Discovery of an arc of particles near Enceladus’ orbit:

a possible key to the origin of the E ring

C. Roddier, F. Roddier, J. E. Graves, M. J. Northcott

Institute for Astronomy, University of Hawaii,2680 Woodlawn Drive, Honolulu, HI 96822, U.S.A.

E-mail: croddier@ifa.hawaii.edu

Abstract: High angular resolution (adaptive optics) images taken on August 12, 1995 between

11:26 and 12:23 (UT) show a faint elongated structure apparently moving away from Saturn. The

structure is consistent with light scattered by an arc of particles on a Keplerian orbit close to that of

Enceladus. The orbit is slightly inclined (1.8˚), and the arc is 76˚ ahead of the satellite. It appears to

be a transient phenomenon since the arc brightness decreased by almost a factor two during the

observations, and no such structure was observed at the same orbital position two days before. A

possible explanation is that a large block of ice previously ejected by Enceladus collided with ice

fragments trapped on the satellite orbit near its L4 Lagrange point. The collision likely occurred

about 6 h before observations started and produced a rapidly expanding cloud of small particles.

We estimate the total mass of particles to be at least 105 kg.

*********

We report here on the discovery of an arc-type structure orbiting Saturn at about the distance of

Enceladus. Observations were made almost two days after the crossing of the ring plane by the

Earth in August 1995. The discovery was done after full processing and analysis of the data. It was

announced in (Roddier et al. 1997).

1. DATA ACQUISITION

Observations were made with the University of Hawaii 1024 x 1024 HgCdTe camera (Hodapp

et al. 1995) on August 12, 1995 between 11:26 and 12:23 (UT). The camera was attached to the

University-of-Hawaii 13-actuator adaptive optics system, producing images with a point spread

function (PSF) full-width-at-half-maximum (FWHM) in the 0.1 to 0.2 arcsec range, and a pixel

size of 0.035 arcsec per pixel. A description of the instrument can be found in Roddier et al. (1991).

The whole instrument was mounted at the f/35 focus of the Canada-France-Hawaii Telescope (CFHT)

on Mauna Kea. Data were taken on the east side of Saturn, using Dione as a guide source. They

consist of a total of sixty 30-s exposures. Twenty exposures were taken in the H band (1.65 µm)

from 11:26 to 11:41, 25 exposures were taken the J band (1.22 µm) between 11:46 and 12:08, and

15 were taken in the H band between 12:13 and 12:23. Evidence for the arc was found on all of the

60 exposures. Similar data were later taken on the west side using Tethys as a guide source, be-

tween 12:30 and 15:30 (UT). By comparison, they do not show such a structure. To avoid detector

saturation, Saturn itself was left outside the 36 x 36 arcsec field of view. To help eliminate camera

artifacts and bad pixels, the image was moved over 12 different locations on the camera detector,

and five exposures were taken at each location.

2. GENERAL DATA PROCESSING

The data reduction described here was made in an attempt to detect the E ring, a challenging

goal in the infrared with such a high magnification. In addition to standard flat-fielding and bad

pixel interpolation, a lot of effort has been put in accurate background estimation and subtraction.

We were particularly careful to identify and eliminate optical and camera artifacts which are impor-

tant at very low light level. In spite of having kept Saturn outside of the field of view, light from the

bright planet scattered by optical surfaces covered most of the field of view making background

estimates from the same set of frames difficult. In addition, known artifacts from the camera elec-

tronics were evident as sets of horizontal or nearly vertical line structures. We were forced to use

sky background images taken about an hour later. However, on these images camera artifacts tend

to differ from those on the object images. Background images were selected that best cancel arti-

facts seen on the object images. Residuals were found to disappear when a median was taken over

the 12 different object positions.

Another difficulty was accurate recentering of all the images in the same reference frame. The

positions of Janus and Epimetheus obtained from Mark Showalter’s ephemeris generator

(@ringside.arc.nasa.gov) were used as references. Both distortions in our camera optics and inac-

curacies in the calculated positions produced errors of the order of ± 2 pixels. We found it possible

to locate the tip of the F ring with a better accuracy and often used its location to further reduce the

uncertainty.

Figure 1 shows a median of all the frames recorded on August 12. The position of Saturn is

indicated by the gray circular area in the center. No data were taken inside the black irregularly

shaped area that surrounds it. Although satellites moved from one frame to another, one can still see

the bright spot and diffraction spikes produced by Dione near the left edge and those produced by

Tethys near the right edge. To reproduce the full dynamic range of the data, growing intensities

have been divided by successive factors of 2 at a camera Analog-to-Digital Unit (ADU) of 30, and

then of 4 at ADU = 60, 15 at ADU = 120, and 250 at ADU = 500. One ADU corresponds to a

detection of seven photo-electrons by the camera. A thin black line running from east to west has

been drawn along the rings. Small spots along this line were produced by Janus on the left (east)

side and by Epimetheus on the right (west) side. Because they were near maximum elongation,

they did not disappear in the median image. Although extremely faint, we believe that the E ring

can be seen on the west side beyond the tip of the black line, whereas on the east side it coincides

with a bright diffraction spike produced by Dione. The arc discussed in this paper is the bright

feature almost horizontally elongated seen on the east side just above the ring plane. Starting from

Saturn’s center, two additional thin black radial lines have been drawn on the west side. They are

located symmetrically on each side of the ring direction and form a small angle with it. These lines

underline straight shadows bordered with brighter regions. Computer simulations have shown that

these structures are diffraction patterns produced by the telescope spider arms (opposite spider

arms are not exactly aligned). They are due to the narrow dark shadow cast by the rings on the

planet’s surface. These diffraction patterns are symmetric about the center of Saturn, and the same

structures are also seen on the east side. On the contrary, the bright feature discussed in this paper

is seen only on the east side where it clearly stands out. It does not correspond to any of the known

“ghost” images produced by internal reflections of the instrument, and was seen only on images

taken on August 12. An unknown camera artifact also seems excluded since the bright line appears

sharper when seeing improves. We are therefore brought to believe that this object is real.

3. OBJECT MOTION

Figure 2 shows the time evolution of the object. From left to right and top to bottom, the first

five images (a to e) are median images calculated with mid-exposure intervals of 10 min. The

bottom right image (f) is a median of all the images (also seen in Fig. 1). Upper row (a, b) are H-

band images, middle row (c, d) are J-band images, and bottom left (e) is an H band image. Janus

and the tip of the F ring are seen on the right side. A dark vertical line has been drawn to coincide

with the sharp right edge of the structure in the upper left image. One clearly observes a regular

eastward motion of the object (away from Saturn). One also notes that as it moves eastward, the

object becomes less elongated.

Because of the fuzzy nature of the object, its exact position is difficult to estimate. The sharp

right edge is the easiest to locate. We have estimated its distance to the center of Saturn as a func-

tion of time. Best results were obtained by using medians taken over sets of five frames that were

rapidly recorded one after the other without moving the object on the detector array. This gave us a

set of 12 distances. Figure 3 shows a plot of these distances versus time. The first 4 points are from

the first H band images. The gap corresponds to the first three J-band images which are the most

affected by the background. For this reason, they were eliminated from the plot. The next two

points are from the last J band images and the last three points from the last H band images.

The full line is a least square fit with a circular Keplerian orbit. The orbit radius is found to be

242 600 ± 4 000 km, a value close to and consistent with that of Enceladus (238 040 km). It seems

therefore likely that we see here particles orbiting Saturn at approximately the distance of Enceladus.

Particles must then be scattered along a portion of the orbit forming an arc. As particles move

eastward closer to their maximum elongation, the arc does indeed appear foreshortened. In Section

4 (Fig. 4) we show that an orbital radius of 242 600 km is in close agreement with the maximum

distance to Saturn of the east arc edge as seen in the images, which is fully consistent with the

proposed interpretation. Comparison with the actual position of Enceladus shows that the sharp

(west) edge of the arc is 7 h ahead of Enceladus, that is about 76˚ ahead, which is not very far from

(but not exactly at) Enceladus Lagrange L4 point. In addition, the arc is seen about 7 300 km above

the ring plane; that is, its orbit is inclined (see Section 5). By comparison, the orbit of Enceladus is

almost in the ring plane.

4. FURTHER DATA PROCESSING

Because the arc is very faint, its brightness distribution is difficult to determine without a good

estimate of the background level. This was done by modeling the background with two-dimen-

sional first and second degree polynomials. Polynomials were locally fitted to the background over

small domains about 10 pixels long covering two bands parallel to the arc and on each side of it.

Each band is about 30 pixels wide. The gap with the arc in the middle is about 60 pixels wide.

Successive domains overlap each other by half the domain area, and each background estimate is

the average of the values obtained from the two overlapping domains. The polynomials, which are

orthogonal over the fitted domain, are generated by a single value decomposition (SVD) routine

(Press et al., 1990). This procedure was also used for the main rings and was found to be quite

effective.

Figure 4 shows images of the arc obtained after subtraction of the above background estimate.

From top to bottom, the first three images are a median of the early H frames (a), of the J frames (b),

and of the late H frames (c). The next image (d) is from a median of 44 frames taken later the same

night on the west side. The same background subtraction was applied to it. We display here the part

of this image that is symmetric of the field-of-view of the upper three images with respect to the

center of Saturn. In this frame, east and west and north and south have been reversed for compari-

son with the upper images (a, b, c). One clearly sees the same large diffraction spike which also

appears in the background of the above images.

To remove this artifact, we computed again median images from series of five background-

subtracted frames. From each of the 12 images, and from the above described artifact image (d), we

have subtracted a constant value over a rectangular area corresponding to the artifact. In each case,

the value of the constant was chosen to make the artifact best disappear outside the arc area. These

13 corrected images were used for our photometric analysis as described in Section 6.

5. ORBIT DETERMINATION

The bottom image (e) in Fig. 4 is a median of all the corrected arc images. The section left

unilluminated after the motion of the tip of the arc was not taken into account in the calculation of

the median in order to produce a bright area along all the regions successively occupied by the arc.

This image was used to fit an orbit. Superimposed on it is a visual fit with an ellipse centered on

Saturn. The vertical line shows the maximum eastward elongation calculated from the observed

orbital motion (242 600 km). It coincides well with the observed maximum eastward extent of the

arc. The fit gives an inclination angle of 1.8˚ ± 0.05˚, which is higher than that of Enceladus (0.02˚).

Taking 44.7˚ as the sub-Earth longitude, the longitude of the ascending node of the fitted arc orbit

is 216˚ ± 3˚.

Figure 5 shows the estimated position of the tip of the arc (sinusoidal curve) as a function of

time (in hours) and distance to Saturn (in units of 105 km). The horizontal dashed line indicates the

ring plane crossing time taken here as the origin. An arrow indicates where and when the arc was

detected. The arc should have also been visible either on our data (hatched areas) or on the Hubble

Space Telescope data (dotted lines) that were taken on August 10 (UT) a few hours before the

crossing time, as the arc was near its maximum western elongation (right side). However, analysis

of these data shows no evidence of an arc. We estimate that the arc must have been at least 10 times

fainter to avoid detection. From this observation, we conclude it is a time variable phenomenon.

6. PHOTOMETRIC ANALYSIS

The 13 corrected images described in Section 5 were scanned photometrically through rectan-

gular apertures, 10 pixels wide and 6 pixels high. The height of the aperture (0.21 arcsec) was

empirically determined to be high enough to avoid any loss of signal, while avoiding adding un-

wanted contribution from the background noise. Each profile was shifted and the origin set at the

sharp arc edge. Figure 6a shows a mean of the first four (H band) profiles (full line) together with

that of the last three (H band) profiles (dotted line). The length of the error bars was set to plus or

minus the standard deviation. A profile of the background image is also shown for comparison. On

this figure, the arc brightness is plotted as observed (projected on the sky), as a function of position

coordinate x (in camera pixels) increasing with the apparent distance to Saturn, i.e. Saturn is on the

left and East is on the right side. Figure 6b shows the same profiles after correction for the fore-

shortening effect; i.e., it shows the brightness of the arc as a function of the orbital longitude

increasing in the direction of the orbital motion. The correction was made by considering the arc as

a bright line with a negligible geometrical thickness (this approximation is shown to be valid in the

Appendix). If this is the case, then is simply related to by the following equation:

, or . (1)

Assuming a circular orbit with radius, the relation becomes (see Fig. 8a)

(2)

where is expressed in radians. Individual profiles were corrected according to Eq. (2) and shifted

by 10.6˚ per hour, assuming an orbital period of 33.8 hours consistent with the radius of 242 600

km estimated above. Figure 6b shows the mean of the first four (H band) corrected profiles (full

line) and the mean of the last three (H band) corrected profiles (dotted line). Clearly, the corrected

profiles shown on Fig. 6b are much more similar in width and shape than the uncorrected ones

shown on Fig. 6a. It again confirms our hypothesis of an arc in orbit.

One can see the gain in signal-to-noise ratio produced by an apparent increase of the arc bright-

I x x J( ) ( )d d= ϕ ϕ J I xx

( ) ( )ϕϕ

= d

d

J I x a x( ) ( )ϕ = −2 2

ness due to the foreshortening effect. Without this effect, the arc would have been even more diffi-

cult to detect since its brightness would have been that given in Fig. 6b, but the noise level would

have been that of Fig. 6a. Comparison between the first and last H band images shows evidence of

a decrease of the arc brightness during the observations. No such decrease being observed on the

peak brightness of Janus, we believe this drop in brightness is real. It confirms that the phenomenon

is time variable. We have estimated the peak brightness of each profile by integrating them over a

width of 7˚ near the maximum. The decay of the brightness is shown in Fig. 7.

Figure 6b also shows a mean of the five J-band profiles (dashed line). The brightness in the J-

band profiles was arbitrarily multiplied by a factor 1.3 to match the first H band profiles in the

wings. An intriguing result is that the brightness peak is sharper in the J band than in the H band.

From Fig. 6, one can estimate the integrated arc brightness at the beginning of the observations.

The integrated H magnitude is found to be of the order of magnitude 18. Taking -28.17 as the

integrated H magnitude of the Sun (Campins et al. 1985), the arc brightness is that of a perfect

Lambert diffuser with a total area of 100 km2. From Fig. 6, at half maximum brightness, the arc

extends over about 15˚ or 64 000 km. An equivalent Lambert diffuser with the same extent would

be only 1.6 m high.

If we assume that, like the E ring, the arc is made of 1 µm ice particles, with a near 100%

reflectivity, then their total volume is about 100 m3, and their mass about 105 kg. This is a lower

bound of the cloud mass, since the arc can also contain larger particles, and the particle reflectivity

can only be lower. Moreover, the arc may have been much brighter before we started the observa-

tions.

7. A POSSIBLE INTERPRETATION

We have shown evidence for particles orbiting Saturn at a distance close to that of Enceladus.

Particles are scattered along the orbit forming an arc with a sharp maximum concentration not very

far from the Enceladus L4 Lagrange point.

The first interpretation which comes to mind is that we have detected particles trapped in tad-

pole orbits around the Enceladus L4 point. Analytical and numerical investigations of the motions

of particles in the E-ring indeed indicate that particles can get trapped along stable horseshoe and

tadpole orbits (Seidelmann et al. 1984, Burns 1997). However, the orbit of the observed arc signifi-

cantly differs from that of Enceladus. Its inclination is higher (1.8˚ instead of 0.02˚) and its radius is

also probably larger (242 600 km instead of 238 000 km). In addition, the maximum concentration

of particles is not exactly at the longitude of the L4 point but 16˚ away from it. Finally, the phenom-

enon appears to be highly time variable. The arc was not seen on August 10, neither on our data nor

on the HST data. Moreover, the arc brightness was found to decrease during the observations.

Therefore, it could be a short-lived event.

It has long been suggested that the E-ring particles somehow emanate from Enceladus (Hill

1984), possibly through volcanism (Pang et al. 1984) or explosive cryovolcanism (Kargel et al.

1996), a variety of volcanism driven by water or a mixture of water, or simply from metoritic

impacts. Enceladus itself is covered with a recently-deposited layer, possibly frozen water drops,

which are reminiscent of E-ring material (Buratti 1988). With an escape velocity of only 0.2 km/s,

a significant fraction of the ejected material could have escaped Enceladus and remained on a

nearby orbit (Hamilton and Burns 1994). The observed orbital inclination of 1.8˚ could be ac-

counted for by combining the Enceladus orbital velocity of 12.6 km/s with an ejection velocity of

0.4 km/s, which is only twice the escape velocity. This value is consistent with ejection velocities

recently calculated for water driven explosions on Mars (Fagents and Wilson 1996). However, the

observed arc longitude implies that such an ejection occurred at least a month before the observa-

tions. It is difficult to imagine how such a cloud of particles could appear in two days other than

from some sort of catastrophic on orbit collision.

Enceladus has also a long history of impact cratering (Kargel and Pozio 1996) a process which

is now known to produce large ejecta with speeds that can exceed a planet escape velocity (Melosh

1995). The analysis of the size-range distribution of secondary craters around large primary craters

on the Moon and Mars (Vickery 1986) has shown that such impacts can produce ejecta up to

several hundred meters wide with velocities mostly exceeding 0.2 km/s. Therefore, impact cratering

on Enceladus must have sent into space ice blocks weighting up to 109 kg. As these blocks accumu-

late on orbits close to that of Enceladus, collisions must occasionally occur and scatter ice frag-

ments along Enceladus’ orbit. Eventually, such fragments may become trapped in tadpole or horse-

shoe orbits with a maximum concentration near the Lagrange points. The large original ice blocks

which are on elliptical or inclined orbits, are the most likely to collide with the scattered fragments.

If such a collision occurs, it will scatter fragments and small particles along the orbit of the original

block 1 that is on a probably inclined or eccentric orbit. Moreover, the cloud is likely to be found

near (but not exactly at) a Lagrangian point where collisions are the most likely to occur. We may

have witnessed the immediate aftermath of such a collision.

8. DISCUSSION

If such a collision occurred, it must have happened when the large ice block was crossing

Enceladus orbit. Given the longitude of the ascending node determined in Section 5, the nearest

possible collision time must have been 5.5 ± 0.3 h (UT), that is 5.9 h before the beginning of our

observations. In experimental collision of ice spheres (Arakawa and Higa 1996), observed ejection

velocities range from 3 to 0.1 the impact velocity. Taking 0.4 km/s as the impact velocity (Section

7) gives a range of ejection velocities of ∆v = 1.2 km/s.

The fragment dispersion can be estimated as follows. For Keplerian orbits, the product of the

orbital period T times the cube of the orbital velocity v is constant

v T3 = constant (3)

or, after taking derivatives,

∆ ∆T

T

v

v= −3 . ( 4)

Since a time t after the collision, a fragment with orbital period T moves in longitude by an amount

1 See for example the fragments of comet Shoemaker Levy SL9.

ϕ π= 2 t T/ , the longitudinal dispersion ∆ϕ of the fragments is given by

∆ ∆ϕϕ

= − T

T(5)

Putting Eq. (4) into Eq. (5) gives

∆ ∆ϕ ϕ= 3v

v(6)

Taking v= 12.4 km/s as the orbital velocity of the ice block, 5.9 hours after collision, the longitude

of the remaining fragments are expected to have moved in longitude by an amount ϕ = 65˚. Assum-

ing ∆v = 1.2 km/s, Eq. (9) gives a longitudinal dispersion of

∆ϕ = 19˚, (7)

in excellent agreement with the full extent of the arc observed on Fig. 6b.

More difficult to understand is the rapid decrease of the arc brightness during the observations.

The arc does not appear to be vertically resolved. Although it may vertically expand during the

observations, the amount of material visible in one pixel should remain the same, unless it evapo-

rates. Laboratory studies of collisions between ice blocks show that it produces a very large number

of fine particles (Kato et al. 1995) forming a cloud of haze (Arakawa et al. 1995). Such a haze

could particularly explain the sharp peak observed in the J band in Fig. 6. We can speculate that the

energy released by the collision is sufficient to melt the ice and form a cloud of water droplets.

These would quickly evaporate, leaving perhaps only tiny ice crystals.

Similar collisions may have already been observed in the past: Baum et al. (1981) found evi-

dence for some bunching of E-ring particles in the vicinity of the Enceladus trailing Lagrangian

point L5. If such collisions regularly occur and produce a large amount of small particles, then the

smallest 1-µm size particles will slowly diffuse according to the process described by Hamilton

(1992) and Horanyi et al. (1992), providing a means to replenish the E ring. The lifetime of the E

ring is known to be much shorter than the age of the solar system (Cheng et al. 1982, Hill and

Mendis 1986). If we take 7 x 108 kg as the E ring mass (Showalter et al. 1991) and 103 years as its

lifetime (Haff et al. 1983), then one must provide almost 106 kg of material per year to keep the E

ring in its current state. Showalter et al. (1991) estimated the necessary mass rate as 2 x 10-4 the

mass of Enceladus over the age of the solar system, which also amounts to about 106 kg/year. If

each collision produces about 105 kg of small particles, then the collision rate must be of the order

of once a month.

9. CONCLUSION

A 2-h sequence of adaptive optics images taken almost two days after the Earth crossed Saturn’s

ring plane revealed a faint elongated structure apparently moving away from Saturn. We have

shown evidence that it is light scattered by an arc of particles on a Keplerian orbit close to that of

Enceladus. We have also shown that it is a transient phenomenon, lasting typically 10 h. Our most

likely explanation is that a large block of ice previously ejected by Enceladus collided with ice

fragments trapped in Enceladus’ orbit near its L4 Lagrange point. The collision produced a rapidly

expanding cloud of small ice particles, with a minimum mass of at least 105 kg. If such collisions

occur at an average rate of once a month, the amount of generated small particles is sufficient to

maintain the E ring in its present state. If this interpretation is correct, then the Cassini orbiter

should definitely avoid orbiting near Enceladus. On the other hand, it should watch carefully for

such events to occur.

APPENDIX

We give here a more rigorous derivation of the relation between the arc brightness I x( ) ob-

served on the sky and its brightness distribution J( )ϕ along the orbit, which takes into account the

width w of the arc. Let I0 be the brightness of a single particle and n the number of particles per

unit volume. Assuming the arc is optically thin, a segment of it with length dϕ will have a bright-

ness (Fig. 8b)

J I na wh( )ϕ ϕ ϕd d= 0 , (A1)

where a is the arc orbital radius and h its thickness (perpendicular to the plane of Fig. 8b). As seen

from the Earth, the arc brightness distribution is

I x x I n xlh( )d d= 0 , (A2)

where l is the geometrical depth of the arc along the line of sight (Fig. 8b). Comparing Eqs. (A1)

and (A2) gives

J x

I x

aw

l

( )

( )= . (A3)

The depth l is a function of x given by

l a w x a x= + − − −( )2 2 2 2 (A4)

or, with ε = w a/ ,

l a x a x= +( ) −[ ] − −[ ]2 2 2 1 2 2 2 1 21 ε

/ / . (A5)

One can safely assume ε << 1, which gives after straightforward development

l a xa

a x≅ −( ) +

−

⎛

⎝⎜⎞

⎠⎟−

⎡

⎣

⎢⎢

⎤

⎦

⎥⎥

2 2 1 2 2

2 2

1 2

1 2 1/

/ε

. (A6)

If we further assume

2 12

2 2a

a x

ε−

<< (A7)

then, Eq. (A6) becomes

law

a x≅

−2 2 . (A8)

Putting Eq. (A8) into Eq. (A3) gives

J x

I xa x

( )

( )= −2 2 , (A9)

which is identical to Eq. (2) of the main text. Approximation (A7) can be rewritten

x a2 2 1 2<< −( )ε or, x a a w<< −( ) = −1 ε (A10)

which means that the approximation is valid except at maximum elongation over a distance of the

order of the arc width. For a wide arc, one would indeed expect to see there an increase of the arc

brightness at the edge due to high values of l. Since such an increase is not observed, we conclude

that the arc width must be narrow (unresolved) and that Eq. (A9) is valid.

ACKNOWLEDGMENTS

The processing and analysis of the data was supported on NASA grants NAGW 4935 and NAG

5-3731. We acknowledge the use of Mark Showalter’s ephemeris generator written under the aus-

pices of the Planetary Data System’s Rings Node, a very convenient tool. We also thank Toby

Owen, Mark Schowalter, and an unknown reviewer for their very useful comments on our original

manuscript.

REFERENCES

ARAKAWA, M., MAENO, N., HIGA, M., IIJIMA, Y-I., and KATO, M. 1995. Ejection velocity of ice impact

fragments. Icarus 118, 341-354.

ARAKAWA, M., and HIGA, M. 1996. Measurements of ejection velocities in collisional disruption of

ice spheres. Planet Space Sci. 44, 901-908.

BAUM, W. A., KREIDL, T., WESTPHAL, J. A., DANIELSON, G. E., SEIDELMANN, P. K., PASCU, D., and

CURRIE, D. G. 1981. Saturn’s E ring. I. CCD observations of March 1980. Icarus 47, 84-96.

BURATTI, B. J. 1988. Enceladus-Implications of its unusual photometric properties. Icarus 75, 113-

126.

BURNS, J. A. 1997. Dynamically cleared regions beyond Saturn’s ring for Cassini’s safe passage.

Bull.–Am. Astron. Soc. 29, 999.

CAMPINS, H., RIEKE, G. H., and LEBOFSKY, M. J. 1985. Absolute calibration of photometry at 1

through 5 µm. Astron. J. 90, 896-899.

CHENG, A., LANZEROTTI, L. J., and PIRRONELLO, V. 1982. On the lifetime of E-ring grains and their

nature. In “Sun and Planetary System”, (Fricke, W. and Teleki, G., eds.), pp. 251-252. Reidel,

Dordrecht.

FAGENTS, S. A., and WILSON, L. 1996. Numerical modeling of ejecta dispersal from transient volca-

nic explosions on Mars. Icarus 123, 284-295.

HAFF, P. K., SISCOE, G. L., and EVIATAR, A. 1983. Ring and plasma. The enigmae of Enceladus.

Icarus 56, 426-438.

HAMILTON, D. P. 1992. Motion of dust in a planetary magnetosphere: orbit-averaged equations for

oblateness, electromagnetic, and radiation forces with application to saturn’s E ring. Icarus

101, 244-264.

HAMILTON, D. P., and BURNS, J. A. 1994. Origin of Saturn’s E ring: self-sustained, naturally. Science

264, 550-553.

HILL, T. W. 1984. Saturn’s E ring. Adv. Space Res. 4, 149-157.

HILL, J. R., and MENDIS, D. A. 1986. The dynamical evolution of Saturn’s E-ring. Earth, Moon,

Planets 36, 11-21.

HODAPP, K.-W., J. L. HORA, D. N. B. HALL, L. L. COWIE, M. METZGER, E. M. IRWIN, T. J. KELLER, K.

VURAL, L. J. KOZLOWSKI, and W. E. KLEINHANS 1995. Astronomical characterization results of

1024 x 1024 HgCdTe HAWAII detector array. SPIE Proc. 2475, 8-14.

HORANYI, M., BURNS, J. A., and HAMILTON, D. P. 1992. The dynamics of Saturn’s E ring particles.

Icarus 97, 248-259.

KARGEL, J. S., and POZIO, S. 1996. The volcanic and tectonic history of Enceladus. Icarus 119, 385-

404.

KATO, N., IIJIMA, Y-I., ARAKAWA, M., OKIMURA, Y., FUJIMURA, A., MAENO, N., and MIZUTANI, H. 1995.

Ice-on-ice impact experiments. Icarus 113, 423-441.

MELOSH, H. J. 1995. Cratering dynamics and the delivery of meteorites to Earth. Meteoritics 30,

545-546.

PANG, K. D., VOGE, C. C., RHOADS, J. W., and AJELLO, J. M. 1984. The E ring of Saturn and

satellite Enceladus. J. Geophys. Res. 89, 9459-9470.

PRESS, W. P., FLANNERY, B. P., TEUKOLSKY, S. A., and VETTERLING, W. T. 1990. Numerical Recipes

in C, p. 60. Cambridge Univ. Press. Cambridge, UK.

RODDIER, C., RODDIER, F., GRAVES, J. E., and NORTHCOTT, M. J. 1997. Saturn. IAU circular No. 6697

(July 10).

RODDIER, F., NORTHCOTT, M., and GRAVES, J. E. 1991. A simple low-order adaptive optics system for

near infrared applications. Publ. Astron. Soc. Pacific. 103, 131-148.

SEIDELMANN, P. K., HARRINGTON, and R. S., SZEBEHELY, Y. 1984. Dynamics of saturn’s E ring. Icarus

58, 169-177.

SCHOWALTER, M. R., CUZZI, J. N., and LARSON, S.M. 1991. Structure and particle properties of

Saturn’s E ring. Icarus 94, 451-473.

VICKERY, A. M. 1986. Size-velocity distribution of large ejecta fragments. Icarus 67, 224-236.

Figure captions

Figure 1. Deep image of the rings of Saturn made from all the data taken on August 12, 1995. The

position of Saturn is indicated by the gray circular area in the center. The bright radial

streak seen on the left (east side) just above the ring plane is the arc discussed in this

paper. See text for other details.

Figure 2. From left to right and top to bottom: the first 5 images (a to d) show a time sequence of

the object motion with 10 minute intervals. The bottom right image (f) is a median of all

the five other images.

Figure 3. Plot showing the distance between the sharp right edge and Saturn as a function of time.

The full line is a least-square fit with a circular Keplerian orbit.

Figure 4. From top to bottom. First three images are median of a) the early H band frames, b) J

band frames, and c) late H band frames. Next image (d) shows an underlying diffraction

artifacts, as it also appears on the west side. Bottom image (e) is a median of images (a),

(b), and (c) with the artifact removed. The black line shows the estimated orbit.

Figure 5. Estimated position of the tip of the arc (sinusoidal curve) as a function of time (in hours)

and distance to Saturn (in units of 105 km). The horizontal dashed line indicates the ring

plane crossing time taken here as the origin. The hatched areas cover our adaptive op-

tics observations. The horizontal dotted lines show the time of HST observations. The

vertical dotted lines indicate Saturn inner and outer ring boundaries. An arrow indicates

where and when the arc was detected.

Figure 6. Photometric profiles of the arc. Intensity scales are on the left the averaged ADU value

and on the right the height in meters of an equivalent perfect Lambert diffuser. Full lines

show a mean of the early H band profiles. Dotted line is a mean of the late H-band

profiles. Dashed line is a mean of the J-band profile (arbitrarily multiplied by a factor

1.3 for comparison with the early H profile). Fig. 6a shows the arc brightness as a

function of distance (in pixels) on the sky starting from the sharp edge. Fig. 6b shows

the arc brightness as a function of longitude expressed in degrees (bottom scale) or in

km along the arc (above scale), starting also from the arc sharp edge.

Figure 7. Time decay of the arc peak brightness. Time is given in hours (UT). Brightness is ex-

pressed in meters as in Fig. 6.

Figure 8. Arc geometry for a negligible (a) or finite (b) arc width w.

Fig.

1

a b

c d

e f

Fig. 2

195

200

205

210

215

220

225

11.4 11.6 11.8 12 12.2 12.4

Dis

tanc

e (1

03 k

m)

Time (hours)

Fig. 3

Distance to Saturn’s center (103 km)

280 190200220230240250260270 210

a

a

b

c

d

e

Fig. 4

Fig. 5

Fig. 6

0.5

1

1.5

2

11.4 11.6 11.8 12 12.2 12.4

Inte

nsit

y (m

eter

s)

UT (hours)

Fig. 7

dϕ

dx ϕx

J(ϕ)

I(x)

a

ϕx

a

w

l

Fig. 8a

Fig. 8b