Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
The Pluto System
Detection and Variability of Nix and Hydra
Generals Paper I
Alessondra Springmann
August 1, 2008
1
1 Introduction
Once thought to be a remote oddball of our outer Solar System, Pluto is now known to be
one of many objects in the Kuiper Belt. Its three moons—previously thought to be unique
for such a small, remote body—make Pluto one of the numerous binary or multiple-body
Kuiper Belt objects (KBOs) that have been discovered in the last 16 years. Up to 20% of
KBOs in various dynamical classes are multiple systems (Noll et al., 2008), implying that it
is not uncommon for a body residing in this region of the Solar System to be a binary or
multiple system.
Pluto’s largest moon Charon was discovered in 1978 (Christy & Harrington, 1978), mak-
ing Pluto a double planet until the 2005 discovery of Pluto’s smallest moons, Nix and Hydra
(Weaver et al., 2005). Recent dynamical simulations involving these two newly-discovered
moons present new constraints on the formation of the Pluto system. The satellites’ circular
orbits, coplanar with the orbit of Charon, suggest that these moons all formed from material
ejected into orbit around Pluto by a giant impact event (Canup, 2005; Stern et al., 2006b).
1.1 Motivation
Nix and Hydra’s recent discovery and their importance in our understanding the Pluto
system make them an important and timely mission goal for New Horizons, set for a 2015
encounter with Pluto. Providing limits on the rotation periods of these moons allows for more
efficient imaging of their surfaces by the New Horizons cameras, in addition to confirming the
period constraints determined by Buie et al. (2007) and allowing for more precise simulations
to be carried out of the Pluto system’s formation and evolution. In this paper, we discuss
attempts to detect these moons and constrain their rotation periods from data obtained with
the 6.5-meter Baade telescope at Las Campanas Observatory in July 2007.
2
2 Pluto System Overview
It is widely accepted that the Pluto system was formed in a giant impact collision (Canup,
2005; Stern et al., 2006b) with little or no exchange of heat or heating of the bodies’ materials
occuring, resulting in the formation of not only Pluto and Charon, but also Hydra and Nix.
This assumption is valid because the angular momentum of the Pluto system is quite high
(see references 9-12 in Stern et al., 2006b), thus Charon must have formed from a collision
with Pluto and was therefore likely not captured into orbit.
Smaller than Eris and red, Pluto has a surfaced varied in both color and albedo (Brown,
2008). The largest KBOs, including Pluto, have methane ice on their surfaces, in addi-
tion to N2, CO, ethane, and perhaps water ice (McKinnon et al., 2008). The density of
the Pluto-Charon-Nix-Hydra quaternary system (2.01 g cm−3, Tholen et al., 2008) implies
that the impactors that formed this system were initially accreted (McKinnon et al., 2008).
These impactors were differentiated by accretion and radiogenic heating, which resulted in
a differentiated Pluto. Taking the argument for Pluto’s being differentiated even further,
Hussmann et al. (2006) suggest the presence of a sub-surface ocean below Pluto’s low-density
ice. Although formation simulations that result in differentiation are accepted, scenarios that
result in subsurface oceans or cryovolcanism are not as widely considered to be as valid.
2.1 Charon
2.1.1 Formation
Charon formed from mantle material leftover from the collision that formed the entire
Pluto system. As this collision resulted in little or no melting of the impactors’ material,
there was little heat exchange, and thus a large portion of the impactors’ mantles remained
water ice and became the major constituents of Charon (Stern et al., 2006b).
3
2.1.2 Mutual Events
The discovery of Charon allowed for more precise mass determination for Pluto. Soon
after the discovery of Charon, its orbit plane was oriented such that it was edge-on as seen
from Earth, resulting in mutual events of Pluto and Charon being visible for approximately
five years (Binzel & Hubbard, 1997), allowing for more precise measurements of color and
albedo of both bodies (Figure 2 shows a color map of Pluto’s surface). While we will not
experience another “season” of Pluto-Charon mutual events again in our lifetimes, improve-
ments in telescope technology and size have allowed us to study the bodies individually and
also through occultations (Noll et al., 2008). The discovery of additional binary or multiple
KBOs showcases the diversity of the Kuiper Belt, and allows us to apply what has been
learned from Pluto to other systems.
2.1.3 Physical Properties
Charon’s surface is dominated by water ice, in addition to a bit of ammonia (see McKinnon
et al., 2008, and references therein); quite a different surface from Pluto. Both Pluto and
Charon are tidally locked, the same faces of these two bodies always facing the companion.
Because of the relative sizes of Pluto and Charon (d\ = 2390 km; dC = 1207 km), these two
bodies are occasionally called a “double planet”.
2.2 Nix & Hydra
Nix and Hydra likely formed in the same collision that produced Charon, probably origi-
nating as debris from this giant impact between two large, differentiated bodies (Stern et al.,
2006b). Their coplanar orbits with Charon also suggest that they formed in the same impact
from which formed Charon. Combining their coplanar orbits with their neutral colors, it
can be assumed that Nix and Hydra are similar in composition to Charon, i.e. that their
4
constituents are mostly water ice.
Another argument for Nix and Hydra forming at the same time as Charon is that they are
larger than the critical size boundary for catastrophic breakup over the age of the solar system
(Durda & Stern, 2000; Stern et al., 2006b). Forming along with Pluto and Charon, Nix and
Hydra are likely the same age as the former two bodies and have withstood disruption
forces that would have torn apart smaller bodies. It is unlikely that these small moons
were captured, because of their coplanar, circular orbits with Charon, and if they had been
captured, these two small moons would be in neither coplanar nor circular orbits, and would
not be orbiting in a common direction (Ward & Canup, 2006).
Formation of these moons alongside Pluto and Charon was likely, as they are presently
in highly circular orbits—e = 0.01504 for Nix and e = 0.00870 for Hydra (Tholen et al.,
2008)—implying that they were not dynamically captured, as the circularization timescales
are greater than the age of the Solar System (Stern et al., 2006b).
Knowing the rotation rates of these moons will help verify whether they are in any sort
of spin-orbit resonances with Pluto, and perhaps place important constraints on the rotation
rates of moons of other KBOs, as mentioned in Section 2.3. Numerous simulations by Canup
(2005); Lissauer (2006); Stern et al. (2006b); Ward & Canup (2006); Lithwick & Wu (2008);
Tholen et al. (2008) have guessed at the specifics of origin of the Pluto system, but without
rotational constraint on Nix and Hydra, knowing the particulars of how they formed and
evolved to their current orbits will elude dynamicists.
Even if Nix and Hydra have resisted disruption by impacts or tidal forces, they may
have lost up to 10% of their masses due to impact erosion (Stern, 2008a,c). Material ejected
from their surfaces by collisions with other KBO fragments could help feed the hypothetical,
tenuous rings or ring arcs in the Pluto system1.
1Binzel (2006) and Stern et al. (2006b) propose the possibility of rings (or ring arcs) in the Pluto system,of time-varying optical depth of approximately τ = 5× 10−6, of the same order of magnitude as the opticaldepth of the tenuous rings around Jupiter (Showalter et al., 1987; Throop et al., 2004). As Pluto is far away
5
The proximity of Nix and Hydra to corotation—more literally, co-orbital—resonances
also implies a common impact origin for these smaller moons, meaning that Nix and Hydra
were driven outward by Charon’s migration (Ward & Canup, 2006). A corotation resonance
occurs where the object’s semi-major axis oscillates, rather than its eccentricity. At some
point, the outward migration of Nix and Hydra ceased, most likely due to the end of Charon’s
outward migration. Nix likely escaped from its corotation resonance with Charon first, before
the end of Charon’s outward semi-major axis migration due to tidal expansion. Ward &
Canup (2006) also show that these particular corotation resonances (4:1 and 6:1) are the
most conducive to moon formation around Pluto, as other corotation resonances are close
to the Hill sphere and other dynamically unstable regions. Nix and Hydra’s probable initial
locations in the 4:1 and 6:1 corotation resonances could have lead to the aggregation of
material at these points and thus helped contribute to the long-term survival of these moons
over the age of the Solar System.
Though the orbits of these two small moons are relatively well known (Tholen et al., 2008),
not much is known about either the exact dimensions of Nix and Hydra or their rotation
rates. Stern et al. (2006b) assume minimal masses for Nix and Hydra and calculate that
these moons are likely not in synchronous rotation with Pluto, unless they formed much
closer to Pluto and migrated outward (Ward & Canup, 2006). Thus, understanding the
rotation rates of Nix and Hydra helps constrain formation models of multiple-body systems,
including other KBOs.
from the perturbing forces of the Sun, the possibility of there existing rings of collisional ejecta from Nixand Hydra in the Pluto system is high, though the arrival of New Horizons will ultimately confirm or denythe prescence of rings. Stern (2008a,b) also suggests the potential resurfacing of Nix and Hydra by impactejecta, up to 70 meters on each small moon if the process is efficient.
6
2.3 Other Kuiper Belt Binaries
Because up to a fifth of Kuiper Belt objects are binary or have multiple satellites
(Stephens & Noll, 2006), studying the multiple system of Pluto allows us to better under-
stand the evolution and current states of not only the Pluto system but also other multiple
Kuiper Belt objects. Although there are plenty of binary KBOs, it is significantly harder to
observe smaller satellites of 100-km class TNOs (Noll et al., 2008), especially for narrowly-
separated objects. Comparing the frequencies of nearly equal-sized binaries with objects
with smaller companions is difficult, due to observational limits from seeing, telescope size,
differences in detectors, etc. Pluto, being one of the brightest KBOs, allows us to most
readily study its multiple satellites, making it an ideal target for New Horizons (Section 3).
3 New Horizons
A collaboration between the John Hopkins Applied Physics Laboratory and the South-
west Research Institute, New Horizons was launched in January 2006 and is expected to
fly by Pluto on July 14 in 2015. New Horizons’ science goals include imaging the surfaces
and thus mapping the compositions of the four main bodies in the Pluto system, studying
Pluto’s atmosphere, searching for an atmosphere around Charon, along with imaging the en-
tire Pluto Hill sphere, searching for presently undiscovered moons and probable rings (Stern
et al., 2006b; Young et al., 2007; Stern, 2008a).
The recent discovery of Nix and Hydra is very timely in allowing target planning and
new science goals for the New Horizons mission. If the rotation rates of these moons are
verified, New Horizons will be able to more optimally image their surfaces.
After visiting Pluto New Horizons will swing by at least one yet-to-be-determined KBO
to further study the other icy bodies of the outer Solar System, linking them with other
dynamical populations such as Centaurs and comets. Hopefully in the process, New Horizons
7
will begin to answer our questions about the origins and compositions of the cold leftovers
of our Solar System’s formation and evolution.
Nix and Hydra’s recent discovery and their importance in understanding of the Pluto
system make them an important and timely mission goal for New Horizons. Until New Hori-
zons arrives at Pluto, a good deal will remain uncertain about the Pluto system, including
the absolute radius of Pluto, the nature of its atmosphere, the rotation rates of Nix and
Hydra, and the masses and composition of these moons.
4 Observations
To determine or constrain potential short period rotation of Nix and Hydra, images of
the Pluto system were obtained with the 6.5-meter Walter Baade Magellan telescope at the
Las Campanas Observatory, Chile. Repeated imaging sequences covering a total of 10 hours
on UT 2007 July 11 and 12 were executed using the Inamori Magellan Areal Camera and
Spectrograph (IMACS) in the Baade f/4 direct imaging mode. A total of 815 images with
integration times of 6 seconds (cadence of 30 seconds) in a Sloan i′ filter were collected under
0.4-0.7 arcsecond seeing. Only Chip 2 out of the eight chips that compose IMACS was used
and was further subrastered to reduce image readout time. A sample image sequence is in
Figure 1, moving 0.44 arcseconds per hour.
5 Data Reduction and Analysis
The data were calibrated using bias and flat frame images. Routines from the IRAF/DAOPHOT
package were used to perform stellar photometry on the Pluto fields.
The point spread functions (PSFs) of Pluto and Charon overwhelm any signal from Nix
and Hydra. Therefore, in order to determine the rotation rates of these small moons, much
8
Figure 1: Six-second integrations of the Pluto system taken 1 hour apart on UT 2007 July11. The Pluto-Charon system is the bright object near the center of the frames that changesposition between exposures and is indicated by the blue circle.
less detect their signal in the science frames, the light from Pluto and Charon must be
modeled and subtracted from the images to reveal the signatures of Nix and Hydra.
This section is separated into three subsections pertaining to the physics of rotational
lightcurves, the astronomy of limiting magnitude calculations, and the mathematics of point
spread function fitting:
1. Rotational lightcurves and detection (Section 5.1)
2. Astrometry/limiting magnitude calculation (Section 5.2)
3. Point spread function modeling and subtraction (Section 5.3)
5.1 Rotational lightcurves
Rotation rates of small Solar System bodies are determined by measuring their changes
in brightness via aperture photometry. As a non-uniform body rotates, it reflect varying
amounts of light due to changes in albedo, surface area, and other surface features, producing
changes in brightness that can be measured by a detector. Plotting the change in brightness
or the magnitude of the object as a function of time produces a lightcurve. Analyzing
9
the lightcurve yields information regarding the rotation period of the object and also more
detailed information regarding surface features and body shape with sufficient data.
To first approximation, a rotating moon such as Nix or Hydra can be approximated by a
triaxial ellipsoid (Figure 2). As the moon rotates, it presents different faces to the Sun and
thus reflects varying amounts of light at the observer, generating a lightcurve, which can be
approximated by a Fourier series of sinusoidal curves. Two cases that would generate flat,
or featureless, lightcurves include a uniformly spherical rotating object, or a rotating object
viewed from its pole. Should either of these cases apply to Nix or Hydra, no discernible
change in lightcurve will be detected from that moon.
Using the standard magnitude equation,
∆m = m2 −m1 = 2.5 log10
(b1b2
)(1)
where m indicates magnitude, b is the luminosity, and t is the exposure time, and then
substituting for b = F/t while assuming that t1 = t2, we obtain:
∆m = 2.5 log10
(F1
F2
). (2)
Assuming that flux is proportional to the projected area of the target, the maximum
points of the lightcurve will correspond to when the largest face of the target is reflecting
light at the observer. Likewise, the minimum points of the lightcurve correspond to when
the object is presenting the smallest amount of reflected light to the observer and thus yield
the smallest amount of flux. Taking the area of an ellipse is π · a · b, the projected area of
the largest face of the ellipsoid will be Al = πac,while the smallest face will have a projected
area of As = πbc. From this, we obtain:
∆m = 2.5 log10
(πac
πbc
)= 2.5 log10
(a
b
).
10
Thus, the change in lightcurve amplitude as the body rotates is proportional to the ratio
of the semi-major over the semi-minor axis. A target body with an axial ratio of 1.25 will
ideally show a change in lightcurve semiamplitude of 0.24 (Figure 3).
Figure 2: A triaxial ellipsoid with axes of different lengths (a 6= b 6= c).
During the observations conducted in July 2007, the phase angle of Pluto was 0◦.07 and
increasing, making Pluto quite close to opposition and thus at the hight of its brightness as
seen from Earth. This makes the Sun, Pluto, and Earth effectively in a line, since the angle
between the Sun, Pluto, and the Earth is less than one degree (Figure 4).
5.2 Limiting Magnitude Calculation
In order to determine if Nix and Hydra are detectable in our science images, we calculated
how many images would need to be coadded to obtain sufficient signal-to-noise on Nix. Nix
was chosen as the limiting case because it is the dimmest (apparent V magnitude of 23.7
(Stern et al., 2006a)) of these moons, to bring them to the same brightness as the dimmest
stars in our exposures.
Using interactive IDL astrom routines from Marc Buie2, we matched the locations of the
stars in our frames to US Naval Observatory stellar locations. Initally, the predicted catalog
2http://www.boulder.swri.edu/~buie/idl/
11
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1!0.5
!0.4
!0.3
!0.2
!0.1
0
0.1
0.2
0.3
0.4
0.5
rotational phase
! m
agni
tude
Figure 3: A synthetic lightcurve of an object with an axial ratio of 1.25 over one rotationperiod. The two maxima and two minima correspond to the two large and two small facesof the object, respectively. The lightcurve is approximated to first order by a sine curve.
Figure 4: The viewing geometry of the Pluto system as seen from Earth on UT July 11 2007.(Note: orbits and body radii not to scale.)
12
positions do not match the positions of the stars on the chip very well. The positions of
the stars on the chip are manually aligned with the predicted catalog positions (Figure 5).
astrom then calculates the vectors between the catalog and chip locations of the stars, then
determines the standard magnitudes of the stars in the exposure.
Figure 5: Alignment between the USNO star catalog and star chip positions. Yellow circlesindicate a match between star catalog and chip positions; red circles indicate that the routinewas unable to find a match between predicted and chip positions.
To calculate how many images of Nix and Hydra need to be coadded, we used the
astrom calculations of the limiting magnitude star in each exposure to determine the dimmest
magnitude visible of a star in our images. The magnitudes we reference from the USNO
catalog magnitudeare in R. We realize this is suboptimal for comparing to photometry
performed in Sloan i′; however, it makes a reasonable estimate.
For exposure 101 with exposure time t0 = 6 seconds) on UT July 11, the sky back-
ground, µ, is 53.19 counts, while the standard deviation, σ, is ±10 counts. The dimmest
star detectable by the photometry routines used would have S1 = µ + 10σ counts, in this
13
case, 153.19 counts above background for a signal to noise ratio of 10σ/σ = 10. For a As a
test to scale the limiting magnitude, we used a star located at RA 17:47:33.56, declination
-16:25:36.39 having a catalog R magnitude of 16.12. This image records a signal, S0, of
10910.3 counts with the sky subtracted. Using the standard magnitude equation (equation
1), where brightness, b, is related to flux, F , and integration time, t, by b = F/t, it is possible
to calculate the difference in magnitudes between the star with signal S0 and a star with
signal S1:
∆m = 2.5 log10
(S0
S1
)= 2.5 log10
(10910.3
100
)= 5.09
Thus, a star with 10910 counts corresponds to a star in the Sloan i′ filter with magnitude
16.12, while a star with 100 counts in the same filter corresponds to a magnitude of 16.12 +
5.09 = 21.21, which is the limiting magnitude for a six-second exposure. The results from
the astrom routines also indicate that the average limiting magnitude for all of our science
frames from UT July 11 2007 is roughly 21.
Using the V magnitudes from Stern et al. (2006a) we estimate the magnitude of Nix as
24.0. In order for Nix to be detected in our science frames, some amount more signal on
these moons will need to be obtained, which will correspond to an increase of n times in
exposure length to obtain a SNR of 10 on this moon.
Using the magnitude equation (Equation 1) again we find that:
∆m = 24.0− 21.21 = 2.79 = 2.5 log10
(m[N,H]
mlimiting
)= 2.5 log10
(mlimiting · nmlimiting
)
2.79
2.5= log10(n); n = 10
2.792.5
n = 13.02
Therefore, Nix needs a factor of 13 more flux for a 10 σ detection of this moon from
the background signal, corresponding to a 65-second exposure, approximately 11 6-second
14
exposures.
5.3 Point Spread Function Models
In order to subtract the signal from Pluto and Charon, models of the point spread
functions (PSFs) that these objects create on the chip are modeled. As the PSF changes
from exposure to exposure, a different PSF must be generated for each image.
To create a PSF for the image, a series of tasks from the DAOPHOT routines are used.
First, the daofind task creates an initial star list by locating all the star-like sources in an
image within a certain width and range of counts, then tvmark marks the stars from the
star list on the displayed FITS file. Using both phot and daofind, the background values
and initial magnitudes for the stars in the star list are calculated.
To generate a PSF from the stars in the image, between seven and nine stars brighter
than Pluto are manually selected with pstselect and an initial PSF is created with the psf
routine, where the type of PSF (Lorentz, Gauss, or Moffat) can be specified.
The nstar task then takes the PSF generated in the previous step and fits it to the rest
of the stars in the field. The PSFs of the fitted PSF stars and their neighbors are then
subtracted from the image using substar. This process is then repeated on the subtracted
images until a clean PSF is obtained, usually after three or four iterations.
5.4 PSF Modeling
Several different point spread function (PSF) types were used to accurately model the light
from Pluto and Charon and thus subtract these two bodies’ PSFs. Three different models
have been used for approximating the point spread functions of the data. Davis (1994) de-
scribe Gauss, Moffat, and Lorentz models for IRAF/DAOPHOT PSF modeling as discussed
in Appendix A. However, the Moffat model best describes the shape of the PSFs obtained
15
with IMACS and thus was used for creating PSFs in this paper.
After the stars have been modeled and subtracted in each frame, approximately 65 sec-
onds (integration time, not clock time) of images are coadded together to obtain sufficient
signal to noise with respect to Nix and Hydra, as described in Section 5.2.
Although the Moffat25 models were better than the Gauss and Lorentz PSF models,
they were not sufficient to completely remove the signal from Pluto, Charon, and other
stars neighboring the locations of Nix and Hydra in our science frames. Thus, coadding 11
exposures did not yield a detection of either Nix or Hydra, discussed further in the next
section.
6 Discussion and Future Work
Our observations yield continuous coverage with integration times of six seconds over a
span of seven hours on the first night of observing. The high sampling frequency should
allow or constrain rotation periods of Nix and Hydra under 24 hours, should these moons
be detected.
Improvements to the PSF fitting routines would most likely yield a positive detection of
these moons, such that photometry could be performed on them to provide limits to their
rotation rates. Pluto does not appear as a point source in the science images; it is resolved
as a disk. Thus, being able to convolve a point spread function into a “disk spread function”
would allow the signal from Pluto to be modeled better than by simply using a Moffat or
other Gaussian model.
The high cadence of observations complements the observing sequence of the Hubble
telescope by Buie et al. (2007), which should constrain the moons’ rotation rates on the
order of days or longer. The Buie team should be able to discern if Nix and Hydra are
variable on a time scale on the order of 12 hours due to the durations of time between their
16
exposures on the Hubble Space Telescope. Between our measurements and those of the Buie
team, the rotation rates of these moons should be discerned in the near future and made
available to the New Horizons team in order to maximize the imaging capabilities of this
spacecraft during its flyby of these moons.
The arrival of New Horizons at Pluto will begin the in situ exploration of the Kuiper
Belt and the icy regions beyond Neptune, helping us understand better the nature of not
only Plutos moons but also their dwarf planet companion and other Kuiper Belt objects.
By imaging the surfaces of these moons and other objects beyond Pluto, we will be able to
determine the compositions of these moons and also their orbital characteristics, giving us
further clues as to their dynamical evolutions as well as those of other Kuiper Belt objects.
Almost 30 years after the realization that Pluto was a double planet, the discovery of
1998 WW31 demonstrated that Pluto was no longer unique in its binary status (Veillet et al.,
2001). According to the frequency predictions of Noll et al. (2008), there could be up to tens
of thousands of Kuiper Belt objects with satellites or that are binaries (Stephens & Noll,
2006). Though we know of only three moons in orbit around Pluto, there could have been
more moons that were disrupted or destabilized throughout the dynamical history of the
Pluto system. In addition to moons that once were, more satellites could be lurking below
the detection thresholds of current telescopes, along with rings or ring arcs, not just around
the tightly-bound Pluto system but around other KBOs. Maybe there exist KBO systems
less tightly-bound than Pluto, such as ones with captured satellites in highly eccentric orbits.
The arrival of New Horizons in the Kuiper Belt will answer numerous questions about the
history of this region, but will likely also raise new ones as the spacecraft’s science instruments
begin to probe the chilly remnants of planet formation. The objects that compose Pluto’s
cohort beyond Neptune should continue to yield more information and also more mysteries
as to their nature, for as Hal Levison says, “. . . the Kuiper belt is very strange.”
17
References
Binzel, R. P. 2006, Nature, 439, 924
Binzel, R. P. & Hubbard, W. B. 1997, Mutual Events and Stellar Occultations (Pluto and
Charon), 85–+
Brown, M. E. 2008, The Largest Kuiper Belt Objects (The Solar System Beyond Neptune),
335–344
Buie, M. W., Grundy, W. M., & Tholen, D. J. 2007, in AAS/Division for Planetary Sci-
ences Meeting Abstracts, Vol. 39, AAS/Division for Planetary Sciences Meeting Abstracts,
#62.08–+
Canup, R. M. 2005, Science, 307, 546
Christy, J. W. & Harrington, R. S. 1978, Astronomy Journal, 83, 1005
Davis, L. E. 1994, A Reference Guide to the IRAF/DAOPHOT Package, IRAF Programming
Group, National Optical Astronomy Observatories
Durda, D. D. & Stern, S. A. 2000, Icarus, 145, 220
Hussmann, H., Sohl, F., & Spohn, T. 2006, Icarus, 185, 258
Lissauer, J. J. 2006, Science, 313, 1054
Lithwick, Y. & Wu, Y. 2008, ArXiv e-prints, 802
McKinnon, W. B., Prialnik, D., Stern, S. A., & Coradini, A. 2008, Structure and Evolution
of Kuiper Belt Objects and Dwarf Planets (The Solar System Beyond Neptune), 213–241
Noll, K. S., Grundy, W. M., Chiang, E. I., Margot, J.-L., & Kern, S. D. 2008, Binaries in
the Kuiper Belt (The Solar System Beyond Neptune), 345–363
18
Showalter, M. R., Burns, J. A., Cuzzi, J. N., & Pollack, J. B. 1987, Icarus, 69, 458
Stephens, D. C. & Noll, K. S. 2006, Astronomy Journal, 131, 1142
Stern, S. A. 2008a, Icarus, 805
Stern, S. A. 2008b, in IAU/Asteroids, Comets, Meteorites Meeting Abstracts
—. 2008c, Space Science Reviews, 4
Stern, S. A., Mutchler, M. J., Weaver, H. A., & Steffl, A. J. 2006a, The Astronomical Journal,
preprint, The Astronomical Journal
Stern, S. A., Weaver, H. A., Steffl, A. J., Mutchler, M. J., Merline, W. J., Buie, M. W.,
Young, E. F., Young, L. A., & Spencer, J. R. 2006b, Nature, 439, 946
Tholen, D. J., Buie, M. W., Grundy, W. M., & Elliott, G. T. 2008, Astronomy Journal, 135,
777
Throop, H. B., Porco, C. C., West, R. A., Burns, J. A., Showalter, M. R., & Nicholson, P. D.
2004, Icarus, 172, 59
Trujillo, I., Aguerri, J. A. L., Cepa, J., & Gutierrez, C. M. 2001, Monthly Notices of the
Royal Astronomical Society, 328, 977
Veillet, C., Doressoundiram, A., Shapiro, J., Kavelaars, J. J., & Morbidelli, A. 2001, IAU
Circular, 7610, 1
Ward, W. R. & Canup, R. M. 2006, Science, 313, 1107
Weaver, H. A., Stern, S. A., Mutchler, M. J., Steffl, A. J., Buie, M. W., Merline, W. J.,
Spencer, J. R., Young, E. F., & Young, L. A. 2005, IAU Circular, 8625, 1
19
Young, L. A., Stern, S. A., Weaver, H. A., Bagenal, F., Binzel, R. P., Buratti, B., Cheng,
A. F., Cruikshank, D., Gladstone, G. R., Grundy, W. M., Hinson, D. P., Horanyi, M.,
Jennings, D. E., Linscott, I. R., McComas, D. J., McKinnon, W. B., McNutt, R., Moore,
J. M., Murchie, S., Porco, C. C., Reitsema, H., Reuter, D. C., Spencer, J. R., Slater, D. C.,
Strobel, D., Summers, M. E., & Tyler, G. L. 2007, Space Science Reviews, 709
Appendix
A Point Spread Function Models
Three different models were used to approximate the point spread functions of the data.
Davis (1994) describe Gaussian, Moffat, and Lorentz models for PSFs as discussed below.
A.1 Moffat25
The Moffat function, as described by Trujillo et al. (2001), is
PSF(r) =β − 1
πα2
[1 +
(r
α
)2]−β
, (3)
with the full width at half maximum, FWHM=2α√
21/β − 1, where PSF(FWHM/2) =
(1/2)PSF(0) and the total flux is normalized to 1. According to Davis (1994), the Mof-
fat family of functions are good for modeling under-sampled ground-based data. For the
DAOPHOT Moffat25 model, β is set as 2.5.
20
A.2 Gauss
The Gauss model for a PSF is a limiting case of the Moffat PSF (when β →∞) (Trujillo
et al., 2001):
PSF(r) = 4(21/β − 1
) β − 1
πF 2
[1 + 4(21/β − 1)
(r
F
)2]−β
. (4)
As β →∞, we can substitute 21/β − 1 with (ln 2)/β, so:
limβ→∞
PSF(r) = limβ→∞
β − 1
β
4 ln 2
πF 2
[1 +
4 ln 2
β
(r
F
)2]−β
. (5)
Using limm→∞(1 + z/m)m = ez we have
limβ→∞
PSF(r) =4 ln 2
πF 2e−
4 ln 2F2 r2 . (6)
Finally, writing F 2 = 8σ2 ln 2 we obtain:
limβ→∞
PSF(r) =1
2πσ2e−
12( rσ )
2
. (7)
The IRAF/DAOPHOT package describes the Gauss model as being the “best choice for
well-sampled, fwhmpsf ≥ 2.5pixels, ground-based images because the interpolation errors
are small and the evaluation is efficient as the function is separable in x and y.”
A.3 Lorentz
The Lorentz, or Cauchy, distribution for a PSF is an elliptical function with extended
wings that can be aligned along an arbitrary position angle. It is described mathematically
as follows:
PSF(r) =
[γ2
(x− x0)2 + γ2
](8)
21
where Γ describes the width of the function. As opposed to the Gauss distribution model,
the Lorentz model’s extended wings do not decrease to zero as rapidly. A large Gauss
distribution is exponentially related to the square of the deviation, whereas a large Lorentz
distribution is inversely related to the square of the deviation.
22