals during which the auroral region is significantly contracted and clearly forming a spiral shape, followed by intervals of reinflation of the auroral region. The timing of the two major auroral contraction events is consistent with the arrival at Saturn of the solar wind shocks observed by Cassini on 17 January and 25 January. The resulting magnetospheric compression may have led to a contraction of the auroral oval, forming a bright spiral around the pole, with intensity decreasing with increasing local time. These episodes of auroral contraction were followed by periods during which the auroral size increased back to the size before the shock. We take that during these relaxation periods, magnetopause reconnection is no longer dominated by tail reconnection, but new open flux gives rise to aurora forming around the expanding open/closed field line boundary, enhanced at dawn compared with dusk. The average brightness appears to be anticorrelated with the size of the auroral region and correlated with the amount of emission in the dawn sector. These observations thus strongly suggest that the auroral morphology responded to the solar wind conditions. In the frame of this suggestion, we propose a scenario in which the rotating auroral structures observed during the full Saturnian rotation result from plasma heating during short (hours) and distinct episodes of tail reconnection. In this scenario the initially expanded ring of auroral emission is a consequence of the major compression event that occurred before the HST campaign on 1–5 January. Consequently, it is suggested that the auroral morphology characterizing the last images of the HST campaign (30 January) evolved into the morphology that prevailed at the beginning of the campaign (8 January), if the field strength fell very low, or into the midcampaign morphology (20 January) if it did not. In conclusion, the present HST campaign revealed a large range of auroral morphologies of Saturn's southern hemisphere, responding to a representative set of solar wind conditions.

Acknowledgments

[37] This work is based on observations with the NASA/ESA Hubble Space Telescope, obtained at the Space Telescope Science Institute (STScI), which is operated by AURA, inc. for NASA under contract NAS5-26555. DG and JCG are supported by the Belgian Fund for Scientific Research (FNRS). Partial funding for this research was provided by the PRODEX program of the European Space Agency (ESA). Work at the University of Leicester is supported by PPARC grant PPA/G/O/2003/00013. EJB is supported by PPARC Postdoctoral Fellowship PPA/P/S/2002/00168. JC acknowledges support from STScI grant GO-10083.01-A to Boston University. We thank F.J. Crary and the Cassini CAPS instrument team for use of the solar wind velocity data employed to generate the voltage values shown in Figure 11. CAPS is supported by a contract with NASA/JPL.

[38] Arthur Richmond thanks the reviewers for their assistance in evaluating this paper.

Ancillary